[Bug 78221] 3.16 RC1: AMD R9 270 GPU locks up on some heavy 2D activity - GPU VM fault occurs. (possibly DMA copying issue strikes back?)

bugzilla-daemon at bugzilla.kernel.org bugzilla-daemon at bugzilla.kernel.org
Tue Aug 5 01:06:47 PDT 2014


--- Comment #16 from t3st3r at mail.ru ---
I have to admit this bug really suxx. I've attempted to bisect 3.15 -> 3.16rc1
several times but these attempts failed so far.

It looks like while I generally found quite fast ways to toggle this bug in
lucky cases, in some cases bug does not toggles for many hours or even can
require a reboot on same kernel version to increase chance bug appears. Bug
also seems to be really picky on previous history of GPU usage (e.g. launching
some 3D game before BfW can screw anything up and bug would not toggle in
literally days,  but can occasionally backstab).

In some cases deciding if kernel is bugged or not turned out to be a really
daunting and time consuming task. My last attempt was also wrong. I bet some of
"good" kernels were not as good as they should. Bad kernels on other hand
supposed to be bad, i.e. GPU crashed.

So last attempt also led me into really strange area, I don't even have
hardware in question so this module is never used.

P.S. and as far as I understand,
http://lists.freedesktop.org/archives/dri-devel/2014-June/062305.html fix
wasn't ported into 3.16 series? So 3.16 keeps failing for me.

And as example, last bisect looked like this:
$ git bisect log
git bisect start
# good: [1860e379875dfe7271c649058aeddffe5afd9d0d] Linux 3.15
git bisect good 1860e379875dfe7271c649058aeddffe5afd9d0d
# bad: [7171511eaec5bf23fb06078f59784a3a0626b38f] Linux 3.16-rc1
git bisect bad 7171511eaec5bf23fb06078f59784a3a0626b38f
# good: [aaeb2554337217dfa4eac2fcc90da7be540b9a73] Merge branch 'v4l_for_linus'
of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media into next
git bisect good aaeb2554337217dfa4eac2fcc90da7be540b9a73
# good: [16b9057804c02e2d351e9c8f606e909b43cbd9e7] Merge branch 'for-linus' of
git bisect good 16b9057804c02e2d351e9c8f606e909b43cbd9e7
# bad: [249c8b8d7e2d1bf9505dc46458537e77326c24fd] i40evf: remove unnecessary
log messages
git bisect bad 249c8b8d7e2d1bf9505dc46458537e77326c24fd
# good: [758bd61aa987e82765bd432f37bd81bd197c4b1a] Merge branch 'master' of
git bisect good 758bd61aa987e82765bd432f37bd81bd197c4b1a
# bad: [9db7cb6901740453a442e598563b576987dd471b] Merge branch 'master' of
git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-next into
git bisect bad 9db7cb6901740453a442e598563b576987dd471b
# bad: [99abe65ff18b6bbac2e55524827b571c3eccfa86] Merge tag 'nfc-next-3.16-1'
of git://git.kernel.org/pub/scm/linux/kernel/git/sameo/nfc-next
git bisect bad 99abe65ff18b6bbac2e55524827b571c3eccfa86
# bad: [75e58071c0c64f331ccc4c0037990a1e50099f7f] Merge branch 'for-linville'
of git://github.com/kvalo/ath
git bisect bad 75e58071c0c64f331ccc4c0037990a1e50099f7f
# bad: [d5738b41e555f97f597b19bc549fa811b516d6b6] Revert "wl1251: enforce
changed hw encryption support on monitor state change"
git bisect bad d5738b41e555f97f597b19bc549fa811b516d6b6
# bad: [0aa7142812c19af25ad21405eefc499e83da2fcc] iwlwifi: mvm: fix sparse
warning when _DEBUGFS isn't set
git bisect bad 0aa7142812c19af25ad21405eefc499e83da2fcc
# bad: [14b485f041e35f60212317017c2127b8a9b6be31] iwlwifi: mvm: prevent nic to
powered up at driver load
git bisect bad 14b485f041e35f60212317017c2127b8a9b6be31
# bad: [1e9551debacdaa044eeb514f4366beac6e18f6d9] iwlwifi: mvm: rs: don't allow
TPC when power save is disabled
git bisect bad 1e9551debacdaa044eeb514f4366beac6e18f6d9
# bad: [cebeb0f1885fa93c44be5d4e0b9b640210ff088c] Merge remote-tracking branch
'wireless-next/master' into iwlwifi-next
git bisect bad cebeb0f1885fa93c44be5d4e0b9b640210ff088c
# bad: [939ecf6b14c46e3448411a934418311b492bfee4] Merge remote-tracking branch
'iwlwifi-fixes/master' into iwlwifi-next
git bisect bad 939ecf6b14c46e3448411a934418311b492bfee4
# first bad commit: [939ecf6b14c46e3448411a934418311b492bfee4] Merge
remote-tracking branch 'iwlwifi-fixes/master' into iwlwifi-next

Obviously iwlwifi haves nothing to do with this bug. I bet I failed to judge
quality of some kernel(s) correctly one more time.

You are receiving this mail because:
You are watching the assignee of the bug.

More information about the dri-devel mailing list