Just came across this WARNING (running latest git kernel):
Mar 29 09:14:39 gentoo kernel: udev[888]: starting version 164 Mar 29 15:45:45 gentoo kernel: ------------[ cut here ]------------ Mar 29 15:45:45 gentoo kernel: WARNING: at drivers/gpu/drm/ttm/ttm_page_alloc.c:758 ttm_put_pages+0x1c6/0x2b0() Mar 29 15:45:45 gentoo kernel: Hardware name: System Product Name Mar 29 15:45:45 gentoo kernel: Pid: 688, comm: kworker/u:6 Not tainted 2.6.38-09065-g89078d5 #23 Mar 29 15:45:45 gentoo kernel: Call Trace: Mar 29 15:45:45 gentoo kernel: [<ffffffff81073fea>] warn_slowpath_common+0x7a/0xb0 Mar 29 15:45:45 gentoo kernel: [<ffffffff8107410a>] warn_slowpath_null+0x1a/0x20 Mar 29 15:45:45 gentoo kernel: [<ffffffff8130fe46>] ttm_put_pages+0x1c6/0x2b0 Mar 29 15:45:45 gentoo kernel: [<ffffffff8130889d>] ttm_tt_free_alloced_pages+0xbd/0x100 Mar 29 15:45:45 gentoo kernel: [<ffffffff8130933c>] ttm_tt_swapout+0x1ac/0x2a0 Mar 29 15:45:45 gentoo kernel: [<ffffffff8130ab85>] ttm_bo_swapout+0x1a5/0x230 Mar 29 15:45:45 gentoo kernel: [<ffffffff812fcefc>] ? drm_mm_kmalloc+0x3c/0xd0 Mar 29 15:45:45 gentoo kernel: [<ffffffff81307d38>] ttm_shrink+0xd8/0x120 Mar 29 15:45:45 gentoo kernel: [<ffffffff8108f101>] ? queue_delayed_work_on+0x91/0x100 Mar 29 15:45:45 gentoo kernel: [<ffffffff81307d80>] ? ttm_shrink+0x120/0x120 Mar 29 15:45:45 gentoo kernel: [<ffffffff81307d99>] ttm_shrink_work+0x19/0x20 Mar 29 15:45:45 gentoo kernel: [<ffffffff8108e0ae>] process_one_work+0x11e/0x4c0 Mar 29 15:45:45 gentoo kernel: [<ffffffff8108ea2e>] worker_thread+0x15e/0x340 Mar 29 15:45:45 gentoo kernel: [<ffffffff8108e8d0>] ? manage_workers.isra.28+0x220/0x220 Mar 29 15:45:45 gentoo kernel: [<ffffffff8109348c>] kthread+0x8c/0xa0 Mar 29 15:45:45 gentoo kernel: [<ffffffff8156e814>] kernel_thread_helper+0x4/0x10 Mar 29 15:45:45 gentoo kernel: [<ffffffff81093400>] ? kthread_flush_work_fn+0x20/0x20 Mar 29 15:45:45 gentoo kernel: [<ffffffff8156e810>] ? gs_change+0xb/0xb Mar 29 15:45:45 gentoo kernel: ---[ end trace d5b07e4231e70291 ]--- Mar 29 15:45:45 gentoo kernel: ------------[ cut here ]------------ ... It's repeated ad nauseam from there on. (/var/log/kern.log is now over 85MB in size)
On 2011.03.29 at 16:56 +0200, Markus Trippelsdorf wrote:
Just came across this WARNING (running latest git kernel):
Mar 29 09:14:39 gentoo kernel: udev[888]: starting version 164 Mar 29 15:45:45 gentoo kernel: ------------[ cut here ]------------ Mar 29 15:45:45 gentoo kernel: WARNING: at drivers/gpu/drm/ttm/ttm_page_alloc.c:758 ttm_put_pages+0x1c6/0x2b0() Mar 29 15:45:45 gentoo kernel: Hardware name: System Product Name
It happened again today.
------------[ cut here ]------------ WARNING: at drivers/gpu/drm/ttm/ttm_page_alloc.c:758 ttm_put_pages+0x1c6/0x2c0() Hardware name: System Product Name Pid: 680, comm: kworker/u:6 Not tainted 2.6.39-rc1-00254-gd297f96 #5 Call Trace: [<ffffffff810736af>] warn_slowpath_common+0x6f/0xa0 [<ffffffff810737fa>] warn_slowpath_null+0x1a/0x20 [<ffffffff812c7726>] ttm_put_pages+0x1c6/0x2c0 [<ffffffff812bff7d>] ttm_tt_free_alloced_pages+0xbd/0x100 [<ffffffff812c0a5c>] ttm_tt_swapout+0x1bc/0x2c0 [<ffffffff812c2365>] ttm_bo_swapout+0x1a5/0x240 [<ffffffff812b5637>] ? drm_mm_kmalloc+0x37/0xe0 [<ffffffff812bf3b8>] ttm_shrink+0xd8/0x140 [<ffffffff8108eb01>] ? queue_work+0x21/0x60 [<ffffffff812bf420>] ? ttm_shrink+0x140/0x140 [<ffffffff812bf439>] ttm_shrink_work+0x19/0x20 [<ffffffff8108dade>] process_one_work+0x11e/0x4e0 [<ffffffff8108e4ce>] worker_thread+0x16e/0x360 [<ffffffff8108e360>] ? manage_workers.isra.28+0x260/0x260 [<ffffffff810932ac>] kthread+0x8c/0xa0 [<ffffffff8152a394>] kernel_thread_helper+0x4/0x10 [<ffffffff81093220>] ? kthread_flush_work_fn+0x20/0x20 [<ffffffff8152a390>] ? gs_change+0xb/0xb ---[ end trace cb7fc1aad109a0a7 ]--- ------------[ cut here ]------------
So its not a easily reproducible problem. My config and a full dmesg are attached, in case anybody is interested.
