From: NeilBrown <neilb@suse.de>
To: Andrew Morton <akpm@osdl.org>
Cc: linux-raid@vger.kernel.org
Subject: [PATCH md 013 of 18] Improve handing of read errors with raid6
Date: Mon, 28 Nov 2005 10:40:43 +1100 [thread overview]
Message-ID: <1051127234043.14889@suse.de> (raw)
In-Reply-To: 20051128102824.14498.patches@notabene
This is a simple port of match functionality across from raid5.
If we get a read error, we don't kick the drive straight away, but
try to over-write with good data first.
Signed-off-by: Neil Brown <neilb@suse.de>
### Diffstat output
./drivers/md/raid6main.c | 70 ++++++++++++++++++++++++++++++++++++++++++++---
1 file changed, 66 insertions(+), 4 deletions(-)
diff ./drivers/md/raid6main.c~current~ ./drivers/md/raid6main.c
--- ./drivers/md/raid6main.c~current~ 2005-11-28 10:12:56.000000000 +1100
+++ ./drivers/md/raid6main.c 2005-11-28 10:12:59.000000000 +1100
@@ -367,8 +367,8 @@ static void shrink_stripes(raid6_conf_t
conf->slab_cache = NULL;
}
-static int raid6_end_read_request (struct bio * bi, unsigned int bytes_done,
- int error)
+static int raid6_end_read_request(struct bio * bi, unsigned int bytes_done,
+ int error)
{
struct stripe_head *sh = bi->bi_private;
raid6_conf_t *conf = sh->raid_conf;
@@ -420,9 +420,35 @@ static int raid6_end_read_request (struc
#else
set_bit(R5_UPTODATE, &sh->dev[i].flags);
#endif
+ if (test_bit(R5_ReadError, &sh->dev[i].flags)) {
+ printk(KERN_INFO "raid6: read error corrected!!\n");
+ clear_bit(R5_ReadError, &sh->dev[i].flags);
+ clear_bit(R5_ReWrite, &sh->dev[i].flags);
+ }
+ if (atomic_read(&conf->disks[i].rdev->read_errors))
+ atomic_set(&conf->disks[i].rdev->read_errors, 0);
} else {
- md_error(conf->mddev, conf->disks[i].rdev);
+ int retry = 0;
clear_bit(R5_UPTODATE, &sh->dev[i].flags);
+ atomic_inc(&conf->disks[i].rdev->read_errors);
+ if (conf->mddev->degraded)
+ printk(KERN_WARNING "raid6: read error not correctable.\n");
+ else if (test_bit(R5_ReWrite, &sh->dev[i].flags))
+ /* Oh, no!!! */
+ printk(KERN_WARNING "raid6: read error NOT corrected!!\n");
+ else if (atomic_read(&conf->disks[i].rdev->read_errors)
+ > conf->max_nr_stripes)
+ printk(KERN_WARNING
+ "raid6: Too many read errors, failing device.\n");
+ else
+ retry = 1;
+ if (retry)
+ set_bit(R5_ReadError, &sh->dev[i].flags);
+ else {
+ clear_bit(R5_ReadError, &sh->dev[i].flags);
+ clear_bit(R5_ReWrite, &sh->dev[i].flags);
+ md_error(conf->mddev, conf->disks[i].rdev);
+ }
}
rdev_dec_pending(conf->disks[i].rdev, conf->mddev);
#if 0
@@ -1079,6 +1105,12 @@ static void handle_stripe(struct stripe_
if (dev->written) written++;
rdev = conf->disks[i].rdev; /* FIXME, should I be looking rdev */
if (!rdev || !test_bit(In_sync, &rdev->flags)) {
+ /* The ReadError flag will just be confusing now */
+ clear_bit(R5_ReadError, &dev->flags);
+ clear_bit(R5_ReWrite, &dev->flags);
+ }
+ if (!rdev || !test_bit(In_sync, &rdev->flags)
+ || test_bit(R5_ReadError, &dev->flags)) {
if ( failed < 2 )
failed_num[failed] = i;
failed++;
@@ -1095,6 +1127,14 @@ static void handle_stripe(struct stripe_
if (failed > 2 && to_read+to_write+written) {
for (i=disks; i--; ) {
int bitmap_end = 0;
+
+ if (test_bit(R5_ReadError, &sh->dev[i].flags)) {
+ mdk_rdev_t *rdev = conf->disks[i].rdev;
+ if (rdev && test_bit(In_sync, &rdev->flags))
+ /* multiple read failures in one stripe */
+ md_error(conf->mddev, rdev);
+ }
+
spin_lock_irq(&conf->device_lock);
/* fail all writes first */
bi = sh->dev[i].towrite;
@@ -1130,7 +1170,8 @@ static void handle_stripe(struct stripe_
}
