projects / ipfire-2.x.git / blame
| Commit | Line | Data |
|---|---|---|
| 2cb7cef9 BS |
1 | From: NeilBrown <neilb@suse.de> |
| 2 | Subject: Allow read error in single drive in raid1 to be passed up | |
| 3 | Patch-mainline: 2.6.29 | |
| 4 | References: bnc#447835 | |
| 5 | ||
| 6 | If a raid1 only has a single working device and gets a read error, | |
| 7 | we choose to simply return that error up to the filesystem (or whatever) | |
| 8 | rather than failing the whol array. | |
| 9 | ||
| 10 | However the codes doesn't quite do that. We attempt a readbalance | |
| 11 | which allocated the same drive, so we retry the read - indefinitely. | |
| 12 | ||
| 13 | Instead: If read_balance in the error case choose the same drive that just | |
| 14 | failed, treat it as a failure and don't retry. | |
| 15 | ||
| 16 | Signed-off-by: Neil Brown <neilb@suse.de> | |
| 17 | ||
| 18 | --- | |
| 19 | drivers/md/raid1.c | 3 ++- | |
| 20 | 1 file changed, 2 insertions(+), 1 deletion(-) | |
| 21 | ||
| 22 | --- a/drivers/md/raid1.c | |
| 23 | +++ b/drivers/md/raid1.c | |
| 24 | @@ -1637,7 +1637,8 @@ static void raid1d(mddev_t *mddev) | |
| 25 | } | |
| 26 | ||
| 27 | bio = r1_bio->bios[r1_bio->read_disk]; | |
| 28 | - if ((disk=read_balance(conf, r1_bio)) == -1) { | |
| 29 | + if ((disk=read_balance(conf, r1_bio)) == -1 || | |
| 30 | + disk == r1_bio->read_disk) { | |
| 31 | printk(KERN_ALERT "raid1: %s: unrecoverable I/O" | |
| 32 | " read error for block %llu\n", | |
| 33 | bdevname(bio->bi_bdev,b), |