On Tue, Jul 16, 2024 at 2:20 PM Gao Xiang <hsiangkao@xxxxxxxxxxxxxxxxx> wrote:
emm, that's the original design of down_read()Sorry, I should provide more details. page_cache_ra_unbounded() will
I don't see why it's an atomic context,
so this patch is incorrect.
call filemap_invalidate_lock_shared(mapping) to ensure the integrity
of page cache during readahead, which will disable preempt.
Why a rwsem sleepable lock disable preemption?
context should be always non-atomic context, which is appliedAFAICT, filemap_fault/read have added the folios of readahead to page
to all kernel filesystems.
cache which means the aops->readahead basically just need to map the
block to this folios and then launch the bio. The erofs is a little
bit different to others as it has to alloc_pages for decompression
when doing this.