busybox/miscutils/inotifyd.c
James Byrne 6937487be7 libbb: reduce the overhead of single parameter bb_error_msg() calls
Back in 2007, commit 0c97c9d437 ("'simple' error message functions by
Loic Grenie") introduced bb_simple_perror_msg() to allow for a lower
overhead call to bb_perror_msg() when only a string was being printed
with no parameters. This saves space for some CPU architectures because
it avoids the overhead of a call to a variadic function. However there
has never been a simple version of bb_error_msg(), and since 2007 many
new calls to bb_perror_msg() have been added that only take a single
parameter and so could have been using bb_simple_perror_message().

This changeset introduces 'simple' versions of bb_info_msg(),
bb_error_msg(), bb_error_msg_and_die(), bb_herror_msg() and
bb_herror_msg_and_die(), and replaces all calls that only take a
single parameter, or use something like ("%s", arg), with calls to the
corresponding 'simple' version.

Since it is likely that single parameter calls to the variadic functions
may be accidentally reintroduced in the future a new debugging config
option WARN_SIMPLE_MSG has been introduced. This uses some macro magic
which will cause any such calls to generate a warning, but this is
turned off by default to avoid use of the unpleasant macros in normal
circumstances.

This is a large changeset due to the number of calls that have been
replaced. The only files that contain changes other than simple
substitution of function calls are libbb.h, libbb/herror_msg.c,
libbb/verror_msg.c and libbb/xfuncs_printf.c. In miscutils/devfsd.c,
networking/udhcp/common.h and util-linux/mdev.c additonal macros have
been added for logging so that single parameter and multiple parameter
logging variants exist.

The amount of space saved varies considerably by architecture, and was
found to be as follows (for 'defconfig' using GCC 7.4):

Arm:     -92 bytes
MIPS:    -52 bytes
PPC:   -1836 bytes
x86_64: -938 bytes

Note that for the MIPS architecture only an exception had to be made
disabling the 'simple' calls for 'udhcp' (in networking/udhcp/common.h)
because it made these files larger on MIPS.

Signed-off-by: James Byrne <james.byrne@origamienergy.com>
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2019-07-02 11:35:03 +02:00

