procps/library/slabinfo.c

1041 lines
36 KiB
C
Raw Normal View History

/*
* slabinfo.c - slab pools related definitions for libproc2
*
* Copyright © 2015-2023 Jim Warner <james.warner@comcast.net>
* Copyright © 2015-2023 Craig Small <csmall@dropbear.xyz>
* Copyright © 2004-2006 Albert Cahalan
* Copyright © 2003 Chris Rivera
* Copyright © 2003 Fabian Frederick
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
2004-07-19 10:26:10 +05:30
#include <ctype.h>
2015-07-04 10:29:59 +05:30
#include <errno.h>
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
#include <fcntl.h>
#include <limits.h>
#include <stdio.h>
#include <stdlib.h>
2015-07-04 10:29:59 +05:30
#include <string.h>
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#include <time.h>
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
#include <unistd.h>
2015-07-04 10:29:59 +05:30
#include <sys/stat.h>
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
#include <sys/types.h>
#include "procps-private.h"
#include "slabinfo.h"
2015-07-04 10:29:59 +05:30
2015-07-04 10:29:59 +05:30
#define SLABINFO_FILE "/proc/slabinfo"
#define SLABINFO_LINE_LEN 2048
#define SLABINFO_NAME_LEN 128
2015-07-04 10:29:59 +05:30
#define STACKS_INCR 128 // amount reap stack allocations grow
/* ---------------------------------------------------------------------------- +
this #define will be used to help ensure that our Item_table is synchronized |
with all the enumerators found in the associated header file. It is intended |
to only be defined locally (and temporarily) at some point prior to release! | */
// #define ITEMTABLE_DEBUG //-------------------------------------------------- |
// ---------------------------------------------------------------------------- +
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
/*
Because 'select' could, at most, return only node[0] values and since 'reap' |
would be forced to duplicate global slabs stuff in every node results stack, |
the following #define can be used to enforce strictly logical return values. |
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
select: allow only SLABINFO & SLABS items
library: refactored some header file items and origins This commit is intended as a refinement of the patches mentioned below, where origins/sources of newlib items were added to the header files for user documentation. However, if those additions are to be truly effective, along with kernel documentation (where available), the following prerequisites must also have been satisfied: . our identifiers closely align with linux field names . our derived items are documented or self-documenting Satisfying those prerequisites prompted this patch and for these changes, kernel sources were emphasized over available documentation (shame on me, it should always have been so). And, while some 'new' fields were found to be conditional, they were included unconditionally. These changes appear more extensive than they actually need be since I have attempted to enforce some spacing conventions. So, I've summarize the significant things in the sections that follow. For a proper perspective, use: 'git diff --ignore-space-change' (good as alias). ___________________________________________ <PIDS> api This api is unique in that there exists many different file/directory origins subordinate to /proc/<pid>. And our item identifiers are sometimes coerced so as to be able to group related or similar enumerators together. So, users needed more help relating our identifiers to an actual documented field. Thus, we will now also add the field names as with 'stat: delayacct_blkio_ticks'. Each item ending with a '_C' now consistently includes both the parent's count/time plus waited for children. That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT since its original name is an implementation artifact. ___________________________________________ <STAT> api The only api change was to correct a typo ('dervied'). _________________________________________ <VMSTAT> api Even ignoring white space, this interface received the largest number of changes. Mostly, this was because of deficiencies in the proc(5) documentation. Recall that this documentation already sorely lacks any substance. Usually, just kernel releases are noted, not contents. When compared to kernel source, that proc(5) contained many non-existent fields and also omitted many others. ________________________________________ <MEMINFO> api Sadly, with this api many of the changes were simply a correction of some earlier 'human error' where several fields where hashed then tracked but never represented with an item enumerator in this meminfo.h header file. _______________________________________ <SLABINFO> api The 'SLABS' (summary) & 'SLABNODE' items were reversed since the former are derived from the separate caches. More significantly, those 'SLABNODE' guys were renamed to 'SLAB' since they concern individual caches and the concept of 'nodes' is really an implementation detail. Also, several enumerators were changed to more closely agree with official slabinfo(5) documentation referred to in what we're treating as a base document: proc(5). Lastly, while those 'SLABS' items are solely a product of our library and not represented in slabinfo(5), the names attempt to parallel those found as 'SLAB' items. ______________________________________ <DISKSTATS> api One enumeration identifier was changed so as to better reflect its relationship to that actual documentation: 'Documentation/iostats.txt', as referenced in proc(5). Reference(s): . 12/2018, item origins added (and commit msg history) commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14 . 01/2019, <stat> origins tweaked commit 201e816b26ddaccc923ec40977c92037cdd0c34e Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
reap: allow only SLABINFO & SLAB items
Without the #define, these functions always return something even if just 0. |
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
get: return only SLABS results, else 0
select: return only SLABINFO & SLABS results, else zero
library: refactored some header file items and origins This commit is intended as a refinement of the patches mentioned below, where origins/sources of newlib items were added to the header files for user documentation. However, if those additions are to be truly effective, along with kernel documentation (where available), the following prerequisites must also have been satisfied: . our identifiers closely align with linux field names . our derived items are documented or self-documenting Satisfying those prerequisites prompted this patch and for these changes, kernel sources were emphasized over available documentation (shame on me, it should always have been so). And, while some 'new' fields were found to be conditional, they were included unconditionally. These changes appear more extensive than they actually need be since I have attempted to enforce some spacing conventions. So, I've summarize the significant things in the sections that follow. For a proper perspective, use: 'git diff --ignore-space-change' (good as alias). ___________________________________________ <PIDS> api This api is unique in that there exists many different file/directory origins subordinate to /proc/<pid>. And our item identifiers are sometimes coerced so as to be able to group related or similar enumerators together. So, users needed more help relating our identifiers to an actual documented field. Thus, we will now also add the field names as with 'stat: delayacct_blkio_ticks'. Each item ending with a '_C' now consistently includes both the parent's count/time plus waited for children. That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT since its original name is an implementation artifact. ___________________________________________ <STAT> api The only api change was to correct a typo ('dervied'). _________________________________________ <VMSTAT> api Even ignoring white space, this interface received the largest number of changes. Mostly, this was because of deficiencies in the proc(5) documentation. Recall that this documentation already sorely lacks any substance. Usually, just kernel releases are noted, not contents. When compared to kernel source, that proc(5) contained many non-existent fields and also omitted many others. ________________________________________ <MEMINFO> api Sadly, with this api many of the changes were simply a correction of some earlier 'human error' where several fields where hashed then tracked but never represented with an item enumerator in this meminfo.h header file. _______________________________________ <SLABINFO> api The 'SLABS' (summary) & 'SLABNODE' items were reversed since the former are derived from the separate caches. More significantly, those 'SLABNODE' guys were renamed to 'SLAB' since they concern individual caches and the concept of 'nodes' is really an implementation detail. Also, several enumerators were changed to more closely agree with official slabinfo(5) documentation referred to in what we're treating as a base document: proc(5). Lastly, while those 'SLABS' items are solely a product of our library and not represented in slabinfo(5), the names attempt to parallel those found as 'SLAB' items. ______________________________________ <DISKSTATS> api One enumeration identifier was changed so as to better reflect its relationship to that actual documentation: 'Documentation/iostats.txt', as referenced in proc(5). Reference(s): . 12/2018, item origins added (and commit msg history) commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14 . 01/2019, <stat> origins tweaked commit 201e816b26ddaccc923ec40977c92037cdd0c34e Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
reap: return any requested, even when duplicated in each cache's stack */
//#define ENFORCE_LOGICAL // ensure only logical items accepted by select/reap
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabs_summ {
unsigned int nr_objs; // number of objects, among all caches
unsigned int nr_active_objs; // number of active objects, among all caches
unsigned int nr_pages; // number of pages consumed by all objects
unsigned int nr_slabs; // number of slabs, among all caches
unsigned int nr_active_slabs; // number of active slabs, among all caches
unsigned int nr_caches; // number of caches
unsigned int nr_active_caches; // number of active caches
unsigned int avg_obj_size; // average object size
unsigned int min_obj_size; // size of smallest object
unsigned int max_obj_size; // size of largest object
unsigned long active_size; // size of all active objects
unsigned long total_size; // size of all objects
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
};
