Bug 1018

Summary:	Segfault in debug_autotest on Alpine
Product:	DPDK	Reporter:	Owen Hilyard (ohilyard)
Component:	other	Assignee:	Thomas Monjalon (thomas)
Status:	UNCONFIRMED ---
Severity:	normal	CC:	david.marchand, thomas
Priority:	Normal
Version:	unspecified
Target Milestone:	---
Hardware:	x86
OS:	Linux

Description Owen Hilyard 2022-05-26 19:37:40 CEST

When running the debug_autotest under meson or as an independent invocation of dpdk_test, there is a segfault inside of rte_dump_stack. This seems to be related to the alpine port of libexecinfo. 

ARM and POWERPC were not tested, since Alpine is currently only run on x86.

Comment 1 Thomas Monjalon 2023-03-09 18:38:31 CET

The libexecinfo support has been completely removed from Alpine, so the test should not crash anymore.

See this commit: bc1db4f45af3 ("build: detect backtrace availability")

Comment 2 David Marchand 2023-03-10 10:51:06 CET

There is another issue.

Running debug_autotest ends up with a timeout.

08:30:00 DPDK_TEST='debug_autotest' /__w/dpdk/dpdk/build/app/test/dpdk-test --no-huge -m 2048 --file-prefix=debug_autotest
--- stdout ---
RTE>>debug_autotest
Child process terminated as expected - Test passed!
--- stderr ---
EAL: Detected CPU lcores: 2
EAL: Detected NUMA nodes: 1
EAL: Detected shared linkage of DPDK
EAL: Multi-process socket /var/run/dpdk/debug_autotest/mp_socket
EAL: Selected IOVA mode 'VA'
EAL: Cannot open VFIO container /dev/vfio/vfio, error 2 (No such file or directory)
EAL: VFIO support could not be initialized
APP: HPET is not enabled, using TSC as default timer
PANIC in test_panic():
Test Debug
test_exit_val


Attaching gdb to it (sorry I lost the output, so this is from memory..), the main thread was waiting for a child process on waitpid(-1).
There may be an issue with musl waitpid..