On Tue, Aug 13, 2024 at 01:55:46PM +0100, Andre Przywara wrote: > On Tue, 13 Aug 2024 12:00:06 +0100 > Mark Brown <broonie@xxxxxxxxxx> wrote: > > Hi broonie, > > > On Mon, Aug 12, 2024 at 03:09:24PM +0100, Andre Przywara wrote: > > > > > + /* Did we find the lowest supported VL? */ > > > + if (use_sme && vq < sve_vq_from_vl(vl)) > > > + break; > > > > We don't need the use_sme check here, SVE is just architecturally > > guaranteed to never trip the && case. Unless you add a warning for > > broken implementations I'd just skip it. > > Ah, thanks, I wasn't sure about that, and wanted to mimic the existing > code as close as possible. Will surely just drop it. > > Thanks, > Andre Maybe at least worth a comment? I was looking at this code the other week, and this puzzled me until I went and looked back at the architecture. Cheers ---Dave >
