Created attachment 105274 [details] dmesg after s3 ==System Environment== -------------------------- Regression: Yes. Good commit: f4be89cecea437aaddd7700d05c6bdb5678041f7 (-fixes branch) drm/i915: Fix crash when failing to parse MIPI VBT Non-working platforms: Broadwell ==kernel== -------------------------- -nightly: d82af52b1766594ece621e427f1604194ca2b415 (failed) -queued: 0f09bb2bf3d51a1e57e8dfc46e3c159170651c29 (failed) drm/i915/bdw: Make sure error capture keeps working with Execlists -fixes: 1a125d8a2c22b11741fc47d4ffcf7a5ffa044dd3 (failed) drm/i915: don't try to retrain a DP link on an inactive CRTC ==Bug detailed description== ----------------------------- System dmesg contains "ERROR"(like below) when machine resumes from s3 . <3>[ 20.508358] [drm:hsw_unclaimed_reg_detect] *ERROR* Unclaimed register dete ==Reproduce steps== ---------------------------- 1. S3 2. check dmesg
I will bisect later
==Bisect results== ---------------------------- Bisect shows: 66bc2cab191816d931f4d660d2f162130dd202b2 is the first bad commit. commit 66bc2cab191816d931f4d660d2f162130dd202b2 Author: Paulo Zanoni <paulo.r.zanoni@intel.com> AuthorDate: Wed Jul 16 17:49:30 2014 -0300 Commit: Daniel Vetter <daniel.vetter@ffwll.ch> CommitDate: Wed Jul 23 07:05:37 2014 +0200 drm/i915: BDW can also detect unclaimed registers By the time I wrote this patch, it allowed me to catch some problems. But due to patch reordering - in order to prevent fake "regression" reports - this patch may be merged after the fixes of the problems identified by this patch.
(In reply to comment #2) > ==Bisect results== > ---------------------------- > Bisect shows: 66bc2cab191816d931f4d660d2f162130dd202b2 is the first bad > commit. > commit 66bc2cab191816d931f4d660d2f162130dd202b2 > Author: Paulo Zanoni <paulo.r.zanoni@intel.com> > AuthorDate: Wed Jul 16 17:49:30 2014 -0300 > Commit: Daniel Vetter <daniel.vetter@ffwll.ch> > CommitDate: Wed Jul 23 07:05:37 2014 +0200 > > drm/i915: BDW can also detect unclaimed registers > > By the time I wrote this patch, it allowed me to catch some problems. > But due to patch reordering - in order to prevent fake "regression" > reports - this patch may be merged after the fixes of the problems > identified by this patch. This is not a "real" regression since the commit above is the one that starts detecting the problem, not the one that introduces the problem.
I have no idea about how I find out the one that introduces the problem. If someone give me a hand, that will be awesome.
(In reply to comment #4) > I have no idea about how I find out the one that introduces the problem. If > someone give me a hand, that will be awesome. At least in theory you could always apply the referenced commit at each step of the bisect. But I suspect you'd only find bdw enabling commits anyway.
Could you please retest with latest -nightly and if error persist could you please boot with i915.mmio_debug=1 than paste new dmesg?
(In reply to comment #6) > Could you please retest with latest -nightly and if error persist could you > please boot with i915.mmio_debug=1 than paste new dmesg? When boot with i915.mmio_debug=1, ERROR is gone. But ERROR exists without i915.mmio_debug=1 .
Created attachment 106888 [details] dmesg-with-i915.mmio_debug=1
Created attachment 106889 [details] dmesg-without-i915.mmio_debug=1
Created attachment 107437 [details] [review] Possible fix Hi Does this patch fix the problem? Thanks, Paulo
Created attachment 107532 [details] dmesg (In reply to Paulo Zanoni from comment #10) > Created attachment 107437 [details] [review] [review] > Possible fix > > Hi > > Does this patch fix the problem? > > Thanks, > Paulo The dmesg error still able to reproduce with this patch. root@x-bdw05:~# echo 0 > /sys/class/rtc/rtc0/wakealarm ; echo +10 > /sys/class/rtc/rtc0/wakealarm; echo mem > /sys/power/state root@x-bdw05:~# root@x-bdw05:~# dmesg -r|egrep "<[1-4]>"|grep drm <3>[ 669.801135] [drm:hsw_unclaimed_reg_detect] *ERROR* Unclaimed register detected. Please use the i915.mmio_debug=1 to debug this problem.
Created attachment 107625 [details] [review] Correct fix Ok, so I retested my original patch and it doesn't fix the problem for me either: it fixes a different problem I was seeing in the past... So I wrote a new patch, and I can confirm this one fixes the problem for me on my BDW machine. Please test this one.
(In reply to Paulo Zanoni from comment #12) > Created attachment 107625 [details] [review] [review] > Correct fix > > Ok, so I retested my original patch and it doesn't fix the problem for me > either: it fixes a different problem I was seeing in the past... > > So I wrote a new patch, and I can confirm this one fixes the problem for me > on my BDW machine. > > Please test this one. The demsg error unable to reproduce on this patch. root@x-bdw05:~# echo 0 > /sys/class/rtc/rtc0/wakealarm ; echo +10 > /sys/class/rtc/rtc0/wakealarm; echo mem > /sys/power/state root@x-bdw05:~# dmesg -r|egrep "<[1-4]>"|grep drm root@x-bdw05:~#
*** Bug 85220 has been marked as a duplicate of this bug. ***
Closing bug. Fixed by: Commit 25ba23c3198315efe6d70f67acd73cf11540ca19 drm/i915: run hsw_disable_pc8() later on resume If the bug still happens, please reopen.
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.