Regression: Bad page map in process xyz

Regression: Bad page map in process xyz

[Jiri Slaby]

Hi,

I'm getting a ton of
BUG: Bad page map in process zypper  pte:676b700029736c6f pmd:44967067
when trying to upgrade the system by:
zypper dup

I bisected that to:
commit afb1c03746aa940374b73a7d5750ee05a2376077
Author: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Date:   Fri Feb 17 10:57:58 2012 +1100

    thp: optimize away unnecessary page table locking

thanks,
-- 
js
suse labs

Re: Regression: Bad page map in process xyz

[Maciej Rutecki]

On poniedziałek, 20 lutego 2012 o 11:02:17 Jiri Slaby wrote:
> Hi,
> 
> I'm getting a ton of
> BUG: Bad page map in process zypper  pte:676b700029736c6f pmd:44967067
> when trying to upgrade the system by:
> zypper dup
> 
> I bisected that to:
> commit afb1c03746aa940374b73a7d5750ee05a2376077
> Author: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
> Date:   Fri Feb 17 10:57:58 2012 +1100
> 
>     thp: optimize away unnecessary page table locking
> 
> thanks,

I created a Bugzilla entry at 
https://bugzilla.kernel.org/show_bug.cgi?id=42820
for your bug/regression report, please add your address to the CC list in 
there, thanks!
-- 
Maciej Rutecki
http://www.mrutecki.pl

Re: Regression: Bad page map in process xyz

[Hugh Dickins]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 1096 bytes --]

On Sun, 26 Feb 2012, Maciej Rutecki wrote:
> On poniedziałek, 20 lutego 2012 o 11:02:17 Jiri Slaby wrote:
> > Hi,
> > 
> > I'm getting a ton of
> > BUG: Bad page map in process zypper  pte:676b700029736c6f pmd:44967067
> > when trying to upgrade the system by:
> > zypper dup
> > 
> > I bisected that to:
> > commit afb1c03746aa940374b73a7d5750ee05a2376077
> > Author: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
> > Date:   Fri Feb 17 10:57:58 2012 +1100
> > 
> >     thp: optimize away unnecessary page table locking
> > 
> > thanks,
> 
> I created a Bugzilla entry at 
> https://bugzilla.kernel.org/show_bug.cgi?id=42820
> for your bug/regression report, please add your address to the CC list in 
> there, thanks!

No, thanks for spotting it, but please remove from the regressions
report: it's not a regression in 3.3-rc but in linux-next - don't take
my word for it, check the commit and you'll not find it in 3.3-rc.

We do still need to get the fix into linux-next: Horiguchi-san, has
akpm put your fix in mm-commits yet?  Please send it again if not.

Hugh

Re: Regression: Bad page map in process xyz

[Maciej Rutecki]

On niedziela, 26 lutego 2012 o 14:10:31 Hugh Dickins wrote:
> On Sun, 26 Feb 2012, Maciej Rutecki wrote:
> > On poniedziałek, 20 lutego 2012 o 11:02:17 Jiri Slaby wrote:
> > > Hi,
> > > 
> > > I'm getting a ton of
> > > BUG: Bad page map in process zypper  pte:676b700029736c6f pmd:44967067
> > > when trying to upgrade the system by:
> > > zypper dup
> > > 
> > > I bisected that to:
> > > commit afb1c03746aa940374b73a7d5750ee05a2376077
> > > Author: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
> > > Date:   Fri Feb 17 10:57:58 2012 +1100
> > > 
> > >     thp: optimize away unnecessary page table locking
> > > 
> > > thanks,
> > 
> > I created a Bugzilla entry at
> > https://bugzilla.kernel.org/show_bug.cgi?id=42820
> > for your bug/regression report, please add your address to the CC list in
> > there, thanks!
> 
> No, thanks for spotting it, but please remove from the regressions
> report: it's not a regression in 3.3-rc but in linux-next - don't take
> my word for it, check the commit and you'll not find it in 3.3-rc.
> 
> We do still need to get the fix into linux-next: Horiguchi-san, has
> akpm put your fix in mm-commits yet?  Please send it again if not.
> 
> Hugh

Thanks for the information. I should check where is commit placed before.

