Re: BUGed to death

[Duncan Sands]

On Monday 14 April 2003 22:19, Martin J. Bligh wrote:
> 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 ;-(
>
> I was pondering CONFIG_RUN_WILD_NAKED_AND_FREE, but maybe we can
> just nail a few of the hottest path ones instead (I think you did
> a couple already recently). I guess that suggestion isn't much
> use without more profile data though ;-)

You would think that the compiler would consider a branch leading to
ud2 (i.e. BUG()) to be "unlikely", but it doesn't seem to.  Maybe some
improvement can be made there.

All the best,

Duncan.

PS: gcc 3.2.3