Intel VAAPI video freeze/hang bug
#61
Then let me try again ...

To simulate the "Use SW Filter for VAAPI" behaviour in helix, the following settings need to be selected

- Video -> Acceleration: Disable Prefer VAAPI Render Method
- While Playing a Video:
Set Deinterlace: Auto
Deinterlacing-Method: BOB (_not_ the VAAPI BOB) or De-Interlace (More cpu load, high quality).
-> Save for all movies

Background:
When watching interlaced Content and MCDI / VAAPI-BOB or MCDI is selected this would implicitely use the VAAPI Render Method, which causes the issues on the chromebox. Don't choose the GPU deinterlacers if you want to workaround those hangs.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
#62
Hi ,

I am also facing similar issues. i upgraded from a zotac atom box to a intel g3258 processor with 4 GB of RAM and running of msata drive in a shuttle ds81 mini pC.

I am facing skipped frames and it hangs. i have seen tearing in 1080p videos and the performance is probably 1/10th of what i had with my zotac box.

please help else i will have to auction off my new box
.
no longer applicable after kernel upgrade

my logs are in haripi2 XBMC logs


i re -read the other threads and suspect it could be because of my olderkernel. updated to 3.18.4 and the issues seem to be minimized. . navigated the seraches through google and did not do my research properly within forum .
Reply
#63
This is still an issue in OE 5.0.8.

Shouldn't the Chromebox wiki be updated?
Reply
#64
No it's not. And without logfiles (dmesg | pastebinit, kodi.log) it won't also make it into the wiki for sure.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
#65
Please see log and dmesg.

http://sprunge.us/TgFe

http://xbmclogs.com/pxycywilo
Reply
#66
No GPU hang at all,

but other funny thing

Quote:10:58:05 T:140110240089856 WARNING: VAAPI::FFGetBuffer - no surface available - dec: 0, render: 0
10:58:07 T:140110240089856 WARNING: Previous line repeats 3 times.
10:58:07 T:140110240089856 ERROR: VAAPI:Big Grinecode - timed out waiting for output message - decoded: 0, proc: 0, has free surface: no
10:58:07 T:140109349385984 NOTICE: COutput::OnExit: Output Thread terminated
10:58:07 T:140109349385984 NOTICE: Thread Vaapi-Output start, auto delete: false
10:58:07 T:140109349385984 NOTICE: COutput::OnStartup: Output Thread created
10:58:07 T:140109752039168 NOTICE: Thread Vaapi-Output start, auto delete: false
10:58:07 T:140109752039168 NOTICE: COutput::OnStartup: Output Thread created
10:58:07 T:140110240089856 WARNING: VAAPI::FFGetBuffer - no surface available - dec: 0, render: 0
10:58:09 T:140110240089856 WARNING: Previous line repeats 3 times.
10:58:09 T:140110240089856 ERROR: VAAPI:Big Grinecode - timed out waiting for output message - decoded: 0, proc: 0, has free surface: no

This i did not see yet. Please make a backup and test with the just released OpenELEC 6 Beta1.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
#67
http://xbmclogs.com/pu6777cdu

http://sprunge.us/SNJM
Reply
#68
Reproduce from local disk / usb disk please. I see a lot of network outage errors.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
#69
It's much hard to get it to hang. I have to jump tracks and it only hangs for a minute or so now.

http://xbmclogs.com/pjwgk7yeu

http://sprunge.us/SJEI
Reply
#70
Sorry - not a Debug Log :-(

But - here you found something:

Quote:[ 37.345406] [drm] GPU HANG: ecode 7:0:0x00200000, in kodi.bin [758], reason: Ring hung, action: reset
[ 37.345411] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[ 37.345414] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
[ 37.345416] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[ 37.345419] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
[ 37.345422] [drm] GPU crash dump saved to /sys/class/drm/card0/error
[ 37.345434] ------------[ cut here ]------------
[ 37.345447] WARNING: CPU: 0 PID: 154 at drivers/gpu/drm/i915/intel_display.c:9603 intel_mmio_flip_work_func+0x51/0x271()
[ 37.345451] WARN_ON(__i915_wait_request(mmio_flip->req, crtc->reset_counter, false, NULL, NULL) != 0)
[ 37.345453] Modules linked in: fuse ath9k ath9k_common ath9k_hw ath mac80211 wl(PO) ir_lirc_codec ir_xmp_decoder lirc_dev ir_mce_kbd_decoder ir_sharp_decoder ir_sanyo_decoder ir_sony_decoder ir_jvc_decoder ir_rc6_decoder ir_rc5_decoder ir_nec_decoder btusb rc_rc6_mce ath3k bluetooth mceusb rc_core intel_powerclamp intel_rapl iosf_mbi snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_codec_generic snd_hda_intel snd_hda_controller cfg80211 snd_hda_codec snd_hwdep snd_pcm
[ 37.345501] CPU: 0 PID: 154 Comm: kworker/0:6 Tainted: P W O 4.0.1 #1
[ 37.345504] Hardware name: Google Panther/Panther, BIOS 4.0-7883-g3652b0d 01/07/2015
[ 37.345510] Workqueue: events intel_mmio_flip_work_func
[ 37.345513] 0000000000000000 0000000000000009 ffffffff816b932d ffff88007569bd78
[ 37.345519] ffffffff81033286 ffff8800756887c0 ffffffff813a9df9 ffff88007569bd48
[ 37.345524] ffff8800755d98b0 ffff880100211b80 ffff8800755d98b0 ffff8800755d9000
[ 37.345529] Call Trace:
[ 37.345541] [<ffffffff816b932d>] ? dump_stack+0x40/0x50
[ 37.345553] [<ffffffff81033286>] ? warn_slowpath_common+0x93/0xab
[ 37.345558] [<ffffffff813a9df9>] ? intel_mmio_flip_work_func+0x51/0x271
[ 37.345565] [<ffffffff810332e3>] ? warn_slowpath_fmt+0x45/0x4a
[ 37.345570] [<ffffffff813a9df9>] ? intel_mmio_flip_work_func+0x51/0x271
[ 37.345577] [<ffffffff81043121>] ? process_one_work+0x142/0x214
[ 37.345581] [<ffffffff8104365f>] ? worker_thread+0x1c3/0x26d
[ 37.345586] [<ffffffff8104349c>] ? rescuer_thread+0x284/0x284
[ 37.345592] [<ffffffff81046cfc>] ? kthread+0xab/0xb3
[ 37.345598] [<ffffffff81046c51>] ? __kthread_parkme+0x5d/0x5d
[ 37.345605] [<ffffffff816c0c88>] ? ret_from_fork+0x58/0x90
[ 37.345611] [<ffffffff81046c51>] ? __kthread_parkme+0x5d/0x5d
[ 37.345615] ---[ end trace f90d2dbdd1d77e8d ]---
[ 37.352528] drm/i915: Resetting chip after gpu hang

