Re: GPF in aio_migratepage

2014-01-21 Thread Kristian Nielsen
Kristian Nielsen writes: > Benjamin LaHaise writes: > >> Linus just pushed out 3.13-rc5 that has changes to aio_migratepage() that >> should make it much more robust, as well as other fixes. Can you please >> give it a spin as well and let me know if it works? Thanks a bunch! > > Ok, will do

Re: GPF in aio_migratepage

2013-12-22 Thread Kristian Nielsen
Benjamin LaHaise writes: > Linus just pushed out 3.13-rc5 that has changes to aio_migratepage() that > should make it much more robust, as well as other fixes. Can you please > give it a spin as well and let me know if it works? Thanks a bunch! Ok, will do. - Kristian. -- To unsubscribe fr

Re: GPF in aio_migratepage

2013-12-22 Thread Benjamin LaHaise
Hi Kristian, On Sun, Dec 22, 2013 at 09:44:45PM +0100, Kristian Nielsen wrote: > There are other changes in that area since 3.13-rc1 though. > > Anyway, I am now running with 3.13-rc4 and will report if I see anything. > Given that I do not have any way to reproduce (I only ever saw this once), >

Re: GPF in aio_migratepage

2013-12-22 Thread Kristian Nielsen
Gu Zheng writes: > This issue seems like a problem that has been fixed yet: > http://article.gmane.org/gmane.linux.kernel.aio.general/3741/match=potential+use+after+free+aio%5fmigratepage > commit 5e9ae2e5da0beb93f8557fc92a8f4fbc05ea448f > aio: fix use-after-free in aio_migratepage > So I think m

Re: GPF in aio_migratepage

2013-12-15 Thread Gu Zheng
Hi Kristian, On 12/16/2013 10:58 AM, Gu Zheng wrote: > Hi Kristian, > On 12/16/2013 05:59 AM, Kristian Nielsen wrote: > >> What is the status of this? >> >> If I understand correctly, the crash I saw is different from what Dave >> saw. Thought the crash you saw is different from Dave's, but as y

Re: GPF in aio_migratepage

2013-12-15 Thread Gu Zheng
Hi Kristian, On 12/16/2013 05:59 AM, Kristian Nielsen wrote: > What is the status of this? > > If I understand correctly, the crash I saw is different from what Dave > saw. > > There was one patched scheduled for inclusion that fixes Dave's crash. But > what about mine? I have been running 3.13-

Re: GPF in aio_migratepage

2013-12-15 Thread Kristian Nielsen
What is the status of this? If I understand correctly, the crash I saw is different from what Dave saw. There was one patched scheduled for inclusion that fixes Dave's crash. But what about mine? I have been running 3.13-rc2 for a couple of weeks now with your other patch, without seeing it again

Re: GPF in aio_migratepage

2013-12-03 Thread Gu Zheng
Hi Dave, According to your analysis and the dump stack, it seems that it tried to migrate aio ring pages in the compact path, but the aio context(mapping->private_data) is invalid (not NULL), so only one case can cause this condition, aio ring file was left and not cleaned up in the fail path of i

Re: GPF in aio_migratepage

2013-12-02 Thread Dave Jones
On Mon, Dec 02, 2013 at 06:10:46PM +0800, Gu Zheng wrote: > Hi Kristian, Dave, > > Could you please help to check whether the following patch can fix this > issue? This introduces some locking bugs.. [ 222.327950] BUG: sleeping function called from invalid context at kernel/locking/mutex

Re: GPF in aio_migratepage

2013-12-02 Thread Kristian Nielsen
Gu Zheng writes: > Hi Kristian, Dave, > > Could you please help to check whether the following patch can fix this issue? > Signed-off-by: Gu Zheng > --- > fs/aio.c | 28 ++-- > 1 files changed, 10 insertions(+), 18 deletions(-) > Ok. I've applied the patch to 3.13-rc

Re: GPF in aio_migratepage

2013-12-02 Thread Gu Zheng
Hi Kristian, Dave, Could you please help to check whether the following patch can fix this issue? Signed-off-by: Gu Zheng --- fs/aio.c | 28 ++-- 1 files changed, 10 insertions(+), 18 deletions(-) diff --git a/fs/aio.c b/fs/aio.c index 08159ed..fc1fd0a 100644 --- a/f

Re: GPF in aio_migratepage

2013-11-30 Thread Kristian Nielsen
Benjamin LaHaise writes: > For Dave: what line is this bug on? Is it the dereference of ctx when > doing spin_lock_irqsave(&ctx->completion_lock, flags); or is the > ctx->ring_pages[idx] = new; ? From the 64 bit splat, I'm thinking the > former, which is quite strange given that the clearing

Re: GPF in aio_migratepage

2013-11-26 Thread Dave Jones
On Tue, Nov 26, 2013 at 10:23:37AM -0500, Benjamin LaHaise wrote: > On Mon, Nov 25, 2013 at 11:19:53PM -0800, Kent Overstreet wrote: > > On Tue, Nov 26, 2013 at 01:01:32AM -0500, Dave Jones wrote: > > > On Mon, Nov 25, 2013 at 10:26:45PM -0500, Dave Jones wrote: > > > > Hi Kent, > > > > >

Re: GPF in aio_migratepage

2013-11-26 Thread Benjamin LaHaise
On Mon, Nov 25, 2013 at 11:19:53PM -0800, Kent Overstreet wrote: > On Tue, Nov 26, 2013 at 01:01:32AM -0500, Dave Jones wrote: > > On Mon, Nov 25, 2013 at 10:26:45PM -0500, Dave Jones wrote: > > > Hi Kent, > > > > > > I hit the GPF below on a tree based on > > 8e45099e029bb6b369b27d8d4920db8ca

Re: GPF in aio_migratepage

2013-11-25 Thread Kent Overstreet
On Tue, Nov 26, 2013 at 01:01:32AM -0500, Dave Jones wrote: > On Mon, Nov 25, 2013 at 10:26:45PM -0500, Dave Jones wrote: > > Hi Kent, > > > > I hit the GPF below on a tree based on > 8e45099e029bb6b369b27d8d4920db8caff5ecce > > which has your commit e34ecee2ae791df674dfb466ce40692ca6218e43 >

Re: GPF in aio_migratepage

2013-11-25 Thread Dave Jones
On Mon, Nov 25, 2013 at 10:26:45PM -0500, Dave Jones wrote: > Hi Kent, > > I hit the GPF below on a tree based on > 8e45099e029bb6b369b27d8d4920db8caff5ecce > which has your commit e34ecee2ae791df674dfb466ce40692ca6218e43 > ("aio: Fix a trinity splat"). Is this another path your patch miss

GPF in aio_migratepage

2013-11-25 Thread Dave Jones
Hi Kent, I hit the GPF below on a tree based on 8e45099e029bb6b369b27d8d4920db8caff5ecce which has your commit e34ecee2ae791df674dfb466ce40692ca6218e43 ("aio: Fix a trinity splat"). Is this another path your patch missed, or a completely different bug to what you were chasing ? Dave gen