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabs_node {
char name[SLABINFO_NAME_LEN+1]; // name of this cache
unsigned long cache_size; // size of entire cache
unsigned int nr_objs; // number of objects in this cache
unsigned int nr_active_objs; // number of active objects
unsigned int obj_size; // size of each object
unsigned int objs_per_slab; // number of objects per slab
unsigned int pages_per_slab; // number of pages per slab
unsigned int nr_slabs; // number of slabs in this cache
unsigned int nr_active_slabs; // number of active slabs
unsigned int use; // percent full: total / active
2015-07-04 10:29:59 +05:30
};
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabs_hist {
struct slabs_summ new;
struct slabs_summ old;
};
struct stacks_extent {
int ext_numstacks;
struct stacks_extent *next;
struct slabinfo_stack **stacks;
};
struct ext_support {
int numitems; // includes 'logical_end' delimiter
enum slabinfo_item *items; // includes 'logical_end' delimiter
struct stacks_extent *extents; // anchor for these extents
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#ifdef ENFORCE_LOGICAL
enum slabinfo_item lowest; // range of allowable enums
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
enum slabinfo_item highest;
#endif
};
struct fetch_support {
struct slabinfo_stack **anchor; // fetch consolidated extents
int n_alloc; // number of above pointers allocated
int n_inuse; // number of above pointers occupied
int n_alloc_save; // last known reap.stacks allocation
struct slabinfo_reaped results; // count + stacks for return to caller
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
};
struct slabinfo_info {
2015-07-04 10:29:59 +05:30
int refcount;
FILE *slabinfo_fp;
int nodes_alloc; // nodes alloc()ed
int nodes_used; // nodes using alloced memory
struct slabs_node *nodes; // first slabnode of this list
struct slabs_hist slabs; // new/old slabs_summ data
struct ext_support select_ext; // supports concurrent select/reap
struct ext_support fetch_ext; // supports concurrent select/reap
struct fetch_support fetch; // support for procps_slabinfo_reap
struct slabs_node nul_node; // used by slabinfo_get/select
struct slabinfo_result get_this; // used by slabinfo_get
time_t sav_secs; // used by slabinfo_get
};
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
// ___ Results 'Set' Support ||||||||||||||||||||||||||||||||||||||||||||||||||
#define setNAME(e) set_ ## e
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#define setDECL(e) static void setNAME(e) \
(struct slabinfo_result *R, struct slabs_hist *S, struct slabs_node *N)
// regular assignment
#define REG_set(e,t,x) setDECL(e) { (void)N; R->result. t = S->new. x; }
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#define NOD_set(e,t,x) setDECL(e) { (void)S; R->result. t = N-> x; }
// delta assignment
#define HST_set(e,t,x) setDECL(e) { (void)N; R->result. t = (signed long)S->new. x - S->old. x; }
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
setDECL(SLABINFO_noop) { (void)R; (void)S; (void)N; }
setDECL(SLABINFO_extra) { (void)S; (void)N; R->result.ul_int = 0; }
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
library: refactored some header file items and origins This commit is intended as a refinement of the patches mentioned below, where origins/sources of newlib items were added to the header files for user documentation. However, if those additions are to be truly effective, along with kernel documentation (where available), the following prerequisites must also have been satisfied: . our identifiers closely align with linux field names . our derived items are documented or self-documenting Satisfying those prerequisites prompted this patch and for these changes, kernel sources were emphasized over available documentation (shame on me, it should always have been so). And, while some 'new' fields were found to be conditional, they were included unconditionally. These changes appear more extensive than they actually need be since I have attempted to enforce some spacing conventions. So, I've summarize the significant things in the sections that follow. For a proper perspective, use: 'git diff --ignore-space-change' (good as alias). ___________________________________________ <PIDS> api This api is unique in that there exists many different file/directory origins subordinate to /proc/<pid>. And our item identifiers are sometimes coerced so as to be able to group related or similar enumerators together. So, users needed more help relating our identifiers to an actual documented field. Thus, we will now also add the field names as with 'stat: delayacct_blkio_ticks'. Each item ending with a '_C' now consistently includes both the parent's count/time plus waited for children. That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT since its original name is an implementation artifact. ___________________________________________ <STAT> api The only api change was to correct a typo ('dervied'). _________________________________________ <VMSTAT> api Even ignoring white space, this interface received the largest number of changes. Mostly, this was because of deficiencies in the proc(5) documentation. Recall that this documentation already sorely lacks any substance. Usually, just kernel releases are noted, not contents. When compared to kernel source, that proc(5) contained many non-existent fields and also omitted many others. ________________________________________ <MEMINFO> api Sadly, with this api many of the changes were simply a correction of some earlier 'human error' where several fields where hashed then tracked but never represented with an item enumerator in this meminfo.h header file. _______________________________________ <SLABINFO> api The 'SLABS' (summary) & 'SLABNODE' items were reversed since the former are derived from the separate caches. More significantly, those 'SLABNODE' guys were renamed to 'SLAB' since they concern individual caches and the concept of 'nodes' is really an implementation detail. Also, several enumerators were changed to more closely agree with official slabinfo(5) documentation referred to in what we're treating as a base document: proc(5). Lastly, while those 'SLABS' items are solely a product of our library and not represented in slabinfo(5), the names attempt to parallel those found as 'SLAB' items. ______________________________________ <DISKSTATS> api One enumeration identifier was changed so as to better reflect its relationship to that actual documentation: 'Documentation/iostats.txt', as referenced in proc(5). Reference(s): . 12/2018, item origins added (and commit msg history) commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14 . 01/2019, <stat> origins tweaked commit 201e816b26ddaccc923ec40977c92037cdd0c34e Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
NOD_set(SLAB_NAME, str, name)
NOD_set(SLAB_NUM_OBJS, u_int, nr_objs)
NOD_set(SLAB_ACTIVE_OBJS, u_int, nr_active_objs)
NOD_set(SLAB_OBJ_SIZE, u_int, obj_size)
NOD_set(SLAB_OBJ_PER_SLAB, u_int, objs_per_slab)
NOD_set(SLAB_NUMS_SLABS, u_int, nr_slabs)
NOD_set(SLAB_ACTIVE_SLABS, u_int, nr_active_slabs)
NOD_set(SLAB_PAGES_PER_SLAB, u_int, pages_per_slab)
NOD_set(SLAB_PERCENT_USED, u_int, use)
NOD_set(SLAB_SIZE_TOTAL, ul_int, cache_size)
REG_set(SLABS_CACHES_TOTAL, u_int, nr_caches)
REG_set(SLABS_CACHES_ACTIVE, u_int, nr_active_caches)
REG_set(SLABS_NUM_OBJS, u_int, nr_objs)
REG_set(SLABS_ACTIVE_OBJS, u_int, nr_active_objs)
REG_set(SLABS_OBJ_SIZE_AVG, u_int, avg_obj_size)
REG_set(SLABS_OBJ_SIZE_MIN, u_int, min_obj_size)
REG_set(SLABS_OBJ_SIZE_MAX, u_int, max_obj_size)
REG_set(SLABS_NUMS_SLABS, u_int, nr_slabs)
REG_set(SLABS_ACTIVE_SLABS, u_int, nr_active_slabs)
REG_set(SLABS_PAGES_TOTAL, u_int, nr_pages)
REG_set(SLABS_SIZE_ACTIVE, ul_int, active_size)
REG_set(SLABS_SIZE_TOTAL, ul_int, total_size)
HST_set(SLABS_DELTA_CACHES_TOTAL, s_int, nr_caches)
HST_set(SLABS_DELTA_CACHES_ACTIVE, s_int, nr_active_caches)
HST_set(SLABS_DELTA_NUM_OBJS, s_int, nr_objs)
HST_set(SLABS_DELTA_ACTIVE_OBJS, s_int, nr_active_objs)
HST_set(SLABS_DELTA_OBJ_SIZE_AVG, s_int, avg_obj_size)
HST_set(SLABS_DELTA_OBJ_SIZE_MIN, s_int, min_obj_size)
HST_set(SLABS_DELTA_OBJ_SIZE_MAX, s_int, max_obj_size)
HST_set(SLABS_DELTA_NUMS_SLABS, s_int, nr_slabs)
HST_set(SLABS_DELTA_ACTIVE_SLABS, s_int, nr_active_slabs)
HST_set(SLABS_DELTA_PAGES_TOTAL, s_int, nr_pages)
HST_set(SLABS_DELTA_SIZE_ACTIVE, s_int, active_size)
HST_set(SLABS_DELTA_SIZE_TOTAL, s_int, total_size)
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#undef setDECL
#undef REG_set
#undef NOD_set
#undef HST_set
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
// ___ Sorting Support ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
struct sort_parms {
int offset;
enum slabinfo_sort_order order;
2015-07-04 10:29:59 +05:30
};
#define srtNAME(t) sort_slabinfo_ ## t
#define srtDECL(t) static int srtNAME(t) \
(const struct slabinfo_stack **A, const struct slabinfo_stack **B, struct sort_parms *P)
srtDECL(u_int) {
const struct slabinfo_result *a = (*A)->head + P->offset; \
const struct slabinfo_result *b = (*B)->head + P->offset; \
if ( a->result.u_int > b->result.u_int ) return P->order > 0 ? 1 : -1; \
if ( a->result.u_int < b->result.u_int ) return P->order > 0 ? -1 : 1; \
return 0;
}
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
srtDECL(ul_int) {
const struct slabinfo_result *a = (*A)->head + P->offset; \
const struct slabinfo_result *b = (*B)->head + P->offset; \
if ( a->result.ul_int > b->result.ul_int ) return P->order > 0 ? 1 : -1; \
if ( a->result.ul_int < b->result.ul_int ) return P->order > 0 ? -1 : 1; \
return 0;
}
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
srtDECL(str) {
const struct slabinfo_result *a = (*A)->head + P->offset;
const struct slabinfo_result *b = (*B)->head + P->offset;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return P->order * strcoll(a->result.str, b->result.str);
}
srtDECL(noop) { \
(void)A; (void)B; (void)P; \
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return 0;
}
#undef srtDECL
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
// ___ Controlling Table ||||||||||||||||||||||||||||||||||||||||||||||||||||||
typedef void (*SET_t)(struct slabinfo_result *, struct slabs_hist *, struct slabs_node *);
#ifdef ITEMTABLE_DEBUG
#define RS(e) (SET_t)setNAME(e), e, STRINGIFY(e)
#else
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#define RS(e) (SET_t)setNAME(e)
#endif
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
typedef int (*QSR_t)(const void *, const void *, void *);
#define QS(t) (QSR_t)srtNAME(t)
#define TS(t) STRINGIFY(t)
#define TS_noop ""
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
/*
* Need it be said?
