Re: [PATCH] Revert "mm/page_isolation: unset migratetype directly for non Buddy page"

Re: [PATCH] Revert "mm/page_isolation: unset migratetype directly for non Buddy page"

[Vlastimil Babka]

On 1/17/22 15:27, Chen Wandun wrote:
> This reverts commit 075782149abff45ee22f27315eced44d02b96779.
> 
> commit 075782149abf ("mm/page_isolation: unset migratetype directly for

That's an old -next commit id, it went to mainline as 721fb891ad0b
("mm/page_isolation: unset migratetype directly for non Buddy page")

> non Buddy page") will result memory that should in buddy disappear by
> mistake. move_freepages_block move all pages in pageblock instead of
> pages indicated by input parameter, so if input pages is not in buddy
> but other pages in pageblock is in buddy, it will result in page out of
> control.
> 
> Reported-by: "kernelci.org bot" <bot@kernelci.org>
> Signed-off-by: Chen Wandun <chenwandun@huawei.com>

Acked-by: Vlastimil Babka <vbabka@suse.cz>

#regzbot ^introduced: 721fb891ad0b

There should better be also:

Fixes: 721fb891ad0b ("mm/page_isolation: unset migratetype directly for non
Buddy page")

Especially as that commit had a (inadequate, IMHO, for an optimization)
Fixes: tag and could end up being backported somewhere without the followup
fix/revert.

> ---
>  mm/page_isolation.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/mm/page_isolation.c b/mm/page_isolation.c
> index 6a0ddda6b3c5..f67c4c70f17f 100644
> --- a/mm/page_isolation.c
> +++ b/mm/page_isolation.c
> @@ -115,7 +115,7 @@ static void unset_migratetype_isolate(struct page *page, unsigned migratetype)
>  	 * onlining - just onlined memory won't immediately be considered for
>  	 * allocation.
>  	 */
> -	if (!isolated_page && PageBuddy(page)) {
> +	if (!isolated_page) {
>  		nr_pages = move_freepages_block(zone, page, migratetype, NULL);
>  		__mod_zone_freepage_state(zone, nr_pages, migratetype);
>  	}

Re: [PATCH] Revert "mm/page_isolation: unset migratetype directly for non Buddy page"

[Chen Wandun]

在 2022/1/26 0:40, Vlastimil Babka 写道:
> On 1/17/22 15:27, Chen Wandun wrote:
>> This reverts commit 075782149abff45ee22f27315eced44d02b96779.
>>
>> commit 075782149abf ("mm/page_isolation: unset migratetype directly for
> That's an old -next commit id, it went to mainline as 721fb891ad0b
> ("mm/page_isolation: unset migratetype directly for non Buddy page")
>
>> non Buddy page") will result memory that should in buddy disappear by
>> mistake. move_freepages_block move all pages in pageblock instead of
>> pages indicated by input parameter, so if input pages is not in buddy
>> but other pages in pageblock is in buddy, it will result in page out of
>> control.
>>
>> Reported-by: "kernelci.org bot" <bot@kernelci.org>
>> Signed-off-by: Chen Wandun <chenwandun@huawei.com>
> Acked-by: Vlastimil Babka <vbabka@suse.cz>
>
> #regzbot ^introduced: 721fb891ad0b
>
> There should better be also:
>
> Fixes: 721fb891ad0b ("mm/page_isolation: unset migratetype directly for non
> Buddy page")
>
> Especially as that commit had a (inadequate, IMHO, for an optimization)
> Fixes: tag and could end up being backported somewhere without the followup
> fix/revert.
thanks, I will send v2
>
>> ---
>>   mm/page_isolation.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/mm/page_isolation.c b/mm/page_isolation.c
>> index 6a0ddda6b3c5..f67c4c70f17f 100644
>> --- a/mm/page_isolation.c
>> +++ b/mm/page_isolation.c
>> @@ -115,7 +115,7 @@ static void unset_migratetype_isolate(struct page *page, unsigned migratetype)
>>   	 * onlining - just onlined memory won't immediately be considered for
>>   	 * allocation.
>>   	 */
>> -	if (!isolated_page && PageBuddy(page)) {
>> +	if (!isolated_page) {
>>   		nr_pages = move_freepages_block(zone, page, migratetype, NULL);
>>   		__mod_zone_freepage_state(zone, nr_pages, migratetype);
>>   	}
> .

