BUG: non-zero nr_pmds on freeing mm: -1

BUG: non-zero nr_pmds on freeing mm: -1

[John David Anglin]

With 4.0.0-rc3+ #1 SMP, I get hundreds of these BUG messages:
BUG: non-zero nr_pmds on freeing mm: -1

However, system does boot.

Dave
--
John David Anglin	dave.anglin@bell.net

Re: BUG: non-zero nr_pmds on freeing mm: -1

[John David Anglin]

On 2015-03-11 7:32 AM, Domenico Andreoli wrote:
> I also do. Related to https://lkml.org/lkml/2015/2/13/431?
More info is here:
http://www.spinics.net/lists/linux-mm/msg84937.html

There doesn't seem to be a patch yet.

Dave

-- 
John David Anglin  dave.anglin@bell.net

Re: BUG: non-zero nr_pmds on freeing mm: -1

[Graham Gower]

I thought this patch was supposed to address the problem?
https://lkml.org/lkml/2015/2/26/230

However, its not fixed it for me either in 4.0.0-rc3.

-Graham

On 12 March 2015 at 00:44, John David Anglin <dave.anglin@bell.net> wrote:
> On 2015-03-11 7:32 AM, Domenico Andreoli wrote:
>>
>> I also do. Related to https://lkml.org/lkml/2015/2/13/431?
>
> More info is here:
> http://www.spinics.net/lists/linux-mm/msg84937.html
>
> There doesn't seem to be a patch yet.
>
>
> Dave
>
> --
> John David Anglin  dave.anglin@bell.net
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: BUG: non-zero nr_pmds on freeing mm: -1

[John David Anglin]

On 2015-03-11, at 7:31 PM, Graham Gower wrote:

> I thought this patch was supposed to address the problem?
> https://lkml.org/lkml/2015/2/26/230
> 
> However, its not fixed it for me either in 4.0.0-rc3.

I doubt the patch is correct for parisc.  I believe the system where I saw the BUG messages
has PT_NLEVELS == 3.

Dave
--
John David Anglin	dave.anglin@bell.net

Re: BUG: non-zero nr_pmds on freeing mm: -1

[Aaro Koskinen]

Hi,

On Wed, Mar 11, 2015 at 08:21:35PM -0400, John David Anglin wrote:
> On 2015-03-11, at 7:31 PM, Graham Gower wrote:
> > I thought this patch was supposed to address the problem?
> > https://lkml.org/lkml/2015/2/26/230
> > 
> > However, its not fixed it for me either in 4.0.0-rc3.
> 
> I doubt the patch is correct for parisc. I believe the system where
> I saw the BUG messages has PT_NLEVELS == 3.

For 32-bit kernels, the fix helped.

A.

Re: BUG: non-zero nr_pmds on freeing mm: -1

[John David Anglin]

On 2015-03-12 3:35 PM, Aaro Koskinen wrote:
> Hi,
>
> On Wed, Mar 11, 2015 at 08:21:35PM -0400, John David Anglin wrote:
>> On 2015-03-11, at 7:31 PM, Graham Gower wrote:
>>> I thought this patch was supposed to address the problem?
>>> https://lkml.org/lkml/2015/2/26/230
>>>
>>> However, its not fixed it for me either in 4.0.0-rc3.
>> I doubt the patch is correct for parisc. I believe the system where
>> I saw the BUG messages has PT_NLEVELS == 3.
> For 32-bit kernels, the fix helped.
32-bit kernels have PT_NLEVELS == 2.  I think the define needs to be 
moved outside #if/#endif
but didn't get a chance to test last night.

Dave

-- 
John David Anglin  dave.anglin@bell.net

Re: BUG: non-zero nr_pmds on freeing mm: -1

[Aaro Koskinen]

Hi,

On Thu, Mar 12, 2015 at 03:51:11PM -0400, John David Anglin wrote:
> On 2015-03-12 3:35 PM, Aaro Koskinen wrote:
> >On Wed, Mar 11, 2015 at 08:21:35PM -0400, John David Anglin wrote:
> >>On 2015-03-11, at 7:31 PM, Graham Gower wrote:
> >>>I thought this patch was supposed to address the problem?
> >>>https://lkml.org/lkml/2015/2/26/230
> >>>
> >>>However, its not fixed it for me either in 4.0.0-rc3.
> >>I doubt the patch is correct for parisc. I believe the system where
> >>I saw the BUG messages has PT_NLEVELS == 3.
> >For 32-bit kernels, the fix helped.
> 32-bit kernels have PT_NLEVELS == 2.  I think the define needs to be moved
> outside #if/#endif
> but didn't get a chance to test last night.