* This table must be kept in the exact same order as
* those *enum slabinfo_item* guys ! */
static struct {
SET_t setsfunc; // the actual result setting routine
#ifdef ITEMTABLE_DEBUG
int enumnumb; // enumerator (must match position!)
char *enum2str; // enumerator name as a char* string
#endif
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
QSR_t sortfunc; // sort cmp func for a specific type
char *type2str; // the result type as a string value
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
} Item_table[] = {
library: refactored some header file items and origins This commit is intended as a refinement of the patches mentioned below, where origins/sources of newlib items were added to the header files for user documentation. However, if those additions are to be truly effective, along with kernel documentation (where available), the following prerequisites must also have been satisfied: . our identifiers closely align with linux field names . our derived items are documented or self-documenting Satisfying those prerequisites prompted this patch and for these changes, kernel sources were emphasized over available documentation (shame on me, it should always have been so). And, while some 'new' fields were found to be conditional, they were included unconditionally. These changes appear more extensive than they actually need be since I have attempted to enforce some spacing conventions. So, I've summarize the significant things in the sections that follow. For a proper perspective, use: 'git diff --ignore-space-change' (good as alias). ___________________________________________ <PIDS> api This api is unique in that there exists many different file/directory origins subordinate to /proc/<pid>. And our item identifiers are sometimes coerced so as to be able to group related or similar enumerators together. So, users needed more help relating our identifiers to an actual documented field. Thus, we will now also add the field names as with 'stat: delayacct_blkio_ticks'. Each item ending with a '_C' now consistently includes both the parent's count/time plus waited for children. That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT since its original name is an implementation artifact. ___________________________________________ <STAT> api The only api change was to correct a typo ('dervied'). _________________________________________ <VMSTAT> api Even ignoring white space, this interface received the largest number of changes. Mostly, this was because of deficiencies in the proc(5) documentation. Recall that this documentation already sorely lacks any substance. Usually, just kernel releases are noted, not contents. When compared to kernel source, that proc(5) contained many non-existent fields and also omitted many others. ________________________________________ <MEMINFO> api Sadly, with this api many of the changes were simply a correction of some earlier 'human error' where several fields where hashed then tracked but never represented with an item enumerator in this meminfo.h header file. _______________________________________ <SLABINFO> api The 'SLABS' (summary) & 'SLABNODE' items were reversed since the former are derived from the separate caches. More significantly, those 'SLABNODE' guys were renamed to 'SLAB' since they concern individual caches and the concept of 'nodes' is really an implementation detail. Also, several enumerators were changed to more closely agree with official slabinfo(5) documentation referred to in what we're treating as a base document: proc(5). Lastly, while those 'SLABS' items are solely a product of our library and not represented in slabinfo(5), the names attempt to parallel those found as 'SLAB' items. ______________________________________ <DISKSTATS> api One enumeration identifier was changed so as to better reflect its relationship to that actual documentation: 'Documentation/iostats.txt', as referenced in proc(5). Reference(s): . 12/2018, item origins added (and commit msg history) commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14 . 01/2019, <stat> origins tweaked commit 201e816b26ddaccc923ec40977c92037cdd0c34e Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
/* setsfunc sortfunc type2str
------------------------------ ----------- ---------- */
{ RS(SLABINFO_noop), QS(noop), TS_noop },
{ RS(SLABINFO_extra), QS(ul_int), TS_noop },
library: refactored some header file items and origins This commit is intended as a refinement of the patches mentioned below, where origins/sources of newlib items were added to the header files for user documentation. However, if those additions are to be truly effective, along with kernel documentation (where available), the following prerequisites must also have been satisfied: . our identifiers closely align with linux field names . our derived items are documented or self-documenting Satisfying those prerequisites prompted this patch and for these changes, kernel sources were emphasized over available documentation (shame on me, it should always have been so). And, while some 'new' fields were found to be conditional, they were included unconditionally. These changes appear more extensive than they actually need be since I have attempted to enforce some spacing conventions. So, I've summarize the significant things in the sections that follow. For a proper perspective, use: 'git diff --ignore-space-change' (good as alias). ___________________________________________ <PIDS> api This api is unique in that there exists many different file/directory origins subordinate to /proc/<pid>. And our item identifiers are sometimes coerced so as to be able to group related or similar enumerators together. So, users needed more help relating our identifiers to an actual documented field. Thus, we will now also add the field names as with 'stat: delayacct_blkio_ticks'. Each item ending with a '_C' now consistently includes both the parent's count/time plus waited for children. That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT since its original name is an implementation artifact. ___________________________________________ <STAT> api The only api change was to correct a typo ('dervied'). _________________________________________ <VMSTAT> api Even ignoring white space, this interface received the largest number of changes. Mostly, this was because of deficiencies in the proc(5) documentation. Recall that this documentation already sorely lacks any substance. Usually, just kernel releases are noted, not contents. When compared to kernel source, that proc(5) contained many non-existent fields and also omitted many others. ________________________________________ <MEMINFO> api Sadly, with this api many of the changes were simply a correction of some earlier 'human error' where several fields where hashed then tracked but never represented with an item enumerator in this meminfo.h header file. _______________________________________ <SLABINFO> api The 'SLABS' (summary) & 'SLABNODE' items were reversed since the former are derived from the separate caches. More significantly, those 'SLABNODE' guys were renamed to 'SLAB' since they concern individual caches and the concept of 'nodes' is really an implementation detail. Also, several enumerators were changed to more closely agree with official slabinfo(5) documentation referred to in what we're treating as a base document: proc(5). Lastly, while those 'SLABS' items are solely a product of our library and not represented in slabinfo(5), the names attempt to parallel those found as 'SLAB' items. ______________________________________ <DISKSTATS> api One enumeration identifier was changed so as to better reflect its relationship to that actual documentation: 'Documentation/iostats.txt', as referenced in proc(5). Reference(s): . 12/2018, item origins added (and commit msg history) commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14 . 01/2019, <stat> origins tweaked commit 201e816b26ddaccc923ec40977c92037cdd0c34e Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
{ RS(SLAB_NAME), QS(str), TS(str) },
{ RS(SLAB_NUM_OBJS), QS(u_int), TS(u_int) },
{ RS(SLAB_ACTIVE_OBJS), QS(u_int), TS(u_int) },
{ RS(SLAB_OBJ_SIZE), QS(u_int), TS(u_int) },
{ RS(SLAB_OBJ_PER_SLAB), QS(u_int), TS(u_int) },
{ RS(SLAB_NUMS_SLABS), QS(u_int), TS(u_int) },
{ RS(SLAB_ACTIVE_SLABS), QS(u_int), TS(u_int) },
{ RS(SLAB_PAGES_PER_SLAB), QS(u_int), TS(u_int) },
{ RS(SLAB_PERCENT_USED), QS(u_int), TS(u_int) },
{ RS(SLAB_SIZE_TOTAL), QS(ul_int), TS(ul_int) },
{ RS(SLABS_CACHES_TOTAL), QS(noop), TS(u_int) },
{ RS(SLABS_CACHES_ACTIVE), QS(noop), TS(u_int) },
{ RS(SLABS_NUM_OBJS), QS(noop), TS(u_int) },
{ RS(SLABS_ACTIVE_OBJS), QS(noop), TS(u_int) },
{ RS(SLABS_OBJ_SIZE_AVG), QS(noop), TS(u_int) },
{ RS(SLABS_OBJ_SIZE_MIN), QS(noop), TS(u_int) },
{ RS(SLABS_OBJ_SIZE_MAX), QS(noop), TS(u_int) },
{ RS(SLABS_NUMS_SLABS), QS(noop), TS(u_int) },
{ RS(SLABS_ACTIVE_SLABS), QS(noop), TS(u_int) },
{ RS(SLABS_PAGES_TOTAL), QS(noop), TS(u_int) },
{ RS(SLABS_SIZE_ACTIVE), QS(noop), TS(ul_int) },
{ RS(SLABS_SIZE_TOTAL), QS(noop), TS(ul_int) },
{ RS(SLABS_DELTA_CACHES_TOTAL), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_CACHES_ACTIVE), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_NUM_OBJS), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_ACTIVE_OBJS), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_OBJ_SIZE_AVG), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_OBJ_SIZE_MIN), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_OBJ_SIZE_MAX), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_NUMS_SLABS), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_ACTIVE_SLABS), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_PAGES_TOTAL), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_SIZE_ACTIVE), QS(noop), TS(s_int) },
{ RS(SLABS_DELTA_SIZE_TOTAL), QS(noop), TS(s_int) },
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
};
/* please note,
* this enum MUST be 1 greater than the highest value of any enum */
enum slabinfo_item SLABINFO_logical_end = MAXTABLE(Item_table);
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#undef setNAME
#undef srtNAME
#undef RS
#undef QS
// ___ Private Functions ||||||||||||||||||||||||||||||||||||||||||||||||||||||
// --- slabnode specific support ----------------------------------------------
2015-07-04 10:29:59 +05:30
/* Alloc up more slabnode memory, if required
*/
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
static int alloc_slabnodes (
struct slabinfo_info *info)
{
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabs_node *new_nodes;
2015-07-04 10:29:59 +05:30
int new_count;
if (info->nodes_used < info->nodes_alloc)
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 1;
2015-07-04 10:29:59 +05:30
/* Increment the allocated number of slabs */
new_count = info->nodes_alloc * 5/4+30;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
new_nodes = realloc(info->nodes, sizeof(struct slabs_node) * new_count);
if (!new_nodes)
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 0;
2015-07-04 10:29:59 +05:30
info->nodes = new_nodes;
info->nodes_alloc = new_count;
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 1;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
} // end: alloc_slabnodes
/*
2015-07-04 10:29:59 +05:30
* get_slabnode - allocate slab_info structures using a free list
*
* In the fast path, we simply return a node off the free list. In the slow
* list, we malloc() a new node. The free list is never automatically reaped,
* both for simplicity and because the number of slab caches is fairly
* constant.
