Re: Suspend 2 merge: 50/51: Device mapper support.

[Alasdair G Kergon <agk@redhat.com>]

On Fri, Dec 03, 2004 at 09:08:18AM +1100, Nigel Cunningham wrote:
> My mistake. The code has been improved and I haven't reverted some of
> the changes in drivers/md to match. I'll do that and make the two
> exports that are needed (dm_io_get and dm_io_put) into an
> include/linux/dm.h.
 
It would be device-mapper.h - or the whole of dm-io.h.
The vmalloc comment also looks wrong BTW - it's extra kmalloc 
GFP_KERNEL memory you're asking for here.

I'd like to understand why you need to call those functions
and how it integrates with the rest of what you're doing:
do you have calls to other dm functions in other patches?

Or is this particular change optional, but you have test results 
showing it to be desirable or necessary in certain cases, maybe 
indicating a shortcoming within the dm-io code which should be 
addressed instead?

Alasdair
-- 
agk@redhat.com