On 2011.04.05 at 18:31 +0200, Markus Trippelsdorf wrote:
On 2011.03.29 at 16:56 +0200, Markus Trippelsdorf wrote:
Just came across this WARNING (running latest git kernel):
Mar 29 09:14:39 gentoo kernel: udev[888]: starting version 164 Mar 29 15:45:45 gentoo kernel: ------------[ cut here ]------------ Mar 29 15:45:45 gentoo kernel: WARNING: at drivers/gpu/drm/ttm/ttm_page_alloc.c:758 ttm_put_pages+0x1c6/0x2b0() Mar 29 15:45:45 gentoo kernel: Hardware name: System Product Name
It happened again today.
------------[ cut here ]------------ WARNING: at drivers/gpu/drm/ttm/ttm_page_alloc.c:758 ttm_put_pages+0x1c6/0x2c0() Hardware name: System Product Name Pid: 680, comm: kworker/u:6 Not tainted 2.6.39-rc1-00254-gd297f96 #5 Call Trace: [<ffffffff810736af>] warn_slowpath_common+0x6f/0xa0 [<ffffffff810737fa>] warn_slowpath_null+0x1a/0x20 [<ffffffff812c7726>] ttm_put_pages+0x1c6/0x2c0 [<ffffffff812bff7d>] ttm_tt_free_alloced_pages+0xbd/0x100 [<ffffffff812c0a5c>] ttm_tt_swapout+0x1bc/0x2c0 [<ffffffff812c2365>] ttm_bo_swapout+0x1a5/0x240 [<ffffffff812b5637>] ? drm_mm_kmalloc+0x37/0xe0 [<ffffffff812bf3b8>] ttm_shrink+0xd8/0x140 [<ffffffff8108eb01>] ? queue_work+0x21/0x60 [<ffffffff812bf420>] ? ttm_shrink+0x140/0x140 [<ffffffff812bf439>] ttm_shrink_work+0x19/0x20 [<ffffffff8108dade>] process_one_work+0x11e/0x4e0 [<ffffffff8108e4ce>] worker_thread+0x16e/0x360 [<ffffffff8108e360>] ? manage_workers.isra.28+0x260/0x260 [<ffffffff810932ac>] kthread+0x8c/0xa0 [<ffffffff8152a394>] kernel_thread_helper+0x4/0x10 [<ffffffff81093220>] ? kthread_flush_work_fn+0x20/0x20 [<ffffffff8152a390>] ? gs_change+0xb/0xb ---[ end trace cb7fc1aad109a0a7 ]--- ------------[ cut here ]------------
Looks like this is caused by f9820a46dd7888b05a36e81166fb1abcc47dcc3f :
ttm: Utilize the DMA API for pages that have TTM_PAGE_FLAG_DMA32 set.
CCing the relevant people.
Dave, Konrad
I think we need to back out this change as soon as possible. I will follow up the discussion with Konrad on how we should attempt to handle vm passthrough as soon as things settle down a bit over here. Hopefully this week.
Thomas
On 04/06/2011 10:54 AM, Markus Trippelsdorf wrote:
On 2011.04.05 at 18:31 +0200, Markus Trippelsdorf wrote:
On 2011.03.29 at 16:56 +0200, Markus Trippelsdorf wrote:
Just came across this WARNING (running latest git kernel):
Mar 29 09:14:39 gentoo kernel: udev[888]: starting version 164 Mar 29 15:45:45 gentoo kernel: ------------[ cut here ]------------ Mar 29 15:45:45 gentoo kernel: WARNING: at drivers/gpu/drm/ttm/ttm_page_alloc.c:758 ttm_put_pages+0x1c6/0x2b0() Mar 29 15:45:45 gentoo kernel: Hardware name: System Product Name
It happened again today.
