CS stops working (no more bb executed, no more MI commands, both LRI and STORE_DWORD) after a libva chained batch buffer. Not the first time we have seen such behaviour (but always in conjunction with libva).

I have these hangs _very_ often, mainly with html5 video.
Now i thought, many others don't seem to have the problem, so I removed my long-year kernel command line options (since 3.2 or something):

i915.enable_rc6=1 i915.enable_fbc=1 i915.powersave=1 i915.modeset=1 i915.lvds_downclock=1

And now the crashes and hangs are gone.

I'd be glad to further help debugging this.

(Probably related to another bug by me then: #92330)

Veeery interesting: I forgot I changed something else except those 4 parameters: intel_iommu=on, it was off before (because of playing with xen).

Now, with the iommu on again, it does not crash even after I enable my beloved 4 parameters again (yea i know, rc6 and the modeset is default now :)

To confirm, I just double-tested it with mmu=off again and keeping the 4 params, and yes it hung very quickly when watching youtube.

=> IOMMU on == good!

Both on Gentoo, Linux 4.3.0-JJ #142 SMP Mon Nov 2 02:43:59 CET 2015 x86_64 Intel(R) Core(TM) i5-2520M CPU @ 2.50GHz GenuineIntel GNU/Linux:

dmesg | grep drm  # with iommu on
[    1.591911] [drm] Initialized drm 1.1.0 20060810
[    9.870699] [drm] Memory usable by graphics device = 2048M
[    9.870780] [drm] VT-d active for gfx access
[    9.870857] [drm] Disabling PPGTT because VT-d is on
[    9.870935] [drm] Replacing VGA console driver
[    9.886575] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
[    9.886582] [drm] Driver supports precise vblank timestamp query.
[    9.886602] [drm] DMAR active, disabling use of stolen memory
[    9.987376] [drm] GMBUS [i915 gmbus dpb] timed out, falling back to bit banging on pin 5
[   10.003780] [drm] Initialized i915 1.6.0 20150731 for 0000:00:02.0 on minor 0
[   10.003797] fbcon: inteldrmfb (fb0) is primary device
[   10.853001] i915 0000:00:02.0: fb0: inteldrmfb frame buffer device
[   11.464863] drm: not enough stolen space for compressed buffer (need 4325376 more bytes), disabling. Hint: you may be able to increase stolen memory size in the BIOS to avoid this.



