lkml.org 
[lkml]   [2022]   [May]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
SubjectRe: [PATCH v4] mm: fix is_pinnable_page against on cma page
From
On 5/23/22 10:16 PM, Minchan Kim wrote:
> On Mon, May 23, 2022 at 07:55:25PM -0700, John Hubbard wrote:
>> On 5/23/22 09:33, Minchan Kim wrote:
>> ...
>>>> So then:
>>>>
>>>> diff --git a/mm/page_alloc.c b/mm/page_alloc.c
>>>> index 0e42038382c1..b404f87e2682 100644
>>>> --- a/mm/page_alloc.c
>>>> +++ b/mm/page_alloc.c
>>>> @@ -482,7 +482,12 @@ unsigned long __get_pfnblock_flags_mask(const struct page *page,
>>>> word_bitidx = bitidx / BITS_PER_LONG;
>>>> bitidx &= (BITS_PER_LONG-1);
>>>>
>>>> - word = bitmap[word_bitidx];
>>>> + /*
>>>> + * This races, without locks, with set_pageblock_migratetype(). Ensure
>>> set_pfnblock_flags_mask would be better?
>>>> + * a consistent (non-tearing) read of the memory array, so that results,
>>>
>>> Thanks for proceeding and suggestion, John.
>>>
>>> IIUC, the load tearing wouldn't be an issue since [1] fixed the issue.
>>
>> Did it? [1] fixed something, but I'm not sure we can claim that that
>> code is now safe against tearing in all possible cases, especially given
>> the recent discussion here. Specifically, having this code do a read,
>> then follow that up with calculations, seems correct. Anything else is
>
> The load tearing you are trying to explain in the comment would be
> solved by [1] since the bits will always align on a word and accessing
> word size based on word aligned address is always atomic so there is
> no load tearing problem IIUC.
>
> Instead of the tearing problem, what we are trying to solve with
> READ_ONCE is to prevent refetching when the function would be
> inlined in the future.
>

I'm perhaps using "tearing" as too broad of a term, maybe just removing
the "(non-tearing)" part would fix up the comment.

>> sketchy...
>>
>>>
>>> The concern in our dicussion was aggressive compiler(e.g., LTO) or code refactoring
>>> to make the code inline in *future* could potentially cause forcing refetching(i.e.,
>>> re-read) tie bitmap[word_bitidx].
>>>
>>> If so, shouldn't the comment be the one you helped before?
>>
>> Well, maybe updated to something like this?
>>
>> /*
>> * This races, without locks, with set_pageblock_migratetype(). Ensure
>
> set_pageblock_migratetype is more upper level function so it would
> be better fit to say set_pfnblock_flags_mask.

OK

>
>> * a consistent (non-tearing) read of the memory array, so that results,
>
> So tearing problem should't already happen by [1] so I am trying to
> explain refetching(or re-read) problem in the comment.
>
>> * even though racy, are not corrupted--even if this function is
>
> The value is already atomic so I don't think it could be corrupted
> even though it would be inlined in the future.
>
> Please correct me if I miss something.
>
>> * refactored and/or inlined.
>> */
>

thanks,

--
John Hubbard
NVIDIA

\
 
 \ /
  Last update: 2022-05-24 08:23    [W:0.096 / U:1.920 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site