2002-02-02 04:17:29 +05:30
|
|
|
/*
|
2011-11-07 21:51:41 +05:30
|
|
|
* Copyright 1998-2002 by Albert Cahalan; all rights resered.
|
2002-02-02 04:17:29 +05:30
|
|
|
* This file may be used subject to the terms and conditions of the
|
2011-11-07 21:51:41 +05:30
|
|
|
* GNU Library General Public License Version 2, or any later version
|
2002-02-02 04:17:29 +05:30
|
|
|
* at your option, as published by the Free Software Foundation.
|
|
|
|
|
* This program 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 Library General Public License for more details.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#ifndef PROCPS_PS_H
|
|
|
|
|
#define PROCPS_PS_H
|
|
|
|
|
|
2012-01-02 12:16:17 +05:30
|
|
|
#include "../include/nls.h"
|
2016-04-16 12:33:57 +05:30
|
|
|
#include <proc/procps.h>
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
|
2015-10-07 10:30:00 +05:30
|
|
|
// --- <pids> interface begin ||||||||||||||||||||||||||||||||||||||||||||
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
// -----------------------------------------------------------------------
|
|
|
|
|
|
|
|
|
|
// hack to minimize code impact
|
|
|
|
|
#undef proc_t
|
|
|
|
|
#define proc_t struct pids_stack
|
|
|
|
|
|
|
|
|
|
/* this is for allocation of the Pids_items and represents a compromise.
|
|
|
|
|
we can't predict how many fields will actually be requested yet there
|
|
|
|
|
are numerous duplicate format_array entries. here are the statistics:
|
|
|
|
|
252 entries in the format_array
|
|
|
|
|
82 of those entries are unique
|
|
|
|
|
60 equals a former proc_t size
|
|
|
|
|
in reality, only a small portion of the stack depth will be occupied,
|
|
|
|
|
and the excess represents storage cost only, not a run-time cpu cost! */
|
|
|
|
|
#define PIDSITEMS 70
|
|
|
|
|
|
|
|
|
|
/* a 'results stack value' extractor macro
|
|
|
|
|
where: E=rel enum, T=data type, S=stack */
|
2016-08-05 10:30:00 +05:30
|
|
|
#define rSv(E,T,S) PIDS_VAL(rel_ ## E, T, S, Pids_info)
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
|
|
|
|
|
#define namREL(e) rel_ ## e
|
|
|
|
|
#define makEXT(e) extern int namREL(e);
|
|
|
|
|
#define makREL(e) int namREL(e) = -1;
|
|
|
|
|
#define chkREL(e) if (namREL(e) < 0) { \
|
2016-07-21 10:30:00 +05:30
|
|
|
Pids_items[Pids_index] = PIDS_ ## e; \
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
namREL(e) = (Pids_index < PIDSITEMS) ? Pids_index++ : rel_noop; }
|
|
|
|
|
|
|
|
|
|
#define setREL1(e) { \
|
|
|
|
|
if (!outbuf) { \
|
|
|
|
|
chkREL(e) \
|
|
|
|
|
return 0; \
|
|
|
|
|
} }
|
|
|
|
|
#define setREL2(e1,e2) { \
|
|
|
|
|
if (!outbuf) { \
|
|
|
|
|
chkREL(e1) chkREL(e2) \
|
|
|
|
|
return 0; \
|
|
|
|
|
} }
|
|
|
|
|
#define setREL3(e1,e2,e3) { \
