[Intel-gfx] vblank count error on i915 using intel master

Adam Lantos hege at playma.org
Sat Feb 28 20:10:21 CET 2009


Hi,

I'm running vanilla linux-2.6.29-rc6, libdrm master, mesa-7.3,
xorg-server-1.6.0, xf86-video-intel master on my i915 (Intel
Corporation Mobile 915GM/GMS/910GML Express Graphics Controller (rev
03)).
I use UXA without KMS and kde4.2 kwin compositing manager, I have
nothing special in my xorg.conf.

2D and 3D rendering appear to be fine, but running x11perf or use VT
switching causes blank screen with mouse pointer and dmesg says that:
[drm:i915_get_vblank_counter] *ERROR* trying to get vblank count for
disabled pipe 0

Sometimes I also have the following kernel WARN:

WARNING: at drivers/gpu/drm/i915/i915_gem.c:3048
i915_gem_idle+0x304/0x330 [i915]()
Hardware name: 1867W6T
Modules linked in: i915 fb drm i2c_algo_bit cfbcopyarea cfbimgblt
cfbfillrect vmnet vmblock vmci vmmon ppp_async ppp_generic slhc option
usbserial usb_storage snd_pcm_oss snd_mixer_oss snd_seq_oss
snd_seq_midi_event snd_seq snd_seq_device hdaps input_polldev
snd_intel8x0 snd_ac97_codec pcmcia ac97_bus snd_pcm snd_timer ehci_hcd
ipw2200 uhci_hcd yenta_socket rsrc_nonstatic tg3 snd libipw
pcmcia_core 8250_pci libphy usbcore snd_page_alloc 8250_pnp lib80211
8250 rtc_cmos serial_core intel_agp agpgart thinkpad_acpi rfkill
backlight led_class [last unloaded: cfbfillrect]
Pid: 13160, comm: X Tainted: G        W  2.6.29-rc6-nibbler #2
Call Trace:
 [<c0125d89>] warn_slowpath+0x99/0xc0
 [<c012f203>] __mod_timer+0xa3/0xe0
 [<f85b2965>] i915_gem_retire_requests+0x115/0x140 [i915]
 [<f85b2965>] i915_gem_retire_requests+0x115/0x140 [i915]
 [<f85b3e64>] i915_gem_idle+0x304/0x330 [i915]
 [<f85b4dc4>] i915_gem_leavevt_ioctl+0x24/0x40 [i915]
 [<f83bc5dc>] drm_ioctl+0xec/0x2c0 [drm]
 [<c012124c>] update_curr+0xbc/0x170
 [<f85b4da0>] i915_gem_leavevt_ioctl+0x0/0x40 [i915]
 [<c01971c8>] vfs_ioctl+0x78/0x90
 [<c01974eb>] do_vfs_ioctl+0x1eb/0x560
 [<c012ee71>] run_timer_softirq+0x191/0x1e0
 [<c0144c1d>] tick_dev_program_event+0x3d/0xc0
 [<c012a966>] __do_softirq+0x96/0x110
 [<c013d04d>] hrtimer_interrupt+0xdd/0x1e0
 [<c019789d>] sys_ioctl+0x3d/0x70
 [<c01031ad>] sysenter_do_call+0x12/0x21
---[ end trace be634558b181665b ]---


Should I file a bug, or are there some patches I should try? Which
part of rendering stack can cause this problem? Is it the 2D driver or
DRM/GEM?


thanks,
 Adam



More information about the Intel-gfx mailing list