Lars Marowsky-Bree <lmb@xxxxxxx> wrote:
>
> READA errors failing with EWOULDBLOCK/EAGAIN do not constitute a valid
> reason for failing the path; this lead to erratic errors on DM multipath
> devices. This error can be safely propagated upwards without failing the
> path.
>
Why do you describe this as a -mm patch? We want this in 2.6.12, no?
> --- linux-2.6.12-rc6-mm1.orig/drivers/md/dm-mpath.c 2005-06-08 12:51:02.741055000 +0200
> +++ linux-2.6.12-rc6-mm1/drivers/md/dm-mpath.c 2005-06-08 12:57:55.757828867 +0200
> @@ -985,6 +985,9 @@
> if (!error)
> return 0; /* I/O complete */
>
> + if ((error == -EWOULDBLOCK) && bio_rw_ahead(bio))
> + return error;
> +
> spin_lock(&m->lock);
> if (!m->nr_valid_paths) {
> if (!m->queue_if_no_path || m->suspended) {
>
>
> --
> High Availability & Clustering
> SUSE Labs, Research and Development
> SUSE LINUX Products GmbH - A Novell Business -- Charles Darwin
> "Ignorance more frequently begets confidence than does knowledge"
