mm: page_alloc: avoid false page outside zone error info

[PATCH] mm: page_alloc: avoid false page outside zone error info

Posted by Miaohe Lin 2 years, 7 months ago

If pfn is outside zone boundaries in the first round, ret will be set
to 1. But if pfn is changed to inside the zone boundaries in zone span
seqretry path, ret is still set to 1 leading to false page outside zone
error info.

Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
---
 mm/page_alloc.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 7d3460c7a480..40a9f3ed25e9 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -459,7 +459,7 @@ void set_pageblock_migratetype(struct page *page, int migratetype)
 #ifdef CONFIG_DEBUG_VM
 static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
 {
-	int ret = 0;
+	int ret;
 	unsigned seq;
 	unsigned long pfn = page_to_pfn(page);
 	unsigned long sp, start_pfn;
@@ -470,6 +470,8 @@ static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
 		sp = zone->spanned_pages;
 		if (!zone_spans_pfn(zone, pfn))
 			ret = 1;
+		else
+			ret = 0;
 	} while (zone_span_seqretry(zone, seq));
 
 	if (ret)
-- 
2.33.0

Re: [PATCH] mm: page_alloc: avoid false page outside zone error info

Posted by Matthew Wilcox 2 years, 7 months ago

On Tue, Jul 04, 2023 at 07:18:23PM +0800, Miaohe Lin wrote:
> @@ -470,6 +470,8 @@ static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
>  		sp = zone->spanned_pages;
>  		if (!zone_spans_pfn(zone, pfn))
>  			ret = 1;
> +		else
> +			ret = 0;

Surely 'ret = zone_spans_pfn(zone, pfn);' ?

Also, did you spot this by inspection or do you have a test-case or bug
report?  Should this have a Fixes: tag?

Re: [PATCH] mm: page_alloc: avoid false page outside zone error info

Posted by Miaohe Lin 2 years, 7 months ago

On 2023/7/4 20:13, Matthew Wilcox wrote:
> On Tue, Jul 04, 2023 at 07:18:23PM +0800, Miaohe Lin wrote:
>> @@ -470,6 +470,8 @@ static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
>>  		sp = zone->spanned_pages;
>>  		if (!zone_spans_pfn(zone, pfn))
>>  			ret = 1;
>> +		else
>> +			ret = 0;
> 
> Surely 'ret = zone_spans_pfn(zone, pfn);' ?

Do you mean 'ret = !zone_spans_pfn(zone, pfn);'?  This format looks fine to me.

> 
> Also, did you spot this by inspection or do you have a test-case or bug
> report?  Should this have a Fixes: tag?

This is from code inspection. The race window should be really small thus hard to trigger
in real world. And yes, it seems Fixes tag is a really ancient commit:

	Fixes: bdc8cb984576 ("[PATCH] memory hotplug locking: zone span seqlock")

Thanks for your comment and reply.

Re: [PATCH] mm: page_alloc: avoid false page outside zone error info

Posted by Andrew Morton 2 years, 7 months ago

On Tue, 4 Jul 2023 20:36:00 +0800 Miaohe Lin <linmiaohe@huawei.com> wrote:

> On 2023/7/4 20:13, Matthew Wilcox wrote:
> > On Tue, Jul 04, 2023 at 07:18:23PM +0800, Miaohe Lin wrote:
> >> @@ -470,6 +470,8 @@ static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
> >>  		sp = zone->spanned_pages;
> >>  		if (!zone_spans_pfn(zone, pfn))
> >>  			ret = 1;
> >> +		else
> >> +			ret = 0;
> > 
> > Surely 'ret = zone_spans_pfn(zone, pfn);' ?
> 
> Do you mean 'ret = !zone_spans_pfn(zone, pfn);'?  This format looks fine to me.
> 
> > 
> > Also, did you spot this by inspection or do you have a test-case or bug
> > report?  Should this have a Fixes: tag?
> 
> This is from code inspection. The race window should be really small thus hard to trigger
> in real world. And yes, it seems Fixes tag is a really ancient commit:
> 
> 	Fixes: bdc8cb984576 ("[PATCH] memory hotplug locking: zone span seqlock")
> 