|
|
|
|
|
if (!outbuf) { \
|
|
|
|
|
chkREL(e1) chkREL(e2) chkREL(e3) \
|
|
|
|
|
return 0; \
|
|
|
|
|
} }
|
|
|
|
|
|
2016-07-21 10:30:00 +05:30
|
|
|
extern struct pids_info *Pids_info;
|
|
|
|
|
extern enum pids_item *Pids_items;
|
|
|
|
|
extern int Pids_index;
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
|
|
|
|
|
// most of these need not be extern, they're unique to output.c
|
|
|
|
|
// (but for future flexibility the easiest path has been taken)
|
|
|
|
|
makEXT(ADDR_END_CODE)
|
|
|
|
|
makEXT(ADDR_KSTK_EIP)
|
|
|
|
|
makEXT(ADDR_KSTK_ESP)
|
|
|
|
|
makEXT(ADDR_START_CODE)
|
|
|
|
|
makEXT(ADDR_START_STACK)
|
2015-10-12 10:30:00 +05:30
|
|
|
makEXT(CGNAME)
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
makEXT(CGROUP)
|
|
|
|
|
makEXT(CMD)
|
|
|
|
|
makEXT(CMDLINE)
|
|
|
|
|
makEXT(ENVIRON)
|
|
|
|
|
makEXT(FLAGS)
|
|
|
|
|
makEXT(FLT_MAJ)
|
|
|
|
|
makEXT(FLT_MAJ_C)
|
|
|
|
|
makEXT(FLT_MIN)
|
|
|
|
|
makEXT(FLT_MIN_C)
|
|
|
|
|
makEXT(ID_EGID)
|
|
|
|
|
makEXT(ID_EGROUP)
|
|
|
|
|
makEXT(ID_EUID)
|
|
|
|
|
makEXT(ID_EUSER)
|
|
|
|
|
makEXT(ID_FGID)
|
|
|
|
|
makEXT(ID_FGROUP)
|
|
|
|
|
makEXT(ID_FUID)
|
|
|
|
|
makEXT(ID_FUSER)
|
2018-02-11 13:52:22 +05:30
|
|
|
makEXT(ID_LOGIN)
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
makEXT(ID_PGRP)
|
|
|
|
|
makEXT(ID_PID)
|
|
|
|
|
makEXT(ID_PPID)
|
|
|
|
|
makEXT(ID_RGID)
|
|
|
|
|
makEXT(ID_RGROUP)
|
|
|
|
|
makEXT(ID_RUID)
|
|
|
|
|
makEXT(ID_RUSER)
|
|
|
|
|
makEXT(ID_SESSION)
|
|
|
|
|
makEXT(ID_SGID)
|
|
|
|
|
makEXT(ID_SGROUP)
|
|
|
|
|
makEXT(ID_SUID)
|
|
|
|
|
makEXT(ID_SUSER)
|
|
|
|
|
makEXT(ID_TGID)
|
|
|
|
|
makEXT(ID_TPGID)
|
|
|
|
|
makEXT(LXCNAME)
|
|
|
|
|
makEXT(NICE)
|
|
|
|
|
makEXT(NLWP)
|
|
|
|
|
makEXT(NS_IPC)
|
|
|
|
|
makEXT(NS_MNT)
|
|
|
|
|
makEXT(NS_NET)
|
|
|
|
|
makEXT(NS_PID)
|
|
|
|
|
makEXT(NS_USER)
|
|
|
|
|
makEXT(NS_UTS)
|
|
|
|
|
makEXT(PRIORITY)
|
|
|
|
|
makEXT(PROCESSOR)
|
2017-05-12 10:35:00 +05:30
|
|
|
makEXT(PROCESSOR_NODE)
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
makEXT(RSS)
|
|
|
|
|
makEXT(RSS_RLIM)
|
|
|
|
|
makEXT(RTPRIO)
|
|
|
|
|
makEXT(SCHED_CLASS)
|
|
|
|
|
makEXT(SD_MACH)
|
|
|
|
|
makEXT(SD_OUID)
|
|
|
|
|
makEXT(SD_SEAT)
|
|
|
|
|
makEXT(SD_SESS)
|
|
|
|
|
makEXT(SD_SLICE)
|
|
|
|
|
makEXT(SD_UNIT)
|
|
|
|
|
makEXT(SD_UUNIT)
|
|
|
|
|
makEXT(SIGBLOCKED)
|
|
|
|
|
makEXT(SIGCATCH)
|
|
|
|
|
makEXT(SIGIGNORE)
|
|
|
|
|
makEXT(SIGNALS)
|
|
|
|
|
makEXT(SIGPENDING)
|
|
|
|
|
makEXT(STATE)
|
|
|
|
|
makEXT(SUPGIDS)
|
|
|
|
|
makEXT(SUPGROUPS)
|
|
|
|
|
makEXT(TICS_ALL)
|
|
|
|
|
makEXT(TICS_ALL_C)
|
|
|
|
|
makEXT(TIME_ALL)
|
|
|
|
|
makEXT(TIME_ELAPSED)
|
|
|
|
|
makEXT(TIME_START)
|
|
|
|
|
makEXT(TTY)
|
|
|
|
|
