Re: [PATCH 10/25] Unionfs: add un/likely conditionals on copyup ops

From: Adrian Bunk
Date: Wed Sep 26 2007 - 14:34:41 EST

On Wed, Sep 26, 2007 at 09:40:20AM -0400, Erez Zadok wrote:
> Also, Auke, if indeed compilers are [sic] likely to do better than
> programmers adding un/likely wrappers, then why do we still support that in
> the kernel? (Working for a company tat produces high-quality compilers, you
> may know the answer better.)
> Personally I'm not too fond of what those wrappers do the code: they make it
> a bit harder to read the code (yet another extra set of parentheses); and
> they cause the code to be indented further to the right, which you sometimes
> have to split to multiple lines to avoid going over 80 chars.

There are some places in generic code where it makes sense, e.g.:
#define BUG_ON(condition) do { if (unlikely(condition)) BUG(); } while(0)

If you run into a BUG() it's anyway game over.

And there are some rare hotpaths in the kernel where it might make
sense, and many other places where the likely/unlikely usage that might
be present doesn't make sense.

Unless you know you need it you simply shouldn't use likely/unlikely.

> Cheers,
> Erez.



"Is there not promise of rain?" Ling Tan asked suddenly out
of the darkness. There had been need of rain for many days.
"Only a promise," Lao Er said.
Pearl S. Buck - Dragon Seed

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at
Please read the FAQ at