Commit 097426f6 authored by NeilBrown's avatar NeilBrown Committed by Linus Torvalds

[PATCH] md: fix possible problem in raid1/raid10 error overwriting

The code to overwrite/reread for addressing read errors in raid1/raid10
currently assumes that the read will not alter the buffer which could be used
to write to the next device.  This is not a safe assumption to make.

So we split the loops into a overwrite loop and a separate re-read loop, so
that the writing is complete before reading is attempted.

Cc: Paul Clements <paul.clements@steeleye.com>
Signed-off-by: default avatarNeil Brown <neilb@suse.de>
Signed-off-by: default avatarAndrew Morton <akpm@osdl.org>
Signed-off-by: default avatarLinus Torvalds <torvalds@osdl.org>
parent 2604b703
...@@ -1253,6 +1253,7 @@ static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio) ...@@ -1253,6 +1253,7 @@ static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio)
} while (!success && d != r1_bio->read_disk); } while (!success && d != r1_bio->read_disk);
if (success) { if (success) {
int start = d;
/* write it back and re-read */ /* write it back and re-read */
set_bit(R1BIO_Uptodate, &r1_bio->state); set_bit(R1BIO_Uptodate, &r1_bio->state);
while (d != r1_bio->read_disk) { while (d != r1_bio->read_disk) {
...@@ -1266,15 +1267,24 @@ static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio) ...@@ -1266,15 +1267,24 @@ static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio)
sect + rdev->data_offset, sect + rdev->data_offset,
s<<9, s<<9,
bio->bi_io_vec[idx].bv_page, bio->bi_io_vec[idx].bv_page,
WRITE) == 0 || WRITE) == 0)
sync_page_io(rdev->bdev, md_error(mddev, rdev);
}
d = start;
while (d != r1_bio->read_disk) {
if (d == 0)
d = conf->raid_disks;
d--;
if (r1_bio->bios[d]->bi_end_io != end_sync_read)
continue;
rdev = conf->mirrors[d].rdev;
if (sync_page_io(rdev->bdev,
sect + rdev->data_offset, sect + rdev->data_offset,
s<<9, s<<9,
bio->bi_io_vec[idx].bv_page, bio->bi_io_vec[idx].bv_page,
READ) == 0) { READ) == 0)
md_error(mddev, rdev); md_error(mddev, rdev);
} }
}
} else { } else {
char b[BDEVNAME_SIZE]; char b[BDEVNAME_SIZE];
/* Cannot read from anywhere, array is toast */ /* Cannot read from anywhere, array is toast */
...@@ -1445,6 +1455,7 @@ static void raid1d(mddev_t *mddev) ...@@ -1445,6 +1455,7 @@ static void raid1d(mddev_t *mddev)
if (success) { if (success) {
/* write it back and re-read */ /* write it back and re-read */
int start = d;
while (d != r1_bio->read_disk) { while (d != r1_bio->read_disk) {
if (d==0) if (d==0)
d = conf->raid_disks; d = conf->raid_disks;
...@@ -1454,14 +1465,25 @@ static void raid1d(mddev_t *mddev) ...@@ -1454,14 +1465,25 @@ static void raid1d(mddev_t *mddev)
test_bit(In_sync, &rdev->flags)) { test_bit(In_sync, &rdev->flags)) {
if (sync_page_io(rdev->bdev, if (sync_page_io(rdev->bdev,
sect + rdev->data_offset, sect + rdev->data_offset,
s<<9, conf->tmppage, WRITE) == 0 || s<<9, conf->tmppage, WRITE) == 0)
sync_page_io(rdev->bdev,
sect + rdev->data_offset,
s<<9, conf->tmppage, READ) == 0) {
/* Well, this device is dead */ /* Well, this device is dead */
md_error(mddev, rdev); md_error(mddev, rdev);
} }
} }
d = start;
while (d != r1_bio->read_disk) {
if (d==0)
d = conf->raid_disks;
d--;
rdev = conf->mirrors[d].rdev;
if (rdev &&
test_bit(In_sync, &rdev->flags)) {
if (sync_page_io(rdev->bdev,
sect + rdev->data_offset,
s<<9, conf->tmppage, READ) == 0)
/* Well, this device is dead */
md_error(mddev, rdev);
}
} }
} else { } else {
/* Cannot read from anywhere -- bye bye array */ /* Cannot read from anywhere -- bye bye array */
......
...@@ -1421,6 +1421,7 @@ static void raid10d(mddev_t *mddev) ...@@ -1421,6 +1421,7 @@ static void raid10d(mddev_t *mddev)
} while (!success && sl != r10_bio->read_slot); } while (!success && sl != r10_bio->read_slot);
if (success) { if (success) {
int start = sl;
/* write it back and re-read */ /* write it back and re-read */
while (sl != r10_bio->read_slot) { while (sl != r10_bio->read_slot) {
int d; int d;
...@@ -1434,16 +1435,29 @@ static void raid10d(mddev_t *mddev) ...@@ -1434,16 +1435,29 @@ static void raid10d(mddev_t *mddev)
if (sync_page_io(rdev->bdev, if (sync_page_io(rdev->bdev,
r10_bio->devs[sl].addr + r10_bio->devs[sl].addr +
sect + rdev->data_offset, sect + rdev->data_offset,
s<<9, conf->tmppage, WRITE) == 0 || s<<9, conf->tmppage, WRITE) == 0)
sync_page_io(rdev->bdev, /* Well, this device is dead */
md_error(mddev, rdev);
}
}
sl = start;
while (sl != r10_bio->read_slot) {
int d;
if (sl==0)
sl = conf->copies;
sl--;
d = r10_bio->devs[sl].devnum;
rdev = conf->mirrors[d].rdev;
if (rdev &&
test_bit(In_sync, &rdev->flags)) {
if (sync_page_io(rdev->bdev,
r10_bio->devs[sl].addr + r10_bio->devs[sl].addr +
sect + rdev->data_offset, sect + rdev->data_offset,
s<<9, conf->tmppage, READ) == 0) { s<<9, conf->tmppage, READ) == 0)
/* Well, this device is dead */ /* Well, this device is dead */
md_error(mddev, rdev); md_error(mddev, rdev);
} }
} }
}
} else { } else {
/* Cannot read from anywhere -- bye bye array */ /* Cannot read from anywhere -- bye bye array */
md_error(mddev, conf->mirrors[r10_bio->devs[r10_bio->read_slot].devnum].rdev); md_error(mddev, conf->mirrors[r10_bio->devs[r10_bio->read_slot].devnum].rdev);
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment