On Thursday 13 March 2008 14:45:39 Christian Kujau wrote: > and so does -git8 and -git9...sigh :( hmm, I can boot any of these -rcX-gitY kernels (but not without some stuck tasks after about a hour or so...) but back: I tried lockdep and ran into this: (debug blocker) pktcdvd: writer pktcdvd0 mapped to sr0 ============================================= [ INFO: possible recursive locking detected ] 2.6.25-rc5-git3 #2 --------------------------------------------- scsi_id/8262 is trying to acquire lock: (&bdev->bd_mutex){--..}, at: [<ffffffff802a8e51>] do_open+0x7b/0x2c0 but task is already holding lock: (&bdev->bd_mutex){--..}, at: [<ffffffff802a8e51>] do_open+0x7b/0x2c0 other info that might help us debug this: 2 locks held by scsi_id/8262: #0: (&bdev->bd_mutex){--..}, at: [<ffffffff802a8e51>] do_open+0x7b/0x2c0 #1: (&ctl_mutex#2){--..}, at: [<ffffffff88231af2>] pkt_open+0x26/0x516 [pktcdvd] stack backtrace: Pid: 8262, comm: scsi_id Not tainted 2.6.25-rc5-git3 #2 Call Trace: [<ffffffff8024ffe4>] __lock_acquire+0x8c9/0xc72 [<ffffffff804bb23e>] ? __mutex_unlock_slowpath+0xf9/0x105 [<ffffffff8025079f>] lock_acquire+0x5e/0x77 [<ffffffff802a8e51>] ? do_open+0x7b/0x2c0 [<ffffffff804bb336>] mutex_lock_nested+0xe1/0x270 [<ffffffff802a8e51>] ? do_open+0x7b/0x2c0 [<ffffffff802a8e51>] do_open+0x7b/0x2c0 [<ffffffff802a9116>] __blkdev_get+0x80/0x92 [<ffffffff802a9133>] blkdev_get+0xb/0xd [<ffffffff88231b6b>] :pktcdvd:pkt_open+0x9f/0x516 [<ffffffff802a8e84>] do_open+0xae/0x2c0 [<ffffffff802a92a4>] ? blkdev_open+0x0/0x6a [<ffffffff804bcbab>] ? _spin_unlock+0x26/0x2a [<ffffffff802a92a4>] ? blkdev_open+0x0/0x6a [<ffffffff802a92db>] blkdev_open+0x37/0x6a [<ffffffff80282571>] __dentry_open+0xe6/0x1c0 [<ffffffff802826e4>] nameidata_to_filp+0x2e/0x40 [<ffffffff8028272f>] do_filp_open+0x39/0x4b [<ffffffff804bcbab>] ? _spin_unlock+0x26/0x2a [<ffffffff8028246f>] ? get_unused_fd_flags+0x10d/0x11c [<ffffffff80282792>] do_sys_open+0x51/0xd9 [<ffffffff80282843>] sys_open+0x1b/0x1d [<ffffffff8020b22b>] system_call_after_swapgs+0x7b/0x80 soooo.....? scsi_id/8262 is trying to acquire lock: (&bdev->bd_mutex){--..}, at: [<ffffffff802a8e51>] do_open+0x7b/0x2c0 but task is already holding lock: (&bdev->bd_mutex){--..}, at: [<ffffffff802a8e51>] do_open+0x7b/0x2c0 sounds a bit strange? Is this a SMP related problem after all? (*disabled pktcddvd for now... to see if still locks) Regards, Chr... -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel