* Josh Poimboeuf <jpoimboe@xxxxxxxxxx> wrote: > > So if this observation of mine is true we could go back to the old code for the > > hotpath, but use RDI for TBL and not reload it in the hotpath. > > Thanks for the excellent breakdown. > > When I looked at the patch again, I came to the same conclusion as your > #4, which is that RDI isn't being used in the inner loops. It *is* used > in the outermost loop, however. > > So v2 of my sha512-avx2-asm.S patch spilled CTX onto the stack, instead > of TBL: > > https://lkml.kernel.org/r/20170913223303.pskmy2v7nto6rvtg@treble Indeed - I should have checked your v2 patch, but somehow missed it. Would have saved me some looking+typing ;-) Thanks, Ingo