[dm-devel] Trying to use dmcache with zram block devices

[Daniel Wheeler]

Hello all,

I am trying to use zram for the dmcache meta and dmcache caching device
to cache a local logical volume.

It works just fine if I use logical volumes from a disk as the caching
device, but using a zram block device as the cache sets up properly but
fails to boot a virtual machine on the resulting volume. Interestingly,
I can use zram as a meta cache device without error, but using zram as
the cache itself is when this fails.

This is how I'm setting up my device map:

0 20758528 cache /dev/zram0 /dev/zram1 /dev/c0v0/vdb 64 1 writethrough mq 6 sequential_threshold 5120 random_threshold 2 migration_threshold 819200

zram0 and zram1 are both 1GB in size so plenty of space here.
I am booting linux 3.11.4 with that version of dmcache.

When I try and boot a VM off of the resulting cache, the VM fails to
write to it's disk (I think), and displays the following error when
booting CentOS 6 in the VM:

mount: wrong fs type, bad option, bad superblock on /dev/mapper/VM_NAME,
 	missing codepage or helper program, or other error
 	In some cases useful info is found in syslog - try
 	dmesg | tail or so


Kernel Panic - not syncing: Attempting to kill init!
Pid: 1, comm: init Not tainted 2.6.32-279.e16.x86_64 #1
Call Trace:
  [<ffffffff814fd11a>] ? panic+0xa0/0x168
  [<ffffffff81070bd2>] ? do_exit+0x862/0x870
  [<ffffffff8117ca75>] ? fput+0x25/0x30
  [<ffffffff81070c38>] ? do_group_exit+0x58/0xd0
  [<ffffffff81070cc7>] ? sys_exit_group+0x17/0x20
  [<ffffffff8100b0f2>] ? system_call_fastpath+0x16/0x1b


Can you think of any reason that zram cannot be a cache for dmcache devices?

If this is a bug, does it reside in dmcache or in zram?

How can I test this further?

-Daniel