Hi all,
I've been trying (unsuccessfully) to have my bcache device automount on
boot as a backing store for libvirt to use.
I was trying to use the loop in the bcache initramfs script and found my
machine was repeatedly hanging. After some testing, I found this was
due to bcache blowing up when I echo'd the device node of an extended
partition to the register sys interface.
Here's what I was doing:
for i in `ls /dev/sd*`; do
echo $i > /sys/fs/bcache/register_quiet
done
It hangs on /dev/sda2. Here's my disk layout for sda:
root@brain:~# fdisk -l /dev/sda
Disk /dev/sda: 1000.2 GB, 1000204886016 bytes
255 heads, 63 sectors/track, 121601 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x000b0e70
Device Boot Start End Blocks Id System
/dev/sda1 * 1 120559 968381440 83 Linux
/dev/sda2 120559 121602 8378369 5 Extended
/dev/sda5 120559 121602 8378368 82 Linux swap / Solaris
and kernel :
Linux brain 3.5.0-rc6+ #1 SMP Thu Jul 19 16:13:20 EDT 2012 x86_64 x86_64
x86_64 GNU/Linux
git commit : a29809551cd847e5588c9ebc843eda5cd3736d8e
I found doing this "echo /dev/sda2 > /sys/fs/bcache/register_quiet"
completely hangs the machine, producing this in my logs:
Aug 2 12:37:12 brain kernel: [ 395.881134] INFO: rcu_sched
self-detected stall on CPU { 3} (t=6000 jiffies)
Aug 2 12:37:12 brain kernel: [ 395.881153] Pid: 1662, comm: bash Not
tainted 3.5.0-rc6+ #1
Aug 2 12:37:12 brain kernel: [ 395.881156] Call Trace:
Aug 2 12:37:12 brain kernel: [ 395.881158] <IRQ> [<ffffffff810e4e96>]
__rcu_pending+0x216/0x4a0
Aug 2 12:37:12 brain kernel: [ 395.881174] [<ffffffff810a6f30>] ?
tick_nohz_handler+0xf0/0xf0
Aug 2 12:37:12 brain kernel: [ 395.881179] [<ffffffff810e514c>]
rcu_pending+0x2c/0x70
Aug 2 12:37:12 brain kernel: [ 395.881184] [<ffffffff810e5991>]
rcu_check_callbacks+0xa1/0x140
Aug 2 12:37:12 brain kernel: [ 395.881189] [<ffffffff81063578>]
update_process_times+0x48/0x90
Aug 2 12:37:12 brain kernel: [ 395.881193] [<ffffffff810a6f94>]
tick_sched_timer+0x64/0xc0
Aug 2 12:37:12 brain kernel: [ 395.881197] [<ffffffff81079a32>]
__run_hrtimer+0x72/0x1d0
Aug 2 12:37:12 brain kernel: [ 395.881203] [<ffffffff8101b9a9>] ?
read_tsc+0x9/0x20
Aug 2 12:37:12 brain kernel: [ 395.881207] [<ffffffff8107a263>]
hrtimer_interrupt+0xe3/0x210
Aug 2 12:37:12 brain kernel: [ 395.881212] [<ffffffff8163d319>]
smp_apic_timer_interrupt+0x69/0x99
Aug 2 12:37:12 brain kernel: [ 395.881219] [<ffffffff8163c08a>]
apic_timer_interrupt+0x6a/0x70
Aug 2 12:37:12 brain kernel: [ 395.881220] <EOI> [<ffffffff8130bb43>]
? radix_tree_lookup_element+0x43/0x90
Aug 2 12:37:12 brain kernel: [ 395.881230] [<ffffffff8130bb9e>]
radix_tree_lookup_slot+0xe/0x10
Aug 2 12:37:12 brain kernel: [ 395.881237] [<ffffffff8111bf6e>]
find_get_page+0x1e/0xa0
Aug 2 12:37:12 brain kernel: [ 395.881240] [<ffffffff8111c1e6>]
find_lock_page+0x26/0x80
Aug 2 12:37:12 brain kernel: [ 395.881246] [<ffffffff811a8704>] ?
__find_get_block+0xa4/0x210
Aug 2 12:37:12 brain kernel: [ 395.881250] [<ffffffff8111cebf>]
find_or_create_page+0x3f/0xb0
Aug 2 12:37:12 brain kernel: [ 395.881254] [<ffffffff811a8ef6>]
__getblk+0xe6/0x280
Aug 2 12:37:12 brain kernel: [ 395.881270] [<ffffffff811a90e3>]
__bread+0x13/0xa0
Aug 2 12:37:12 brain kernel: [ 395.881285] [<ffffffffa001a5d2>]
register_bcache+0x192/0xa00 [bcache]
Aug 2 12:37:12 brain kernel: [ 395.881292] [<ffffffff8115d761>] ?
alloc_pages_current+0xb1/0x120
Aug 2 12:37:12 brain kernel: [ 395.881298] [<ffffffff81308987>]
kobj_attr_store+0x17/0x20
Aug 2 12:37:12 brain kernel: [ 395.881304] [<ffffffff811e6281>]
sysfs_write_file+0xd1/0x160
Aug 2 12:37:12 brain kernel: [ 395.881311] [<ffffffff81177c06>]
vfs_write+0xc6/0x180
Aug 2 12:37:12 brain kernel: [ 395.881315] [<ffffffff81177f21>]
sys_write+0x51/0x90
Aug 2 12:37:12 brain kernel: [ 395.881320] [<ffffffff8163b5e9>]
system_call_fastpath+0x16/0x1b
Any thoughts on this ?
What is the correct way to pull bcache devices up prior to mounting, so
they can be automounted via fstab entries ?
Please let me know if I can provide more / better info.
Thanks,
Brook
--
To unsubscribe from this list: send the line "unsubscribe linux-bcache" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
