Bug 36392

Summary: 2.6.39, 3.0-rc2 Intel DRI Regression: Black Screen after trying to play video
Product: Drivers Reporter: Dennis Jansen (dennis.jansen)
Component: Video(DRI - Intel)Assignee: drivers_video-dri-intel (drivers_video-dri-intel)
Status: CLOSED UNREPRODUCIBLE    
Severity: high CC: akpm, bugzilla-kernel, chris, florian, maciej.rutecki, rjw
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 2.6.39,3.0-rc3 Tree: Mainline
Regression: Yes
Bug Depends on:    
Bug Blocks: 32012    
Attachments: xorg log
/var/log/messages excerpt
error_state second recorded crash
second crash xorg.log
3: error_state
4: error_state in 3.0-rc2
4: xorg.log in 3.0-rc2
4: dmesg in 3.0-rc2
5: var/log/messages in 3.0-rc3

Description Dennis Jansen 2011-06-01 04:41:57 UTC
It hangs, then the screen is black with backlight on, then backlight turns off. The system still responds then but no screen. Dmesg and xorg log to follow.

[drm] capturing error event; look for more information in /debug/dri/0/i915_error_state
Jun  1 06:28:46 notebook kernel: [35653.482419] [drm] Changing LVDS panel from (+hsync, +vsync) to (-hsync, -vsync)
Jun  1 06:28:46 notebook kernel: [35653.482574] ------------[ cut here ]------------
Jun  1 06:28:46 notebook kernel: [35653.482617] WARNING: at /home/kernel-ppa/COD/linux/drivers/gpu/drm/i915/intel_display.c:1079 assert_pll+0xc3/0xd0 [i915]()
Jun  1 06:28:46 notebook kernel: [35653.482621] Hardware name: Extensa 5220                   
Jun  1 06:28:46 notebook kernel: [35653.482623] PLL state assertion failure (expected on, current off)
Jun  1 06:28:46 notebook kernel: [35653.482626] Modules linked in: nls_iso8859_1 nls_cp437 vfat fat isofs usb_storage uas autofs4 binfmt_misc ppdev deflate zlib_deflate ctr twofish_generic twofish_i586 twofish_common camellia serpent blowfish cast5 des_generic cryptd aes_i586 aes_generic xcbc rmd160 sha512_generic sha256_generic sha1_generic crypto_null af_key snd_hda_codec_realtek snd_hda_intel snd_hda_codec snd_hwdep snd_pcm snd_seq_midi arc4 snd_rawmidi snd_seq_midi_event iwlagn snd_seq joydev mac80211 usbserial snd_timer snd_seq_device pcmcia cfg80211 snd ati_remote2 tifm_7xx1 nsc_ircc soundcore psmouse snd_page_alloc irda tifm_core yenta_socket crc_ccitt pcmcia_rsrc pcmcia_core sdhci_pci sdhci serio_raw acer_wmi sparse_keymap lp parport usbhid hid ahci libahci tg3 i915 drm_kms_helper drm i2c_algo_bit video
Jun  1 06:28:46 notebook kernel: [35653.482690] Pid: 18737, comm: kworker/u:31 Not tainted 2.6.39-020639-generic #201105190911
Jun  1 06:28:46 notebook kernel: [35653.482693] Call Trace:
Jun  1 06:28:46 notebook kernel: [35653.482723]  [<f82d7d83>] ? assert_pll+0xc3/0xd0 [i915]
Jun  1 06:28:46 notebook kernel: [35653.482737]  [<c104ffd1>] warn_slowpath_common+0x81/0xa0
Jun  1 06:28:46 notebook kernel: [35653.482768]  [<f82d7d83>] ? assert_pll+0xc3/0xd0 [i915]
Jun  1 06:28:46 notebook kernel: [35653.482778]  [<c1050093>] warn_slowpath_fmt+0x33/0x40
Jun  1 06:28:46 notebook kernel: [35653.482809]  [<f82d7d83>] assert_pll+0xc3/0xd0 [i915]
Jun  1 06:28:46 notebook kernel: [35653.482859]  [<f82d9ead>] intel_enable_pipe+0x3d/0x160 [i915]
Jun  1 06:28:46 notebook kernel: [35653.482897]  [<f82e64e7>] intel_crtc_mode_set+0xc27/0x2960 [i915]
Jun  1 06:28:46 notebook kernel: [35653.482931]  [<f82b25af>] ? i915_get_crtc_scanoutpos+0x4f/0x360 [i915]
Jun  1 06:28:46 notebook kernel: [35653.482937]  [<c1078500>] ? getnstimeofday+0x50/0x110
Jun  1 06:28:46 notebook kernel: [35653.482941]  [<c1078616>] ? do_gettimeofday+0x16/0x40
Jun  1 06:28:46 notebook kernel: [35653.482971]  [<f8054184>] ? drm_calc_vbltimestamp_from_scanoutpos+0x3b4/0x3e0 [drm]
Jun  1 06:28:46 notebook kernel: [35653.482991]  [<c128313d>] ? format_decode+0x2fd/0x360
Jun  1 06:28:46 notebook kernel: [35653.482996]  [<c1328150>] ? register_con_driver+0x170/0x170
Jun  1 06:28:46 notebook kernel: [35653.483001]  [<c1050447>] ? __call_console_drivers+0x77/0x90
Jun  1 06:28:46 notebook kernel: [35653.483043]  [<f82e147e>] ? i8xx_enable_fbc+0x1ee/0x3d0 [i915]
Jun  1 06:28:46 notebook kernel: [35653.483086]  [<f82e9fe5>] ? i965_update_wm+0x165/0x430 [i915]
Jun  1 06:28:46 notebook kernel: [35653.483111]  [<f8114108>] drm_crtc_helper_set_mode+0x298/0x410 [drm_kms_helper]
Jun  1 06:28:46 notebook kernel: [35653.483122]  [<c1528d0f>] ? _raw_spin_lock_irqsave+0x2f/0x50
Jun  1 06:28:46 notebook kernel: [35653.483143]  [<f8114764>] drm_helper_resume_force_mode+0x84/0x130 [drm_kms_helper]
Jun  1 06:28:46 notebook kernel: [35653.483171]  [<f82aae83>] i915_reset+0x193/0x210 [i915]
Jun  1 06:28:46 notebook kernel: [35653.483329]  [<f82af4fc>] i915_error_work_func+0xac/0xf0 [i915]
Jun  1 06:28:46 notebook kernel: [35653.483360]  [<c106806a>] process_one_work+0xda/0x340
Jun  1 06:28:46 notebook kernel: [35653.483383]  [<f82af450>] ? i915_get_vblank_timestamp+0xb0/0xb0 [i915]
Jun  1 06:28:46 notebook kernel: [35653.483388]  [<c1069883>] worker_thread+0xb3/0x210
Jun  1 06:28:46 notebook kernel: [35653.483392]  [<c10697d0>] ? manage_workers+0x90/0x90
Jun  1 06:28:46 notebook kernel: [35653.483396]  [<c106d465>] kthread+0x75/0x80
Jun  1 06:28:46 notebook kernel: [35653.483401]  [<c106d3f0>] ? kthread_bind+0x80/0x80
Jun  1 06:28:46 notebook kernel: [35653.483406]  [<c15305be>] kernel_thread_helper+0x6/0x10
Jun  1 06:28:46 notebook kernel: [35653.483409] ---[ end trace 17dfbc8d86d40af6 ]---
Jun  1 06:28:46 notebook kernel: [35653.592134] ------------[ cut here ]------------
Comment 1 Dennis Jansen 2011-06-01 04:42:24 UTC
Created attachment 60332 [details]
xorg log
Comment 2 Dennis Jansen 2011-06-01 04:43:13 UTC
Created attachment 60342 [details]
/var/log/messages excerpt

