On 1:50 12/12, Christoph Hellwig wrote:
> On Wed, Dec 11, 2019 at 06:30:39PM -0600, Goldwyn Rodrigues wrote:
> > From: Goldwyn Rodrigues <rgoldwyn@xxxxxxxx>
> >
> > Filesystems such as btrfs can perform direct I/O without holding the
> > inode->i_rwsem in some of the cases like writing within i_size.
>
> How is that safe?
This (inode_lock release) is only done for writes within i_size.
We only have to safeguard write against truncates, which is done by
inode_dio_wait() call in the truncate sequence (I had mistakenly removed
it in patch 8/8, I shall reinstate that). The commit that introduced this
optimization is 38851cc19adb ("Btrfs: implement unlocked dio write")
>
> > + lockdep_assert_held(&file_inode(file)->i_rwsem);
>
> Having the asserts in the callers is pointless. The assert is inside
> the iomap helper to ensure the expected calling conventions, as the
> code is written under the assumption that we have i_rwsem.
Hmm, conflicting opinions from you and Dave. Anyways, I have removed it
in individual filesystems.
--
Goldwyn