Thanks.  I updated the changelog:

: If pfn is outside zone boundaries in the first round, ret will be set to
: 1.  But if pfn is changed to inside the zone boundaries in zone span
: seqretry path, ret is still set to 1 leading to false page outside zone
: error info.
: 
: This is from code inspection.  The race window should be really small thus
: hard to trigger in real world.
: 
: Link: https://lkml.kernel.org/r/20230704111823.940331-1-linmiaohe@huawei.com
: Fixes: bdc8cb984576 ("[PATCH] memory hotplug locking: zone span seqlock")

and I made the change suggested by Matthew:

--- a/mm/page_alloc.c~mm-page_alloc-avoid-false-page-outside-zone-error-info-fix
+++ a/mm/page_alloc.c
@@ -468,10 +468,7 @@ static int page_outside_zone_boundaries(
 		seq = zone_span_seqbegin(zone);
 		start_pfn = zone->zone_start_pfn;
 		sp = zone->spanned_pages;
-		if (!zone_spans_pfn(zone, pfn))
-			ret = 1;
-		else
-			ret = 0;
+		ret = !zone_spans_pfn(zone, pfn);
 	} while (zone_span_seqretry(zone, seq));
 
 	if (ret)
_

Re: [PATCH] mm: page_alloc: avoid false page outside zone error info

Posted by Miaohe Lin 2 years, 7 months ago

On 2023/7/5 0:38, Andrew Morton wrote:
> On Tue, 4 Jul 2023 20:36:00 +0800 Miaohe Lin <linmiaohe@huawei.com> wrote:
> 
>> On 2023/7/4 20:13, Matthew Wilcox wrote:
>>> On Tue, Jul 04, 2023 at 07:18:23PM +0800, Miaohe Lin wrote:
>>>> @@ -470,6 +470,8 @@ static int page_outside_zone_boundaries(struct zone *zone, struct page *page)
>>>>  		sp = zone->spanned_pages;
>>>>  		if (!zone_spans_pfn(zone, pfn))
>>>>  			ret = 1;
>>>> +		else
>>>> +			ret = 0;
>>>
>>> Surely 'ret = zone_spans_pfn(zone, pfn);' ?
>>
>> Do you mean 'ret = !zone_spans_pfn(zone, pfn);'?  This format looks fine to me.
>>
>>>
>>> Also, did you spot this by inspection or do you have a test-case or bug
>>> report?  Should this have a Fixes: tag?
>>
>> This is from code inspection. The race window should be really small thus hard to trigger
>> in real world. And yes, it seems Fixes tag is a really ancient commit:
>>
>> 	Fixes: bdc8cb984576 ("[PATCH] memory hotplug locking: zone span seqlock")
>>
> 
> Thanks.  I updated the changelog:
> 
> : If pfn is outside zone boundaries in the first round, ret will be set to
> : 1.  But if pfn is changed to inside the zone boundaries in zone span
> : seqretry path, ret is still set to 1 leading to false page outside zone
> : error info.
> : 
> : This is from code inspection.  The race window should be really small thus
> : hard to trigger in real world.
> : 
> : Link: https://lkml.kernel.org/r/20230704111823.940331-1-linmiaohe@huawei.com
> : Fixes: bdc8cb984576 ("[PATCH] memory hotplug locking: zone span seqlock")
> 
> and I made the change suggested by Matthew:
> 
> --- a/mm/page_alloc.c~mm-page_alloc-avoid-false-page-outside-zone-error-info-fix
> +++ a/mm/page_alloc.c
> @@ -468,10 +468,7 @@ static int page_outside_zone_boundaries(
>  		seq = zone_span_seqbegin(zone);
>  		start_pfn = zone->zone_start_pfn;
>  		sp = zone->spanned_pages;
> -		if (!zone_spans_pfn(zone, pfn))
> -			ret = 1;
> -		else
> -			ret = 0;
> +		ret = !zone_spans_pfn(zone, pfn);
>  	} while (zone_span_seqretry(zone, seq));

These changes look good to me. Thanks for doing this.