Re: [RFC 2/6] mm/migrate_pages: split unmap_and_move() to _unmap() and _move()

[John Hubbard <jhubbard@nvidia.com>]

On 9/27/22 18:49, Yang Shi wrote:
> On Tue, Sep 27, 2022 at 6:45 PM John Hubbard <jhubbard@nvidia.com> wrote:
>>
>> On 9/27/22 18:41, Huang, Ying wrote:
>>>>>> I also agree that we cannot make any rules such as "do not lock > 1 page
>>>>>> at the same time, elsewhere in the kernel", because it is already
>>>>>> happening, for example in page-writeback.c, which locks PAGEVEC_SIZE
>>>>>> (15) pages per batch [1].
>>>>
>>>> That's not really the case though. The inner loop of write_cache_page()
>>>> only ever locks one page at a time, either directly via the
>>>> unlock_page() on L2338 (those goto's are amazing) or indirectly via
>>>> (*writepage)() on L2359.
>>>>
>>>> So there's no deadlock potential there because unlocking any previously
>>>> locked page(s) doesn't depend on obtaining the lock for another page.
>>>> Unless I've missed something?
>>>
>>> Yes.  This is my understanding too after checking ext4_writepage().
>>>
>>
>> Yes, I missed the ".writepage() shall unlock the page" design point. Now
>> it seems much more reasonable and safer. :)
> 
> .writepage is deprecated (see
> https://lore.kernel.org/linux-fsdevel/20220719041311.709250-1-hch@lst.de/),
> write back actually uses .writepages.

write_cache_pages() seems to directly call it, though:

generic_writepages()
  write_cache_pages(__writepage)
    __writepage()
      mapping->a_ops->writepage(page, wbc)

So it seems like it's still alive and well. And in any case, it is definitely
passing one page at a time from write_cache_pages(), right?


 thanks,

-- 
John Hubbard
NVIDIA