Latest working kernel version: Earliest failing kernel version: Distribution: Debian Sid amd64 Hardware Environment: Thinkpad T61 8897 (GM965) Software Environment: drm: 2.4.3+git+20090105+a8c5480-1 xserver: 1.5.99.901 xf86-video-intel: 2.6.0 Problem Description: Before 2.6.29-rc2, I couldn't even run Xorg without having some garbled fonts and gtk widgets (like http://molly.corsac.net/~corsac/screen.png) Starting 2.6.29-rc2 (and the drm-next merge I guess) I can know run Xorg, but I still have various problems: - X cursor is garbled (can make a screenshot later if needed) - at start, I have some WARNINGs: [ 1.152254] ------------[ cut here ]------------ [ 1.152260] WARNING: at drivers/gpu/drm/drm_crtc.c:213 drm_mode_object_get+0x 46/0xac [drm]() [ 1.152265] Hardware name: 8897CTO [ 1.152268] drm_mode_object_get called w/o mode_config lock [ 1.152272] Modules linked in: i915(+) fb drm i2c_algo_bit cfbcopyarea cfbimg blt cfbfillrect acpi_cpufreq processor [ 1.152291] Pid: 493, comm: modprobe Not tainted 2.6.29-rc2 #45 [ 1.152295] Call Trace: [ 1.152308] [<ffffffff8023b15d>] warn_slowpath+0xd8/0x112 [ 1.152318] [<ffffffff80230cbc>] dequeue_entity+0xf/0x102 [ 1.152325] [<ffffffff8023198d>] dequeue_task_fair+0x1c/0x90 [ 1.152336] [<ffffffff804d67a3>] thread_return+0x3d/0xc6 [ 1.152344] [<ffffffff80243c3a>] lock_timer_base+0x26/0x4b [ 1.152350] [<ffffffff80243cb1>] try_to_del_timer_sync+0x52/0x5b [ 1.152359] [<ffffffff8020cb8e>] reschedule_interrupt+0xe/0x20 [ 1.152368] [<ffffffff80361fd9>] delay_tsc+0x0/0x3a [ 1.152403] [<ffffffffa0035009>] drm_mode_object_get+0x46/0xac [drm] [ 1.152437] [<ffffffffa0035157>] drm_mode_connector_update_edid_property+0xe8/0x163 [drm] [ 1.152469] [<ffffffffa0039922>] drm_get_edid+0x176/0x1d8 [drm] [ 1.152499] [<ffffffffa0075015>] intel_ddc_get_modes+0x2d/0x4b [i915] [ 1.152523] [<ffffffffa0073566>] intel_lvds_init+0x10c/0x2ce [i915] [ 1.152547] [<ffffffffa0072232>] intel_modeset_init+0x339/0x609 [i915] [ 1.152569] [<ffffffffa00680f4>] i915_driver_load+0x770/0x7ca [i915] [ 1.152604] [<ffffffffa0031628>] drm_get_dev+0x313/0x3d3 [drm] [ 1.152635] [<ffffffffa002cdbe>] drm_init+0x6e/0xa5 [drm] [ 1.152656] [<ffffffffa008c000>] i915_init+0x0/0x4c [i915] [ 1.152662] [<ffffffff80209051>] _stext+0x51/0x120 [ 1.152669] [<ffffffff8035fc12>] __up_read+0x13/0x8d [ 1.152679] [<ffffffff802506e7>] __blocking_notifier_call_chain+0x51/0x5f [ 1.152687] [<ffffffff8025e7cb>] sys_init_module+0xa0/0x1ba [ 1.152693] [<ffffffff8020bedb>] system_call_fastpath+0x16/0x1b [ 1.152698] ---[ end trace 8f4f96a97eeb629e ]--- [ 1.152734] ------------[ cut here ]------------ And trying to restart Xorg for some reason, it fails (the screen is blocked at the moment) and I have in Xorg.0.log: Jan 17 16:51:47 hidalgo kernel: [ 533.088113] [drm:i915_wait_irq] *ERROR* EBUSY -- rec: 1 emitted: 2 with in dmesg: [ 533.088113] [drm:i915_wait_irq] *ERROR* EBUSY -- rec: 1 emitted: 2 and no vt switch is possible atm. If you need more info, please ask. xorg.conf follows: Section "Device" Driver "intel" BusId "PCI:0:2:0" Identifier "intel" Option "FramebufferCompression" "on" Option "Legacy3D" "false" Option "AccelMethod" "UXA" Option "XvPreferOverlay" "on" Option "XvMC" "on" EndSection I'll try without any xorg.conf.
No xorg.conf doesn't change anything Option "SWCursor" "on" doesn't change anything either
Why did you mark the bug as a regression? What was the last working kernel?
Well, I didn't really marked the bug as a regression. Well, the global behavior is a regression (there's no cursor problem in 2.6.28 or 2.6.28-drm-intel-2.6.28 or even previous 2.6.29 ones, but I can't really recall which one, I guess a git bisect could help). And there's no WARNING either. But I know that KMS code is really new, so it's related I guess.
Created attachment 19913 [details] Correct cursor on i965 with KMS
Ok, the cursor problem is fixed using a patch from Jesse Barnes (I'm attaching it, I didn't find it on a public ML). But I still have the WARNING, which is a different issue I guess. Not sure if I just change the title, clone the bug or something else?
Hmhm, afaik this is fixed, I don't have that problem anymore on Debian sid.