Re: missed optimization compiling naive get_unaligned_le32 on x86

[Axel Freyn <axel-freyn at gmx dot de>]

Hi Zeev,
On Fri, Apr 15, 2011 at 07:24:42PM +0300, Zeev Tarantov wrote:
> When compiling this code:
> 
> unsigned int get_le32(unsigned char *p)
> {
>   return p[0] | p[1] << 8 | p[2] << 16 | p[3] << 24;
> }
> 
> On gcc 4.6.0 rev. 172266 for x86-64, I get:
> 
>         movzbl  1(%rdi), %eax
>         movzbl  2(%rdi), %edx
>         sall    $8, %eax
>         sall    $16, %edx
>         orl     %edx, %eax
>         movzbl  (%rdi), %edx
>         orl     %edx, %eax
>         movzbl  3(%rdi), %edx
>         sall    $24, %edx
>         orl     %edx, %eax
>         ret
> 
> I hoped for much better code. I hoped to avoid ifdef's depending on
> endianess, but this means I can't.
> Am I missing something obvious that precludes the compiler from
> optimizing the expression?
> This is not a regression and other compilers didn't do any better, so
> I hope I'm just missing something.
I don't know to to solve your problem exactly, however rewriting your
code solves it (I think my code does the same as yours...):
unsigned int get_le32(unsigned char *p)
{
  unsigned char p1[4];
  for(int i = 0; i < 3; ++i)
    p1[i] = p[i] << i*8;
  return p1[0] | p1[1] | p1[2] | p1[3];
}
gets compiled into two lines:
	movzbl	(%rdi), %eax
	ret

(this is both gcc 4.7 from the trunk and gcc 4.4.5, also x86-64)

Axel