Re: [PATCH] Revert "mm/page_isolation: unset migratetype directly for non Buddy page"

[Thorsten Leemhuis]

Hi, this is your Linux kernel regression tracker speaking.

On 26.01.22 03:18, Chen Wandun wrote:
> 在 2022/1/26 0:40, Vlastimil Babka 写道:
>> On 1/17/22 15:27, Chen Wandun wrote:
>>> This reverts commit 075782149abff45ee22f27315eced44d02b96779.
>>>
>>> commit 075782149abf ("mm/page_isolation: unset migratetype directly for
>> That's an old -next commit id, it went to mainline as 721fb891ad0b
>> ("mm/page_isolation: unset migratetype directly for non Buddy page")
>>
>>> non Buddy page") will result memory that should in buddy disappear by
>>> mistake. move_freepages_block move all pages in pageblock instead of
>>> pages indicated by input parameter, so if input pages is not in buddy
>>> but other pages in pageblock is in buddy, it will result in page out of
>>> control.
>>>
>>> Reported-by: "kernelci.org bot" <bot@kernelci.org>
>>> Signed-off-by: Chen Wandun <chenwandun@huawei.com>
>> Acked-by: Vlastimil Babka <vbabka@suse.cz>
>>
>> #regzbot ^introduced: 721fb891ad0b

Thx for getting regzbot involved, much appreciated.

>> There should better be also:
>>
>> Fixes: 721fb891ad0b ("mm/page_isolation: unset migratetype directly
>> for non
>> Buddy page")
>>
>> Especially as that commit had a (inadequate, IMHO, for an optimization)
>> Fixes: tag and could end up being backported somewhere without the
>> followup
>> fix/revert.
>
> thanks, I will send v2

Just wondering:

Andrew, I don't see this patch in the list of changes you sent to Linus
yesterday. Shouldn't it be in there due to above reason alone? And there
is another reason: it seems to be an issue people actually hit, as
Guenter mentioned in this thread. It thus complicates things for those
that want to help testing pre-releases.

Or am I missing something/asking too much in this case?

BTW, let's tell regzbot about v2:

#regzbot monitor:
https://lore.kernel.org/all/20220126024436.13921-1-chenwandun@huawei.com/

Ciao, Thorsten

>>> ---
>>>   mm/page_isolation.c | 2 +-
>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> diff --git a/mm/page_isolation.c b/mm/page_isolation.c
>>> index 6a0ddda6b3c5..f67c4c70f17f 100644
>>> --- a/mm/page_isolation.c
>>> +++ b/mm/page_isolation.c
>>> @@ -115,7 +115,7 @@ static void unset_migratetype_isolate(struct page
>>> *page, unsigned migratetype)
>>>        * onlining - just onlined memory won't immediately be
>>> considered for
>>>        * allocation.
>>>        */
>>> -    if (!isolated_page && PageBuddy(page)) {
>>> +    if (!isolated_page) {
>>>           nr_pages = move_freepages_block(zone, page, migratetype,
>>> NULL);
>>>           __mod_zone_freepage_state(zone, nr_pages, migratetype);
>>>       }
>> .
> 
> 
> 

Re: [PATCH] Revert "mm/page_isolation: unset migratetype directly for non Buddy page"

[Andrew Morton]

On Sun, 30 Jan 2022 10:24:43 +0100 Thorsten Leemhuis <regressions@leemhuis.info> wrote:

> Andrew, I don't see this patch in the list of changes you sent to Linus
> yesterday. Shouldn't it be in there due to above reason alone? And there
> is another reason: it seems to be an issue people actually hit, as
> Guenter mentioned in this thread. It thus complicates things for those
> that want to help testing pre-releases.

