Hole punch races in CIFS

Hole punch races in CIFS

[Jan Kara]

Hello,

I was looking into what protects CIFS from races between hole punching and
operations like page fault or readahead and AFAICT there's nothing to
prevent a race like:

CPU1						CPU2
smb3_fallocate()
  smb3_punch_hole()
    truncate_pagecache_range()
						filemap_fault()
						  - loads old data into the
						    page cache
    SMB2_ioctl(..., FSCTL_SET_ZERO_DATA, ...)

And now we have stale data in the page cache and if the page gets later
dirtied and written out, even persistent data corruption. Is there anything
I'm missing?

								Honza
-- 
Jan Kara <jack@suse.com>
SUSE Labs, CR