makEXT(TTY_NAME)
|
|
|
|
|
makEXT(TTY_NUMBER)
|
|
|
|
|
makEXT(VM_DATA)
|
2016-05-14 10:30:00 +05:30
|
|
|
makEXT(VM_RSS_LOCKED)
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
makEXT(VM_RSS)
|
|
|
|
|
makEXT(VM_SIZE)
|
|
|
|
|
makEXT(VM_STACK)
|
|
|
|
|
makEXT(VSIZE_PGS)
|
|
|
|
|
makEXT(WCHAN_NAME)
|
|
|
|
|
makEXT(extra)
|
|
|
|
|
makEXT(noop)
|
|
|
|
|
// -----------------------------------------------------------------------
|
|
|
|
|
// --- <pids> interface end ||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|
|
|
2002-02-02 04:17:29 +05:30
|
|
|
|
|
|
|
|
#if 0
|
2012-02-26 04:46:44 +05:30
|
|
|
#define trace(...) printf(## __VA_ARGS__)
|
2002-02-02 04:17:29 +05:30
|
|
|
#else
|
2012-02-26 04:46:44 +05:30
|
|
|
#define trace(...)
|
2002-02-02 04:17:29 +05:30
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/***************** GENERAL DEFINE ********************/
|
|
|
|
|
|
|
|
|
|
/* selection list */
|
|
|
|
|
#define SEL_RUID 1
|
|
|
|
|
#define SEL_EUID 2
|
|
|
|
|
#define SEL_SUID 3
|
|
|
|
|
#define SEL_FUID 4
|
|
|
|
|
#define SEL_RGID 5
|
|
|
|
|
#define SEL_EGID 6
|
|
|
|
|
#define SEL_SGID 7
|
|
|
|
|
#define SEL_FGID 8
|
|
|
|
|
#define SEL_PGRP 9
|
|
|
|
|
#define SEL_PID 10
|
|
|
|
|
#define SEL_TTY 11
|
|
|
|
|
#define SEL_SESS 12
|
|
|
|
|
#define SEL_COMM 13
|
2003-02-09 13:01:11 +05:30
|
|
|
#define SEL_PPID 14
|
2014-07-11 00:44:02 +05:30
|
|
|
#define SEL_PID_QUICK 15
|
2002-02-02 04:17:29 +05:30
|
|
|
|
|
|
|
|
/* Since an enum could be smashed by a #define, it would be bad. */
|
|
|
|
|
#define U98 0 /* Unix98 standard */ /* This must be 0 */
|
|
|
|
|
#define XXX 1 /* Common extension */
|
|
|
|
|
#define DEC 2 /* Digital Unix */
|
|
|
|
|
#define AIX 3 /* AIX */
|
|
|
|
|
#define SCO 4 /* SCO */
|
|
|
|
|
#define LNX 5 /* Linux original :-) */
|
|
|
|
|
#define BSD 6 /* FreeBSD and OpenBSD */
|
|
|
|
|
#define SUN 7 /* SunOS 5 (Solaris) */
|
|
|
|
|
#define HPU 8 /* HP-UX */
|
|
|
|
|
#define SGI 9 /* Irix */
|
|
|
|
|
#define SOE 10 /* IBM's S/390 OpenEdition */
|
2004-10-12 05:42:25 +05:30
|
|
|
#define TST 11 /* test code */
|
2002-02-02 04:17:29 +05:30
|
|
|
|
|
|
|
|
/*
|
2002-11-27 05:54:01 +05:30
|
|
|
* Try not to overflow the output buffer:
|
2002-02-02 04:17:29 +05:30
|
|
|
* 32 pages for env+cmd
|
2002-11-27 05:54:01 +05:30
|
|
|
* 64 kB pages on IA-64
|
2005-01-27 09:38:10 +05:30
|
|
|
* 4 chars for "\377", or 1 when mangling to '?' (ESC_STRETCH)
|
2002-02-02 04:17:29 +05:30
|
|
|
* plus some slack for other stuff
|
2002-11-27 05:54:01 +05:30
|
|
|
* That is about 8.5 MB on IA-64, or 0.6 MB on i386
|
2005-01-27 09:38:10 +05:30
|
|
|
*
|
|
|
|
|
* Sadly, current kernels only supply one page of env/command data.
