On Thu, Jun 23, 2022 at 01:23:02PM -0700, Stefan Roesch wrote:
>
>
> On 6/23/22 1:18 PM, Darrick J. Wong wrote:
> > On Thu, Jun 23, 2022 at 10:51:49AM -0700, Stefan Roesch wrote:
> >> If iomap_write_iter() encounters -EAGAIN, return -EAGAIN to the caller.
> >>
> >> Signed-off-by: Stefan Roesch <shr@xxxxxx>
> >> ---
> >> fs/iomap/buffered-io.c | 8 +++++++-
> >> 1 file changed, 7 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/fs/iomap/buffered-io.c b/fs/iomap/buffered-io.c
> >> index 83cf093fcb92..f2e36240079f 100644
> >> --- a/fs/iomap/buffered-io.c
> >> +++ b/fs/iomap/buffered-io.c
> >> @@ -830,7 +830,13 @@ static loff_t iomap_write_iter(struct iomap_iter *iter, struct iov_iter *i)
> >> length -= status;
> >> } while (iov_iter_count(i) && length);
> >>
> >> - return written ? written : status;
> >> + if (status == -EAGAIN) {
> >> + iov_iter_revert(i, written);
> >> + return -EAGAIN;
> >> + }
> >> + if (written)
> >> + return written;
> >> + return status;
> >
> > Any particular reason for decomposing the ternary into this? It still
> > looks correct, but it doesn't seem totally necessary...
> >
>
> Do you prefer this version?
>
> + if (status == -EAGAIN) {
> + iov_iter_revert(i, written);
> + return -EAGAIN;
> + }
> return written ? written : status;
Yes, because it /does/ make it a lot more obvious that the only change
is intercepting EAGAIN to rewind the iov_iter. :)
--D
>
>
> > Reviewed-by: Darrick J. Wong <djwong@xxxxxxxxxx>
> >
> > --D
> >
> >> }
> >>
> >> ssize_t
> >> --
> >> 2.30.2
> >>
