On Thu, 2009-10-01 at 08:07 +0800, Laurent Pinchart wrote:
> On Tuesday 29 September 2009 07:57:32 Huang Ying wrote:
> > Hi, Joe,
> >
> > Thanks for the patch. I think that is a good idea.
> >
> > For your patch. I think you need a changelog for each patch.
> >
> > It seems that the binary representation of UUID can be little-endian
> > (used by most kernel components) or big-endian (defined by RFC4122, used
> > in network?). Maybe we should consider about that.
>
> I think that's what the 'r' option is supposed to handle. Maybe you could use
> 'b' and 'l' options instead, to specify the endianness explicitly.
Yes. Thanks for you reminding.
> > In fact, I find there are many different UUID/GUID definitions in
> > kernel, such as that in efi, many file systems, some drivers, etc. It
> > seems that every kernel components need UUID/GUID has its own
> > definition, so I think we should unify all the UUID/GUID definitions in
> > kernel too. The file attached is a draft unified UUID/GUID definition,
> > with byte-order issue in mind.
> >
> > Any comment?
>
> Not much, it would help if you could show where those functions would be used.
>
> I would also use the 'le' and 'be' prefixes (as in le16, be32) instead of 'l'
> and 'b', making it leuuid and beuuid, or maybe uuid_le and uuid_be.
Yes. It seems 'be' and 'le' is better than 'b' an 'l'.
Best Regards,
Huang Ying
|