Re: [PATCH] winecrt0: Do not expect jmp_buf to be aligned in __wine_longjmp on x86_64.

[Paul Gofman]

On 7/14/20 19:01, Gabriel Ivăncescu wrote:
> You mean it's not forcefully aligned, right? If so, I think that's
> normal since the MS ABI mandates that it is 16 byte aligned.

I am not sure this is normal if you explicitly use
__attribute__((force_align_arg_pointer)) on function, without even a
warning with -Wall. Looks like probably gcc or mingw bug to me.
> For this patch, just for future reference, you should be using movdqu,
> which anyway it's just as fast as movdqa unless the processor is very
> old. I don't know if it's still necessary though.

movdqa / movaps can (or could) work on unaligned addresses on some CPUs,
but seems to fault on most. FWIW I generally prefer mov{a|u}ps for SSE
regs, even if just to not mess them up with movq like I did.