2009-07-08 13:19:16 -07:00
|
|
|
/*
|
2013-04-23 13:41:32 -07:00
|
|
|
* Copyright (c) 2008, 2009, 2010, 2011, 2013 Nicira, Inc.
|
2009-07-08 13:19:16 -07:00
|
|
|
*
|
2009-06-15 15:11:30 -07:00
|
|
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
|
|
|
* you may not use this file except in compliance with the License.
|
|
|
|
* You may obtain a copy of the License at:
|
2009-07-08 13:19:16 -07:00
|
|
|
*
|
2009-06-15 15:11:30 -07:00
|
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
*
|
|
|
|
* Unless required by applicable law or agreed to in writing, software
|
|
|
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
|
|
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
|
|
* See the License for the specific language governing permissions and
|
|
|
|
* limitations under the License.
|
2009-07-08 13:19:16 -07:00
|
|
|
*/
|
|
|
|
|
|
|
|
#include <config.h>
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
#include <errno.h>
|
|
|
|
#include <fcntl.h>
|
2014-03-13 15:28:54 -07:00
|
|
|
#include <inttypes.h>
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
#include <string.h>
|
|
|
|
#include <unistd.h>
|
2011-05-13 11:55:22 -07:00
|
|
|
|
2009-07-08 13:19:16 -07:00
|
|
|
#include "backtrace.h"
|
2014-12-15 14:10:38 +01:00
|
|
|
#include "openvswitch/vlog.h"
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
#include "util.h"
|
2014-03-13 15:28:54 -07:00
|
|
|
|
|
|
|
VLOG_DEFINE_THIS_MODULE(backtrace);
|
2011-05-13 11:55:22 -07:00
|
|
|
|
|
|
|
#ifdef HAVE_BACKTRACE
|
|
|
|
#include <execinfo.h>
|
|
|
|
void
|
|
|
|
backtrace_capture(struct backtrace *b)
|
|
|
|
{
|
2023-05-30 09:34:12 +02:00
|
|
|
b->n_frames = backtrace(b->frames, BACKTRACE_MAX_FRAMES);
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
backtrace_format(struct ds *ds, const struct backtrace *bt,
|
|
|
|
const char *delimiter)
|
|
|
|
{
|
|
|
|
if (bt->n_frames) {
|
|
|
|
char **symbols = backtrace_symbols(bt->frames, bt->n_frames);
|
|
|
|
|
|
|
|
if (!symbols) {
|
|
|
|
return;
|
|
|
|
}
|
2011-05-13 11:55:22 -07:00
|
|
|
|
2023-05-30 09:34:12 +02:00
|
|
|
for (int i = 0; i < bt->n_frames - 1; i++) {
|
|
|
|
ds_put_format(ds, "%s%s", symbols[i], delimiter);
|
|
|
|
}
|
|
|
|
|
|
|
|
ds_put_format(ds, "%s", symbols[bt->n_frames - 1]);
|
|
|
|
|
|
|
|
free(symbols);
|
2011-05-13 11:55:22 -07:00
|
|
|
}
|
|
|
|
}
|
2014-03-13 15:28:54 -07:00
|
|
|
|
2013-04-23 13:41:32 -07:00
|
|
|
#else
|
2011-05-13 11:55:22 -07:00
|
|
|
void
|
|
|
|
backtrace_capture(struct backtrace *backtrace)
|
|
|
|
{
|
2010-12-16 11:04:14 -08:00
|
|
|
backtrace->n_frames = 0;
|
2009-07-08 13:19:16 -07:00
|
|
|
}
|
2014-03-13 15:28:54 -07:00
|
|
|
|
2023-05-30 09:34:12 +02:00
|
|
|
void
|
|
|
|
backtrace_format(struct ds *ds, const struct backtrace *bt OVS_UNUSED,
|
|
|
|
const char *delimiter OVS_UNUSED)
|
2014-03-13 15:28:54 -07:00
|
|
|
{
|
2023-05-30 09:34:12 +02:00
|
|
|
ds_put_cstr(ds, "backtrace() is not supported!\n");
|
2014-03-13 15:28:54 -07:00
|
|
|
}
|