Can you please file a bug upstream with the intel people? Like the dmesg tells to do?
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
#71
And if we are at it:

Can you try:

Video -> Acceleration: Prefer VAAPI Render Method set to disabled?

Same issue or does it get better?
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply
#72
Ok.
Reply
#73
(2015-05-08, 07:43)fritsch Wrote: Sorry - not a Debug Log :-(

But - here you found something:

Quote:[ 37.345406] [drm] GPU HANG: ecode 7:0:0x00200000, in kodi.bin [758], reason: Ring hung, action: reset
[ 37.345411] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[ 37.345414] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
[ 37.345416] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[ 37.345419] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
[ 37.345422] [drm] GPU crash dump saved to /sys/class/drm/card0/error
[ 37.345434] ------------[ cut here ]------------
[ 37.345447] WARNING: CPU: 0 PID: 154 at drivers/gpu/drm/i915/intel_display.c:9603 intel_mmio_flip_work_func+0x51/0x271()
[ 37.345451] WARN_ON(__i915_wait_request(mmio_flip->req, crtc->reset_counter, false, NULL, NULL) != 0)
[ 37.345453] Modules linked in: fuse ath9k ath9k_common ath9k_hw ath mac80211 wl(PO) ir_lirc_codec ir_xmp_decoder lirc_dev ir_mce_kbd_decoder ir_sharp_decoder ir_sanyo_decoder ir_sony_decoder ir_jvc_decoder ir_rc6_decoder ir_rc5_decoder ir_nec_decoder btusb rc_rc6_mce ath3k bluetooth mceusb rc_core intel_powerclamp intel_rapl iosf_mbi snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_codec_generic snd_hda_intel snd_hda_controller cfg80211 snd_hda_codec snd_hwdep snd_pcm
[ 37.345501] CPU: 0 PID: 154 Comm: kworker/0:6 Tainted: P W O 4.0.1 #1
[ 37.345504] Hardware name: Google Panther/Panther, BIOS 4.0-7883-g3652b0d 01/07/2015
[ 37.345510] Workqueue: events intel_mmio_flip_work_func
[ 37.345513] 0000000000000000 0000000000000009 ffffffff816b932d ffff88007569bd78
[ 37.345519] ffffffff81033286 ffff8800756887c0 ffffffff813a9df9 ffff88007569bd48
[ 37.345524] ffff8800755d98b0 ffff880100211b80 ffff8800755d98b0 ffff8800755d9000
[ 37.345529] Call Trace:
[ 37.345541] [<ffffffff816b932d>] ? dump_stack+0x40/0x50
[ 37.345553] [<ffffffff81033286>] ? warn_slowpath_common+0x93/0xab
[ 37.345558] [<ffffffff813a9df9>] ? intel_mmio_flip_work_func+0x51/0x271
[ 37.345565] [<ffffffff810332e3>] ? warn_slowpath_fmt+0x45/0x4a
[ 37.345570] [<ffffffff813a9df9>] ? intel_mmio_flip_work_func+0x51/0x271
[ 37.345577] [<ffffffff81043121>] ? process_one_work+0x142/0x214
[ 37.345581] [<ffffffff8104365f>] ? worker_thread+0x1c3/0x26d
[ 37.345586] [<ffffffff8104349c>] ? rescuer_thread+0x284/0x284
[ 37.345592] [<ffffffff81046cfc>] ? kthread+0xab/0xb3
[ 37.345598] [<ffffffff81046c51>] ? __kthread_parkme+0x5d/0x5d
[ 37.345605] [<ffffffff816c0c88>] ? ret_from_fork+0x58/0x90
[ 37.345611] [<ffffffff81046c51>] ? __kthread_parkme+0x5d/0x5d
[ 37.345615] ---[ end trace f90d2dbdd1d77e8d ]---
[ 37.352528] drm/i915: Resetting chip after gpu hang

Can you please file a bug upstream with the intel people? Like the dmesg tells to do?

https://bugs.freedesktop.org/show_bug.cgi?id=90376
Reply
#74
(2015-05-08, 09:08)fritsch Wrote: And if we are at it:

Can you try:

Video -> Acceleration: Prefer VAAPI Render Method set to disabled?

Same issue or does it get better?

Worse.

http://sprunge.us/AFYD

http://xbmclogs.com/pf7xixvkg
Reply
#75
Oki - again a kernel hang :-(

Let's hope upstream cares.
First decide what functions / features you expect from a system. Then decide for the hardware. Don't waste your money on crap.
Reply

Logout Mark Read Team Forum Stats Members Help
Intel VAAPI video freeze/hang bug0