Re: Using CPU flags in C

On Wed, 19 Nov 2025 at 08:55, jh--- via Gcc-help <[email protected]> wrote:
>
> Thank you for your answer.
> I'm not sure I understand it completely, could you give me some details?
> Le 2025-11-18 19:42, Florian Weimer a écrit :
> > * jh:
> >
> >> I don't really know where to look for this information.
> >> Is it possible to use the CPU status flags to communicate information
> >> between inline assembly and the normal C control flow?
> >> For example, an asm statement would set the Z flag and then the C code
> >> would jump (for example with beq or bne in the case of ARM) without
> >> using any additional variable and comparison instruction. Using a
> >> dummy variable is no issue if the generated code is just the
> >> conditional branch instruction.
> >
> > Do you mean something like this?
> >
> > commit 119d658ac2aad88e306b4a66c1717e5ebf86c73f
> > Author: Uros Bizjak <[email protected]>
> > Date:   Fri Aug 29 09:05:23 2025 +0200
> >
> >     x86: Use flag output operands for inline asm in atomic-machine.h
> >
> >     Use the flag output constraints feature available in gcc 6+
> >     ("=@cc<cond>") instead of explicitly setting a boolean variable
> >     with SETcc instruction.  This approach decouples the instruction
> >     that sets the flags from the code that consumes them, allowing
> >     the compiler to create better code when working with flags users.
> Does this "=@cc<cond>" also work for ARM? Is the usage of the conditions
> documented somewhere (like Z is ==0, N==V is signed >=0). I know where
> to find the ARM definitions, but is there a place where I can check
> gcc's interpretation (i.e. which C code generates which flag access)?

Yes, of course it's documented:
https://gcc.gnu.org/onlinedocs/gcc/Extended-Asm.html#Flag-Output-Operands



> >     Instead of e.g.:
> >
> >        lock add %esi,(%rdi)
> >        sets   %sil
> >        test   %sil,%sil
> >        jne    <...>
> >
> >     the compiler now generates:
> >
> >        lock add %esi,(%rdi)
> >        js     <...>
> How is this code integrated into C? Would it be used like this? But can
> I get the -6 value at the same time?
> if(asm_add(36, -42)<0){
>      printf("negative\n");
> }else{
>      printf("positive\n");
> }
> If yes, is it possible to at the same time set a register value and the
> flags and how would that be written?
> For example, I would like to be able to set a register to any value, but
> when I set it to 0, I would like to set the Z flag (to 1 or 0) and I
> would like the compiler to jump depending on my Z flag, while being able
> to use the register value later in the C code (not going through memory
> as is done in the commit).
> >
> >     No functional changes intended.
> >
> >     Signed-off-by: Uros Bizjak <[email protected]>
> >     Cc: H.J.Lu <[email protected]>
> >     Cc: Florian Weimer <[email protected]>
> >     Cc: Carlos O'Donell <[email protected]>
> >     Reviewed-by: Florian Weimer <[email protected]>
> >
> > <https://sourceware.org/cgit/glibc/commit/?id=119d658ac2aad88e306b4a66c1717e5ebf86c73f>
> >
> > Thanks,
> > Florian
> Best regards,
> JH