The "dmesg" is quite long. A lot is happening at the end.
Comment 3 Andrew Morton 2011-06-08 20:03:53 UTC
I assume that 2.6.38 was OK, and that this is a 2.6.38->2.6.39 regression?
Comment 4 Dennis Jansen 2011-06-09 09:35:39 UTC
Yes, this is a .38->.39 regression. There was a different bug in .38, though, more rare, but also related to a black screen (https://bugzilla.kernel.org/show_bug.cgi?id=27312#c8).
Comment 5 Dennis Jansen 2011-06-09 17:56:16 UTC
Created attachment 61372 [details]
error_state second recorded crash

I've logged the error state from debug/dri this time.
Comment 6 Dennis Jansen 2011-06-09 17:59:27 UTC
Created attachment 61382 [details]
second crash xorg.log

btw. This time the backlight didn't turn off, no idea why. And I see lots of iwlagn problems in demsg. let me know if you want the full and I'll extract it from /var/log
Comment 7 Dennis Jansen 2011-06-11 06:16:26 UTC
Still there in 3.0-rc2
Comment 8 Dennis Jansen 2011-06-11 06:17:06 UTC
Created attachment 61542 [details]
3: error_state
Comment 9 Dennis Jansen 2011-06-11 06:17:36 UTC
Created attachment 61552 [details]
4: error_state in 3.0-rc2
Comment 10 Dennis Jansen 2011-06-11 06:18:13 UTC
Created attachment 61562 [details]
4: xorg.log in 3.0-rc2
Comment 11 Dennis Jansen 2011-06-11 06:19:49 UTC
Created attachment 61572 [details]
4: dmesg in 3.0-rc2

By the way: The backlight doesn't turn off (anymore?). (Maybe I confused that with the other bug?)