------------[ cut here ]------------ WARNING: at drivers/gpu/drm/ttm/ttm_page_alloc.c:758 ttm_put_pages+0x1c6/0x2c0() Hardware name: System Product Name Pid: 680, comm: kworker/u:6 Not tainted 2.6.39-rc1-00254-gd297f96 #5 Call Trace: [<ffffffff810736af>] warn_slowpath_common+0x6f/0xa0 [<ffffffff810737fa>] warn_slowpath_null+0x1a/0x20 [<ffffffff812c7726>] ttm_put_pages+0x1c6/0x2c0 [<ffffffff812bff7d>] ttm_tt_free_alloced_pages+0xbd/0x100 [<ffffffff812c0a5c>] ttm_tt_swapout+0x1bc/0x2c0 [<ffffffff812c2365>] ttm_bo_swapout+0x1a5/0x240 [<ffffffff812b5637>] ? drm_mm_kmalloc+0x37/0xe0 [<ffffffff812bf3b8>] ttm_shrink+0xd8/0x140 [<ffffffff8108eb01>] ? queue_work+0x21/0x60 [<ffffffff812bf420>] ? ttm_shrink+0x140/0x140 [<ffffffff812bf439>] ttm_shrink_work+0x19/0x20 [<ffffffff8108dade>] process_one_work+0x11e/0x4e0 [<ffffffff8108e4ce>] worker_thread+0x16e/0x360 [<ffffffff8108e360>] ? manage_workers.isra.28+0x260/0x260 [<ffffffff810932ac>] kthread+0x8c/0xa0 [<ffffffff8152a394>] kernel_thread_helper+0x4/0x10 [<ffffffff81093220>] ? kthread_flush_work_fn+0x20/0x20 [<ffffffff8152a390>] ? gs_change+0xb/0xb ---[ end trace cb7fc1aad109a0a7 ]--- ------------[ cut here ]------------
Looks like this is caused by f9820a46dd7888b05a36e81166fb1abcc47dcc3f :
ttm: Utilize the DMA API for pages that have TTM_PAGE_FLAG_DMA32 set.
CCing the relevant people.
On Wed, Apr 06, 2011 at 11:13:59AM +0200, Thomas Hellstrom wrote:
Dave, Konrad
I think we need to back out this change as soon as possible. I will follow up the discussion with Konrad on how we should attempt to handle vm passthrough as soon as things settle down a bit over here.
Ok. I believe just reverting this particular git commit
69a07f0b117a40fcc1a479358d8e1f41793617f2
does it. Doing it on top of v2.6.39-rc1:
ated: 2427 deleted: 1229), skipping inexact rename detection Finished one revert. [stable/ttm.pci-api-rc2 97bbc38] Revert "ttm: Utilize the DMA API for pages that have TTM_PAGE_FLAG_DMA32 set." 1 files changed, 3 insertions(+), 23 deletions(-)
shows it to being a nice easy backout.
Dave, I stuck the revert on
git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git stable/ttm.pci-api-rc2
Please pull for rc2.
Hopefully this week.
Thomas, waiting for your input on http://mid.gmane.org/20110331154954.GA7695@dumpdata.com (ignore point 1, please).
.. snip..
Looks like this is caused by f9820a46dd7888b05a36e81166fb1abcc47dcc3f :
Wrong git commit.
ttm: Utilize the DMA API for pages that have TTM_PAGE_FLAG_DMA32 set.
It is 69a07f0b117a40fcc1a479358d8e1f41793617f2
CCing the relevant people.
How did you make the machine to do this? Did you just echo XX > to the shrinker?
On 2011.04.06 at 08:27 -0400, Konrad Rzeszutek Wilk wrote:
On Wed, Apr 06, 2011 at 11:13:59AM +0200, Thomas Hellstrom wrote:
Dave, Konrad
I think we need to back out this change as soon as possible. I will follow up the discussion with Konrad on how we should attempt to handle vm passthrough as soon as things settle down a bit over here.
Ok. I believe just reverting this particular git commit
69a07f0b117a40fcc1a479358d8e1f41793617f2
does it. Doing it on top of v2.6.39-rc1:
ated: 2427 deleted: 1229), skipping inexact rename detection Finished one revert. [stable/ttm.pci-api-rc2 97bbc38] Revert "ttm: Utilize the DMA API for pages that have TTM_PAGE_FLAG_DMA32 set." 1 files changed, 3 insertions(+), 23 deletions(-)
shows it to being a nice easy backout.
Dave, I stuck the revert on
git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen.git stable/ttm.pci-api-rc2
Please pull for rc2.
.. snip..
Looks like this is caused by f9820a46dd7888b05a36e81166fb1abcc47dcc3f :
Wrong git commit.
ttm: Utilize the DMA API for pages that have TTM_PAGE_FLAG_DMA32 set.
It is 69a07f0b117a40fcc1a479358d8e1f41793617f2
Oops, you're right. (copy&paste error)
CCing the relevant people.
How did you make the machine to do this? Did you just echo XX > to the shrinker?
No, I've just used the machine normally (browsing the net, viewing pdfs, etc.) and triggered the warning two times in the last couple of days.
dri-devel@lists.freedesktop.org
-
Konrad Rzeszutek Wilk -
Markus Trippelsdorf -
Thomas Hellstrom