|
|
|
|
|
* The buffer is now protected with a guard page, and via other means
|
|
|
|
|
* to avoid hitting the guard page.
|
2002-02-02 04:17:29 +05:30
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/* output buffer size */
|
2005-01-27 09:38:10 +05:30
|
|
|
#define OUTBUF_SIZE (2 * 64*1024 * ESC_STRETCH)
|
2002-02-02 04:17:29 +05:30
|
|
|
|
|
|
|
|
/******************* PS DEFINE *******************/
|
|
|
|
|
|
2003-09-28 22:53:29 +05:30
|
|
|
// Column flags
|
|
|
|
|
// Justification control for flags field comes first.
|
|
|
|
|
#define CF_JUST_MASK 0x0f
|
|
|
|
|
// CF_AIXHACK 0
|
|
|
|
|
#define CF_USER 1 // left if text, right if numeric
|
|
|
|
|
#define CF_LEFT 2
|
|
|
|
|
#define CF_RIGHT 3
|
|
|
|
|
#define CF_UNLIMITED 4
|
|
|
|
|
#define CF_WCHAN 5 // left if text, right if numeric
|
|
|
|
|
#define CF_SIGNAL 6 // right in 9, or 16 if screen_cols>107
|
|
|
|
|
// Then the other flags
|
2004-04-17 08:43:56 +05:30
|
|
|
#define CF_PIDMAX 0x00000010 // react to pid_max
|
2003-09-29 09:39:52 +05:30
|
|
|
// Only one allowed; use separate bits to catch errors.
|
2003-09-28 22:53:29 +05:30
|
|
|
#define CF_PRINT_THREAD_ONLY 0x10000000
|
|
|
|
|
#define CF_PRINT_PROCESS_ONLY 0x20000000
|
|
|
|
|
#define CF_PRINT_EVERY_TIME 0x40000000
|
|
|
|
|
#define CF_PRINT_AS_NEEDED 0x80000000 // means we have no clue, so assume EVERY TIME
|
2003-09-29 09:39:52 +05:30
|
|
|
#define CF_PRINT_MASK 0xf0000000
|
2003-09-28 22:53:29 +05:30
|
|
|
|
2003-09-20 13:59:55 +05:30
|
|
|
/* thread_flags */
|
2003-09-30 09:30:50 +05:30
|
|
|
#define TF_B_H 0x0001
|
|
|
|
|
#define TF_B_m 0x0002
|
|
|
|
|
#define TF_U_m 0x0004
|
|
|
|
|
#define TF_U_T 0x0008
|
|
|
|
|
#define TF_U_L 0x0010
|
|
|
|
|
#define TF_show_proc 0x0100 // show the summary line
|
|
|
|
|
#define TF_show_task 0x0200 // show the per-thread lines
|
2003-10-07 08:42:50 +05:30
|
|
|
#define TF_show_both 0x0400 // distinct proc/task format lists
|
|
|
|
|
#define TF_loose_tasks 0x0800 // let sorting break up task groups (BSDish)
|
|
|
|
|
#define TF_no_sort 0x1000 // don't know if thread-grouping should survive a sort
|
|
|
|
|
#define TF_no_forest 0x2000 // don't see how to do threads w/ forest option
|
|
|
|
|
#define TF_must_use 0x4000 // options only make sense if LWP/SPID column added
|
2003-09-20 13:59:55 +05:30
|
|
|
|
2002-02-02 04:17:29 +05:30
|
|
|
/* personality control flags */
|
|
|
|
|
#define PER_BROKEN_o 0x0001
|
|
|
|
|
#define PER_BSD_h 0x0002
|
|
|
|
|
#define PER_BSD_m 0x0004
|
2004-04-17 08:43:56 +05:30
|
|
|
#define PER_IRIX_l 0x0008
|
2002-02-02 04:17:29 +05:30
|
|
|
#define PER_FORCE_BSD 0x0010
|
|
|
|
|
