Re: [PATCH v2 2/2] mm, sl[aou]b: guarantee natural alignment for kmalloc(power-of-two)

[Vlastimil Babka]

On 9/4/19 8:40 AM, Ming Lei wrote:
> On Wed, Sep 04, 2019 at 07:19:33AM +0200, Christoph Hellwig wrote:
>> On Tue, Sep 03, 2019 at 01:53:12PM -0700, Matthew Wilcox wrote:
>>>> Its enabled in all full debug session as far as I know. Fedora for
>>>> example has been running this for ages to find breakage in device drivers
>>>> etc etc.
>>>
>>> Are you telling me nobody uses the ramdisk driver on fedora?  Because
>>> that's one of the affected drivers.
>>
>> For pmem/brd misaligned memory alone doesn't seem to be the problem.
>> Misaligned memory that cross a page barrier is.  And at least XFS
>> before my log recovery changes only used kmalloc for smaller than
>> page size allocation, so this case probably didn't hit.
> 
> BTW, does sl[aou]b guarantee that smaller than page size allocation via kmalloc()
> won't cross page boundary any time?

AFAIK no, it's the same as with the alignment. After this patch, that
guarantee would follow from the alignment one for power of two sizes,
but sizes 96 and 192 could still cross page boundary.

> Thanks,
> Ming
>