I prefer to give fixes some time in linux-next before upstreaming them.
At least a day, worst case.  So I have this patch in this week's pile.

Re: [PATCH] Revert "mm/page_isolation: unset migratetype directly for non Buddy page"

[Thorsten Leemhuis]

For the record:

#regzbot fixed-by: a85468b766d3bc17c8b17ed23a36ef6469340bb2

On 30.01.22 10:24, Thorsten Leemhuis wrote:
> Hi, this is your Linux kernel regression tracker speaking.
> 
> On 26.01.22 03:18, Chen Wandun wrote:
>> 在 2022/1/26 0:40, Vlastimil Babka 写道:
>>> On 1/17/22 15:27, Chen Wandun wrote:
>>>> This reverts commit 075782149abff45ee22f27315eced44d02b96779.
>>>>
>>>> commit 075782149abf ("mm/page_isolation: unset migratetype directly for
>>> That's an old -next commit id, it went to mainline as 721fb891ad0b
>>> ("mm/page_isolation: unset migratetype directly for non Buddy page")
>>>
>>>> non Buddy page") will result memory that should in buddy disappear by
>>>> mistake. move_freepages_block move all pages in pageblock instead of
>>>> pages indicated by input parameter, so if input pages is not in buddy
>>>> but other pages in pageblock is in buddy, it will result in page out of
>>>> control.
>>>>
>>>> Reported-by: "kernelci.org bot" <bot@kernelci.org>
>>>> Signed-off-by: Chen Wandun <chenwandun@huawei.com>
>>> Acked-by: Vlastimil Babka <vbabka@suse.cz>
>>>
>>> #regzbot ^introduced: 721fb891ad0b
> 
> Thx for getting regzbot involved, much appreciated.
> 
>>> There should better be also:
>>>
>>> Fixes: 721fb891ad0b ("mm/page_isolation: unset migratetype directly
>>> for non
>>> Buddy page")
>>>
>>> Especially as that commit had a (inadequate, IMHO, for an optimization)
>>> Fixes: tag and could end up being backported somewhere without the
>>> followup
>>> fix/revert.
>>
>> thanks, I will send v2
> 
> Just wondering:
> 
> Andrew, I don't see this patch in the list of changes you sent to Linus
> yesterday. Shouldn't it be in there due to above reason alone? And there
> is another reason: it seems to be an issue people actually hit, as
> Guenter mentioned in this thread. It thus complicates things for those
> that want to help testing pre-releases.
> 
> Or am I missing something/asking too much in this case?
> 
> BTW, let's tell regzbot about v2:
> 
> #regzbot monitor:
> https://lore.kernel.org/all/20220126024436.13921-1-chenwandun@huawei.com/
> 
> Ciao, Thorsten
> 
>>>> ---
>>>>   mm/page_isolation.c | 2 +-
>>>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>>>
>>>> diff --git a/mm/page_isolation.c b/mm/page_isolation.c
>>>> index 6a0ddda6b3c5..f67c4c70f17f 100644
>>>> --- a/mm/page_isolation.c
>>>> +++ b/mm/page_isolation.c
>>>> @@ -115,7 +115,7 @@ static void unset_migratetype_isolate(struct page
>>>> *page, unsigned migratetype)
>>>>        * onlining - just onlined memory won't immediately be
>>>> considered for
>>>>        * allocation.
>>>>        */
>>>> -    if (!isolated_page && PageBuddy(page)) {
>>>> +    if (!isolated_page) {
>>>>           nr_pages = move_freepages_block(zone, page, migratetype,
>>>> NULL);
>>>>           __mod_zone_freepage_state(zone, nr_pages, migratetype);
>>>>       }
>>> .
>>
>>
>>
>