[dm-devel] [2.6.22-rc4-mm2 PATCH 8/11] dm-raid1-handle-recovery-read-failures.patch

[Jonathan Brassow]

add header!

 brassow

This patch adds the ability detect read failures during
recovery and flag the device as failed.

We add a call to fail_mirror() when 'read_err' is set
upon recovery completion.  This sets the error_count
on the device.  It maintains backwards compatibility
(does nothing) if the DM_RAID1_HANDLE_ERRORS flag is
not set during mirror device creation.

Using the same code put in place to handle write failures,
it ensures that the region is marked as 'not-in-sync'
and that this is reflected in the log.

Signed-off-by: Jonathan Brassow <jbrassow at redhat.com>

Index: linux-2.6.22-rc4-mm2/drivers/md/dm-raid1.c
===================================================================
--- linux-2.6.22-rc4-mm2.orig/drivers/md/dm-raid1.c
+++ linux-2.6.22-rc4-mm2/drivers/md/dm-raid1.c
@@ -673,9 +673,11 @@ static void recovery_complete(int read_e
 	unsigned long write_err_ulong = (unsigned long)write_err;
 	int m, bit = 0;
 
-	if (read_err)
+	if (read_err) {
 		/* Read error means the failure of default mirror. */
 		DMERR_LIMIT("Unable to read primary mirror during recovery");
+		fail_mirror(ms->default_mirror);
+	}
 
 	if (!write_err)
 		goto out;