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.biesheu...@linaro.org>
> ---
> 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