#define PER_GOOD_o 0x0020
|
|
|
|
|
#define PER_OLD_m 0x0040
|
|
|
|
|
#define PER_NO_DEFAULT_g 0x0080
|
|
|
|
|
#define PER_ZAP_ADDR 0x0100
|
|
|
|
|
#define PER_SANE_USER 0x0200
|
2004-04-26 06:17:31 +05:30
|
|
|
#define PER_HPUX_x 0x0400
|
|
|
|
|
#define PER_SVR4_x 0x0800
|
2004-08-18 07:52:13 +05:30
|
|
|
#define PER_BSD_COLS 0x1000
|
|
|
|
|
#define PER_UNIX_COLS 0x2000
|
2002-02-02 04:17:29 +05:30
|
|
|
|
|
|
|
|
/* Simple selections by bit mask */
|
|
|
|
|
#define SS_B_x 0x01
|
|
|
|
|
#define SS_B_g 0x02
|
|
|
|
|
#define SS_U_d 0x04
|
|
|
|
|
#define SS_U_a 0x08
|
|
|
|
|
#define SS_B_a 0x10
|
|
|
|
|
|
|
|
|
|
/* predefined format flags such as: -l -f l u s -j */
|
|
|
|
|
#define FF_Uf 0x0001 /* -f */
|
|
|
|
|
#define FF_Uj 0x0002 /* -j */
|
|
|
|
|
#define FF_Ul 0x0004 /* -l */
|
|
|
|
|
#define FF_Bj 0x0008 /* j */
|
|
|
|
|
#define FF_Bl 0x0010 /* l */
|
|
|
|
|
#define FF_Bs 0x0020 /* s */
|
|
|
|
|
#define FF_Bu 0x0040 /* u */
|
|
|
|
|
#define FF_Bv 0x0080 /* v */
|
|
|
|
|
#define FF_LX 0x0100 /* X */
|
|
|
|
|
#define FF_Lm 0x0200 /* m */ /* overloaded: threads, sort, format */
|
2002-09-30 12:41:30 +05:30
|
|
|
#define FF_Fc 0x0400 /* --context */ /* Flask security context format */
|
2002-02-02 04:17:29 +05:30
|
|
|
|
|
|
|
|
/* predefined format modifier flags such as: -l -f l u s -j */
|
|
|
|
|
#define FM_c 0x0001 /* -c */
|
|
|
|
|
#define FM_j 0x0002 /* -j */ /* only set when !sysv_j_format */
|
|
|
|
|
#define FM_y 0x0004 /* -y */
|
2003-10-07 08:42:50 +05:30
|
|
|
//#define FM_L 0x0008 /* -L */
|
2002-02-02 04:17:29 +05:30
|
|
|
#define FM_P 0x0010 /* -P */
|
|
|
|
|
#define FM_M 0x0020 /* -M */
|
2003-10-07 08:42:50 +05:30
|
|
|
//#define FM_T 0x0040 /* -T */
|
2002-02-02 04:17:29 +05:30
|
|
|
#define FM_F 0x0080 /* -F */ /* -F also sets the regular -f flags */
|
|
|
|
|
|
|
|
|
|
/* sorting & formatting */
|
|
|
|
|
/* U,B,G is Unix,BSD,Gnu and then there is the option itself */
|
|
|
|
|
#define SF_U_O 1
|
|
|
|
|
#define SF_U_o 2
|
|
|
|
|
#define SF_B_O 3
|
|
|
|
|
#define SF_B_o 4
|
|
|
|
|
#define SF_B_m 5 /* overloaded: threads, sort, format */
|
|
|
|
|
#define SF_G_sort 6
|
|
|
|
|
#define SF_G_format 7
|
|
|
|
|
|
|
|
|
|
/* headers */
|
|
|
|
|
#define HEAD_SINGLE 0 /* default, must be 0 */
|
|
|
|
|
#define HEAD_NONE 1
|
|
|
|
|
#define HEAD_MULTI 2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/********************** GENERAL TYPEDEF *******************/
|
|
|
|
|
|
|
|
|
|
/* Other fields that might be useful:
|
|
|
|
|
*
|
|
|
|
|
* char *name; user-defined column name (format specification)
|
|
|
|
|
* int reverse; sorting in reverse (sort specification)
