I think this was an oversight and the rh_in_sync call should be allowed to block in do_reads. Any reason why it is not coded that way? brassow do_reads happens in the mirror worker thread and operations there are allowed to block. The call to rh_in_sync should be allowed to block. This is especially important for cluster mirroring. Since the cluster log must ensure the region is in_sync cluster-wide and must do network operations to ensure that, it will always return -EWOULDBLOCK if 0 is given for the blocking argument. The way the code is written, a cluster mirror would never be allowed to 'choose_mirror'. Index: linux-2.6.21-rc5-mm4/drivers/md/dm-raid1.c =================================================================== --- linux-2.6.21-rc5-mm4.orig/drivers/md/dm-raid1.c +++ linux-2.6.21-rc5-mm4/drivers/md/dm-raid1.c @@ -741,7 +741,7 @@ static void do_reads(struct mirror_set * /* * We can only read balance if the region is in sync. */ - if (rh_in_sync(&ms->rh, region, 0)) + if (rh_in_sync(&ms->rh, region, 1)) m = choose_mirror(ms, bio->bi_sector); else m = ms->default_mirror; -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