2023-05-30 09:34:12 +02:00
|
|
|
#endif
|
2014-03-13 15:28:54 -07:00
|
|
|
|
|
|
|
void
|
|
|
|
log_backtrace_at(const char *msg, const char *where)
|
|
|
|
{
|
|
|
|
struct backtrace b;
|
|
|
|
struct ds ds = DS_EMPTY_INITIALIZER;
|
|
|
|
|
|
|
|
backtrace_capture(&b);
|
|
|
|
if (msg) {
|
|
|
|
ds_put_format(&ds, "%s ", msg);
|
|
|
|
}
|
|
|
|
|
|
|
|
ds_put_cstr(&ds, where);
|
2023-05-30 09:34:12 +02:00
|
|
|
ds_put_cstr(&ds, " backtrace:\n");
|
|
|
|
backtrace_format(&ds, &b, "\n");
|
|
|
|
VLOG_ERR("%s", ds_cstr_ro(&ds));
|
2014-03-13 15:28:54 -07:00
|
|
|
|
|
|
|
ds_destroy(&ds);
|
|
|
|
}
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
|
2023-05-30 09:34:12 +02:00
|
|
|
#if defined(HAVE_UNWIND) || defined(HAVE_BACKTRACE)
|
|
|
|
static bool
|
|
|
|
read_received_backtrace(int fd, void *dest, size_t len)
|
|
|
|
{
|
|
|
|
VLOG_DBG("%s fd %d", __func__, fd);
|
|
|
|
fcntl(fd, F_SETFL, O_NONBLOCK);
|
|
|
|
memset(dest, 0, len);
|
|
|
|
|
|
|
|
int byte_read = read(fd, dest, len);
|
|
|
|
if (byte_read < 0) {
|
|
|
|
VLOG_ERR("Read fd %d failed: %s", fd, ovs_strerror(errno));
|
|
|
|
}
|
|
|
|
|
|
|
|
return byte_read > 0;;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
static bool
|
|
|
|
read_received_backtrace(int fd OVS_UNUSED, void *dest OVS_UNUSED,
|
|
|
|
size_t len OVS_UNUSED)
|
|
|
|
{
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
#ifdef HAVE_UNWIND
|
|
|
|
void
|
2023-05-30 09:34:12 +02:00
|
|
|
log_received_backtrace(int fd)
|
|
|
|
{
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
struct unw_backtrace backtrace[UNW_MAX_DEPTH];
|
|
|
|
|
2023-05-30 09:34:12 +02:00
|
|
|
if (read_received_backtrace(fd, backtrace, UNW_MAX_BUF)) {
|
|
|
|
struct ds ds = DS_EMPTY_INITIALIZER;
|
|
|
|
|
|
|
|
ds_put_cstr(&ds, BACKTRACE_DUMP_MSG);
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
|
|
|
|
for (int i = 0; i < UNW_MAX_DEPTH; i++) {
|
|
|
|
if (backtrace[i].func[0] == 0) {
|
|
|
|
break;
|
|
|
|
}
|
2023-05-30 09:34:12 +02:00
|
|
|
ds_put_format(&ds, "0x%016"PRIxPTR" <%s+0x%"PRIxPTR">\n",
|
|
|
|
backtrace[i].ip,
|
|
|
|
backtrace[i].func,
|
|
|
|
backtrace[i].offset);
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
}
|
2023-05-30 09:34:12 +02:00
|
|
|
|
|
|
|
VLOG_WARN("%s", ds_cstr_ro(&ds));
|
|
|
|
|
|
|
|
ds_destroy(&ds);
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
}
|
|
|
|
}
|
2023-05-30 09:34:12 +02:00
|
|
|
#elif HAVE_BACKTRACE
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
void
|
2023-05-30 09:34:12 +02:00
|
|
|
log_received_backtrace(int fd)
|
|
|
|
{
|
|
|
|
struct backtrace bt;
|
|
|
|
|
|
|
|
if (read_received_backtrace(fd, &bt, sizeof bt)) {
|
|
|
|
struct ds ds = DS_EMPTY_INITIALIZER;
|
|
|
|
|
|
|
|
bt.n_frames = MIN(bt.n_frames, BACKTRACE_MAX_FRAMES);
|
|
|
|
|
|
|
|
ds_put_cstr(&ds, BACKTRACE_DUMP_MSG);
|
|
|
|
backtrace_format(&ds, &bt, "\n");
|
|
|
|
VLOG_WARN("%s", ds_cstr_ro(&ds));
|
|
|
|
|
|
|
|
ds_destroy(&ds);
|
|
|
|
}
|
fatal-signal: Catch SIGSEGV and print backtrace.