dmesg | grep drm  # with iommu off
[    1.586329] [drm] Initialized drm 1.1.0 20060810
[    9.244689] [drm] Memory usable by graphics device = 2048M
[    9.244774] [drm] Replacing VGA console driver
[    9.250961] [drm] Supports vblank timestamp caching Rev 2 (21.10.2013).
[    9.250968] [drm] Driver supports precise vblank timestamp query.
[    9.354597] [drm] GMBUS [i915 gmbus dpb] timed out, falling back to bit banging on pin 5
[    9.371987] fbcon: inteldrmfb (fb0) is primary device
[    9.372597] [drm] Initialized i915 1.6.0 20150731 for 0000:00:02.0 on minor 0
[   10.227585] i915 0000:00:02.0: fb0: inteldrmfb frame buffer device
(...) # another hang I just had:
[  253.551252] ------------[ cut here ]------------
[  253.551341] WARNING: CPU: 2 PID: 973 at drivers/gpu/drm/i915/intel_display.c:3926 intel_crtc_wait_for_pending_flips+0xf8/0x1f1 [i915]()
[  253.551346] WARN_ON(wait_event_timeout(dev_priv->pending_flip_queue, !intel_crtc_has_pending_flip(crtc), 60*HZ) == 0)
[  253.551350] Modules linked in:
[  253.551354]  nf_conntrack_ipv4 nf_defrag_ipv4 iptable_filter ip_tables bnep nf_conntrack_ipv6 nf_defrag_ipv6 xt_conntrack nf_conntrack ip6table_filter ip6_tables x_tables sch_fq_codel snd_hda_codec_hdmi iTCO_wdt uvcvideo videobuf2_vmalloc videobuf2_memops videobuf2_core snd_hda_codec_generic v4l2_common videodev joydev media btusb btrtl btbcm btintel cmac iwldvm mac80211 bluetooth coretemp kvm_intel snd_hda_intel kvm iwlwifi snd_hda_codec snd_hwdep cfg80211 sdhci_pci thinkpad_acpi snd_hda_core nvram microcode psmouse sdhci snd_pcm i915 serio_raw mmc_core pcspkr rfkill snd_timer i2c_i801 lpc_ich snd evdev soundcore
[  253.551444] CPU: 2 PID: 973 Comm: X Not tainted 4.3.0-JJ #142
[  253.551447] Hardware name: LENOVO 4296CTO/4296CTO, BIOS 8DET70WW (1.40 ) 05/14/2015
[  253.551452]  0000000000000000 0000000033858303 ffff880401ddfa80 ffffffff8444d8e4
[  253.551459]  ffff880401ddfac8 ffff880401ddfab8 ffffffff8409bae6 ffffffffc04bbc52
[  253.551466]  ffff880409d5e000 0000000000000000 ffff8804092b4800 ffff880409d28ee0
[  253.551473] Call Trace:
[  253.551487]  [<ffffffff8444d8e4>] dump_stack+0x44/0x55
[  253.551498]  [<ffffffff8409bae6>] warn_slowpath_common+0x99/0xb2
[  253.551555]  [<ffffffffc04bbc52>] ? intel_crtc_wait_for_pending_flips+0xf8/0x1f1 [i915]
[  253.551563]  [<ffffffff8409bb56>] warn_slowpath_fmt+0x57/0x73
[  253.551612]  [<ffffffffc04bbc52>] intel_crtc_wait_for_pending_flips+0xf8/0x1f1 [i915]
[  253.551620]  [<ffffffff840cea0a>] ? wait_woken+0x72/0x72
[  253.551667]  [<ffffffffc04bcaf6>] intel_pre_plane_update+0xa7/0x101 [i915]
[  253.551713]  [<ffffffffc04bdcab>] intel_atomic_commit+0xddc/0xed4 [i915]
[  253.551761]  [<ffffffffc04c12dc>] ? intel_atomic_check+0x8a2/0xc15 [i915]
[  253.551808]  [<ffffffffc04a9f00>] ? intel_crtc_duplicate_state+0x3c/0x7a [i915]
[  253.551819]  [<ffffffff8454b256>] ? __drm_atomic_helper_crtc_duplicate_state+0x2f/0x41
[  253.551826]  [<ffffffff845698b5>] ? drm_atomic_check_only+0x168/0x45b
[  253.551832]  [<ffffffff84569bf5>] drm_atomic_commit+0x4d/0x52
[  253.551839]  [<ffffffff8454a8b5>] drm_atomic_helper_disable_plane+0xcb/0x116
[  253.551846]  [<ffffffff8455f066>] __setplane_internal+0x39/0x2aa
[  253.551852]  [<ffffffff8455f715>] drm_mode_setplane+0x145/0x166
[  253.551859]  [<ffffffff84553067>] drm_ioctl+0x23d/0x374
[  253.551865]  [<ffffffff8455f5d0>] ? drm_mode_cursor_common+0x14d/0x14d
[  253.551874]  [<ffffffff841c5307>] do_vfs_ioctl+0x3a2/0x42c
[  253.551880]  [<ffffffff841b766f>] ? __sb_end_write+0x1d/0x1f
[  253.551885]  [<ffffffff841b573b>] ? vfs_write+0x154/0x162
[  253.551892]  [<ffffffff841c53e8>] SyS_ioctl+0x57/0x79
[  253.551901]  [<ffffffff8478662e>] entry_SYSCALL_64_fastpath+0x12/0x71
