Re: BUGed to death

[Martin J. Bligh]

>  > It is questionable.  Since even in core kernel code there are
>  > many places with
>  > 	if (cond)
>  > 		BUG();
>  > rather than
>  > 	BUG_ON(cond);
>  > it may be worth seeing if converting them makes a difference
>  > (increases code size though).
> 
> The spinlock code sticks out as a possible good target.
> Any takers for benchmarking ?

I can try that ... However, do I have to use gcc > 3 to see anything?
I had some vague recollection that unlikely() doesn't do much for gcc 2.95,
is that correct?

M.