On Tue, 2016-01-05 at 11:44 +0100, David Sterba wrote:
> We have a full 32 bit number space, so multiples of power of 2 are
> also
> possible if that makes sense.
Hmm that would make a maximum of 4GiB RAID chunks...
perhaps we should reserve some of the higher bits for a multiplier, in
case 4GiB would ever become too little O;-)


>  In general we don't need to set additional
> limitations besides minimum, maximum and "minimal step".
And that can/should be done in the userland.


> > Are there any concerns/constraints with too small/too big chunks
> > when
> > these play together with lower block layers (I'd guess not).
> 
> I don't think so.

Well I was mainly thinking about dm-crypt, that uses 512B blocks and in
fact that size wouldn't be easy to change, as (IIRC) larger block sizes
make XTS less secure.
Obviously *this* isn't anything that btrfs would have to worry about,
especially as we're anyway on a higher block layer level,.. but it just
reminded me that there can be cases where too large / too small may
actually cause issues.


Cheers,
Chris.