[  253.551907] ---[ end trace 1fc1877966ffb925 ]---
[  253.551911] ------------[ cut here ]------------
[  253.551959] WARNING: CPU: 2 PID: 973 at drivers/gpu/drm/i915/intel_display.c:3931 intel_crtc_wait_for_pending_flips+0x132/0x1f1 [i915]()
[  253.551962] Removing stuck page flip
[  253.551965] Modules linked in: nf_conntrack_ipv4 nf_defrag_ipv4 iptable_filter ip_tables bnep nf_conntrack_ipv6 nf_defrag_ipv6 xt_conntrack nf_conntrack ip6table_filter ip6_tables x_tables sch_fq_codel snd_hda_codec_hdmi iTCO_wdt uvcvideo videobuf2_vmalloc videobuf2_memops videobuf2_core snd_hda_codec_generic v4l2_common videodev joydev media btusb btrtl btbcm btintel cmac iwldvm mac80211 bluetooth coretemp kvm_intel snd_hda_intel kvm iwlwifi snd_hda_codec snd_hwdep cfg80211 sdhci_pci thinkpad_acpi snd_hda_core nvram microcode psmouse sdhci snd_pcm i915 serio_raw mmc_core pcspkr rfkill snd_timer i2c_i801 lpc_ich snd evdev soundcore
[  253.552042] CPU: 2 PID: 973 Comm: X Tainted: G        W       4.3.0-JJ #142
[  253.552047] Hardware name: LENOVO 4296CTO/4296CTO, BIOS 8DET70WW (1.40 ) 05/14/2015
[  253.552050]  0000000000000000 0000000033858303 ffff880401ddfa80 ffffffff8444d8e4
[  253.552056]  ffff880401ddfac8 ffff880401ddfab8 ffffffff8409bae6 ffffffffc04bbc8c
[  253.552062]  ffff880409d5e000 ffff8804092b49a8 ffff8804092b4800 ffff880409d28ee0
[  253.552068] Call Trace:
[  253.552076]  [<ffffffff8444d8e4>] dump_stack+0x44/0x55
[  253.552083]  [<ffffffff8409bae6>] warn_slowpath_common+0x99/0xb2
[  253.552128]  [<ffffffffc04bbc8c>] ? intel_crtc_wait_for_pending_flips+0x132/0x1f1 [i915]
[  253.552136]  [<ffffffff8409bb56>] warn_slowpath_fmt+0x57/0x73
[  253.552179]  [<ffffffffc04bbc8c>] intel_crtc_wait_for_pending_flips+0x132/0x1f1 [i915]
[  253.552185]  [<ffffffff840cea0a>] ? wait_woken+0x72/0x72
[  253.552228]  [<ffffffffc04bcaf6>] intel_pre_plane_update+0xa7/0x101 [i915]
[  253.552271]  [<ffffffffc04bdcab>] intel_atomic_commit+0xddc/0xed4 [i915]
[  253.552316]  [<ffffffffc04c12dc>] ? intel_atomic_check+0x8a2/0xc15 [i915]
[  253.552361]  [<ffffffffc04a9f00>] ? intel_crtc_duplicate_state+0x3c/0x7a [i915]
[  253.552370]  [<ffffffff8454b256>] ? __drm_atomic_helper_crtc_duplicate_state+0x2f/0x41
[  253.552376]  [<ffffffff845698b5>] ? drm_atomic_check_only+0x168/0x45b
[  253.552382]  [<ffffffff84569bf5>] drm_atomic_commit+0x4d/0x52
[  253.552389]  [<ffffffff8454a8b5>] drm_atomic_helper_disable_plane+0xcb/0x116
[  253.552395]  [<ffffffff8455f066>] __setplane_internal+0x39/0x2aa
[  253.552401]  [<ffffffff8455f715>] drm_mode_setplane+0x145/0x166
[  253.552407]  [<ffffffff84553067>] drm_ioctl+0x23d/0x374
[  253.552413]  [<ffffffff8455f5d0>] ? drm_mode_cursor_common+0x14d/0x14d
[  253.552420]  [<ffffffff841c5307>] do_vfs_ioctl+0x3a2/0x42c
[  253.552427]  [<ffffffff841b766f>] ? __sb_end_write+0x1d/0x1f
[  253.552432]  [<ffffffff841b573b>] ? vfs_write+0x154/0x162
[  253.552438]  [<ffffffff841c53e8>] SyS_ioctl+0x57/0x79
[  253.552445]  [<ffffffff8478662e>] entry_SYSCALL_64_fastpath+0x12/0x71
[  253.552450] ---[ end trace 1fc1877966ffb926 ]---

(In reply to Jonas Jelten from comment #2)
> I have these hangs _very_ often, mainly with html5 video.
> Now i thought, many others don't seem to have the problem, so I removed my
> long-year kernel command line options (since 3.2 or something):
> 
> i915.enable_rc6=1 i915.enable_fbc=1 i915.powersave=1 i915.modeset=1
> i915.lvds_downclock=1
> 
> And now the crashes and hangs are gone.
> 
> I'd be glad to further help debugging this.

Sorry for neglecting this bug. We don't really support changing said parameters from their per platform defaults. If it works without, we're happy. Glad you found another way to fix it too (comment #3).

Closing.

Closing >1 year old resolved+fixed.