Created attachment 107037 [details] dmesg ==System Environment== -------------------------- Regression: Yes. Good commit on -next-queued: 9c787942907face82da505c2c5493998b56cfc5a(2014_09_08) Non-working platforms: IVB ==kernel== -------------------------- origin/drm-intel-nightly:7101d84020f63f1da7e0c5d021cdd6be4d515de5(fails) drm-intel-nightly: 2014y-09m-25d-15h-14m-02s UTC integration manifest origin/drm-intel-next-queued:7ca5a41f4da201371e131fc0641033652f76bf30(fails) drm/i915: Avoid re-configure panel on every PSR re-enable. origin/drm-intel-fixes: c84db77010877da6c5da119868ed54c43d59e726(works) drm/i915/hdmi: fix hdmi audio state readout ==Bug detailed description== ----------------------------- igt/gem_reset_stats/ban-blt costs long time to execute Output: [root@x-ivb9 tests]# time ./gem_reset_stats --run-subtest ban-blt IGT-Version: 1.8-g32a0308 (x86_64) (Linux: 3.17.0-rc6_drm-intel-nightly_7101d8_20140928_debug+ x86_64) retrying for ban (9) retrying for ban (8) retrying for ban (7) retrying for ban (6) retrying for ban (5) retrying for ban (4) retrying for ban (3) retrying for ban (2) retrying for ban (1) retrying for ban (0) Test assertion failure function test_ban, file gem_reset_stats.c:572: Failed assertion: h4 == -EIO Subtest ban-blt: FAIL (1414.204s) Test requirement not met in function gem_require_ring, file ioctl_wrappers.c:880: Test requirement: gem_has_vebox(fd) real 24m48.268s user 0m0.002s sys 0m0.027s ==Reproduce steps== ---------------------------- 1. ./gem_reset_stats --run-subtest ban-blt
Looks like full-ppgtt. Try testing with i915.enable_ppgtt=1
(In reply to comment #1) > Looks like full-ppgtt. Try testing with i915.enable_ppgtt=1 Test passes with i915.enable_ppgtt=1 [root@x-ivb9 tests]# time ./gem_reset_stats --run-subtest ban-blt IGT-Version: 1.8-g32a0308 (x86_64) (Linux: 3.17.0-rc6_drm-intel-nightly_7101d8_20140929+ x86_64) Subtest ban-blt: SUCCESS (13.517s) Test requirement not met in function gem_require_ring, file ioctl_wrappers.c:880: Test requirement: gem_has_vebox(fd) real 0m13.542s user 0m0.001s sys 0m0.006s [root@x-ivb9 tests]# echo $? 0
The ban takes long time because the ring head is pointing to outside of the batch that is being tested. So the hang recovery works, in time as the 'not progress' limit gets reached. The underlying problem is that with full-ppgtt mode we somehow setup something with hw (page tables?) incorrectly.
This needs to be tracked in a JIRA.
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.