[Mesa-dev] [PATCH] util: Implement assume() for clang.

[Jordan Justen]

On 2014-11-21 15:53:35, Matt Turner wrote:
> On Fri, Nov 21, 2014 at 3:47 PM, Jordan Justen
> <jordan.l.justen at intel.com> wrote:
> > On 2014-11-21 15:17:00, Matt Turner wrote:
> >> ---
> >>  src/util/macros.h | 12 +++++++++++-
> >>  1 file changed, 11 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/src/util/macros.h b/src/util/macros.h
> >> index da5daff..b67596d 100644
> >> --- a/src/util/macros.h
> >> +++ b/src/util/macros.h
> >> @@ -29,6 +29,10 @@
> >>  #  define ARRAY_SIZE(x) (sizeof(x) / sizeof(*(x)))
> >>  #endif
> >>
> >> +/* For compatibility with Clang's __has_builtin() */
> >> +#ifndef __has_builtin
> >> +#  define __has_builtin(x) 0
> >> +#endif
> >>
> >>  /**
> >>   * __builtin_expect macros
> >> @@ -85,7 +89,13 @@ do {                        \
> >>   * Assume macro. Useful for expressing our assumptions to the compiler,
> >>   * typically for purposes of silencing warnings.
> >>   */
> >> -#ifdef HAVE___BUILTIN_UNREACHABLE
> >> +#if __has_builtin(__builtin_assume)
> >> +#define assume(expr)             \
> >> +do {                             \
> >> +   assert(!"assumption failed"); \
> >
> > Did you mean to remove this assert?
> >
> > With that removed
> > Reviewed-by: Jordan Justen <jordan.l.justen at intel.com>
> 
> No. Clang's documentation says that if the expression you're assuming
> is false, behavior is undefined.

Worth noting in the commit message or a comment?

> The assertion is to help us catch
> cases where that happens.

So ...  assert(expr) instead?

-Jordan