Re: [dm-devel] [RFC][PATCH] Add checks for correct hardsectboundary alignment

[Christophe Saout <christophe saout de>]

Am Di, den 27.01.2004 schrieb Joe Thornber um 15:31:

> On Tue, Jan 20, 2004 at 04:53:11PM +0100, Christophe Saout wrote:
> > I think we should add additional checks to ensure that we don't run
> > into trouble with devices that have a non-512-byte sector size.
> 
> You're correct that this is indeed needed.  However your solution
> makes some assumptions about how the target is going to use the
> device.  eg. you do the wrong thing for snapshot exception stores, or
> mirror logs.

How exactly are my assumptions wrong?

The only thing where I think it might do the wrong thing is when the
log/cow store is on a disk with a bigger hardsect size than the rest and
it prevents split_io from being smaller than the hardsect size of that
disk. And I'm not even sure if that is always wrong.

> I think the correct solution would be to add some
> utility functions to dm-table.c in order to make it easy for the
> targets constructors themselves to do the checking.

I had this first and then I figured out (or at least I thought I'd
figured out) that this isn't needed.