On Wed, 2017-06-14 at 10:36 +0100, Chris Wilson wrote: > Quoting Paul Kocialkowski (2017-06-12 16:21:45) > > On Mon, 2017-06-12 at 17:39 +0300, Paul Kocialkowski wrote: > > > This adds a call to close the DRM file descriptor. It is reauired as IGT > > > will attempt to become DRM master after running the test, resulting in a > > > failure. > > > > This should fix (at least) 3 of the failures seen on the CI machine that > > runs > > chamelium tests. > > This would be papering over the bug in the atexit handlers if they > assume they can become drm master and worse cause a failure it they > don't. I was not aware that the core made that presumption, so more > details would be useful. Here is the log from running the dp-hpd-storm test: IGT-Version: 1.19-geeeb304a (x86_64) (Linux: 4.12.0-rc4-ARCH+ x86_64) Subtest dp-hpd-storm: SUCCESS (8.917s) Test requirement not met in function require_connector_present, file chamelium.c:99: Test requirement: found No port of type HDMI-A was found Test requirement not met in function require_connector_present, file chamelium.c:99: Test requirement: found No port of type VGA was found Test requirement not met in function drm_open_driver_master, file drmtest.c:391: Test requirement: drmSetMaster(fd) == 0 Can't become DRM master, please check if no other DRM client is running. Last errno: 22, Invalid argument chamelium: igt_core.c:955: igt_skip: Assertion `in_fixture' failed. ./chamelium.sh: line 3: 8566 Aborted (core dumped) /usr/local/libexec/intel-gpu-tools/chamelium "$@" Which is actually called by igt_hpd_storm_exit_handler in order to call igt_hpd_storm_reset with the drm fd. Thinking about it, I should have made it clear in the commit message that this is very specific to the hpd storm feature testing. -- Paul Kocialkowski <paul.kocialkowski@xxxxxxxxxxxxxxx> Intel Finland Oy - BIC 0357606-4 - Westendinkatu 7, 02160 Espoo _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx