[SOLVED] i915 regression in -current

NicePics13 · 08-25-2015, 11:48 AM

Running August 25th -current, experiencing screen corruption after browsing the web for a while that require a restart of the X server.

Code:

[ 1741.004033] [drm] stuck on render ring
[ 1741.005028] [drm] GPU HANG: ecode 3:0:0x70efffc1, in Xorg [890], reason: Ring hung, action: reset
[ 1741.005031] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[ 1741.005033] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
[ 1741.005035] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[ 1741.005036] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
[ 1741.005038] [drm] GPU crash dump saved to /sys/class/drm/card0/error
[ 1741.315597] drm/i915: Resetting chip after gpu hang

It seems to be this bug: https://bugs.freedesktop.org/show_bug.cgi?id=89334

ponce · 08-25-2015, 12:21 PM

if it's that, in the bug report it says that you shouldn't have problems if you force UXA (last post).
if so, try this as root (create a config file specifying UXA as AccelMethod)

Code:

cat >> /etc/X11/xorg.conf.d/20-intel.conf << "EOF"
Section "Device"
        Identifier "Intel Graphics"
        Driver "intel"
        Option "AccelMethod" "uxa"
EndSection
EOF

then restart X.

NicePics13 · 08-25-2015, 12:41 PM

Quote:

Originally Posted by ponce

if it's that, in the bug report it says that you shouldn't have problems if you force UXA

Remember having to force EXA

UXA seems to be working, borks video overlays but who uses those anymore.
Thanks!

NicePics13 · 08-25-2015, 03:41 PM

The new intel driver still crashes frequently in the background with or without UXA:

Code:

[ 5866.054642] WARNING: CPU: 1 PID: 2576 at drivers/gpu/drm/i915/i915_gem.c:4504 i915_gem_free_object+0x28b/0x2c0 [i915]()
[ 5866.054645] WARN_ON(obj->frontbuffer_bits)
[ 5866.054648] Modules linked in:
[ 5866.054651]  ipv6 cpufreq_ondemand speedstep_lib lp fuse hid_generic usbhid hid btusb btbcm btintel bluetooth joydev pcmcia i915 coretemp drm_kms_helper drm e1000e yenta_socket uhci_hcd i2c_dev i2c_i801 ehci_pci acpi_cpufreq i2c_algo_bit lpc_ich pcmcia_rsrc ptp pps_core pcmcia_core ehci_hcd iwl3945 iwlegacy i2c_core processor video thermal thinkpad_acpi thermal_sys psmouse intel_agp intel_gtt battery snd_hda_codec_analog mac80211 snd_hda_codec_generic agpgart snd_hda_intel snd_hda_controller ac button nsc_ircc snd_hda_codec irda tpm_tis serio_raw evdev tpm nvram ppdev snd_hda_core parport_pc hwmon parport cfg80211 rfkill kvm_intel snd_hwdep snd_pcm snd_timer snd soundcore 8250_fintek crc_ccitt kvm microcode loop
[ 5866.054735] CPU: 1 PID: 2576 Comm: Xorg Not tainted 4.1.6-smp #2
[ 5866.054738] Hardware name: LENOVO 1951A47/1951A47, BIOS 79ETE7WW (2.27 ) 03/21/2011
[ 5866.054742]  00000000 00000000 ee8fdd94 c1bc2d2b ee8fddd8 ee8fddc8 c104d5eb f9c5eefc
[ 5866.054751]  ee8fddf4 00000a10 f9c5e74c 00001198 f9bd4aeb 00001198 f9bd4aeb c2b94a00
[ 5866.054759]  00000000 c2b94a1c ee8fdde0 c104d653 00000009 ee8fddd8 f9c5eefc ee8fddf4
[ 5866.054767] Call Trace:
[ 5866.054777]  [<c1bc2d2b>] dump_stack+0x41/0x52
[ 5866.054783]  [<c104d5eb>] warn_slowpath_common+0x8b/0xc0
[ 5866.054806]  [<f9bd4aeb>] ? i915_gem_free_object+0x28b/0x2c0 [i915]
[ 5866.054828]  [<f9bd4aeb>] ? i915_gem_free_object+0x28b/0x2c0 [i915]
[ 5866.054832]  [<c104d653>] warn_slowpath_fmt+0x33/0x40
[ 5866.054854]  [<f9bd4aeb>] i915_gem_free_object+0x28b/0x2c0 [i915]
[ 5866.054867]  [<f903e1d7>] ? drm_vma_node_revoke+0x57/0x60 [drm]
[ 5866.054877]  [<f9024d3f>] drm_gem_object_free+0x1f/0x30 [drm]
[ 5866.054887]  [<f9025207>] drm_gem_object_handle_unreference_unlocked+0xf7/0x110 [drm]
[ 5866.054896]  [<f90252a4>] drm_gem_handle_delete+0x84/0xc0 [drm]
[ 5866.054906]  [<f9025850>] ? drm_gem_handle_create+0x40/0x40 [drm]
[ 5866.054916]  [<f902586d>] drm_gem_close_ioctl+0x1d/0x30 [drm]
[ 5866.054925]  [<f9026108>] drm_ioctl+0x198/0x570 [drm]
[ 5866.054931]  [<c113428f>] ? unmap_single_vma+0x42f/0x5b0
[ 5866.054940]  [<f9025850>] ? drm_gem_handle_create+0x40/0x40 [drm]
[ 5866.054946]  [<c113912f>] ? __remove_shared_vm_struct+0x2f/0x40
[ 5866.054950]  [<c113988d>] ? unlink_file_vma+0x3d/0x50
[ 5866.054955]  [<c114ff35>] ? kfree+0x135/0x140
[ 5866.054966]  [<f902aecc>] ? drm_vm_close_locked+0x6c/0xb0 [drm]
[ 5866.054976]  [<f902aecc>] ? drm_vm_close_locked+0x6c/0xb0 [drm]
[ 5866.054986]  [<f9025f70>] ? drm_getmap+0xc0/0xc0 [drm]
[ 5866.054992]  [<c1175de2>] do_vfs_ioctl+0x322/0x540
[ 5866.054996]  [<c113932a>] ? remove_vma+0x4a/0x50
[ 5866.055000]  [<c113b22e>] ? do_munmap+0x1de/0x350
[ 5866.055027]  [<c1176060>] SyS_ioctl+0x60/0x90
[ 5866.055033]  [<c1bca825>] syscall_call+0x7/0x7

USUARIONUEVO · 08-25-2015, 03:44 PM

Yea , i report this in the xorg crash thread... but no one ear me.

Xorg reset , and go to system login , in web broser using more of one flash page ... fullscreen to normal , some times, and crash.

I rebuild my xorg package , with the archlinux "collection" , of patchs (bugs confirmed in freedesktop, and put out the patches), because arch , have xorg 1.17.2 amount of moths, and see bugs first , slackware arrives now , and no see the bugs (at the momment).

But timt to time

With the archlinux patches collection , my xorg no need .conf forced , no crash , and stable as a rock... intel and 290x crossfire machines, no more issues.

Read please , open the patches , and read descriptions.
https://projects.archlinux.org/svnto...es/xorg-server

if ponce read this thread , and have time to this, please , put a patched xorg-server pacakge for the people test , and see then... no more issues.

personally i aplied all the *.patch files of arch on my build... for the order of patches read PKGBUILD , to see.

NicePics13 · 08-25-2015, 04:37 PM

Just willy nilly?

I kinda like to know more what a patch does before applying it.

USUARIONUEVO · 08-25-2015, 05:10 PM

Quote:

Originally Posted by NicePics13

Just willy nilly?

I kinda like to know more what a patch does before applying it.

I say , ... if some one want know , open the patch files, inside have an explication , and the number bug asigend in freedesktop.

One is for fix some glamor problems , other prevent the xorg reset to sustem login ... this last i can reproduce in slackware when i open multiple flash pages , xorg go crash.

after patches, no config files need , because code are fixed , and no more freezes web brosing with flash and hardware acceleration enabled on browser.

hitest · 08-26-2015, 05:33 PM

Quote:

Originally Posted by ponce

if it's that, in the bug report it says that you shouldn't have problems if you force UXA (last post).
if so, try this as root (create a config file specifying UXA as AccelMethod)

Code:

cat >> /etc/X11/xorg.conf.d/20-intel.conf << "EOF"
Section "Device"
        Identifier "Intel Graphics"
        Driver "intel"
        Option "AccelMethod" "uxa"
EndSection
EOF

then restart X.

Thanks, ponce. My PC started having x crash shortly after installing the new intel graphics driver. I implemented your fix yesterday and so far so good today. No x crashes.

NicePics13 · 08-27-2015, 10:32 AM

Quote:

Originally Posted by hitest

No x crashes.