*/
static int get_slabnode (
struct slabinfo_info *info,
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabs_node **node)
{
2015-07-04 10:29:59 +05:30
if (info->nodes_used == info->nodes_alloc) {
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
if (!alloc_slabnodes(info))
return 0; // here, errno was set to ENOMEM
2015-07-04 10:29:59 +05:30
}
*node = &(info->nodes[info->nodes_used++]);
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 1;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
} // end: get_slabnode
2015-07-04 10:29:59 +05:30
/* parse_slabinfo20:
*
* Actual parse routine for slabinfo 2.x (2.6 kernels)
* Note: difference between 2.0 and 2.1 is in the ": globalstat" part where version 2.1
* has extra column <nodeallocs>. We don't use ": globalstat" part in both versions.
*
* Formats (we don't use "statistics" extensions)
*
* slabinfo - version: 2.1
* # name <active_objs> <num_objs> <objsize> <objperslab> <pagesperslab> \
* : tunables <batchcount> <limit> <sharedfactor> \
* : slabdata <active_slabs> <num_slabs> <sharedavail>
*
* slabinfo - version: 2.1 (statistics)
* # name <active_objs> <num_objs> <objsize> <objperslab> <pagesperslab> \
* : tunables <batchcount> <limit> <sharedfactor> \
* : slabdata <active_slabs> <num_slabs> <sharedavail> \
* : globalstat <listallocs> <maxobjs> <grown> <reaped> <error> <maxfreeable> <freelimit> <nodeallocs> \
* : cpustat <allochit> <allocmiss> <freehit> <freemiss>
*
* slabinfo - version: 2.0
* # name <active_objs> <num_objs> <objsize> <objperslab> <pagesperslab> \
* : tunables <batchcount> <limit> <sharedfactor> \
* : slabdata <active_slabs> <num_slabs> <sharedavail>
*
* slabinfo - version: 2.0 (statistics)
* # name <active_objs> <num_objs> <objsize> <objperslab> <pagesperslab> \
* : tunables <batchcount> <limit> <sharedfactor> \
* : slabdata <active_slabs> <num_slabs> <sharedavail> \
* : globalstat <listallocs> <maxobjs> <grown> <reaped> <error> <maxfreeable> <freelimit> \
* : cpustat <allochit> <allocmiss> <freehit> <freemiss>
*/
static int parse_slabinfo20 (
struct slabinfo_info *info)
{
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabs_node *node;
2015-07-04 10:29:59 +05:30
char buffer[SLABINFO_LINE_LEN];
int page_size = getpagesize();
struct slabs_summ *slabs = &(info->slabs.new);
2015-07-04 10:29:59 +05:30
slabs->min_obj_size = INT_MAX;
slabs->max_obj_size = 0;
2015-07-04 10:29:59 +05:30
while (fgets(buffer, SLABINFO_LINE_LEN, info->slabinfo_fp )) {
if (buffer[0] == '#')
continue;
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
if (!get_slabnode(info, &node))
return 1; // here, errno was set to ENOMEM
2015-07-04 10:29:59 +05:30
if (sscanf(buffer,
"%" STRINGIFY(SLABINFO_NAME_LEN) "s" \
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
"%u %u %u %u %u : tunables %*u %*u %*u : slabdata %u %u %*u",
node->name,
2015-07-04 10:29:59 +05:30
&node->nr_active_objs, &node->nr_objs,
&node->obj_size, &node->objs_per_slab,
&node->pages_per_slab, &node->nr_active_slabs,
&node->nr_slabs) < 8) {
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
errno = ERANGE;
return 1;
2015-07-04 10:29:59 +05:30
}
if (!node->name[0])
snprintf(node->name, sizeof(node->name), "%s", "unknown");
if (node->obj_size < slabs->min_obj_size)
slabs->min_obj_size = node->obj_size;
if (node->obj_size > slabs->max_obj_size)
slabs->max_obj_size = node->obj_size;
2015-07-04 10:29:59 +05:30
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
node->cache_size = (unsigned long)node->nr_slabs * node->pages_per_slab * page_size;
2015-07-04 10:29:59 +05:30
if (node->nr_objs) {
node->use = (unsigned int)(100 * ((float)node->nr_active_objs / node->nr_objs));
slabs->nr_active_caches++;
2015-07-04 10:29:59 +05:30
} else
node->use = 0;
slabs->nr_objs += node->nr_objs;
slabs->nr_active_objs += node->nr_active_objs;
slabs->total_size += (unsigned long)node->nr_objs * node->obj_size;
slabs->active_size += (unsigned long)node->nr_active_objs * node->obj_size;
slabs->nr_pages += node->nr_slabs * node->pages_per_slab;
slabs->nr_slabs += node->nr_slabs;
slabs->nr_active_slabs += node->nr_active_slabs;
slabs->nr_caches++;
2015-07-04 10:29:59 +05:30
}
if (slabs->nr_objs)
slabs->avg_obj_size = slabs->total_size / slabs->nr_objs;
2015-07-04 10:29:59 +05:30
return 0;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
} // end: parse_slabinfo20
2015-07-04 10:29:59 +05:30
/* slabinfo_read_failed():
2015-07-04 10:29:59 +05:30
*
* Read the data out of /proc/slabinfo putting the information
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
* into the supplied info container
2015-07-04 10:29:59 +05:30
*
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
* Returns: 0 on success, 1 on error
*/
static int slabinfo_read_failed (
struct slabinfo_info *info)
2015-07-04 10:29:59 +05:30
{
char line[SLABINFO_LINE_LEN];
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
int major, minor;
2015-07-04 10:29:59 +05:30
memcpy(&info->slabs.old, &info->slabs.new, sizeof(struct slabs_summ));
memset(&(info->slabs.new), 0, sizeof(struct slabs_summ));
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
if (!alloc_slabnodes(info))
return 1; // here, errno was set to ENOMEM
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
memset(info->nodes, 0, sizeof(struct slabs_node)*info->nodes_alloc);
info->nodes_used = 0;
2015-07-04 10:29:59 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if (NULL == info->slabinfo_fp
&& (info->slabinfo_fp = fopen(SLABINFO_FILE, "r")) == NULL)
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 1;
2015-07-04 10:29:59 +05:30
if (fseek(info->slabinfo_fp, 0L, SEEK_SET) < 0)
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 1;
2015-07-04 10:29:59 +05:30
/* Parse the version string */
if (!fgets(line, SLABINFO_LINE_LEN, info->slabinfo_fp))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 1;
2015-07-04 10:29:59 +05:30
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
if (2 != sscanf(line, "slabinfo - version: %d.%d", &major, &minor)
|| (major != 2)) {
errno = ERANGE;
return 1;
}
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return parse_slabinfo20(info);
} // end: slabinfo_read_failed
2015-07-04 10:29:59 +05:30
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
// ___ Private Functions ||||||||||||||||||||||||||||||||||||||||||||||||||||||
// --- generalized support ----------------------------------------------------
2015-07-04 10:29:59 +05:30
static inline void slabinfo_assign_results (
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabinfo_stack *stack,
struct slabs_hist *summ,
struct slabs_node *node)
2015-07-04 10:29:59 +05:30
{
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabinfo_result *this = stack->head;
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
for (;;) {
enum slabinfo_item item = this->item;
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
if (item >= SLABINFO_logical_end)
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
break;
Item_table[item].setsfunc(this, summ, node);
++this;
2015-07-04 10:29:59 +05:30
}
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return;
} // end: slabinfo_assign_results
2015-07-04 10:29:59 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
static void slabinfo_extents_free_all (
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct ext_support *this)
{
while (this->extents) {
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct stacks_extent *p = this->extents;
this->extents = this->extents->next;
free(p);
};
} // end: slabinfo_extents_free_all
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
static inline struct slabinfo_result *slabinfo_itemize_stack (
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabinfo_result *p,
int depth,
enum slabinfo_item *items)
{
struct slabinfo_result *p_sav = p;
int i;
for (i = 0; i < depth; i++) {
p->item = items[i];
++p;
2015-07-04 10:29:59 +05:30
}
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return p_sav;
} // end: slabinfo_itemize_stack
2015-07-04 10:29:59 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
static inline int slabinfo_items_check_failed (
struct ext_support *this,
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
enum slabinfo_item *items,
int numitems)
{
int i;
2015-07-04 10:29:59 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
/* if an enum is passed instead of an address of one or more enums, ol' gcc
* will silently convert it to an address (possibly NULL). only clang will
* offer any sort of warning like the following:
*
* warning: incompatible integer to pointer conversion passing 'int' to parameter of type 'enum slabinfo_item *'
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
* my_stack = procps_slabinfo_select(info, SLABINFO_noop, num);
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
* ^~~~~~~~~~~~~~~~
*/
if (numitems < 1
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
|| (void *)items < (void *)(unsigned long)(2 * SLABINFO_logical_end))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 1;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
for (i = 0; i < numitems; i++) {
#ifdef ENFORCE_LOGICAL
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
if (items[i] == SLABINFO_noop
|| (items[i] == SLABINFO_extra))
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
continue;
if (items[i] < this->lowest
|| (items[i] > this->highest))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 1;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#else
// a slabinfo_item is currently unsigned, but we'll protect our future
if (items[i] < 0)
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 1;
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
if (items[i] >= SLABINFO_logical_end)
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return 1;
(void)this;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#endif
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
}
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return 0;
} // end: slabinfo_items_check_failed
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
/*
* slabinfo_stacks_alloc():
*
* Allocate and initialize one or more stacks each of which is anchored in an
* associated context structure.