|
|
|
|
|
*
|
|
|
|
|
* name in place of u
|
|
|
|
|
* reverse in place of n
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
typedef union sel_union {
|
|
|
|
|
pid_t pid;
|
2003-02-09 13:01:11 +05:30
|
|
|
pid_t ppid;
|
2002-02-02 04:17:29 +05:30
|
|
|
uid_t uid;
|
|
|
|
|
gid_t gid;
|
|
|
|
|
dev_t tty;
|
2003-02-18 09:21:03 +05:30
|
|
|
char cmd[16]; /* this is _not_ \0 terminated */
|
2002-02-02 04:17:29 +05:30
|
|
|
} sel_union;
|
|
|
|
|
|
|
|
|
|
typedef struct selection_node {
|
|
|
|
|
struct selection_node *next;
|
|
|
|
|
sel_union *u; /* used if selection type has a list of values */
|
|
|
|
|
int n; /* used if selection type has a list of values */
|
|
|
|
|
int typecode;
|
|
|
|
|
} selection_node;
|
|
|
|
|
|
|
|
|
|
typedef struct sort_node {
|
|
|
|
|
struct sort_node *next;
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
enum pids_item sr;
|
|
|
|
|
int (*xe)(char *, proc_t *); // special format_node 'pr' guy
|
|
|
|
|
enum pids_sort_order reverse;
|
2002-02-02 04:17:29 +05:30
|
|
|
int typecode;
|
|
|
|
|
} sort_node;
|
|
|
|
|
|
|
|
|
|
typedef struct format_node {
|
|
|
|
|
struct format_node *next;
|
|
|
|
|
char *name; /* user can override default name */
|
2002-11-27 15:22:19 +05:30
|
|
|
int (*pr)(char *restrict const outbuf, const proc_t *restrict const pp); // print function
|
2002-02-02 04:17:29 +05:30
|
|
|
int width;
|
|
|
|
|
int vendor; /* Vendor that invented this */
|
|
|
|
|
int flags;
|
|
|
|
|
int typecode;
|
|
|
|
|
} format_node;
|
|
|
|
|
|
|
|
|
|
typedef struct format_struct {
|
|
|
|
|
const char *spec; /* format specifier */
|
|
|
|
|
const char *head; /* default header in the POSIX locale */
|
2002-11-27 15:22:19 +05:30
|
|
|
int (* const pr)(char *restrict const outbuf, const proc_t *restrict const pp); // print function
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
enum pids_item sr;
|
2002-02-02 04:17:29 +05:30
|
|
|
const int width;
|
|
|
|
|
const int vendor; /* Where does this come from? */
|
|
|
|
|
const int flags;
|
|
|
|
|
} format_struct;
|
|
|
|
|
|
|
|
|
|
/* though ps-specific, needed by general file */
|
|
|
|
|
typedef struct macro_struct {
|
|
|
|
|
const char *spec; /* format specifier */
|
|
|
|
|
const char *head; /* default header in the POSIX locale */
|
|
|
|
|
} macro_struct;
|
|
|
|
|
|
|
|
|
|
/**************** PS TYPEDEF ***********************/
|
|
|
|
|
|
|
|
|
|
typedef struct aix_struct {
|
|
|
|
|
const int desc; /* 1-character format code */
|
|
|
|
|
const char *spec; /* format specifier */
|
|
|
|
|
const char *head; /* default header in the POSIX locale */
|
|
|
|
|
} aix_struct;
