FYI, we noticed the below changes on
commit ad46cb533d586fdb256855437af876617c6cf609 ("drm/i915: Prevent recursive deadlock on releasing a busy userptr")
v3.17-rc4 ad46cb533d586fdb256855437a testbox/testcase/testparams
---------------- -------------------------- ---------------------------
fail:runs %reproduction fail:runs
| | |
:5 100% 5:5 lkp-t410/piglit/performance-igt-002
:5 100% 5:5 TOTAL piglit.igt/gem_userptr_blits/coherency-sync.crash
v3.17-rc4 ad46cb533d586fdb256855437a
---------------- --------------------------
:5 100% 5:5 lkp-t410/piglit/performance-igt-002
:5 100% 5:5 TOTAL dmesg.Out_of_memory:Kill_process
<6>[ 16.764414] BTRFS: device fsid 8886adc6-3306-411a-bbf8-6638ebb74746 devid 1 transid 57 /dev/sda6
<6>[ 18.558946] intel ips 0000:00:1f.6: i915 driver attached, reenabling gpu turbo
<6>[ 69.352751] Purging GPU memory, 0 bytes freed, 12288 bytes still pinned.
<4>[ 69.363044] gem_userptr_bli invoked oom-killer: gfp_mask=0xa00d2, order=0, oom_score_adj=1000
<6>[ 69.373682] gem_userptr_bli cpuset=/ mems_allowed=0
<4>[ 69.380617] CPU: 1 PID: 3058 Comm: gem_userptr_bli Not tainted 3.17.0-dirty #1
<4>[ 69.390574] Hardware name: LENOVO 25222AU/25222AU, BIOS 6IET52WW (1.12 ) 02/16/2010
<4>[ 69.399759] 0000000000000000 ffff8800749d37d8 ffffffff81870560 ffff880074a00000
<4>[ 69.409950] ffff8800749d3840 ffffffff8186d528 ffff880000000015 0000000000000000
<4>[ 69.419535] 00000000ffffffff ffff880076ffe0c0 0000000000000206 ffffffff81cf5440
<4>[ 69.429544] Call Trace:
<4>[ 69.435415] [<ffffffff81870560>] dump_stack+0x4d/0x66
<4>[ 69.442549] [<ffffffff8186d528>] dump_header+0x9e/0x225
<4>[ 69.448393] [<ffffffff81163bed>] oom_kill_process+0x20d/0x380
<4>[ 69.451840] [<ffffffff81388a65>] ? security_capable_noaudit+0x15/0x20
<4>[ 69.456166] [<ffffffff81164419>] out_of_memory+0x4f9/0x540
<4>[ 69.459434] [<ffffffff8116ad58>] __alloc_pages_nodemask+0xab8/0xb00
<4>[ 69.463816] [<ffffffff811b130f>] alloc_pages_vma+0xaf/0x1c0
<4>[ 69.467235] [<ffffffff8117847f>] shmem_alloc_page+0x5f/0x90
<4>[ 69.470444] [<ffffffff81194cd9>] ? __vm_enough_memory+0x29/0x150
<4>[ 69.474756] [<ffffffff8117b490>] shmem_getpage_gfp+0x550/0x8e0
<4>[ 69.478142] [<ffffffff8117c371>] shmem_read_mapping_page_gfp+0x31/0x60
<4>[ 69.481728] [<ffffffffa01e830c>] i915_gem_object_get_pages_gtt+0x2dc/0x3b0 [i915]
<4>[ 69.486341] [<ffffffff811badec>] ? kmem_cache_alloc_trace+0x4c/0x210
<4>[ 69.490795] [<ffffffffa00d20da>] ? drm_vma_node_allow+0x2a/0xe0 [drm]
<4>[ 69.494314] [<ffffffffa01e3ca0>] i915_gem_object_get_pages+0x50/0xb0 [i915]
<4>[ 69.498890] [<ffffffffa01ea789>] i915_gem_pwrite_ioctl+0x369/0x800 [i915]
<4>[ 69.503486] [<ffffffffa01e5d28>] ? i915_gem_create+0x58/0xe0 [i915]
<4>[ 69.508875] [<ffffffffa00b9df7>] drm_ioctl+0x197/0x680 [drm]
<4>[ 69.513217] [<ffffffff810a0928>] ? __enqueue_entity+0x78/0x80
<4>[ 69.517707] [<ffffffff810a2745>] ? set_next_entity+0x95/0xb0
<4>[ 69.521943] [<ffffffff811ed620>] do_vfs_ioctl+0x2f0/0x4f0
<4>[ 69.527080] [<ffffffff8138e7c6>] ? file_has_perm+0x86/0xa0
<4>[ 69.531373] [<ffffffff811ed8a1>] SyS_ioctl+0x81/0xa0
<4>[ 69.535560] [<ffffffff81879229>] system_call_fastpath+0x16/0x1b
<4>[ 69.539938] Mem-Info:
<4>[ 69.543345] Node 0 DMA per-cpu:
<4>[ 69.547209] CPU 0: hi: 0, btch: 1 usd: 0
--yliu