Terrible experience with 7.x on Meteor Lake iGPU
Having a ThinkPad P1 Gen 7 with Meteor Lake, I’m seeing crashes with both i915 and xe. I’ve reported this upstream to the kernel/DRM folks.
Kernel 6.19.14 was rock solid here. Is anyone else seeing regressions this severe?
With xe, the system can crash within ~30 minutes, which is already bad enough. But now I’m also seeing i915 blow up? WTF?
Relevant part of the trace:
[ 3113.696521] Call Trace:
[ 3113.696526] <TASK>
[ 3113.696528] ? intel_fbc_activate+0x36/0x100 [i915]
[ 3113.696787] intel_atomic_commit_tail+0x1087/0x18d0 [i915]
[ 3113.697193] ? _raw_spin_unlock_irqrestore+0x27/0x50
[ 3113.697201] intel_atomic_commit+0x23d/0x280 [i915]
[ 3113.697495] drm_atomic_commit+0xb1/0xe0 [drm]
[ 3113.697548] ? __pfx___drm_printfn_info+0x10/0x10 [drm]
[ 3113.697582] drm_atomic_connector_commit_dpms+0x109/0x120 [drm]
[ 3113.697657] drm_mode_obj_set_property_ioctl+0x1ba/0x3e0 [drm]
[ 3113.697838] ? __pfx_drm_connector_property_set_ioctl+0x10/0x10 [drm]
[ 3113.697927] drm_connector_property_set_ioctl+0x3c/0x60 [drm]
[ 3113.698054] drm_ioctl_kernel+0xae/0x100 [drm]
[ 3113.698115] drm_ioctl+0x283/0x530 [drm]
[ 3113.698175] ? __pfx_drm_connector_property_set_ioctl+0x10/0x10 [drm]
[ 3113.698232] __x64_sys_ioctl+0x96/0xe0
[ 3113.698237] ? do_syscall_64+0x14c/0x1520
[ 3113.698242] do_syscall_64+0x10f/0x1520
[ 3113.698247] ? fpregs_assert_state_consistent+0x34/0x60
[ 3113.698252] ? do_syscall_64+0x14c/0x1520
[ 3113.698257] ? fpregs_assert_state_consistent+0x34/0x60
[ 3113.698261] ? do_syscall_64+0x14c/0x1520
[ 3113.698265] ? fpregs_assert_state_consistent+0x34/0x60
[ 3113.698274] ? irqentry_exit+0x7b/0x580
[ 3113.698275] entry_SYSCALL_64_after_hwframe+0x76/0x7e