|
|
|
|
|
|
|
|
|
|
typedef struct shortsort_struct {
|
|
|
|
|
const int desc; /* 1-character format code */
|
|
|
|
|
const char *spec; /* format specifier */
|
|
|
|
|
} shortsort_struct;
|
|
|
|
|
|
|
|
|
|
/* Save these options for later: -o o -O O --format --sort */
|
|
|
|
|
typedef struct sf_node {
|
|
|
|
|
struct sf_node *next; /* next arg */
|
|
|
|
|
format_node *f_cooked; /* convert each arg alone, then merge */
|
|
|
|
|
sort_node *s_cooked; /* convert each arg alone, then merge */
|
|
|
|
|
char *sf;
|
|
|
|
|
int sf_code;
|
|
|
|
|
} sf_node;
|
|
|
|
|
|
|
|
|
|
/********************* UNDECIDED GLOBALS **************/
|
|
|
|
|
|
|
|
|
|
/* output.c */
|
2003-09-28 22:53:29 +05:30
|
|
|
extern void show_one_proc(const proc_t *restrict const p, const format_node *restrict fmt);
|
2002-02-02 04:17:29 +05:30
|
|
|
extern void print_format_specifiers(void);
|
|
|
|
|
extern const aix_struct *search_aix_array(const int findme);
|
|
|
|
|
extern const shortsort_struct *search_shortsort_array(const int findme);
|
|
|
|
|
extern const format_struct *search_format_array(const char *findme);
|
|
|
|
|
extern const macro_struct *search_macro_array(const char *findme);
|
|
|
|
|
extern void init_output(void);
|
2003-09-29 09:39:52 +05:30
|
|
|
extern int pr_nop(char *restrict const outbuf, const proc_t *restrict const pp);
|
2002-02-02 04:17:29 +05:30
|
|
|
|
|
|
|
|
/* global.c */
|
|
|
|
|
extern void reset_global(void);
|
|
|
|
|
|
|
|
|
|
/* global.c */
|
|
|
|
|
extern int all_processes;
|
2002-10-12 09:55:57 +05:30
|
|
|
extern const char *bsd_j_format;
|
|
|
|
|
extern const char *bsd_l_format;
|
|
|
|
|
extern const char *bsd_s_format;
|
|
|
|
|
extern const char *bsd_u_format;
|
|
|
|
|
extern const char *bsd_v_format;
|
2002-02-02 04:17:29 +05:30
|
|
|
extern int bsd_c_option;
|
|
|
|
|
extern int bsd_e_option;
|
|
|
|
|
extern uid_t cached_euid;
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
extern int cached_tty;
|
2002-02-02 04:17:29 +05:30
|
|
|
extern char forest_prefix[4 * 32*1024 + 100];
|
|
|
|
|
extern int forest_type;
|
|
|
|
|
extern unsigned format_flags; /* -l -f l u s -j... */
|
|
|
|
|
extern format_node *format_list; /* digested formatting options */
|
|
|
|
|
extern unsigned format_modifiers; /* -c -j -y -P -L... */
|
|
|
|
|
extern int header_gap;
|
|
|
|
|
extern int header_type; /* none, single, multi... */
|
|
|
|
|
extern int include_dead_children;
|
|
|
|
|
extern int lines_to_next_header;
|
|
|
|
|
extern int max_line_width;
|
|
|
|
|
extern int negate_selection;
|
2002-11-27 05:54:01 +05:30
|
|
|
extern int page_size; // "int" for math reasons?