*
* All such stacks will have their result structures properly primed with
* 'items', while the result itself will be zeroed.
*
* Returns a stacks_extent struct anchoring the 'heads' of each new stack.
*/
static struct stacks_extent *slabinfo_stacks_alloc (
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct ext_support *this,
int maxstacks)
2015-07-04 10:29:59 +05:30
{
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct stacks_extent *p_blob;
struct slabinfo_stack **p_vect;
struct slabinfo_stack *p_head;
size_t vect_size, head_size, list_size, blob_size;
void *v_head, *v_list;
int i;
vect_size = sizeof(void *) * maxstacks; // size of the addr vectors |
vect_size += sizeof(void *); // plus NULL addr delimiter |
head_size = sizeof(struct slabinfo_stack); // size of that head struct |
list_size = sizeof(struct slabinfo_result)*this->numitems; // any single results stack |
blob_size = sizeof(struct stacks_extent); // the extent anchor itself |
blob_size += vect_size; // plus room for addr vects |
blob_size += head_size * maxstacks; // plus room for head thing |
blob_size += list_size * maxstacks; // plus room for our stacks |
/* note: all of our memory is allocated in one single blob, facilitating a later free(). |
as a minimum, it is important that those result structures themselves always be |
contiguous within each stack since they are accessed through relative position. | */
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if (NULL == (p_blob = calloc(1, blob_size)))
return NULL;
p_blob->next = this->extents; // push this extent onto... |
this->extents = p_blob; // ...some existing extents |
p_vect = (void *)p_blob + sizeof(struct stacks_extent); // prime our vector pointer |
p_blob->stacks = p_vect; // set actual vectors start |
v_head = (void *)p_vect + vect_size; // prime head pointer start |
v_list = v_head + (head_size * maxstacks); // prime our stacks pointer |
2015-07-04 10:29:59 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
for (i = 0; i < maxstacks; i++) {
p_head = (struct slabinfo_stack *)v_head;
p_head->head = slabinfo_itemize_stack((struct slabinfo_result *)v_list, this->numitems, this->items);
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
p_blob->stacks[i] = p_head;
v_list += list_size;
v_head += head_size;
}
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
p_blob->ext_numstacks = maxstacks;
return p_blob;
} // end: slabinfo_stacks_alloc
2015-07-04 10:29:59 +05:30
static int slabinfo_stacks_fetch (
struct slabinfo_info *info)
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
{
#define n_alloc info->fetch.n_alloc
#define n_inuse info->fetch.n_inuse
#define n_saved info->fetch.n_alloc_save
struct stacks_extent *ext;
// initialize stuff -----------------------------------
if (!info->fetch.anchor) {
if (!(info->fetch.anchor = calloc(sizeof(void *), STACKS_INCR)))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return -1;
n_alloc = STACKS_INCR;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
}
if (!info->fetch_ext.extents) {
if (!(ext = slabinfo_stacks_alloc(&info->fetch_ext, n_alloc)))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return -1; // here, errno was set to ENOMEM
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
memcpy(info->fetch.anchor, ext->stacks, sizeof(void *) * n_alloc);
}
// iterate stuff --------------------------------------
n_inuse = 0;
while (n_inuse < info->nodes_used) {
if (!(n_inuse < n_alloc)) {
n_alloc += STACKS_INCR;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if ((!(info->fetch.anchor = realloc(info->fetch.anchor, sizeof(void *) * n_alloc)))
|| (!(ext = slabinfo_stacks_alloc(&info->fetch_ext, STACKS_INCR))))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return -1; // here, errno was set to ENOMEM
memcpy(info->fetch.anchor + n_inuse, ext->stacks, sizeof(void *) * STACKS_INCR);
}
slabinfo_assign_results(info->fetch.anchor[n_inuse], &info->slabs, &info->nodes[n_inuse]);
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
++n_inuse;
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
}
2015-07-04 10:29:59 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
// finalize stuff -------------------------------------
/* note: we go to this trouble of maintaining a duplicate of the consolidated |
extent stacks addresses represented as our 'anchor' since these ptrs |
are exposed to a user (um, not that we don't trust 'em or anything). |
plus, we can NULL delimit these ptrs which we couldn't do otherwise. | */
if (n_saved < n_inuse + 1) {
n_saved = n_inuse + 1;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if (!(info->fetch.results.stacks = realloc(info->fetch.results.stacks, sizeof(void *) * n_saved)))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return -1;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
}
memcpy(info->fetch.results.stacks, info->fetch.anchor, sizeof(void *) * n_inuse);
info->fetch.results.stacks[n_inuse] = NULL;
info->fetch.results.total = n_inuse;
return n_inuse;
#undef n_alloc
#undef n_inuse
#undef n_saved
} // end: slabinfo_stacks_fetch
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
static int slabinfo_stacks_reconfig_maybe (
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct ext_support *this,
enum slabinfo_item *items,
int numitems)
2015-07-04 10:29:59 +05:30
{
if (slabinfo_items_check_failed(this, items, numitems))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return -1;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
/* is this the first time or have things changed since we were last called?
