Bug #2848
closedi915: *ERROR* pipe is still busy with an old pageflip
0%
Description
Process Xorg stuck in state "lwe" with kernel saying:
WARNING ({ int start_jiffies, elapsed_jiffies, remaining_jiffies, ret; _Bool timeout_expired = 0; _Bool interrupted = 0; long retval; start_jiffies = ticks; lockmgr(&dev_priv->pending_flip_queue.lock, 0x00000002); while (1) { if (!intel_crtc_has_pending_flip(crtc)) break; ret = lksleep(&dev_priv->pending_flip_queue, &dev_priv->pending_flip_queue.lock, 0, "lwe", 60*hz); if (ret 4 || ret (1)) { interrupted = 1; break; } if (ret 35) { timeout_expired = 1; break; } } lockmgr(&dev_priv>pending_flip_queue.lock, 0x00000006); elapsed_jiffies = ticks - start_jiffies; remaining_jiffies = 60*hz - elapsed_jiffies; if (remaining_jiffies <= 0) remaining_jiffies = 1; if (timeout_expired) retval = 0; else if (interrupted) retval = -512; else if (60*hz > 0) retval = remaining_jiffies; else retval = 1; retval; }) 0 failed at /usr/src/sys/dev/drm/i915/intel_display.c:3348
error: [drm:pid1018:intel_pipe_set_base] ERROR pipe is still busy with an old pageflip
These two messages then repeat in 1 minute interval. Switching to console works (but is delayed up to that 1 minute), so the system can be rebooted cleanly.
The machine runs on Intel Celeron J1900 (ie. "<ValleyView/Baytrail> on vgapci0") with software from current dports (xorg-server-1.17.2,1 + xf86-video-intel-2.99.2015.09.09,1).
PS: I suspect this to be some race condition as I've been running "DragonFly desktop" on this machine for several days with no problem so far.
Updated by deef almost 10 years ago
- Status changed from New to Closed
I suspect Francois Tigeot may has fixed this with commit 1b13d1905 ("drm/i915: Update to Linux 3.18"). According to commit message, it introduced "more paranoid pageflips, with added error detection and recovery logic". And really, from this update the box is running rock-stable for more than a month.
Closing the ticket. Thank you Francois, and sorry for the noise.