Not sure if that would be correct. The fix disabled PMD accounting
on 32-bit kernels because that level does not exist, but on
64-bit you probably would need it? Maybe your bug is different.
Does it occur on every process exit, or just occasionally?
(On 32-bit every exit triggered it.)

A.

Re: BUG: non-zero nr_pmds on freeing mm: -1

[John David Anglin]

On 2015-03-12 4:15 PM, Aaro Koskinen wrote:
> Not sure if that would be correct. The fix disabled PMD accounting
> on 32-bit kernels because that level does not exist, but on
> 64-bit you probably would need it? Maybe your bug is different.
> Does it occur on every process exit, or just occasionally?
> (On 32-bit every exit triggered it.)
I would guess it occurs on every process exit.  Note it is mm -1 which I 
didn't see in any
of the other messages that I looked at.

Dave

-- 
John David Anglin  dave.anglin@bell.net

Re: BUG: non-zero nr_pmds on freeing mm: -1

[Kirill A. Shutemov]

On Thu, Mar 12, 2015 at 03:51:11PM -0400, John David Anglin wrote:
> On 2015-03-12 3:35 PM, Aaro Koskinen wrote:
> >Hi,
> >
> >On Wed, Mar 11, 2015 at 08:21:35PM -0400, John David Anglin wrote:
> >>On 2015-03-11, at 7:31 PM, Graham Gower wrote:
> >>>I thought this patch was supposed to address the problem?
> >>>https://lkml.org/lkml/2015/2/26/230
> >>>
> >>>However, its not fixed it for me either in 4.0.0-rc3.
> >>I doubt the patch is correct for parisc. I believe the system where
> >>I saw the BUG messages has PT_NLEVELS == 3.
> >For 32-bit kernels, the fix helped.
> 32-bit kernels have PT_NLEVELS == 2.  I think the define needs to be moved
> outside #if/#endif
> but didn't get a chance to test last night.

There's hack in pgd_alloc() on parisc to initialize one pmd, which is not
accounted and we get underflow on exit. We need to adjust accounting for
that pmd.

Could you try the patch below?

diff --git a/arch/parisc/include/asm/pgalloc.h b/arch/parisc/include/asm/pgalloc.h
index f213f5b4c423..8ee9a0bdc468 100644
--- a/arch/parisc/include/asm/pgalloc.h
+++ b/arch/parisc/include/asm/pgalloc.h
@@ -38,6 +38,7 @@ static inline pgd_t *pgd_alloc(struct mm_struct *mm)
 		/* The first pmd entry also is marked with _PAGE_GATEWAY as
 		 * a signal that this pmd may not be freed */
 		__pgd_val_set(*pgd, PxD_FLAG_ATTACHED);
+		mm_inc_nr_pmds(mm);
 #endif
 	}
 	return actual_pgd;
-- 
 Kirill A. Shutemov

Re: BUG: non-zero nr_pmds on freeing mm: -1

[Kirill A. Shutemov]

On Thu, Mar 12, 2015 at 08:41:37PM -0400, John David Anglin wrote:
> On 2015-03-12, at 6:07 PM, Kirill A. Shutemov wrote:
> 
> > There's hack in pgd_alloc() on parisc to initialize one pmd, which is not
> > accounted and we get underflow on exit. We need to adjust accounting for
> > that pmd.
> > 
> > Could you try the patch below?
> > 
> > diff --git a/arch/parisc/include/asm/pgalloc.h b/arch/parisc/include/asm/pgalloc.h
> > index f213f5b4c423..8ee9a0bdc468 100644
> > --- a/arch/parisc/include/asm/pgalloc.h
> > +++ b/arch/parisc/include/asm/pgalloc.h
> > @@ -38,6 +38,7 @@ static inline pgd_t *pgd_alloc(struct mm_struct *mm)
> > 		/* The first pmd entry also is marked with _PAGE_GATEWAY as
> > 		 * a signal that this pmd may not be freed */
> > 		__pgd_val_set(*pgd, PxD_FLAG_ATTACHED);
> > +		mm_inc_nr_pmds(mm);
> > #endif
> > 	}
> > 	return actual_pgd;
> 
> The patch fixes the BUG messages.
> 
> Helge, the #ifdef in pgd_alloc doesn't seem correct.  We only have a L2/L3 scheme when
> "defined(CONFIG_64BIT) && defined(CONFIG_PARISC_PAGE_SIZE_4KB)".

Looks like you're correct. But that's separate issue. Please, post a
patch.

-- 
 Kirill A. Shutemov