[PATCH v2] MD: raid, fix BUG caused by flags handling

From: Jiri Slaby
Date: Thu Aug 12 2010 - 08:31:53 EST


Commit 74450be1 (block: unify flags for struct bio and struct request)
added direct test of flags in the & form:
const bool do_sync = (bio->bi_rw & REQ_SYNC);
But this doesn't fit into bool with my compiler (gcc 4.5). So change
the type to ulong to avoid the bug.

The BUG looks like:
EXT3-fs (md1): using internal journal
------------[ cut here ]------------
kernel BUG at drivers/scsi/scsi_lib.c:1113!
...
Pid: 879, comm: md3_raid1 Tainted: G W 2.6.35-rc5-mm1_64+ #1265
RIP: 0010:[<ffffffff813312d6>] [<ffffffff813312d6>] scsi_setup_fs_cmnd+0x96/0xd0
Process md3_raid1 (pid: 879, threadinfo ffff8801c4716000, task ffff8801cbd5a6a0)
...
Call Trace:
[<ffffffff81339a78>] sd_prep_fn+0xa8/0x800
[<ffffffff812252e9>] ? cfq_dispatch_request+0x49/0xb0
[<ffffffff8121a24a>] blk_peek_request+0xca/0x1a0
[<ffffffff81330a56>] scsi_request_fn+0x56/0x400
[<ffffffff81219dad>] __generic_unplug_device+0x2d/0x40
[<ffffffff8121a059>] generic_unplug_device+0x29/0x40
[<ffffffff81217682>] blk_unplug+0x12/0x20
[<ffffffff813f14c8>] unplug_slaves+0x78/0xc0
[<ffffffff813f3ecb>] raid1d+0x37b/0x420
[<ffffffff813fb6e3>] md_thread+0x53/0x120
...
Code: e8 d0 fe ff ff 5b 41 5c c9 c3 0f 1f 00 4c 89 e7 be 20 00 00 00 e8 db 9f ff ff 48 89 c7 48 85 c0 74 35 48 89 83 c8 00 00 00 eb a3 <0f> 0b 48 8b 00 48 85 c0 74 83 48 8b 40 48 48 85 c0 0f 84 76 ff
RIP [<ffffffff813312d6>] scsi_setup_fs_cmnd+0x96/0xd0

Signed-off-by: Jiri Slaby <jslaby@xxxxxxx>
Cc: Christoph Hellwig <hch@xxxxxx>
Cc: Neil Brown <neilb@xxxxxxx>
---
drivers/md/raid1.c | 10 ++++++----
drivers/md/raid10.c | 5 +++--
2 files changed, 9 insertions(+), 6 deletions(-)

diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
index 73cc74f..4bfebce 100644
--- a/drivers/md/raid1.c
+++ b/drivers/md/raid1.c
@@ -787,8 +787,8 @@ static int make_request(mddev_t *mddev, struct bio * bio)
struct bio_list bl;
struct page **behind_pages = NULL;
const int rw = bio_data_dir(bio);
- const bool do_sync = (bio->bi_rw & REQ_SYNC);
- bool do_barriers;
+ const unsigned long do_sync = (bio->bi_rw & REQ_SYNC);
+ unsigned long do_barriers;
mdk_rdev_t *blocked_rdev;

/*
@@ -1640,7 +1640,8 @@ static void raid1d(mddev_t *mddev)
* We already have a nr_pending reference on these rdevs.
*/
int i;
- const bool do_sync = (r1_bio->master_bio->bi_rw & REQ_SYNC);
+ const unsigned long do_sync =
+ (r1_bio->master_bio->bi_rw & REQ_SYNC);
clear_bit(R1BIO_BarrierRetry, &r1_bio->state);
clear_bit(R1BIO_Barrier, &r1_bio->state);
for (i=0; i < conf->raid_disks; i++)
@@ -1696,7 +1697,8 @@ static void raid1d(mddev_t *mddev)
(unsigned long long)r1_bio->sector);
raid_end_bio_io(r1_bio);
} else {
- const bool do_sync = r1_bio->master_bio->bi_rw & REQ_SYNC;
+ const unsigned long do_sync =
+ r1_bio->master_bio->bi_rw & REQ_SYNC;
r1_bio->bios[r1_bio->read_disk] =
mddev->ro ? IO_BLOCKED : NULL;
r1_bio->read_disk = disk;
diff --git a/drivers/md/raid10.c b/drivers/md/raid10.c
index a88aeb5..6513c3c 100644
--- a/drivers/md/raid10.c
+++ b/drivers/md/raid10.c
@@ -799,7 +799,7 @@ static int make_request(mddev_t *mddev, struct bio * bio)
int i;
int chunk_sects = conf->chunk_mask + 1;
const int rw = bio_data_dir(bio);
- const bool do_sync = (bio->bi_rw & REQ_SYNC);
+ const unsigned long do_sync = (bio->bi_rw & REQ_SYNC);
struct bio_list bl;
unsigned long flags;
mdk_rdev_t *blocked_rdev;
@@ -1734,7 +1734,8 @@ static void raid10d(mddev_t *mddev)
raid_end_bio_io(r10_bio);
bio_put(bio);
} else {
- const bool do_sync = (r10_bio->master_bio->bi_rw & REQ_SYNC);
+ const unsigned long do_sync =
+ (r10_bio->master_bio->bi_rw & REQ_SYNC);
bio_put(bio);
rdev = conf->mirrors[mirror].rdev;
if (printk_ratelimit())
--
1.7.2.1


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/