Created attachment 108932 [details] dmesg ==System Environment== -------------------------- Regression: not sure Non-working platforms: IVB ==kernel== -------------------------- drm-intel-nightly/f56717bdf29eb7d6a9ec15f095ea2a6858895473 ==Bug detailed description== It sporadically causes "*ERROR* uncleared pch fifo underrun on pch transcoder A", fail rate: 3/5 output: IGT-Version: 1.8-ge622850 (x86_64) (Linux: 3.18.0-rc3_drm-intel-nightly_f56717_20141105+ x86_64) Using monotonic timestamps Beginning wf_vblank-interruptible on crtc 8, connector 18 1024x768 60 1024 1048 1184 1344 768 771 777 806 0xa 0x40 65000 ............................................................ wf_vblank-interruptible on crtc 8, connector 18: PASSED Beginning wf_vblank-interruptible on crtc 12, connector 18 1024x768 60 1024 1048 1184 1344 768 771 777 806 0xa 0x40 65000 ............................................................ wf_vblank-interruptible on crtc 12, connector 18: PASSED Beginning wf_vblank-interruptible on crtc 16, connector 18 1024x768 60 1024 1048 1184 1344 768 771 777 806 0xa 0x40 65000 ............................................................ wf_vblank-interruptible on crtc 16, connector 18: PASSED Subtest wf_vblank-interruptible: SUCCESS (30.349s) ==Reproduce steps== ---------------------------- 1. ./kms_flip --run-subtest wf_vblank-interruptible
subcase rcs-wf_vblank-vs-modeset also has this issue. [root@x-ivb9 tests]# ./kms_flip --run-subtest rcs-wf_vblank-vs-modeset IGT-Version: 1.8-ge34240d (x86_64) (Linux: 3.18.0-rc3_drm-intel-nightly_e6b3eb_20141107_debug+ x86_64) Using monotonic timestamps Beginning rcs-wf_vblank-vs-modeset on crtc 8, connector 18 1024x768 60 1024 1048 1184 1344 768 771 777 806 0xa 0x40 65000 ....... rcs-wf_vblank-vs-modeset on crtc 8, connector 18: PASSED Beginning rcs-wf_vblank-vs-modeset on crtc 12, connector 18 1024x768 60 1024 1048 1184 1344 768 771 777 806 0xa 0x40 65000 ...... rcs-wf_vblank-vs-modeset on crtc 12, connector 18: PASSED Beginning rcs-wf_vblank-vs-modeset on crtc 16, connector 18 1024x768 60 1024 1048 1184 1344 768 771 777 806 0xa 0x40 65000 ...... rcs-wf_vblank-vs-modeset on crtc 16, connector 18: PASSED Subtest rcs-wf_vblank-vs-modeset: SUCCESS (65.045s) You have new mail in /var/spool/mail/root [root@x-ivb9 tests]# dmesg -r|egrep "<[1-4]>"|grep drm <3>[ 7658.287754] [drm:intel_set_pch_fifo_underrun_reporting [i915]] *ERROR* uncleared pch fifo underrun on pch transcoder A <3>[ 7658.287769] [drm:intel_pch_fifo_underrun_irq_handler [i915]] *ERROR* PCH transcoder A FIFO underrun
It also impacts HSW.
I wonder whether we should tune down underrun warnings on the PCH, seems to happen why too often in random circumstances ...
(In reply to Daniel Vetter from comment #3) > I wonder whether we should tune down underrun warnings on the PCH, seems to > happen why too often in random circumstances ... IIRC I saw something like this on my IVB, but only with the non-existing VGA connector. I didn't post the patch since I couldn't be sure if it was just specific to that machine. As I recall, the cure involved rearranging the PCH underrun reporting disable/enable points a bit. I can see if I can still reproduce it here and maybe resurrect that patch, or recreate it if I can't find it anymore.
(In reply to Ville Syrjala from comment #4) > (In reply to Daniel Vetter from comment #3) > > I wonder whether we should tune down underrun warnings on the PCH, seems to > > happen why too often in random circumstances ... > > IIRC I saw something like this on my IVB, but only with the non-existing VGA > connector. I didn't post the patch since I couldn't be sure if it was just > specific to that machine. As I recall, the cure involved rearranging the PCH > underrun reporting disable/enable points a bit. I can see if I can still > reproduce it here and maybe resurrect that patch, or recreate it if I can't > find it anymore. Well we also have reports that the pch underruns at boot-up/takeover. So something is definitely still amiss and it should be easy to get a few tested-bys from reporters if you have a patch.
Please test this patch: http://patchwork.freedesktop.org/patch/37391/
Created attachment 109986 [details] dmesg with patch Test this patch 5 cycles, it works well.
commit b68362278af94e1171f5be9d4e44988601fb0439 Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Mon Nov 24 17:02:45 2014 +0100 drm/i915: More cautious with pch fifo underruns
Closing verified+fixed.
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.