On Thu, May 15, 2014 at 05:16:00PM +0200, Jakub Jelinek wrote:
> On Thu, May 15, 2014 at 05:08:10PM +0200, Marek Polacek wrote:
> > On Thu, May 15, 2014 at 11:39:26AM +0100, Ramana Radhakrishnan wrote:
> > > What's the overhead with bootstrap-ubsan ?
> >
> > I timed normal bootstrap and bootstrap-ubsan on x86_64, 24 cores,
> > Intel(R) Xeon(R) CPU X5670 @ 2.93GHz (aka cfarm 20) and the results:
> >
> > --enable-languages=all
> >
> > real 35m10.419s
> > user 204m5.613s
> > sys 6m15.615s
> >
> > --enable-languages=all --with-build-config=bootstrap-ubsan
> >
> > real 71m39.338s
> > user 347m53.409s
> > sys 7m44.281s
>
> Note that doesn't mean -fsanitize=undefined generated code is twice as slow
> as code compiled without it, guess most of the extra overhead is actually
> compile time (mostly larger cfg due to all the extra checks).
And I think the main offender will be the NULL checking with its extra bbs. So
to
alleviate the cfg size one can use -fsanitize=undefined -fno-sanitize=null - I'd
expect this to make a significant difference.
Marek
