NFS: swap IO handling is slightly different for O_DIRECT IO
authorNeilBrown <neilb@suse.de>
Sun, 6 Mar 2022 23:41:44 +0000 (10:41 +1100)
committerTrond Myklebust <trond.myklebust@hammerspace.com>
Sun, 13 Mar 2022 16:59:35 +0000 (12:59 -0400)
commit64158668ac8b31626a8ce48db4cad08496eb8340
tree7c2704c41a6f575b5a02dc8ef75479a8439fe6a6
parent4dc73c679114a2f408567e2e44770ed934190db2
NFS: swap IO handling is slightly different for O_DIRECT IO

1/ Taking the i_rwsem for swap IO triggers lockdep warnings regarding
   possible deadlocks with "fs_reclaim".  These deadlocks could, I believe,
   eventuate if a buffered read on the swapfile was attempted.

   We don't need coherence with the page cache for a swap file, and
   buffered writes are forbidden anyway.  There is no other need for
   i_rwsem during direct IO.  So never take it for swap_rw()

2/ generic_write_checks() explicitly forbids writes to swap, and
   performs checks that are not needed for swap.  So bypass it
   for swap_rw().

Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
fs/nfs/direct.c
fs/nfs/file.c
include/linux/nfs_fs.h