#2819 CI: test_ipa_subdomains_server failed on rhel6 + --coverage (FAIL: test_ipa_subdom_server)
Closed: Fixed None Opened 3 years ago by lslebodn.

Currently, we ignore failure in make check for COVERAGE BUILD.
http://sssd-ci.duckdns.org/logs/job/29/37/rhel6/ci.html

Link to loutput from make check
http://sssd-ci.duckdns.org/logs/job/29/37/rhel6/ci-build-coverage/ci-make-check.log


I will own this ticket and fix in 1.13.3, because it's my test that's failing.

owner: somebody => jhrozek

Tests don't need BZ clone

rhbz: => 0

Fields changed

milestone: NEEDS_TRIAGE => SSSD 1.13.3
priority: major => minor

As I thought; the bug is in gcc on rhel6.
The test "test_ipa_subdom_server" passes on rhel6 with defaul CFLAGS. It DOES NOT pass only with "--coverage"

I tried to bisect it and the bug is fixed in upstream gcc-4.6.0
{{{
commit 7959b7e646b493f48a2ea7228fbf1c43f84bedea
Author: rth rth@138bc75d-0d04-0410-961f-82ee72b054a4
Date: Wed Jul 21 18:50:02 2010 +0000

   Eliminate the computed goto in x64 varargs.

   Use a simple test and branch vs 0 instead.  Which also allows us 
   to avoid funny code emission problems as well.

   git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@162384 138bc75d-0d04-0410-961f-82ee72b054a4

This is a reason why there is not a problem with test_ipa_subdom_server
on newer distributions.

There is a reasonably workaround. The crash is fixed if sssd is compiled with optimisations and with "--coverage". I would like to avoid setting
"-O1" for all distributions in coverage build because optimisations "obfuscate" a little bit code coverage. So we might enable it only for rhel6.

Fields changed

owner: jhrozek => lslebodn
patch: 0 => 1
status: new => assigned

master:

sssd-1-13:

milestone: SSSD 1.13.3 => SSSD 1.13.2
resolution: => fixed
status: assigned => closed

Metadata Update from @lslebodn:
- Issue assigned to lslebodn
- Issue set to the milestone: SSSD 1.13.2

2 years ago

Login to comment on this ticket.

Metadata