(back from vacation)
On 2015/10/4 12:26, Cal Peake wrote:
On Thu, 1 Oct 2015, Steven Rostedt wrote:
I merged 3.4.109 into 3.4-rt, and it bugged. I then booted 3.4.109
vanilla and it bugged too. 3.4.108 is fine.
I guess this is caused by the following commit, which has already been
reverted in mainline kernel. I'll fix it in 3.4.110.
drm/i915: Don't skip request retirement if the active list is empty
commit 0aedb1626566efd72b369c01992ee7413c82a0c5 upstream.
I'm getting a similar type bug here. I've bisected it down to this commit:
commit 961bd13539b9e7ca5d2e667668141496b7a1d6bc
Author: Michel Dänzer <michel.daenzer@xxxxxxx>
Date: Thu Apr 16 11:17:27 2015 +0900
drm/radeon: Use drm_calloc_ab for CS relocs
commit b421ed15d2c3039eb724680e4de1e4b2bd196a9a upstream.
The number of relocs is passed in by userspace and can be large. It has
been observed to cause kcalloc failures in the wild.
Backing it out of vanilla 3.4.109 has so far eliminated the problem.
As you and Satoshi-san have already found out the culprit, I'll just revert
it in 3.4.110.
There are other 2 commits in drivers/gpu/drm/radeon betwwen 3.4.108 and 3.4.109,
and "drm/radeon: fix VM_CONTEXT*_PAGE_TABLE_END_ADDR handling" has been partially
reverted in mainline kernel, so I'll fix this too.
Steven, you look to be using i915 graphics instead of radeon, so it seems
unlikely to me that we're hitting the same problem. Here's my oops for
comparison though:
...
[<ffffffff810d4b93>] ? do_select+0x333/0x5f0
[<ffffffffa03d90b2>] ? r600_cs_packet_parse+0x42/0x140 [radeon]
[<ffffffff810d4500>] ? __pollwait+0x110/0x110
Oct 3 23:24:38 lancer last message repeated 7 times
[<ffffffff810ba216>] ? kmem_cache_free+0x86/0x90
[<ffffffff81038d92>] ? __dequeue_signal+0x102/0x190
[<ffffffff810d505c>] ? core_sys_select+0x20c/0x380
[<ffffffff8103b608>] ? set_current_blocked+0x38/0x60
[<ffffffff8103b6ec>] ? block_sigmask+0x3c/0x50
[<ffffffff81001c84>] ? do_signal+0x1d4/0x620
[<ffffffff8105b1ad>] ? ktime_get_ts+0x6d/0xe0
[<ffffffff810d5212>] ? sys_select+0x42/0x110
[<ffffffff81288982>] ? system_call_fastpath+0x16/0x1b
--
To unsubscribe from this list: send the line "unsubscribe stable" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html