fix ufs write vs readpage race when writing into a hole
authorAl Viro <viro@ZenIV.linux.org.uk>
Wed, 9 Sep 2015 09:16:39 +0000 (10:16 +0100)
committerLinus Torvalds <torvalds@linux-foundation.org>
Wed, 9 Sep 2015 17:43:12 +0000 (10:43 -0700)
commitbd2843fe1f29f6145c623ae9390f144cefaa611d
tree78627367006fa23ea40a53adc0ddb6c528b43005
parentdaf0e1ed578f65e8395102549e135887e6661860
fix ufs write vs readpage race when writing into a hole

Followup to the UFS series - with the way we clear the new blocks (via
buffer cache, possibly on more than a page worth of file) we really
should not insert a reference to new block into inode block tree until
after we'd cleared it.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
fs/ufs/balloc.c