On Fri, Apr 01, 2022 at 05:01:14AM -0700, Christoph Hellwig wrote:
> > if (XFS_TEST_ERROR(error, log->l_mp, XFS_ERRTAG_IODONE_IOERR)) {
> > xfs_alert(log->l_mp, "log I/O error %d", error);
> > - xfs_force_shutdown(log->l_mp, SHUTDOWN_LOG_IO_ERROR);
> > + xlog_force_shutdown(log, SHUTDOWN_LOG_IO_ERROR);
>
> The SHUTDOWN_LOG_IO_ERROR as an API to xlog_force_shutdown looks
> really weird now. It s only used to do the xfs_log_force at the
> very beginning of the function. I'd suggest to drop the argument
> and just do that manually in the two callers that do not have
> SHUTDOWN_LOG_IO_ERROR set unconditionally. (This is a perfectly
> fine follow on patch, though).
Yeah, I noticed that as well, but deceided not to do anything about
it as it wasn't directly related to fixing the bugs that I was
tripping over. I think there's a few other cleanups w.r.t. shutdown
error types that can be made, too - I'll get to them when I'm stuck
on hard stuff and need to do somethign realtively easy to keep
things moving...
> Otherwise looks good:
>
> Reviewed-by: Christoph Hellwig <hch@xxxxxx>
Thanks!
-Dave.
--
Dave Chinner
david@xxxxxxxxxxxxx