/* fail any reads if this device is non-operational */
- if (!test_bit(R5_Insync, &sh->dev[i].flags)) {
+ if (!test_bit(R5_Insync, &sh->dev[i].flags) ||
+ test_bit(R5_ReadError, &sh->dev[i].flags)) {
bi = sh->dev[i].toread;
sh->dev[i].toread = NULL;
if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
@@ -1457,6 +1498,27 @@ static void handle_stripe(struct stripe_
clear_bit(STRIPE_SYNCING, &sh->state);
}
+ /* If the failed drives are just a ReadError, then we might need
+ * to progress the repair/check process
+ */
+ if (failed <= 2 && ! conf->mddev->ro)
+ for (i=0; i<failed;i++) {
+ dev = &sh->dev[failed_num[i]];
+ if (test_bit(R5_ReadError, &dev->flags)
+ && !test_bit(R5_LOCKED, &dev->flags)
+ && test_bit(R5_UPTODATE, &dev->flags)
+ ) {
+ if (!test_bit(R5_ReWrite, &dev->flags)) {
+ set_bit(R5_Wantwrite, &dev->flags);
+ set_bit(R5_ReWrite, &dev->flags);
+ set_bit(R5_LOCKED, &dev->flags);
+ } else {
+ /* let's read it back */
+ set_bit(R5_Wantread, &dev->flags);
+ set_bit(R5_LOCKED, &dev->flags);
+ }
+ }
+ }
spin_unlock(&sh->lock);
while ((bi=return_bi)) {
next prev parent reply other threads:[~2005-11-27 23:40 UTC|newest]
Thread overview: 33+ messages / expand[flat|nested] mbox.gz Atom feed top
2005-11-27 23:39 [PATCH md 000 of 18] Introduction NeilBrown
2005-11-27 23:39 ` [PATCH md 001 of 18] Improve read speed to raid10 arrays using 'far copies' NeilBrown
2005-11-27 23:39 ` [PATCH md 002 of 18] Fix locking problem in r5/r6 NeilBrown
2005-11-27 23:39 ` [PATCH md 003 of 18] Fix problem with raid6 intent bitmap NeilBrown
2005-11-27 23:39 ` [PATCH md 004 of 18] Set default_bitmap_offset properly in set_array_info NeilBrown
2005-11-27 23:40 ` [PATCH md 005 of 18] Fix --re-add for raid1 and raid6 NeilBrown
2005-11-27 23:40 ` [PATCH md 006 of 18] Improve raid1 "IO Barrier" concept NeilBrown
2005-11-27 23:40 ` [PATCH md 007 of 18] Improve raid10 " NeilBrown
2005-11-27 23:40 ` [PATCH md 008 of 18] Small cleanups for raid5 NeilBrown
2005-11-27 23:40 ` [PATCH md 010 of 18] Move bitmap_create to after md array has been initialised NeilBrown
2005-11-27 23:40 ` [PATCH md 011 of 18] Write intent bitmap support for raid10 NeilBrown
2005-11-27 23:40 ` [PATCH md 012 of 18] Fix raid6 resync check/repair code NeilBrown
2005-11-27 23:40 ` NeilBrown [this message]
2005-11-30 22:33 ` [PATCH md 013 of 18] Improve handing of read errors with raid6 Carlos Carvalho
2005-12-01 2:54 ` Neil Brown
2005-11-27 23:40 ` [PATCH md 014 of 18] Attempt to auto-correct read errors in raid1 NeilBrown
2005-11-29 16:38 ` Paul Clements
2005-11-29 23:21 ` Neil Brown
2005-11-27 23:40 ` [PATCH md 015 of 18] Tidyup some issues with raid1 resync and prepare for catching read errors NeilBrown
2005-11-27 23:40 ` [PATCH md 016 of 18] Better handling for read error in raid1 during resync NeilBrown
2005-11-27 23:41 ` [PATCH md 017 of 18] Handle errors when read-only NeilBrown
2005-12-10 6:41 ` Yanggun
2005-12-10 6:59 ` raid1 mysteriously switching to read-only Neil Brown
2005-12-10 7:50 ` Yanggun
2005-12-10 8:02 ` Neil Brown
2005-12-10 8:10 ` Yanggun
2005-12-10 12:10 ` Neil Brown
2005-12-11 13:04 ` Yanggun
2005-12-11 14:14 ` Patrik Jonsson
2005-12-11 14:29 ` Yanggun
2005-12-11 17:13 ` Ross Vandegrift
2005-12-11 23:28 ` Yanggun
2005-11-27 23:41 ` [PATCH md 018 of 18] Fix up some rdev rcu locking in raid5/6 NeilBrown
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1051127234043.14889@suse.de \
--to=neilb@suse.de \
--cc=akpm@osdl.org \
--cc=linux-raid@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).