if so, gotta' redo all of our stacks stuff ... */
if (this->numitems != numitems + 1
|| memcmp(this->items, items, sizeof(enum slabinfo_item) * numitems)) {
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
// allow for our SLABINFO_logical_end
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if (!(this->items = realloc(this->items, sizeof(enum slabinfo_item) * (numitems + 1))))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return -1;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
memcpy(this->items, items, sizeof(enum slabinfo_item) * numitems);
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
this->items[numitems] = SLABINFO_logical_end;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
this->numitems = numitems + 1;
slabinfo_extents_free_all(this);
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return 1;
}
return 0;
} // end: slabinfo_stacks_reconfig_maybe
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
// ___ Public Functions |||||||||||||||||||||||||||||||||||||||||||||||||||||||
2015-07-04 10:29:59 +05:30
library: more tweaks for code and/or comments, 3rd gen Following is a summary of significant changes (if any) to each of these now upgraded 3rd gen library modules. <meminfo> ............................................ . eliminated duplicate decl of 'struct procps_meminfo' . standardized/normalized results struct union members . added 'std' & 'var' dividers in .c file, like <pids> . how did i miss relocating all these friggin' #undefs . cleanup 'get' return logic (remove a redundant 'if') <pids> ............................................... . repositioned the procps_pidsinfo structure in header . removed the extra trailing comma from enum pids_item . standardized/normalized results struct union members <slabinfo> ........................................... . corrected comment typo (jeeze, in an 'aligned' para) . standardized/normalized results struct union members . added 'std' & 'var' dividers in .c file, like <pids> . removed an obsolete #undef from procps_slabinfo_sort . cleanup 'get' return logic (remove a redundant 'if') <stat> ............................................... . how did i miss relocating all these friggin' #undefs . corrected an initialization fencepost used with numa <=== see Craig, here's a bug fix . removed the extra trailing comma from enum stat_item . standardized/normalized results struct union members . added 'std' & 'var' dividers in .c file, like <pids> . strengthen those parm checks in procps_stat_get func . cleanup 'get' return logic (remove a redundant 'if') <vmstat> ............................................. . standardized/normalized results struct union members . added 'std' & 'var' dividers in .c file, like <pids> . cleanup 'get' return logic (remove a redundant 'if') [ virtually all of these tweaks reflect the author's ] [ continuing pursuit of an unreasonable goal -- that ] [ of a 'perfect' (plus 'pretty') C language program! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-14 10:30:00 +05:30
// --- standard required functions --------------------------------------------
/*
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
* procps_slabinfo_new():
*
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
* @info: location of returned new structure
*
* Returns: < 0 on failure, 0 on success along with
* a pointer to a new context struct
*/
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
PROCPS_EXPORT int procps_slabinfo_new (
struct slabinfo_info **info)
2015-07-04 10:29:59 +05:30
{
struct slabinfo_info *p;
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
#ifdef ITEMTABLE_DEBUG
int i, failed = 0;
for (i = 0; i < MAXTABLE(Item_table); i++) {
if (i != Item_table[i].enumnumb) {
fprintf(stderr, "%s: enum/table error: Item_table[%d] was %s, but its value is %d\n"
, __FILE__, i, Item_table[i].enum2str, Item_table[i].enumnumb);
failed = 1;
}
}
if (failed) _Exit(EXIT_FAILURE);
#endif
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
if (info == NULL || *info != NULL)
return -EINVAL;
if (!(p = calloc(1, sizeof(struct slabinfo_info))))
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return -ENOMEM;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#ifdef ENFORCE_LOGICAL
library: refactored some header file items and origins This commit is intended as a refinement of the patches mentioned below, where origins/sources of newlib items were added to the header files for user documentation. However, if those additions are to be truly effective, along with kernel documentation (where available), the following prerequisites must also have been satisfied: . our identifiers closely align with linux field names . our derived items are documented or self-documenting Satisfying those prerequisites prompted this patch and for these changes, kernel sources were emphasized over available documentation (shame on me, it should always have been so). And, while some 'new' fields were found to be conditional, they were included unconditionally. These changes appear more extensive than they actually need be since I have attempted to enforce some spacing conventions. So, I've summarize the significant things in the sections that follow. For a proper perspective, use: 'git diff --ignore-space-change' (good as alias). ___________________________________________ <PIDS> api This api is unique in that there exists many different file/directory origins subordinate to /proc/<pid>. And our item identifiers are sometimes coerced so as to be able to group related or similar enumerators together. So, users needed more help relating our identifiers to an actual documented field. Thus, we will now also add the field names as with 'stat: delayacct_blkio_ticks'. Each item ending with a '_C' now consistently includes both the parent's count/time plus waited for children. That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT since its original name is an implementation artifact. ___________________________________________ <STAT> api The only api change was to correct a typo ('dervied'). _________________________________________ <VMSTAT> api Even ignoring white space, this interface received the largest number of changes. Mostly, this was because of deficiencies in the proc(5) documentation. Recall that this documentation already sorely lacks any substance. Usually, just kernel releases are noted, not contents. When compared to kernel source, that proc(5) contained many non-existent fields and also omitted many others. ________________________________________ <MEMINFO> api Sadly, with this api many of the changes were simply a correction of some earlier 'human error' where several fields where hashed then tracked but never represented with an item enumerator in this meminfo.h header file. _______________________________________ <SLABINFO> api The 'SLABS' (summary) & 'SLABNODE' items were reversed since the former are derived from the separate caches. More significantly, those 'SLABNODE' guys were renamed to 'SLAB' since they concern individual caches and the concept of 'nodes' is really an implementation detail. Also, several enumerators were changed to more closely agree with official slabinfo(5) documentation referred to in what we're treating as a base document: proc(5). Lastly, while those 'SLABS' items are solely a product of our library and not represented in slabinfo(5), the names attempt to parallel those found as 'SLAB' items. ______________________________________ <DISKSTATS> api One enumeration identifier was changed so as to better reflect its relationship to that actual documentation: 'Documentation/iostats.txt', as referenced in proc(5). Reference(s): . 12/2018, item origins added (and commit msg history) commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14 . 01/2019, <stat> origins tweaked commit 201e816b26ddaccc923ec40977c92037cdd0c34e Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
p->select_ext.lowest = SLABS_CACHES_TOTAL;
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
p->select_ext.highest = SLABS_DELTA_SIZE_TOTAL;
library: refactored some header file items and origins This commit is intended as a refinement of the patches mentioned below, where origins/sources of newlib items were added to the header files for user documentation. However, if those additions are to be truly effective, along with kernel documentation (where available), the following prerequisites must also have been satisfied: . our identifiers closely align with linux field names . our derived items are documented or self-documenting Satisfying those prerequisites prompted this patch and for these changes, kernel sources were emphasized over available documentation (shame on me, it should always have been so). And, while some 'new' fields were found to be conditional, they were included unconditionally. These changes appear more extensive than they actually need be since I have attempted to enforce some spacing conventions. So, I've summarize the significant things in the sections that follow. For a proper perspective, use: 'git diff --ignore-space-change' (good as alias). ___________________________________________ <PIDS> api This api is unique in that there exists many different file/directory origins subordinate to /proc/<pid>. And our item identifiers are sometimes coerced so as to be able to group related or similar enumerators together. So, users needed more help relating our identifiers to an actual documented field. Thus, we will now also add the field names as with 'stat: delayacct_blkio_ticks'. Each item ending with a '_C' now consistently includes both the parent's count/time plus waited for children. That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT since its original name is an implementation artifact. ___________________________________________ <STAT> api The only api change was to correct a typo ('dervied'). _________________________________________ <VMSTAT> api Even ignoring white space, this interface received the largest number of changes. Mostly, this was because of deficiencies in the proc(5) documentation. Recall that this documentation already sorely lacks any substance. Usually, just kernel releases are noted, not contents. When compared to kernel source, that proc(5) contained many non-existent fields and also omitted many others. ________________________________________ <MEMINFO> api Sadly, with this api many of the changes were simply a correction of some earlier 'human error' where several fields where hashed then tracked but never represented with an item enumerator in this meminfo.h header file. _______________________________________ <SLABINFO> api The 'SLABS' (summary) & 'SLABNODE' items were reversed since the former are derived from the separate caches. More significantly, those 'SLABNODE' guys were renamed to 'SLAB' since they concern individual caches and the concept of 'nodes' is really an implementation detail. Also, several enumerators were changed to more closely agree with official slabinfo(5) documentation referred to in what we're treating as a base document: proc(5). Lastly, while those 'SLABS' items are solely a product of our library and not represented in slabinfo(5), the names attempt to parallel those found as 'SLAB' items. ______________________________________ <DISKSTATS> api One enumeration identifier was changed so as to better reflect its relationship to that actual documentation: 'Documentation/iostats.txt', as referenced in proc(5). Reference(s): . 12/2018, item origins added (and commit msg history) commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14 . 01/2019, <stat> origins tweaked commit 201e816b26ddaccc923ec40977c92037cdd0c34e Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
p->fetch_ext.lowest = SLAB_NAME;
p->fetch_ext.highest = SLAB_SIZE_TOTAL;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
#endif
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
p->refcount = 1;
/* do a priming read here for the following potential benefits: |
1) see if that caller's permissions were sufficient (root) |
2) make delta results potentially useful, even if 1st time |
3) elimnate need for history distortions 1st time 'switch' | */
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
