When running tests, we should probably accept any help we can get when
it comes to detecting issues early or making them more debuggable. We
have seen a few cases where a test_progs_noalu32 run, for example,
encountered a soft lockup and stopped making progress. It was only
interrupted once we hit the overall test timeout [0]. We can not and do
not want to necessarily rely on test timeouts, because those rely on
infrastructure provided by the environment we run in (and which is not
present in tools/testing/selftests/bpf/vmtest.sh, for example).
To that end, let's enable panics on soft as well as hard lockups to fail
fast should we encounter one. That's happening in the configuration
indented to be used for selftests (including when using vmtest.sh or
when running in BPF CI).
[0] https://github.com/kernel-patches/bpf/runs/
7844499997
Signed-off-by: Daniel Müller <deso@posteo.net>
Link: https://lore.kernel.org/r/20221025231546.811766-1-deso@posteo.net
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
CONFIG_BLK_DEV_LOOP=y
+CONFIG_BOOTPARAM_HARDLOCKUP_PANIC=y
+CONFIG_BOOTPARAM_SOFTLOCKUP_PANIC=y
CONFIG_BPF=y
CONFIG_BPF_EVENTS=y
CONFIG_BPF_JIT=y
CONFIG_BLK_DEV_RAM_SIZE=16384
CONFIG_BLK_DEV_THROTTLING=y
CONFIG_BONDING=y
-CONFIG_BOOTPARAM_HARDLOCKUP_PANIC=y
CONFIG_BOOTTIME_TRACING=y
CONFIG_BPF_JIT_ALWAYS_ON=y
CONFIG_BPF_KPROBE_OVERRIDE=y