After used sync way to handle METADATA_UPDATED msg, a deadlock could appear if stop a resyncing array. betalinux244:~ # ps aux|grep md|grep D root 17164 0.0 0.0 0 0 ? D Jan09 0:00 [md0_cluster_rec] root 18151 0.0 0.1 19852 2008 ? Ds Jan09 0:00 /sbin/mdadm -Ssq betalinux244:~ # cat /proc/17164/stack [<ffffffffa06a7395>] recv_daemon+0x1f5/0x590 [md_cluster] [<ffffffffa067be20>] md_thread+0x130/0x150 [md_mod] [<ffffffff810995ed>] kthread+0xbd/0xe0 [<ffffffff815e96bf>] ret_from_fork+0x3f/0x70 [<ffffffff81099530>] kthread+0x0/0xe0 [<ffffffffffffffff>] 0xffffffffffffffff betalinux244:~ # cat /proc/18151/stack [<ffffffff81099879>] kthread_stop+0x59/0x130 [<ffffffffa067c566>] md_unregister_thread+0x46/0x80 [md_mod] [<ffffffffa06a6e71>] leave+0x81/0x120 [md_cluster] [<ffffffffa0684e94>] md_cluster_stop+0x14/0x30 [md_mod] [<ffffffffa06858b6>] bitmap_free+0x126/0x130 [md_mod] [<ffffffffa0682d06>] do_md_stop+0x356/0x5f0 [md_mod] [<ffffffffa0683cbe>] md_ioctl+0x6fe/0x1680 [md_mod] [<ffffffff812ed158>] blkdev_ioctl+0x258/0x920 [<ffffffff8122f81d>] block_ioctl+0x3d/0x40 [<ffffffff8120ac0d>] do_vfs_ioctl+0x2cd/0x4a0 [<ffffffff8120ae54>] SyS_ioctl+0x74/0x80 [<ffffffff815e936e>] entry_SYSCALL_64_fastpath+0x12/0x6d [<ffffffffffffffff>] 0xffffffffffffffff Since md_unregister_thread(&cinfo->recv_thread) is blocked by recv_daemon -> process_recvd_msg -> process_metadata_update. To resolve the issue, MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD is also need to be set before unregister thread. Reviewed-by: NeilBrown <neilb@xxxxxxxx> Signed-off-by: Guoqing Jiang <gqjiang@xxxxxxxx> --- drivers/md/md-cluster.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/md/md-cluster.c b/drivers/md/md-cluster.c index 0aad477d1b20..5e2c54be6f30 100644 --- a/drivers/md/md-cluster.c +++ b/drivers/md/md-cluster.c @@ -932,6 +932,7 @@ static int join(struct mddev *mddev, int nodes) return 0; err: + set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state); md_unregister_thread(&cinfo->recovery_thread); md_unregister_thread(&cinfo->recv_thread); lockres_free(cinfo->message_lockres); @@ -987,6 +988,7 @@ static int leave(struct mddev *mddev) if (cinfo->slot_number > 0 && mddev->recovery_cp != MaxSector) resync_bitmap(mddev); + set_bit(MD_CLUSTER_HOLDING_MUTEX_FOR_RECVD, &cinfo->state); md_unregister_thread(&cinfo->recovery_thread); md_unregister_thread(&cinfo->recv_thread); lockres_free(cinfo->message_lockres); -- 2.6.2 -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
