tests: bluetooth: tester: Print backtrace on crash in native-sim build #97135
Conversation
|
when figured out on how to pass |
zephyrbot
added
area: Bluetooth Qualification
sjanc
force-pushed
the
tester_backtrace
branch
2 times, most recently
from
61abb6e to
bff0cbc
Compare
tests/bluetooth/tester/src/main.c
Outdated
| } | ||
| LOG_INF("=================================="); | ||
|
|
||
| free(strings); |
There was a problem hiding this comment.
Given that we do a free here, I guess backtrace_symbols is doing a malloc?
There was a problem hiding this comment.
yes, however I could use backtrace_symbols_fd() instead which doesn't use heap
sjanc
force-pushed
the
tester_backtrace
branch
from
bff0cbc to
ffa5d45
Compare
| int nptrs; | ||
|
|
||
| LOG_ERR("SEGMENTATION FAULT (address %p)", si->si_addr); | ||
| nptrs = backtrace(buffer, BACKTRACE_BUF_SIZE); |
There was a problem hiding this comment.
Consider adding a check if nptrs == BACKTRACE_BUF_SIZE which may indicate that the output was truncated, and we could add a log entry or something to let us know that BACKTRACE_BUF_SIZE may need to be increased
There was a problem hiding this comment.
yeah, I'll add this, however BACKTRACE_BUF_SIZE is 100 :)
tests/bluetooth/tester/src/main.c
Outdated
|
|
||
| LOG_ERR("SEGMENTATION FAULT (address %p)", si->si_addr); | ||
| nptrs = backtrace(buffer, BACKTRACE_BUF_SIZE); | ||
| backtrace_symbols_fd(buffer, nptrs, 2); |
There was a problem hiding this comment.
| backtrace_symbols_fd(buffer, nptrs, 2); | |
| backtrace_symbols_fd(buffer, nptrs, STDERR_FILENO); |
There was a problem hiding this comment.
May need to inlcude unistd.h for this
tests/bluetooth/tester/src/main.c
Outdated
| sigemptyset(&sa.sa_mask); | ||
| sa.sa_flags = SA_SIGINFO; | ||
|
|
||
| (void)sigaction(SIGSEGV, &sa, NULL); |
There was a problem hiding this comment.
sigaction may return an error; suggest to actually verify this and LOG if it happens
We may also want to return that error to main and immediately stop the program if it fails to register?
There was a problem hiding this comment.
ERRORS
EFAULT act or oldact points to memory which is not a valid part of the process address space.
EINVAL An invalid signal was specified. This will also be generated if an attempt is made to change the action for SIGKILL or SIGSTOP, which cannot be caught or ignored.
so sigaction() won't fail here, and even if it would fail I don't think we should abort (it is just debug code after all)
There was a problem hiding this comment.
Might be worth having an assert for it. Our coding guidelines require us to test all error codes (not sure if the coding guidelines actually apply to tests/test applications as well)
There was a problem hiding this comment.
I'll add warning log for this
sjanc
force-pushed
the
tester_backtrace
branch
from
ffa5d45 to
6c2bc0b
Compare
If Zephyr process was terminated due to segmentation fault print note and backtrace for debugging purposes. *** Booting Zephyr OS build v4.2.0-5260-g9a778a6beff7 *** [00:00:00.000,000] <dbg> bttester: tester_init: Initializing tester [00:00:00.000,000] <err> bttester_main: SEGMENTATION FAULT (address (nil)) zephyr/build/zephyr/zephyr.exe() [0x8049987] linux-gate.so.1(__kernel_rt_sigreturn+0x0) [0xf7f695b0] zephyr/build/zephyr/zephyr.exe() [0x804a8dc] zephyr/build/zephyr/zephyr.exe() [0x8049fc5] zephyr/build/zephyr/zephyr.exe() [0x80499fb] zephyr/build/zephyr/zephyr.exe() [0x80b2c24] zephyr/build/zephyr/zephyr.exe() [0x8060166] zephyr/build/zephyr/zephyr.exe() [0x80658ea] zephyr/build/zephyr/zephyr.exe() [0x80b9a45] /lib/libc.so.6(+0x72b20) [0xf7cd2b20] /lib/libc.so.6(+0xf8c58) [0xf7d58c58] Signed-off-by: Szymon Janc <[email protected]>
sjanc
force-pushed
the
tester_backtrace
branch
from
6c2bc0b to
9f21fc4
Compare
|
3 participants
If Zephyr process was terminated due to segmentation fault print note and backtrace for debugging purposes.