Re: asm register constraint. Was: [PATCH v2 bpf-next 2/5] bpf: Introduce "volatile compare" macro

[Eduard Zingerman <eddyz87@gmail.com>]

On Tue, 2024-01-16 at 09:47 -0800, Alexei Starovoitov wrote:

[...]

> 2.
> case 'w':
> if (Size == 32 && HasAlu32)
> 
> This is probably unnecessary as well.
> When bpf programmer specifies 'w' constraint, llvm should probably use it
> regardless of alu32 flag.
> 
> aarch64 has this comment:
>     case 'x':
>     case 'w':
>       // For now assume that the person knows what they're
>       // doing with the modifier.
>       return true;
> 
> I'm reading it as the constraint is a final decision.
> Inline assembly shouldn't change with -mcpu flags.

llvm-mc -arch=bpf -mcpu=v2 compiles the asm below w/o warnings:

	.text
	.file	"test-asm-shifts.c"
	.globl	foo
	.p2align	3
	.type	foo,@function
foo:
	call bar
	w0 += 1
	exit
.Lfunc_end0:
	.size	foo, .Lfunc_end0-foo

However, clang would complain about "w" constraint when used
for inline assembly when CPUv2 is used:

  error: couldn't allocate input reg for constraint 'w'

As Yonghong notes in the sibling thread, this would probably require
enabling 32-bit sub-registers for CPUv2.
Will need some time to investigate.