Re: [PATCH] mm/slab.c remove impossible <0 check - size_t is not signed - patch is against 2.6.1-rc1-mm2

[Jens Axboe <axboe@suse.de>]

On Thu, Jan 08 2004, Paul Jackson wrote:
> Jason asked:
> > Well, anything wrong in cleaning them [unsigned compare warnings] up?
> 
> It's more important that we write code that will fit in our limited
> human brains than that we write code that will avoid spurious warnings
> from gcc ('spurious' meaning warnings for code that gcc will correctly
> compile anyway).
> 
> Or, see a couple months ago, in a thread with the Subject of:
> 
>   [PATCH] irda: fix type of struct irda_ias_set.attribute.irda_attrib_string.len
> 
> in which Linus wrote:
> > That's why I hate the "sign compare" warning of gcc so much - it warns 
> > about things that you CANNOT sanely write in any other way. That makes 
> > that particular warning _evil_, since it encourages people to write crap 
> > code.

That's fine and has its place, no doubt about that. It doesn't cover the
patch in this thread though. The check is dead code. It's a cosmetic
problem though, gcc should not generate the code checking for < 0.

-- 
Jens Axboe