Richard Henderson <richard.hender...@linaro.org> writes:
> For v1, I had split this into 4 logically distinct parts. But > apparently there are minor interdependencies, because the later > sets would not apply standalone, says Alex. > > Rather than tease them apart, and then have to undo that work > in order to actually apply them later, I'll just lump them. > > So: > > Part 1, patches 1-7, tcg_gen_gvec_dup_imm, is reviewed. > > Part 2, patch 8, vector tail clearing, is reviewed, and I have > moved the target/arm patches into a different queue. > > Part 3, patches 9-25, TYPE_CONST temporaries, is mostly unreviewed. > > Part 4, patch 26, load_dest for GVecGen2, a support patch for SVE2. > > Part 5, patches 27-36, add vector rotate patterns, is brand new. > I include two demonstrators for target/ppc and target/s390x. I've done my review pass for now. I made it through all the core code but my brain was too frazzled to look at the back end generation code so I'll have another go at that on the next revision when the sparc regression is figured out. -- Alex Bennée