Have you seen in dmesg if the driver itself keeps crashing/restarting without bringing down X in the process?

hitest · 08-27-2015, 10:54 AM

Quote:

Originally Posted by NicePics13

Have you seen in dmesg if the driver itself keeps crashing/restarting without bringing down X in the process?

I spoke too soon. I may need to apply the extra patches that ponce posted. I was multi-tasking and x crashed. ***Sigh***

NicePics13 · 09-13-2015, 12:44 PM

No crashes anymore when web browsing and watching videos (intel textured video) in -current since September, even with SNA acceleration enabled

Video overlays on the other hand are very choppy and will crash the video driver:

Code:

[25899.249584] WARNING: CPU: 1 PID: 1002 at drivers/gpu/drm/i915/i915_gem.c:5087 i915_gem_track_fb+0xf2/0x160 [i915]()
[25899.249589] WARN_ON(new->frontbuffer_bits & frontbuffer_bits)
[25899.249591] Modules linked in:
[25899.249595]  hid_generic usbhid hid pci_stub vboxpci(O) vboxnetadp(O) vboxnetflt(O) vboxdrv(O) rfcomm bnep ipv6 xt_tcpudp nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack iptable_filter ip_tables x_tables cpufreq_ondemand lp fuse btusb btbcm btintel bluetooth joydev snd_hda_codec_analog snd_hda_codec_generic i2c_dev pcmcia thinkpad_acpi snd_hda_intel snd_hda_controller i2c_i801 ppdev snd_hda_codec coretemp parport_pc i915 e1000e iwl3945 yenta_socket acpi_cpufreq nsc_ircc snd_hda_core pcmcia_rsrc iwlegacy parport psmouse mac80211 drm_kms_helper pcmcia_core serio_raw tpm_tis nvram tpm video processor drm i2c_algo_bit i2c_core thermal evdev cfg80211 lpc_ich snd_hwdep rfkill thermal_sys 8250_fintek hwmon battery ac snd_pcm intel_agp snd_timer intel_gtt agpgart uhci_hcd ehci_pci snd soundcore
[25899.249682]  kvm_intel ehci_hcd button kvm ptp pps_core irda microcode crc_ccitt loop
[25899.249699] CPU: 1 PID: 1002 Comm: Xorg Tainted: G           O    4.1.6 #2
[25899.249703] Hardware name: LENOVO 1951A47/1951A47, BIOS 79ETE7WW (2.27 ) 03/21/2011
[25899.249706]  ffffffffc063a2d0 ffff880071d53c08 ffffffff81cc0611 0000000000000007
[25899.249712]  ffff880071d53c58 ffff880071d53c48 ffffffff81087c4a ffff880071d53c68
[25899.249716]  0000000000000080 ffff8800643f7c00 ffff8800643f68c0 ffff8800643f7c00
[25899.249721] Call Trace:
[25899.249734]  [<ffffffff81cc0611>] dump_stack+0x45/0x57
[25899.249741]  [<ffffffff81087c4a>] warn_slowpath_common+0x8a/0xc0
[25899.249745]  [<ffffffff81087cc6>] warn_slowpath_fmt+0x46/0x50
[25899.249764]  [<ffffffffc05a1fc2>] i915_gem_track_fb+0xf2/0x160 [i915]
[25899.249789]  [<ffffffffc05f08f2>] intel_overlay_put_image+0xea2/0x12d0 [i915]
[25899.249807]  [<ffffffffc03179c5>] drm_ioctl+0x1a5/0x6a0 [drm]
[25899.249814]  [<ffffffff810b7779>] ? update_curr+0xd9/0x160
[25899.249821]  [<ffffffff810e43ce>] ? enqueue_hrtimer+0x2e/0x90
[25899.249825]  [<ffffffff810bd0e8>] ? pick_next_task_fair+0x778/0x980
[25899.249831]  [<ffffffff811c7460>] do_vfs_ioctl+0x2e0/0x4e0
[25899.249836]  [<ffffffff81cc4500>] ? __schedule+0x2e0/0x980
[25899.249840]  [<ffffffff811c76e1>] SyS_ioctl+0x81/0xa0
[25899.249845]  [<ffffffff81cc869b>] system_call_fastpath+0x16/0x6e

hitest · 09-13-2015, 02:27 PM

I've had good results so far with full screen video in Youtube and in Netflix; no crashes to report.

The recent updates are working well for me.

NicePics13 · 09-13-2015, 05:36 PM

Spoke too soon. Stuck on render ring when opening a browser tab. Back to UXA.