On Tue, Feb 26, 2013 at 03:02:01PM +0800, Liu Bo wrote: > On Sun, Feb 24, 2013 at 06:55:46PM -0800, Marc MERLIN wrote: > > Is this useful to anyone? > > > > Hi Marc, > > Thanks for the report, of course they're useful. Thanks. I wasn't sure since I haven't seen the real problem of crashes during mount due to unexpected state in replay logs being improved over the last 4 kernel versions, and I wasn't sure if they are being worked on. > Could you please also show us your workloads and it'd be better to know how to > reproduce this? Sure thing. Workload is a simple laptop, where something tyipcally dies during writes and I get a full system hang because linux is unable to flush its disk queues, so in the end I power cycle. Details here: http://www.mail-archive.com/linux-kernel@xxxxxxxxxxxxxxx/msg391505.html As mentioned in my other mail, I do run on top of dmcrypt since I can't have an unencrypted laptop, and ecryptfs is way too slow while not supporting encryption of long filenames. The next question is: are your write requests getting there in the order they should. My crashes and your BUG() that get triggered indicate that maybe not. Currently, I have: gandalfthegreat:~# cat /sys/block/sda/queue/scheduler noop [deadline] cfq That could be a problem I guess, so I'll change it to noop, just in case. My boot is like this: /vmlinuz-3.7.8-amd64-preempt-20130222 root=/dev/mapper/cryptroot ro rootflags=subvol=root cryptopts=source=/dev/sda4,keyscript=/sbin/cryptgetpw,discard btrfs is mounted like so: LABEL=btrfs_pool1 / btrfs subvol=root,defaults,compress=lzo,discard,nossd,space_cache,noatime Is there anything else I can give? (I'll answer on the other thread with the fsimage) Marc -- "A mouse is a device used to point at the xterm you want to type in" - A.S.R. Microsoft is to operating systems .... .... what McDonalds is to gourmet cooking Home page: http://marc.merlins.org/ -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
