[Spice-devel] qxl dri related kernel oops

[David Mansfield]

Hi All,

One of our users is experiencing a recurring kernel oops that seems to 
be qxl related.  It only occurs when the user is "on the road" 
(literally) using spice through a mobile tether and thus has highly 
variable bandwidth and latency at the time when the oopsen occur. System 
is F20, running on a Centos 7 host.  Here is one (I have some more but 
they are basically identical):

[117395.526064] CPU: 1 PID: 1679 Comm: Xorg Not tainted 
3.15.10-200.fc20.x86_64 #1
[117395.526065] Hardware name: Red Hat KVM, BIOS Bochs 01/01/2011
[117395.526066]  0000000000000000 00000000ed51ec7f ffff880513f378a8 
ffffffff816ef9c8
[117395.526067]  ffff880513f378f0 ffff880513f378e0 ffffffff8108931d 
0000000000000001
[117395.526069]  ffff8804af232740 0000000000000001 ffff8804af232758 
0000000000000000
[117395.526070] Call Trace:
[117395.526074]  [<ffffffff816ef9c8>] dump_stack+0x45/0x56
[117395.526076]  [<ffffffff8108931d>] warn_slowpath_common+0x7d/0xa0
[117395.526078]  [<ffffffff8108939c>] warn_slowpath_fmt+0x5c/0x80
[117395.526081]  [<ffffffff816f2c03>] ? schedule_hrtimeout_range+0x13/0x20
[117395.526084]  [<ffffffffa00d7012>] qxl_sync_obj_wait+0x182/0x220 [qxl]
[117395.526089]  [<ffffffffa008e78f>] ttm_bo_wait+0xaf/0x1b0 [ttm]
[117395.526093]  [<ffffffffa00900c9>] ttm_bo_evict+0x59/0x350 [ttm]
[117395.526097]  [<ffffffffa00904dd>] ttm_mem_evict_first+0x11d/0x190 [ttm]
[117395.526100]  [<ffffffffa00907e0>] ttm_bo_mem_space+0x290/0x340 [ttm]
[117395.526103]  [<ffffffffa0090ce2>] ttm_bo_validate+0x1d2/0x250 [ttm]
[117395.526107]  [<ffffffffa0090fc5>] ttm_bo_init+0x265/0x420 [ttm]
[117395.526111]  [<ffffffffa00d90bf>] qxl_bo_create+0x13f/0x1a0 [qxl]
[117395.526114]  [<ffffffffa00d8e40>] ? qxl_fbdev_qobj_is_fb+0x30/0x30 [qxl]
[117395.526117]  [<ffffffffa00d9819>] qxl_gem_object_create+0x69/0x110 [qxl]
[117395.526121]  [<ffffffffa00d991c>] 
qxl_gem_object_create_with_handle+0x5c/0x120 [qxl]
[117395.526124]  [<ffffffffa00dcb36>] qxl_alloc_ioctl+0x46/0xc0 [qxl]
[117395.526132]  [<ffffffffa0048a3c>] drm_ioctl+0x1ec/0x660 [drm]
[117395.526134]  [<ffffffff811a6dad>] ? mmap_region+0x19d/0x5d0
[117395.526136]  [<ffffffff811a74d5>] ? do_mmap_pgoff+0x2f5/0x3c0
[117395.526138]  [<ffffffff812e9f63>] ? file_map_prot_check+0x63/0xd0
[117395.526140]  [<ffffffff811f9ac0>] do_vfs_ioctl+0x2e0/0x4a0
[117395.526142]  [<ffffffff811f9d01>] SyS_ioctl+0x81/0xa0
[117395.526144]  [<ffffffff816ffae9>] system_call_fastpath+0x16/0x1b
[117395.526145] ---[ end trace afddd0edc2a1f412 ]---
[117395.526146] [TTM] Failed to expire sync object before buffer eviction
[117395.526154] qxl 0000:00:02.0: object_init failed for (262144, 
0x00000001)
[117395.526155] [drm:qxl_gem_object_create] *ERROR* Failed to allocate 
GEM object (261140, 1, 4096, -12)
[117395.526156] [drm:qxl_alloc_ioctl] *ERROR* qxl_alloc_ioctl: failed to 
create gem ret=-12
[117398.799075] vlc[24330]: segfault at 6536c4f5200 ip 000000329b23942d 
sp 00007f84fba7fa00 error 4 in libc-2.18.so[329b200000+1b4000]

The process that gets the segfault varies.

Has this been seen before?  After this, Xorg / GDM restarts ok and 
system works fine without a reboot.

-- 
Thanks,
David Mansfield
Cobite, INC.