On Wed 20-01-21 16:20:01, Christoph Hellwig wrote:

On Wed, Jan 20, 2021 at 05:06:10PM +0100, Jan Kara wrote:

quoted

Provide an address_space operation for filling pages needed for read
into page cache. Filesystems can use this operation to seriealize
page cache filling with e.g. hole punching properly.

Besides the impending rewrite of the area - having another indirection
here is just horrible for performance.  If we want locking in this area
it should be in core code and common for multiple file systems.

This would mean pulling i_mmap_sem out from ext4/XFS/F2FS private inode
into the VFS inode. Which is fine by me but it would grow struct inode for
proc / tmpfs / btrfs (although for btrfs I'm not convinced it isn't
actually prone to the race and doesn't need similar protection as xfs /
ext4) so some people may object.

								Honza
-- 
Jan Kara [off-list ref]
SUSE Labs, CR