225 lines
6.5 KiB
C

/* vi: set sw=4 ts=4: */
/*
* simple inotify daemon
* reports filesystem changes via userspace agent
*
* Copyright (C) 2008 by Vladimir Dronnikov <dronnikov@gmail.com>
*
* Licensed under GPLv2, see file LICENSE in this source tree.
*/
/*
* Use as follows:
* # inotifyd /user/space/agent dir/or/file/being/watched[:mask] ...
*
* When a filesystem event matching the specified mask is occurred on specified file (or directory)
* a userspace agent is spawned and given the following parameters:
* $1. actual event(s)
* $2. file (or directory) name
* $3. name of subfile (if any), in case of watching a directory
*
* E.g. inotifyd ./dev-watcher /dev:n
*
* ./dev-watcher can be, say:
* #!/bin/sh
* echo "We have new device in here! Hello, $3!"
*
* See below for mask names explanation.
*/
//config:config INOTIFYD
//config: bool "inotifyd (3.6 kb)"
//config: default n # doesn't build on Knoppix 5
//config: help
//config: Simple inotify daemon. Reports filesystem changes. Requires
//config: kernel >= 2.6.13
//applet:IF_INOTIFYD(APPLET(inotifyd, BB_DIR_SBIN, BB_SUID_DROP))
//kbuild:lib-$(CONFIG_INOTIFYD) += inotifyd.o
//usage:#define inotifyd_trivial_usage
//usage: "PROG FILE1[:MASK]..."
//usage:#define inotifyd_full_usage "\n\n"
//usage: "Run PROG on filesystem changes."
//usage: "\nWhen a filesystem event matching MASK occurs on FILEn,"
//usage: "\nPROG ACTUAL_EVENTS FILEn [SUBFILE] is run."
//usage: "\nIf PROG is -, events are sent to stdout."
//usage: "\nEvents:"
//usage: "\n a File is accessed"
//usage: "\n c File is modified"
//usage: "\n e Metadata changed"
//usage: "\n w Writable file is closed"
//usage: "\n 0 Unwritable file is closed"
//usage: "\n r File is opened"
//usage: "\n D File is deleted"
//usage: "\n M File is moved"
//usage: "\n u Backing fs is unmounted"
//usage: "\n o Event queue overflowed"
//usage: "\n x File can't be watched anymore"
//usage: "\nIf watching a directory:"
//usage: "\n y Subfile is moved into dir"
//usage: "\n m Subfile is moved out of dir"
//usage: "\n n Subfile is created"
//usage: "\n d Subfile is deleted"
//usage: "\n"
//usage: "\ninotifyd waits for PROG to exit."
//usage: "\nWhen x event happens for all FILEs, inotifyd exits."
#include "libbb.h"
#include "common_bufsiz.h"
#include <sys/inotify.h>
static const char mask_names[] ALIGN1 =
"a" // 0x00000001 File was accessed
"c" // 0x00000002 File was modified
"e" // 0x00000004 Metadata changed
"w" // 0x00000008 Writable file was closed
"0" // 0x00000010 Unwritable file closed
"r" // 0x00000020 File was opened
"m" // 0x00000040 File was moved from X
"y" // 0x00000080 File was moved to Y
"n" // 0x00000100 Subfile was created
"d" // 0x00000200 Subfile was deleted
"D" // 0x00000400 Self was deleted
"M" // 0x00000800 Self was moved
"\0" // 0x00001000 (unused)
// Kernel events, always reported:
"u" // 0x00002000 Backing fs was unmounted
"o" // 0x00004000 Event queued overflowed
"x" // 0x00008000 File is no longer watched (usually deleted)
;
enum {
MASK_BITS = sizeof(mask_names) - 1
};
int inotifyd_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
int inotifyd_main(int argc, char **argv)
{
int n;
unsigned mask;
struct pollfd pfd;
char **watches; // names of files being watched
const char *args[5];
// sanity check: agent and at least one watch must be given
if (!argv[1] || !argv[2])
bb_show_usage();
argv++;
// inotify_add_watch will number watched files
// starting from 1, thus watches[0] is unimportant,
// and 1st file name is watches[1].
watches = argv;
args[0] = *argv;
args[4] = NULL;
argc -= 2; // number of files we watch
// open inotify
pfd.fd = inotify_init();
if (pfd.fd < 0)
bb_simple_perror_msg_and_die("no kernel support");
// setup watches
while (*++argv) {
char *path = *argv;
char *masks = strchr(path, ':');
mask = 0x0fff; // assuming we want all non-kernel events
// if mask is specified ->
if (masks) {
*masks = '\0'; // split path and mask
// convert mask names to mask bitset
mask = 0;
while (*++masks) {
const char *found;
found = memchr(mask_names, *masks, MASK_BITS);
if (found)
mask |= (1 << (found - mask_names));
}
}
// add watch
n = inotify_add_watch(pfd.fd, path, mask);
if (n < 0)
bb_perror_msg_and_die("add watch (%s) failed", path);
//bb_error_msg("added %d [%s]:%4X", n, path, mask);
}
// setup signals
bb_signals(BB_FATAL_SIGS, record_signo);
// do watch
pfd.events = POLLIN;
while (1) {
int len;
void *buf;
struct inotify_event *ie;
again:
if (bb_got_signal)
break;
n = poll(&pfd, 1, -1);
// Signal interrupted us?
if (n < 0 && errno == EINTR)
goto again;
// Under Linux, above if() is not necessary.
// Non-fatal signals, e.g. SIGCHLD, when set to SIG_DFL,
// are not interrupting poll().
// Thus we can just break if n <= 0 (see below),
// because EINTR will happen only on SIGTERM et al.
// But this might be not true under other Unixes,
// and is generally way too subtle to depend on.
if (n <= 0) // strange error?
break;
// read out all pending events
// (NB: len must be int, not ssize_t or long!)
#define eventbuf bb_common_bufsiz1
setup_common_bufsiz();
xioctl(pfd.fd, FIONREAD, &len);
ie = buf = (len <= COMMON_BUFSIZE) ? eventbuf : xmalloc(len);
len = full_read(pfd.fd, buf, len);
// process events. N.B. events may vary in length
while (len > 0) {
int i;
// cache relevant events mask
unsigned m = ie->mask & ((1 << MASK_BITS) - 1);
if (m) {
char events[MASK_BITS + 1];
char *s = events;
for (i = 0; i < MASK_BITS; ++i, m >>= 1) {
if ((m & 1) && (mask_names[i] != '\0'))
*s++ = mask_names[i];
}
*s = '\0';
if (LONE_CHAR(args[0], '-')) {
/* "inotifyd - FILE": built-in echo */
printf(ie->len ? "%s\t%s\t%s\n" : "%s\t%s\n", events,
watches[ie->wd],
ie->name);
fflush(stdout);
} else {
// bb_error_msg("exec %s %08X\t%s\t%s\t%s", args[0],
// ie->mask, events, watches[ie->wd], ie->len ? ie->name : "");
args[1] = events;
args[2] = watches[ie->wd];
args[3] = ie->len ? ie->name : NULL;
spawn_and_wait((char **)args);
}
// we are done if all files got final x event
if (ie->mask & 0x8000) {
if (--argc <= 0)
goto done;
inotify_rm_watch(pfd.fd, ie->wd);
}
}
// next event
i = sizeof(struct inotify_event) + ie->len;
len -= i;
ie = (void*)((char*)ie + i);
}
if (eventbuf != buf)
free(buf);
} // while (1)
done:
return bb_got_signal;
}