|
2002-02-02 04:17:29 +05:30
|
|
|
extern unsigned personality;
|
|
|
|
|
extern int prefer_bsd_defaults;
|
|
|
|
|
extern int running_only;
|
|
|
|
|
extern int screen_cols;
|
|
|
|
|
extern int screen_rows;
|
|
|
|
|
extern selection_node *selection_list;
|
|
|
|
|
extern unsigned simple_select;
|
|
|
|
|
extern sort_node *sort_list;
|
2002-10-12 09:55:57 +05:30
|
|
|
extern const char *sysv_f_format;
|
|
|
|
|
extern const char *sysv_fl_format;
|
|
|
|
|
extern const char *sysv_j_format;
|
|
|
|
|
extern const char *sysv_l_format;
|
2003-09-20 13:59:55 +05:30
|
|
|
extern unsigned thread_flags;
|
2002-02-02 04:17:29 +05:30
|
|
|
extern int unix_f_option;
|
|
|
|
|
extern int user_is_number;
|
|
|
|
|
extern int wchan_is_number;
|
2012-01-02 13:03:34 +05:30
|
|
|
extern const char *the_word_help;
|
2002-02-02 04:17:29 +05:30
|
|
|
|
|
|
|
|
/************************* PS GLOBALS *********************/
|
|
|
|
|
|
2012-01-02 13:03:34 +05:30
|
|
|
/* display.c */
|
|
|
|
|
extern char *myname;
|
|
|
|
|
|
2002-02-02 04:17:29 +05:30
|
|
|
/* sortformat.c */
|
|
|
|
|
extern int defer_sf_option(const char *arg, int source);
|
2015-10-26 05:48:52 +05:30
|
|
|
extern const char *process_sf_options();
|
2002-02-02 04:17:29 +05:30
|
|
|
extern void reset_sortformat(void);
|
|
|
|
|
|
|
|
|
|
/* select.c */
|
|
|
|
|
extern int want_this_proc(proc_t *buf);
|
|
|
|
|
extern const char *select_bits_setup(void);
|
|
|
|
|
|
|
|
|
|
/* help.c */
|
ps: exploit those new <pids> task/threads capabilities
This commit represents the ps transition to the <pids>
'stacks' interface. While an effort to minimize impact
on existing code was made (as with a disguised proc_t)
the changes were still extensive. Along the way, a few
modifications beyond simply conversion were also made.
------------------------------------------------------
Here's a brief overview the design of this conversion:
. The need to satisfy relative enum requirements could
not easily have been made table driven since any entry
in the format_array might require several <pids> items
in support. So I decided to allow every print function
to contribute its own relative enums once the decision
as to exactly what will be printed had been finalized.
. A similar approach was taken for sorting, since it's
possible to have sort keys that will not be displayed.
Here, I relied on the existing print extensions above.
. In summary, just prior to printing ps walks thru two
lists one time (the format_list & sort_list) and calls
each print function. That function does not print, but
sets its required enum if necessary. Later, when those
same functions are called repeatedly for every printed
line, the only overhead will be an if test and branch.
------------------------------------------------------
Below is a summary of major changes beyond conversion:
. Sorts are now the responsibility of the library. And
therefore the total # of sortable fields substantially
increased without effort. Additionally, several quirky
fields remain as sortable, even though they can't ever
be printed(?). Surely that must make sense to someone.
[ while on this subject of sort, please do *not* try ]
[ to sort old ps on 'args'. or better yet, if you do ]
[ try that sort, see if you can determine his order, ]
[ without peeking at the source. that one hurts yet! ]
. All logic dealing with the old openproc flags and ps
struct members known as 'need' have been whacked since
that entire area was solely the new library's concern.
. Remaining malloc/calloc calls to stdlib were changed
to xmalloc/xcalloc from our own include/xalloc.h file.
None of the replaced calls ever checked return values.
[ be aware that 2 minor potential memory leaks exist ]
[ depending on command line arguments. no attempt is ]
[ made to free dynamically acquired format/sort node ]
[ structures upon return; a conscious design choice. ]
Signed-off-by: Jim Warner <james.warner@comcast.net>
2015-10-03 10:30:00 +05:30
|
|
|
extern void __attribute__ ((__noreturn__)) do_help(const char *opt, int rc);
|
2002-02-02 04:17:29 +05:30
|
|
|
|
|
|
|
|
/* global.c */
|
|
|
|
|
extern void self_info(void);
|
2012-01-11 02:17:11 +05:30
|
|
|
extern void catastrophic_failure(const char *filename, unsigned int linenum,
|
|
|
|
|
const char *message);
|
2002-02-02 04:17:29 +05:30
|
|
|
|
|
|
|
|
/* parser.c */
|
|
|
|
|
extern int arg_parse(int argc, char *argv[]);
|
|
|
|
|
|
|
|
|
|
#endif
|