Regards
-- 
Maciej Rutecki
http://www.mrutecki.pl

Re: Regression: Bad page map in process xyz

[Dave Jones]

On Sun, Feb 26, 2012 at 05:10:31AM -0800, Hugh Dickins wrote:
 > On Sun, 26 Feb 2012, Maciej Rutecki wrote:
 > > On poniedziałek, 20 lutego 2012 o 11:02:17 Jiri Slaby wrote:
 > > > Hi,
 > > > 
 > > > I'm getting a ton of
 > > > BUG: Bad page map in process zypper  pte:676b700029736c6f pmd:44967067
 > > > when trying to upgrade the system by:
 > > > zypper dup
 > > > 
 > > > I bisected that to:
 > > > commit afb1c03746aa940374b73a7d5750ee05a2376077
 > > > Author: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
 > > > Date:   Fri Feb 17 10:57:58 2012 +1100
 > > > 
 > > >     thp: optimize away unnecessary page table locking
 > > > 
 > > > thanks,
 > > 
 > > I created a Bugzilla entry at 
 > > https://bugzilla.kernel.org/show_bug.cgi?id=42820
 > > for your bug/regression report, please add your address to the CC list in 
 > > there, thanks!
 > 
 > No, thanks for spotting it, but please remove from the regressions
 > report: it's not a regression in 3.3-rc but in linux-next - don't take
 > my word for it, check the commit and you'll not find it in 3.3-rc.
 > 
 > We do still need to get the fix into linux-next: Horiguchi-san, has
 > akpm put your fix in mm-commits yet?  Please send it again if not.

FWIW, we've been getting a bunch of these since 3.0 in Fedora.
I've been trying to come up with some way of trying to reproduce them
myself, without any luck. Some of our users seem to hit them surprisingly easily.

At first I started wondering if it was just bad hardware, but the frequency
that we've been getting reports seems to suggest something more screwed up.

https://bugzilla.redhat.com/buglist.cgi?bug_status=NEW&bug_status=ASSIGNED&bug_status=MODIFIED&bug_status=ON_DEV&bug_status=ON_QA&bug_status=VERIFIED&bug_status=RELEASE_PENDING&bug_status=POST&classification=Fedora&component=kernel&product=Fedora&query_format=advanced&short_desc=bug+page+map&short_desc_type=allwordssubstr&version=15&version=16&version=rawhide&order=bug_id&query_based_on=

Until last week, we only had reports up until 3.2, but now that the F17 alpha
is getting tested, people are starting to hit it on 3.3rc too

I'll try a test build with that commit backed out for our users to try out
next week.

	Dave

Re: Regression: Bad page map in process xyz

[Naoya Horiguchi]

Hi Hugh,

On Sun, Feb 26, 2012 at 05:10:31AM -0800, Hugh Dickins wrote:
> On Sun, 26 Feb 2012, Maciej Rutecki wrote:
> > On poniedziaナＦk, 20 lutego 2012 o 11:02:17 Jiri Slaby wrote:
> > > Hi,
> > > 
> > > I'm getting a ton of
> > > BUG: Bad page map in process zypper  pte:676b700029736c6f pmd:44967067
> > > when trying to upgrade the system by:
> > > zypper dup
> > > 
> > > I bisected that to:
> > > commit afb1c03746aa940374b73a7d5750ee05a2376077
> > > Author: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
> > > Date:   Fri Feb 17 10:57:58 2012 +1100
> > > 
> > >     thp: optimize away unnecessary page table locking
> > > 
> > > thanks,
> > 
> > I created a Bugzilla entry at 
> > https://bugzilla.kernel.org/show_bug.cgi?id=42820
> > for your bug/regression report, please add your address to the CC list in 
> > there, thanks!
> 
> No, thanks for spotting it, but please remove from the regressions
> report: it's not a regression in 3.3-rc but in linux-next - don't take
> my word for it, check the commit and you'll not find it in 3.3-rc.
> 
> We do still need to get the fix into linux-next: Horiguchi-san, has
> akpm put your fix in mm-commits yet?  Please send it again if not.

Sorry for late reply.
And yes, this fix is in mm-commits now.

Thanks,
Naoya

Re: Regression: Bad page map in process xyz

[Jiri Slaby]

On 02/27/2012 05:45 AM, Naoya Horiguchi wrote:
>> We do still need to get the fix into linux-next: Horiguchi-san, has
>> akpm put your fix in mm-commits yet?  Please send it again if not.
>
> Sorry for late reply.
> And yes, this fix is in mm-commits now.

And in -next too as of today:
commit 57a2e0ac358d580399a63b54fe4275632bbf63f5
Author: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Date:   Sat Feb 25 12:28:03 2012 +1100

     fix mremap bug of failing to split thp

thanks,
-- 
js