selftests/powerpc: Don't rely on segfault to rerun the test

Browse source

The test case tm-signal-context-force-tm expects a segfault to happen
on returning from signal handler, and then does a setcontext() to run
the test again. However, the test doesn't always segfault, causing the
test to run a single time.

This patch fixes the test by putting it within a loop and jumping, via
setcontext, just prior to the loop in case it segfaults. This way we
get the desired behavior (run the test COUNT_MAX times) regardless if
it segfaults or not. This also reduces the use of setcontext for
control flow logic, keeping it only in the segfault handler.

Also, since 'count' is changed within the signal handler, it is
declared as volatile to prevent any compiler optimization getting
confused with asynchronous changes.

Signed-off-by: Gustavo Luiz Duarte <gustavold@linux.ibm.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20200211033831.11165-3-gustavold@linux.ibm.com

...

This commit is contained in:

Gustavo Luiz Duarte 2020-02-11 00:38:31 -03:00 • committed by Michael Ellerman

parent 915b7f6f9a

commit 0f8f554e52

1 changed files with 35 additions and 39 deletions

Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL

Download patch file Download diff file Expand all files Collapse all files

78

tools/testing/selftests/powerpc/tm/tm-signal-context-force-tm.c

View file