if (slabinfo_read_failed(p)) {
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
procps_slabinfo_unref(&p);
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return -errno;
}
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
*info = p;
return 0;
} // end: procps_slabinfo_new
2015-07-04 10:29:59 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
PROCPS_EXPORT int procps_slabinfo_ref (
struct slabinfo_info *info)
2015-07-04 10:29:59 +05:30
{
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if (info == NULL)
return -EINVAL;
info->refcount++;
return info->refcount;
} // end: procps_slabinfo_ref
2015-07-04 10:29:59 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
PROCPS_EXPORT int procps_slabinfo_unref (
struct slabinfo_info **info)
2015-07-04 10:29:59 +05:30
{
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if (info == NULL || *info == NULL)
return -EINVAL;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
(*info)->refcount--;
if ((*info)->refcount < 1) {
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
int errno_sav = errno;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if ((*info)->slabinfo_fp) {
fclose((*info)->slabinfo_fp);
(*info)->slabinfo_fp = NULL;
}
if ((*info)->select_ext.extents)
slabinfo_extents_free_all((&(*info)->select_ext));
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if ((*info)->select_ext.items)
free((*info)->select_ext.items);
if ((*info)->fetch.anchor)
free((*info)->fetch.anchor);
if ((*info)->fetch.results.stacks)
free((*info)->fetch.results.stacks);
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if ((*info)->fetch_ext.extents)
slabinfo_extents_free_all(&(*info)->fetch_ext);
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if ((*info)->fetch_ext.items)
free((*info)->fetch_ext.items);
free((*info)->nodes);
free(*info);
*info = NULL;
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
errno = errno_sav;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return 0;
}
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return (*info)->refcount;
} // end: procps_slabinfo_unref
2015-07-04 10:29:59 +05:30
library: more tweaks for code and/or comments, 3rd gen Following is a summary of significant changes (if any) to each of these now upgraded 3rd gen library modules. <meminfo> ............................................ . eliminated duplicate decl of 'struct procps_meminfo' . standardized/normalized results struct union members . added 'std' & 'var' dividers in .c file, like <pids> . how did i miss relocating all these friggin' #undefs . cleanup 'get' return logic (remove a redundant 'if') <pids> ............................................... . repositioned the procps_pidsinfo structure in header . removed the extra trailing comma from enum pids_item . standardized/normalized results struct union members <slabinfo> ........................................... . corrected comment typo (jeeze, in an 'aligned' para) . standardized/normalized results struct union members . added 'std' & 'var' dividers in .c file, like <pids> . removed an obsolete #undef from procps_slabinfo_sort . cleanup 'get' return logic (remove a redundant 'if') <stat> ............................................... . how did i miss relocating all these friggin' #undefs . corrected an initialization fencepost used with numa <=== see Craig, here's a bug fix . removed the extra trailing comma from enum stat_item . standardized/normalized results struct union members . added 'std' & 'var' dividers in .c file, like <pids> . strengthen those parm checks in procps_stat_get func . cleanup 'get' return logic (remove a redundant 'if') <vmstat> ............................................. . standardized/normalized results struct union members . added 'std' & 'var' dividers in .c file, like <pids> . cleanup 'get' return logic (remove a redundant 'if') [ virtually all of these tweaks reflect the author's ] [ continuing pursuit of an unreasonable goal -- that ] [ of a 'perfect' (plus 'pretty') C language program! ] Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-14 10:30:00 +05:30
// --- variable interface functions -------------------------------------------
PROCPS_EXPORT struct slabinfo_result *procps_slabinfo_get (
struct slabinfo_info *info,
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
enum slabinfo_item item)
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
{
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
time_t cur_secs;
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
errno = EINVAL;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if (info == NULL)
return NULL;
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
if (item < 0 || item >= SLABINFO_logical_end)
return NULL;
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
errno = 0;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
/* we will NOT read the slabinfo file with every call - rather, we'll offer
a granularity of 1 second between reads ... */
cur_secs = time(NULL);
if (1 <= cur_secs - info->sav_secs) {
if (slabinfo_read_failed(info))
return NULL;
info->sav_secs = cur_secs;
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
}
info->get_this.item = item;
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
// with 'get', we must NOT honor the usual 'noop' guarantee
info->get_this.result.ul_int = 0;
Item_table[item].setsfunc(&info->get_this, &info->slabs, &info->nul_node);
return &info->get_this;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
} // end: procps_slabinfo_get
/* procps_slabinfo_reap():
*
library: refactored some header file items and origins This commit is intended as a refinement of the patches mentioned below, where origins/sources of newlib items were added to the header files for user documentation. However, if those additions are to be truly effective, along with kernel documentation (where available), the following prerequisites must also have been satisfied: . our identifiers closely align with linux field names . our derived items are documented or self-documenting Satisfying those prerequisites prompted this patch and for these changes, kernel sources were emphasized over available documentation (shame on me, it should always have been so). And, while some 'new' fields were found to be conditional, they were included unconditionally. These changes appear more extensive than they actually need be since I have attempted to enforce some spacing conventions. So, I've summarize the significant things in the sections that follow. For a proper perspective, use: 'git diff --ignore-space-change' (good as alias). ___________________________________________ <PIDS> api This api is unique in that there exists many different file/directory origins subordinate to /proc/<pid>. And our item identifiers are sometimes coerced so as to be able to group related or similar enumerators together. So, users needed more help relating our identifiers to an actual documented field. Thus, we will now also add the field names as with 'stat: delayacct_blkio_ticks'. Each item ending with a '_C' now consistently includes both the parent's count/time plus waited for children. That 'RTPRIO' guy was renamed/relocated as PRIORITY_RT since its original name is an implementation artifact. ___________________________________________ <STAT> api The only api change was to correct a typo ('dervied'). _________________________________________ <VMSTAT> api Even ignoring white space, this interface received the largest number of changes. Mostly, this was because of deficiencies in the proc(5) documentation. Recall that this documentation already sorely lacks any substance. Usually, just kernel releases are noted, not contents. When compared to kernel source, that proc(5) contained many non-existent fields and also omitted many others. ________________________________________ <MEMINFO> api Sadly, with this api many of the changes were simply a correction of some earlier 'human error' where several fields where hashed then tracked but never represented with an item enumerator in this meminfo.h header file. _______________________________________ <SLABINFO> api The 'SLABS' (summary) & 'SLABNODE' items were reversed since the former are derived from the separate caches. More significantly, those 'SLABNODE' guys were renamed to 'SLAB' since they concern individual caches and the concept of 'nodes' is really an implementation detail. Also, several enumerators were changed to more closely agree with official slabinfo(5) documentation referred to in what we're treating as a base document: proc(5). Lastly, while those 'SLABS' items are solely a product of our library and not represented in slabinfo(5), the names attempt to parallel those found as 'SLAB' items. ______________________________________ <DISKSTATS> api One enumeration identifier was changed so as to better reflect its relationship to that actual documentation: 'Documentation/iostats.txt', as referenced in proc(5). Reference(s): . 12/2018, item origins added (and commit msg history) commit 96d59cbf46b3ff687bd29fad4708074a0e1cea14 . 01/2019, <stat> origins tweaked commit 201e816b26ddaccc923ec40977c92037cdd0c34e Signed-off-by: Jim Warner <james.warner@comcast.net>
2019-03-12 11:30:00 +05:30
* Harvest all the requested SLAB (individual nodes) information
* providing the result stacks along with the total number of nodes.
*
* Returns: pointer to a slabinfo_reaped struct on success, NULL on error.
*/
PROCPS_EXPORT struct slabinfo_reaped *procps_slabinfo_reap (
struct slabinfo_info *info,
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
enum slabinfo_item *items,
int numitems)
2015-07-04 10:29:59 +05:30
{
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
errno = EINVAL;
if (info == NULL || items == NULL)
return NULL;
if (0 > slabinfo_stacks_reconfig_maybe(&info->fetch_ext, items, numitems))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return NULL; // here, errno may be overridden with ENOMEM
errno = 0;
if (slabinfo_read_failed(info))
return NULL;
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
if (0 > slabinfo_stacks_fetch(info))
return NULL;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return &info->fetch.results;
} // end: procps_slabinfo_reap
2015-07-04 10:29:59 +05:30
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
/* procps_slabinfo_select():
2015-07-04 10:29:59 +05:30
*
* Obtain all the requested SLABS (global) information then return
* it in a single library provided results stack.
2015-07-04 10:29:59 +05:30
*
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
* Returns: pointer to a slabinfo_stack struct on success, NULL on error.
2015-07-04 10:29:59 +05:30
*/
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
PROCPS_EXPORT struct slabinfo_stack *procps_slabinfo_select (
struct slabinfo_info *info,
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
enum slabinfo_item *items,
int numitems)
2015-07-04 10:29:59 +05:30
{
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
errno = EINVAL;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if (info == NULL || items == NULL)
return NULL;
if (0 > slabinfo_stacks_reconfig_maybe(&info->select_ext, items, numitems))
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
return NULL; // here, errno may be overridden with ENOMEM
errno = 0;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if (!info->select_ext.extents
&& (!slabinfo_stacks_alloc(&info->select_ext, 1)))
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return NULL;
if (slabinfo_read_failed(info))
return NULL;
slabinfo_assign_results(info->select_ext.extents->stacks[0], &info->slabs, &info->nul_node);
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
return info->select_ext.extents->stacks[0];
} // end: procps_slabinfo_select
2015-07-04 10:29:59 +05:30
/*
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
* procps_slabinfo_sort():
2015-07-04 10:29:59 +05:30
*
* Sort stacks anchored in the passed stack pointers array
* based on the designated sort enumerator and specified order.
*
* Returns those same addresses sorted.
*
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
* Note: all of the stacks must be homogeneous (of equal length and content).
