Re: [PATCH] btrfs: write-protect folios during data writeback

在 2026/7/10 02:48, Boris Burkov 写道:
> On Thu, Jul 09, 2026 at 06:16:55PM +0930, Qu Wenruo wrote:
[...]
>>
>> In that case, as I mentioned in another thread, I'd prefer not to call
>> folio_mkclean() too early.
>>
>> I'd prefer to call folio_mkclean() inside extent_range_clear_dirty_for_io()
>> instead.
> 
> The idea is the zoned folios submitted by run_delalloc_range with
> pages_dirty=true. Only the one main folio will be cleaned in
> extent_writepage but several get submitted in this loop while locked
> (which is asserted by the pages_dirty check).
> 
> So assuming we want to clean those folios (which I think we do)
> then we need:
> extent_writepage (main one covers usual large folios and inline)
> extent_range_clear_dirty_for_io (covers compression stability)
> extent_write_locked_range pages_dirty = true (zoned batch)
> 
> lock_delalloc_folios() covers the last two in a single more
> principled way, IMO.

Oh I forgot zoned cases. In that case indeed lock_delalloc_folios() is 
the location that covers all.

Then the code looks good to me now.

Thanks,
Qu

> 
> I am still making sure all of this reproduces exactly as I understand
> it, though, to be as convincing as possible :)
>