On Wed, Dec 06, 2017 at 07:43:36PM +0000, Ard Biesheuvel wrote: > We are going to add code to all the NEON crypto routines that will > turn them into non-leaf functions, so we need to manage the stack > frames. To make this less tedious and error prone, add some macros > that take the number of callee saved registers to preserve and the > extra size to allocate in the stack frame (for locals) and emit > the ldp/stp sequences. > > Signed-off-by: Ard Biesheuvel <ard.biesheuvel@xxxxxxxxxx> > --- > arch/arm64/include/asm/assembler.h | 60 ++++++++++++++++++++ > 1 file changed, 60 insertions(+) > > diff --git a/arch/arm64/include/asm/assembler.h b/arch/arm64/include/asm/assembler.h > index aef72d886677..5f61487e9f93 100644 > --- a/arch/arm64/include/asm/assembler.h > +++ b/arch/arm64/include/asm/assembler.h > @@ -499,6 +499,66 @@ alternative_else_nop_endif > #endif > .endm > > + /* > + * frame_push - Push @regcount callee saved registers to the stack, > + * starting at x19, as well as x29/x30, and set x29 to > + * the new value of sp. Add @extra bytes of stack space > + * for locals. > + */ > + .macro frame_push, regcount:req, extra > + __frame st, \regcount, \extra > + .endm > + > + /* > + * frame_pop - Pop @regcount callee saved registers from the stack, > + * starting at x19, as well as x29/x30. Also pop @extra > + * bytes of stack space for locals. > + */ > + .macro frame_pop, regcount:req, extra > + __frame ld, \regcount, \extra > + .endm > + > + .macro __frame, op, regcount:req, extra=0 > + .ifc \op, st > + stp x29, x30, [sp, #-((\regcount + 3) / 2) * 16 - \extra]! > + mov x29, sp > + .endif > + .if \regcount < 0 || \regcount > 10 > + .error "regcount should be in the range [0 ... 10]" > + .endif > + .if (\extra % 16) != 0 > + .error "extra should be a multiple of 16 bytes" > + .endif > + .if \regcount > 1 > + \op\()p x19, x20, [sp, #16] > + .if \regcount > 3 > + \op\()p x21, x22, [sp, #32] > + .if \regcount > 5 > + \op\()p x23, x24, [sp, #48] > + .if \regcount > 7 > + \op\()p x25, x26, [sp, #64] > + .if \regcount > 9 > + \op\()p x27, x28, [sp, #80] Can the _for thing I introduced in fpsimdmacros.h be any use here? Alternatively, the following could replace that .if-slide, providing the calling macro does .altmacro .. .noaltmacro somewhere. .macro _pushpop2 op, n1, n2, offset \op x\n1, x\n2, [sp, #\offset] .endm .macro _pushpop op, first, last, offset .if \first < \last _pushpop2 \op\()p, \first, %\first + 1, \offset _pushpop \op, %\first + 2, \last, %\offset + 16 .elseif \first == \last \op\()r x\first, [sp, #\offset] .endif .endm Also, I wonder whether it would be more readable at the call site to specify the first and last reg numbers instead of the reg count, e.g.: frame_push first_reg=19, last_reg=23 (or whatever). Just syntactic sugar though. [...] Cheers ---Dave
