Skip to content
This repository has been archived by the owner on Jan 20, 2022. It is now read-only.

[LTP] accept02 from LTP fails, but is not reported #1937

Closed
boryspoplawski opened this issue Nov 4, 2020 · 0 comments · Fixed by #1943 or #1949
Closed

[LTP] accept02 from LTP fails, but is not reported #1937

boryspoplawski opened this issue Nov 4, 2020 · 0 comments · Fixed by #1943 or #1949
Assignees

Comments

@boryspoplawski
Copy link
Contributor

Description of the problem

accept02 from LTP test suite fails, but make regression (hence CI) does not report it.

Steps to reproduce

Run ./pal_loader ./accept02 in LibOS/shim/test/ltp/install/testcases/bin - it does not work:
TBROK: setsockopt() failed unexpectedly: ENOPROTOOPT (92)
Run make regression in LibOS/shim/test/ltp/ - the above failure is not reported.

pwmarcz added a commit to pwmarcz/graphene that referenced this issue Nov 4, 2020
Currently, a thread calling exit_group() will cause all other
threads to also exit, but the exit code for these threads will be
0. As a result, another process calling wait() might receive 0
instead of the right exit code.

Instead, make sure that all other threads exit with the same
exit_code and term_signal.

It looks like this regression was introduced in
50ca6c4. I noticed it because
LTP test accept02 incorrectly returned exit code 0 before failing
(see gramineproject#1937).

The previous commit strengthens a test (exit_group) that triggers
the described scenario.
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants