On Tue, Aug 22, 2017 at 04:03:20PM +0100, Alex Bennée wrote: > > Dave Martin <Dave.Martin@xxxxxxx> writes: > > > On Tue, Aug 22, 2017 at 02:53:49PM +0100, Alex Bennée wrote: [...] > >> + > >> +#define SVE_VQ_BITS 128 /* 128 bits in one quadword */ > >> +#define SVE_VQ_BYTES (SVE_VQ_BITS / 8) > >> + > > > > I was trying to keep extraneous #defines to a minimum, since this is a > > uapi header, and people may depend on anything defined here. > > > > I think SVE_VQ_BYTES is reasonable to have, and this allows us to > > rewrite a few hard-coded 0x10s and 16s symbolically which is probably a > > good idea -- I'll add this. > > > > SVE_VQ_BITS looks redundant to me though. It wouldn't be used for any > > purpose other than defining SVE_VQ_BYTES. > > Yeah I was more concerned with getting rid of the magic 0x10's than > showing exactly how many bits something is. OK, I'll take SVE_VQ_BYTES and use it in the appropriate places. There are a few 0x10s/16s in the series that can use this instead of being open-coded. Cheers ---Dave _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
