>> >> Seems all these bug checks are fairly expensive. I can get 1%
>> >> back on system time for kernel compiles by changing BUG to
>> >> "do {} while (0)" to make them all compile away. Profiles aren't
>> >> very revealing though ... seems to be within experimental error ;-(
>> >
>> > What happens if you just turn BUG_ON into "do {} while (0)"?
>>
>> I believe I already did that by turning BUG() into a null expression.
>>
>> # define BUG_ON(condition)
>> do { if (unlikely((condition)!=0)) BUG(); } while(0)
>>
>> The compiler should be smart enough to optimise that away, methinks.
>
> No, I meant what happens if BUG() is non-trivial and BUG_ON is a no-op.
> I thought it might give an indication of whether time was being lost
> evaluating the condition (occurs with BUG and BUG_ON), or mispredicting
> the branch (only occurs with BUG).
Mmmm. wouldn't that just optimise away the BUG_ONs, but not the BUGs ?
Which will tell you something, but I'm not sure anything interesting.
My impression is that if I do:
if (foo)
do {} while (0);
the compiler will just ditch the whole thing, and not evaluate foo.
I'll admit to not having checked that, but still ....
M.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Tue Apr 15 2003 - 22:00:36 EST