Post-increment constraint in inline assembly (SuperH)

[Segher Boessenkool]

Hi!

On Sun, Jan 28, 2018 at 02:32:56PM +0100, Sébastien Michelland wrote:
> I am trying to generate a call to the SH-4A unaligned move "movua.l 
> @rm+, r0" instruction. Currently my inline-assembly approach using
> the '>' constraint is rejected by the compiler:
> 
> Â  #include <stdint.h>
> 
> Â  void f(void)
> Â  {
> Â Â Â Â Â Â Â Â Â  uint32_t longword, *src;
> Â Â Â Â Â Â Â Â Â  __asm__(
> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â  "movua.l %1, %0"
> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â  : "=z" (longword)
> Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â  : ">" (src)
> Â Â Â Â Â Â Â Â Â  );
> Â  }
> 
> Â  % sh3eb-elf-gcc -S movua.c -o movua.s -ffreestanding
> Â  movua.c: In function 'f':
> Â  movua.c:6:2: warning: asm operand 1 probably doesn't match constraints
> Â Â Â  __asm__(
> Â Â Â  ^~~~~~~
> Â  movua.c:6:2: error: impossible constraint in 'asm'
> 
> I have tried a lot of variations to no avail.
> 
> Am I using the '>' constraint the wrong way? What would be the proper 
> way to generate this instruction?

You should use "m>", or the compiler will not know what to do in most
(all?) cases.

You should use "m>"(*src) (note the star) because this should be a
memory operand, not the pointer.

You need to use "longword" or the whole asm can be optimised away.

You need to arrange for src+1 to be used somehow, or the compiler will
not choose to do a post-increment.

This works as expected:

===
long bla;
long *p;

void f(long *src)
{
	long longword;
	__asm__("movua.l %1, %0" : "=z" (longword) : "m>" (*src));
	bla = longword;
	p = src+1;
}
===

resulting in

===
f:
        mov.l   .L3,r1
#APP
        movua.l @r4+, r0
#NO_APP
        mov.l   r0,@r1
        mov.l   .L4,r1
        rts
        mov.l   r4,@r1
===


Segher