The patch catches the SIGSEGV signal and prints the backtrace
using libunwind at the monitor daemon. This makes debugging easier
when there is no debug symbol package or gdb installed on production
systems.
The patch works when the ovs-vswitchd compiles even without debug symbol
(no -g option), because the object files still have function symbols.
For example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <fatal_signal_handler+0x52>
|daemon_unix(monitor)|WARN|0x00007fb4900734b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007fb49013974d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <time_poll+0x108>
|daemon_unix(monitor)|WARN|0x00000000005153ec <poll_block+0x8c>
|daemon_unix(monitor)|WARN|0x000000000058630a <clean_thread_main+0x1aa>
|daemon_unix(monitor)|WARN|0x00000000004ffd1d <ovsthread_wrapper+0x7d>
|daemon_unix(monitor)|WARN|0x00007fb490b3b6ba <start_thread+0xca>
|daemon_unix(monitor)|WARN|0x00007fb49014541d <clone+0x6d>
|daemon_unix(monitor)|ERR|1 crashes: pid 122849 died, killed \
(Segmentation fault), core dumped, restarting
However, if the object files' symbols are stripped, then we can only
get init function plus offset value. This is still useful when trying
to see if two bugs have the same root cause, Example:
|daemon_unix(monitor)|WARN|SIGSEGV detected, backtrace:
|daemon_unix(monitor)|WARN|0x0000000000482752 <_init+0x7d68a>
|daemon_unix(monitor)|WARN|0x00007f5f7c8cf4b0 <killpg+0x40>
|daemon_unix(monitor)|WARN|0x00007f5f7c99574d <__poll+0x2d>
|daemon_unix(monitor)|WARN|0x000000000052b348 <_init+0x126280>
|daemon_unix(monitor)|WARN|0x00000000005153ec <_init+0x110324>
|daemon_unix(monitor)|WARN|0x0000000000407439 <_init+0x2371>
|daemon_unix(monitor)|WARN|0x00007f5f7c8ba830 <__libc_start_main+0xf0>
|daemon_unix(monitor)|WARN|0x0000000000408329 <_init+0x3261>
|daemon_unix(monitor)|ERR|1 crashes: pid 106155 died, killed \
(Segmentation fault), core dumped, restarting
Most C library functions are not async-signal-safe, meaning that
it is not safe to call them from a signal handler, for example
printf() or fflush(). To be async-signal-safe, the handler only
collects the stack info using libunwind, which is signal-safe, and
issues 'write' to the pipe, where the monitor thread reads and
prints to ovs-vswitchd.log.
Tested-at: https://travis-ci.org/williamtu/ovs-travis/builds/590503433
Signed-off-by: William Tu <u9012063@gmail.com>
Signed-off-by: Ben Pfaff <blp@ovn.org>
2019-09-27 10:22:55 -07:00
|
|
|
}
|
2023-05-30 09:34:12 +02:00
|
|
|
#else
|
|
|
|
void
|
|
|
|
log_received_backtrace(int daemonize_fd OVS_UNUSED)
|
|
|
|
{
|
|
|
|
VLOG_WARN("Backtrace using libunwind or backtrace() is not supported.");
|
|
|
|
}
|
|
|
|
#endif
|