Re: [PATCH 2/3] objtool: Ignore retpoline alternatives

[Peter Zijlstra <peterz@infradead.org>]

On Thu, Jan 11, 2018 at 09:29:48AM -0800, Linus Torvalds wrote:

> Secondly, we use lots of the the "small numbers for local labels" both
> in inline asm and in *.S files.
> 
> I think doing
> 
>        jne 1f
>   ...
>   1:
> 
> is a _hell_ of a lot more legible than
> 
>         jne .LPrefix_%
>     ...
>   .LPrefix_%
> 
> unless you have some *major* reason to use an explicit label name.

Small number maybe; but the value at hand was a random 999 or so, which
is not a small number.

But I find a descriptive label ever so much better than a random number.

> Sure, if you grew up writing perl, and think that an illegible mess of
> random characters is a requirement for programming, then the ".L%"
> format looks natural.

I grew up on BASIC and have bad memories of random big number goto. I'll
take those random trailing character any day if it includes human
readable bits before.

> But if you're an actual human, the "small numbers as labels" is fine.

I find descriptive labels much nicer than random numbers, I'll take some
crazy characters if so required.


Consider the retpoline thing:

 call .Lset_up_target
.Lcapture_spec:
  pause
  jmp .Lcapture_spec
.Lset_up_target:
  mov %r11, (%rsp);
  ret;


over:

 call 2f
1:
 pause
 jmp 1b
2:
 mov %r11, (%rsp)
 ret


give me the first any day of the week.