[FFmpeg-devel] [PATCH] Add log2 check to configure and equivalent to internal.h

[Diego Biurrun]

On Tue, Jul 14, 2009 at 06:47:03PM -0300, Ramiro Polla wrote:
> On Sat, Jul 11, 2009 at 6:46 AM, M?ns Rullg?rd<mans at mansr.com> wrote:
> > Reimar D?ffinger <Reimar.Doeffinger at gmx.de> writes:
> >> On Sat, Jul 11, 2009 at 12:18:03AM +0100, M?ns Rullg?rd wrote:
> >>> Alex Converse <alex.converse at gmail.com> writes:
> >>> > If someone on a system without log2 could test this, that would be great.
> >>> >
> >>> > Regards,
> >>> > Alex Converse
> >>> >
> >>> > diff --git a/configure b/configure
> >>> > index 7998273..48630f1 100755
> >>> > --- a/configure
> >>> > +++ b/configure
> >>> > @@ -911,6 +911,7 @@ HAVE_LIST="
> >>> > ? ? ?libdc1394_1
> >>> > ? ? ?libdc1394_2
> >>> > ? ? ?llrint
> >>> > + ? ?log2
> >>> > ? ? ?loongson
> >>> > ? ? ?lrint
> >>> > ? ? ?lrintf
> >>> > @@ -2078,7 +2079,7 @@ check_lib math.h sin -lm
> >>> > ?check_lib va/va.h vaInitialize -lva
> >>> >
> >>> > ?# test for C99 functions in math.h
> >>> > -for func in llrint lrint lrintf round roundf truncf; do
> >>> > +for func in log2 llrint lrint lrintf round roundf truncf; do
> >>> > ? ? ?check_exec <<EOF && enable $func || disable $func
> >>> > ?#include <math.h>
> >>> > ?int main(void) { return ($func(3.999f) > 0)?0:1; }
> >>>
> >>> What's with that weird runtime check anyway? ?Looks pretty useless to
> >>> me.
> >>
> >> Not sure if I understand you right, but if the compiler optimizes away
> >> the function call it wouldn't be a good way to check if the function is
> >> available...
> 
> This happens on cygwin with gcc 4.3.3. It uses a built-in log2 for the
> check to pass, and fails to link aaccoder.o because of undefined
> _log2.
> 
> > Why are these functions not checked with check_func, which seems to
> > work for everything else?
> 
> I don't know why check_exec is being used, but I can confirm that
> using check_func for log2 makes it work on cygwin gcc 4.3.3.

'svn blame' doesn't really know either, so change it to check_func.

Diego