On Wed, Nov 23, 2016 at 08:50:56PM -0500, Ian Sutton wrote: > On Wed, Nov 23, 2016 at 05:48:27PM -0700, Theo de Raadt wrote: > > > I am assuming that arch/aarch64 > > > > There should be no such thing. > > > > This game of calling things 5 names has to stop. > > I'm equally irritated and would like to reach a consensus on a single, > final name. I see three candidates: > > * armv8 > * aarch64 > * arm64 > > armv8 is the umbrella name for architecture, and fits our previous > precedent (armv7). armv8 processors allow execution of 32 bit binaries > however, which conflicts with our intention to have everything 64-bit on > this new arch > > aarch64 would eliminate this conflict but doesn't really refer to an > architecture the way that names in arch/ should > > arm64 is the precedent set by other bsd's. We have a generalized > arch/arm directory that doesn't support specific boards but holds shared > arm sources, suggesting arm64 would be a similar directory for > generalized 64-bit arm code (it wouldn't) > > Let's decide this soon and avoid bikeshedding. I like armv8.
There are 32 bit only armv8 implementations, that name doesn't make sense. I don't see the point in adding a directory without a pmap or toolchain, but if/when it is added it should probably be arm64.