*/
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
PROCPS_EXPORT struct slabinfo_stack **procps_slabinfo_sort (
struct slabinfo_info *info,
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabinfo_stack *stacks[],
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
int numstacked,
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
enum slabinfo_item sortitem,
enum slabinfo_sort_order order)
{
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
struct slabinfo_result *p;
struct sort_parms parms;
int offset;
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
errno = EINVAL;
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
if (info == NULL || stacks == NULL)
return NULL;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
// a slabinfo_item is currently unsigned, but we'll protect our future
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
if (sortitem < 0 || sortitem >= SLABINFO_logical_end)
return NULL;
library: removed all the 'PROCPS_' enumerator prefixes Many of our item enumerator identifiers are very long, especially in that <VMSTAT> module. Additionally, they all contain the exact same universal 'PROCPS_' prefix. The origins for this are likely found in the desire to avoid name clashes with other potential include files. But with procps-ng newlib, we've probably gone way too far. Did 'PROCPS_PIDS_TICS_SYSTEM' actually offer more protection against clash than 'PIDS_TICS_SYSTEM' does? I don't think so. Besides, no matter how big that name becomes, one can never guarantee they'll never be some clash. And, conversely, extremely short names will not always create conflict. Of course, in either case when some clash occurs, one can always #undef that problem. Thus, this commit will eliminate that 'PROCPS_' prefix making all of those enum identifiers a little shorter. And, we'll still be well above some ridiculously short (criminally short) names found in some common headers: - - - - - - - - - - <term.h> - 'tab', 'TTY', etc - - - - - - - - - - - - - - - - <search.h> - 'ENTER', ENTRY', 'FIND', etc ------------------------------------------------------ Finally, with this as a last of the wholesale changes, we will have established the naming conventions below: . only functions will begin with that 'procps_' prefix . exposed structures begin with the module/header name . item enumerators begin like structs, but capitalized . other enumerators work exactly like item enumerators . macros and constants begin just like the enumerators ------------------------------------------------------ Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-07-21 10:30:00 +05:30
if (order != SLABINFO_SORT_ASCEND && order != SLABINFO_SORT_DESCEND)
return NULL;
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
if (numstacked < 2)
return stacks;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
offset = 0;
p = stacks[0]->head;
for (;;) {
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
if (p->item == sortitem)
break;
++offset;
if (p->item >= SLABINFO_logical_end)
return NULL;
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
++p;
}
library: improve and/or standardize 'errno' management With older library logic having been modified to avoid using those potentially deadly alloc.h routines, while improving 'errno' handling, we're ready to standardize and enhance newlib's approach to any potential errors. In so doing, we'll establish the following objectives: . . . . . . . . . . . . . functions returning an 'int' . an error will be indicated by a negative number that is always the inverse of some well known errno.h value . . . . . . . . . . . functions returning an 'address' . any error will be indicated by a NULL return pointer with the actual reason found in the formal errno value And, when errno is manipulated directly we will strive to do so whenever possible within those routines which have been declared with PROCPS_EXPORT. In other words, in the user callable functions defined in source last. [ But, that won't always be possible. In particular, ] [ all the 'read_failed' functions will sometimes set ] [ 'errno' so that they can serve callers returning a ] [ NULL or an int without duplicating a lot of logic. ] [ Also, that includes one subordinate function which ] [ was called by 'read_failed' in the <slabinfo> API. ] ------------------------------------------------------ Along the way, several additional miscellaneous issues were addressed. They're listed here now for posterity. . the '-1' return value passed outside the library was eliminated since it would erroneously equate to -EPERM . the stacks_fetch functions in <diskstats> and <stat> weren't checked for their possible minus return values . hash create was not checked in <meminfo> or <vmstat> . fixed 'new' function faulty parm check in <slabinfo> Signed-off-by: Jim Warner <james.warner@comcast.net>
2017-11-18 10:30:00 +05:30
errno = 0;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
parms.offset = offset;
parms.order = order;
qsort_r(stacks, numstacked, sizeof(void *), (QSR_t)Item_table[p->item].sortfunc, &parms);
library: slab is redesigned to use 'stack' vs. 'chain' In addition to that text shown below the line which is common to several commit messages, this patch contains several minor changes with lessor impact upon the API: . A 'read' was added to function procps_slabnode_count (but only when necessary, i.e. info->nodes_used == 0). . The #include header files are ordered alphabetically now, with all those <sys/??> types separately grouped. ------------------------------------------------------ . The former 'chains' have now become 'stacks' without the 'next' pointer in each result struct. The pointers initially seemed to offer some flexibility with memory allocations and benefits for the library access logic. However, user access was always via displacement and a a statically allocated chain was cumbersome to define. . An enumerator ending in '_noop' will no longer serve as a fencepost delimiter. Rather, it has become a much more important and flexible user oriented tool. Adding one or more such 'items' in any items list passed into the library becomes the means of extending the 'stack' to also include user (not just library) data. Any such data is guaranteed to never be altered by the library. . Anticipating PID support, where many different types must be represented in a result structure, we'll adopt a common naming standard. And, while not every results structure currently needs to reflect disparate types a union will be employed so the same dot qualifier ('.') can be used consistently when accessing all such data. Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-07-21 10:30:00 +05:30
return stacks;
library: normalize/standardize the i/f, <SLABINFO> api Before this major redesign, the slabs interface likely was our messiest 2nd generation attempt at opaqueness. Beyond the standard 'new', 'ref' & 'unref', there were a total of 12 exported functions. Now, there are four. The 1st step was to remove several of those functions. These were quick to go since they were not used (yet): . procps_slabnode_count . procps_slabnode_getname . procps_slabnode_getstack Then, the following were internalized so users needn't be burdened with implementation details in the future: . procps_slabinfo_read (renamed: read_slabinfo_failed) . procps_slabnode_stacks_alloc (renamed: stacks_alloc) Still others evolved into the minimal interface we had strived for in the other upgraded 3rd generation APIs: . procps_slabnode_get -----------> procps_slabinfo_get . separate stack_alloc/fill --> procps_slabinfo_select . separate stacks_alloc/fill ---> procps_slabinfo_reap . procps_slabnode_stacks_sort --> procps_slabinfo_sort ------------------------------------------------------ Beyond those reductions, the major modifications were: . This API tries to be as forgiving as possible and as such won't throw errors when a caller request makes no sense. For example, if a 'get' or 'select' requested a SLABNODE item (with no current means to id that node), results will be zero. By the same token, should 'reap' include a global SLABS item (meaning those values will be duplicated in *every* node stack) it'll be allowed. . If the above behavior is undesired, a new #define of ENFORCE_LOGICAL can be used to restrict certain items. . Permission problems will now be caught at 'new' time thanks to a priming 'read' call. That read also serves to make DELTA values potentially useful at 1st access. . Separate slab/slabnode enumerators were consolidated into one, simplifying validation & the results struct. . Several internal parameter checks were relaxed since they were already checked by the caller. Besides if we cannot trust our own code we might as well hang it up. . That sort provision was made more efficient and will offer the ascending choice, in addition to descending. ------------------------------------------------------ Lastly, some additional thoughts regarding the future: . It would not be difficult to expand 'select' to also accept a nodeid, or to clone it as 'select_node'. And, should the same be extended to 'get', a results struct could be returned instead of signed long accommodating the extra data type(s) like a node name (string data). . The 'get' function is not currently affected by that define ENFORCE_LOGICAL. However, at some future point perhaps -EINVAL would be more appropriate than a zero. Signed-off-by: Jim Warner <james.warner@comcast.net>
2016-06-05 10:30:00 +05:30
} // end: procps_slabinfo_sort
// --- special debugging function(s) ------------------------------------------
/*
* The following isn't part of the normal programming interface. Rather,
* it exists to validate result types referenced in application programs.
*
* It's used only when:
* 1) the 'XTRA_PROCPS_DEBUG' has been defined, or
* 2) an #include of 'xtra-procps-debug.h' is used
*/
PROCPS_EXPORT struct slabinfo_result *xtra_slabinfo_get (
struct slabinfo_info *info,
enum slabinfo_item actual_enum,
const char *typestr,
const char *file,
int lineno)
{
struct slabinfo_result *r = procps_slabinfo_get(info, actual_enum);
if (actual_enum < 0 || actual_enum >= SLABINFO_logical_end) {
fprintf(stderr, "%s line %d: invalid item = %d, type = %s\n"
, file, lineno, actual_enum, typestr);
}
if (r) {
char *str = Item_table[r->item].type2str;
if (str[0]
&& (strcmp(typestr, str)))
fprintf(stderr, "%s line %d: was %s, expected %s\n", file, lineno, typestr, str);
}
return r;
} // end: xtra_slabinfo_get_
PROCPS_EXPORT struct slabinfo_result *xtra_slabinfo_val (
int relative_enum,
const char *typestr,
const struct slabinfo_stack *stack,
struct slabinfo_info *info,
const char *file,
int lineno)
{
char *str;
int i;
for (i = 0; stack->head[i].item < SLABINFO_logical_end; i++)
;
if (relative_enum < 0 || relative_enum >= i) {
fprintf(stderr, "%s line %d: invalid relative_enum = %d, valid range = 0-%d\n"
, file, lineno, relative_enum, i-1);
return NULL;
}
str = Item_table[stack->head[relative_enum].item].type2str;
if (str[0]
&& (strcmp(typestr, str))) {
fprintf(stderr, "%s line %d: was %s, expected %s\n", file, lineno, typestr, str);
}
return &stack->head[relative_enum];
(void)info;
} // end: xtra_slabinfo_val