And it always happened with mplayer, never with flash video.
Comment 12 Dennis Jansen 2011-06-11 11:48:48 UTC
Ok, it's actually much worse now, it crashes really hard on VT switch between X and console sometimes now. I can't say any more than that because it's a bit harder to reproduce and its a very hard freeze: no reisub, no response in any way. wifi still blinking but otherwise unresponsive. And as a make was going on, I had to replace the entire source tree because several files were cut off somewhere or filled with junk data.
Comment 13 Dennis Jansen 2011-06-11 13:40:20 UTC
this might be related to 
"[drm:drm_mode_getfb] *ERROR* invalid framebuffer id"
which shows up each time I switch out of X and back in.
Comment 14 Rafael J. Wysocki 2011-06-13 17:22:38 UTC
On Monday, June 13, 2011, Dennis Jansen wrote:
> On Sun, Jun 12, 2011 at 11:12 PM, Rafael J. Wysocki <rjw@sisk.pl> wrote:
> > This message has been generated automatically as a part of a report
> > of regressions introduced between 2.6.38 and 2.6.39.
> >
> > The following bug entry is on the current list of known regressions
> > introduced between 2.6.38 and 2.6.39.  Please verify if it still should
> > be listed and let the tracking team know (either way).
> 
> Yes it should. And there's nobody actively working on it yet it seems.
> >
> >
> > Bug-Entry       : http://bugzilla.kernel.org/show_bug.cgi?id=36392
> > Subject         : 2.6.39, 3.0-rc2 Intel DRI Regression: Black Screen after
> trying to play video
> > Submitter       : Dennis Jansen <dennis.jansen@web.de>
> > Date            : 2011-06-01 04:41 (12 days old)
Comment 15 Dennis Jansen 2011-06-15 18:58:55 UTC
On Wed, Jun 15, 2011 at 8:52 PM, Jesse Barnes <jbarnes@virtuousgeek.org> wrote:
> On Sun, 12 Jun 2011 23:12:11 +0200 (CEST)
> "Rafael J. Wysocki" <rjw@sisk.pl> wrote:
>
>> This message has been generated automatically as a part of a report
>> of regressions introduced between 2.6.38 and 2.6.39.
>>
>> The following bug entry is on the current list of known regressions
>> introduced between 2.6.38 and 2.6.39.  Please verify if it still should
>> be listed and let the tracking team know (either way).
>>
>>
>> Bug-Entry     : http://bugzilla.kernel.org/show_bug.cgi?id=36392
>> Subject               : 2.6.39, 3.0-rc2 Intel DRI Regression: Black Screen
>> after trying to play video
>> Submitter     : Dennis Jansen <dennis.jansen@web.de>
>> Date          : 2011-06-01 04:41 (12 days old)
>
> Did you ever bisect this issue?
>
> --
> Jesse Barnes, Intel Open Source Technology Center
>


No. Do you mean the black screen or the framebuffer id error? Because the black screen problem is not really bisectable. I don't know how exactly to provoke it. It doesn't happen immediately after boot, but after at least a day or so and one or more suspends I think. And not knowing what exactly is need to provoke it, it would take *quite a while* to get there. I'm currently trying -rc3 to see if things have improved.
Comment 16 Chris Wilson 2011-06-15 19:14:12 UTC
The GPU hang is due to the broken xf86-video-intel DDX. The blank screen is due to failed reset on the chip. The "invalid framebuffer id" error is relatively harmless and due to a silly Ubuntu patch.
Comment 17 Dennis Jansen 2011-06-16 21:24:56 UTC
I think this is fixed in 3.0-rc3.
Comment 18 Dennis Jansen 2011-06-19 06:13:12 UTC
No, it wasn't. It took longer this time than before. But that could just be random of course.
Comment 19 Dennis Jansen 2011-06-19 06:19:02 UTC
Created attachment 62752 [details]
5: var/log/messages in 3.0-rc3

attachded messages from 3.0-rc3.
First you see resume from standby, then the crash, then a new clean boot.
As always, it happened when trying to playback a video.
Comment 20 Dennis Jansen 2011-06-19 06:46:46 UTC
So Chris, are you saying that the crash is due to the driver being "too old"?

If so, is this a Won't fix? Do I have to upgrade my distribution or at least X driver to use and keep testing new releases?
Comment 21 Dennis Jansen 2011-07-26 08:35:58 UTC
I haven't noticed this anymore after the last -rc upgrade. Hoping it's not just accident or less frequent, I'm closing the bug.
Comment 22 Florian Mickler 2011-08-04 19:44:25 UTC
Thanks for following up on this. 
If the fix is not known, we better close it as unreproducible. 

Regards,
Flo