> On Tue, May 8, 2012 at 10:37 AM, Jan Hubicka <hubi...@ucw.cz> wrote:
> >
> > > Hi Honza,
> > >
> > > I added L_gcov_reset and L_gcov_dump for the new interfaces, and also
> > > added a description into the gcov man page. Let me know if it looks
> > > ok now.
> > >
> > > Bootstrapped and tested on x86_64-unknown-linux-gnu.
> > >
> > > Thanks,
> > > Teresa
> > >
> > > 2012-05-08 Teresa Johnson <tejohn...@google.com>
> > >
> > > * libgcc/libgcov.c (gcov_clear, __gcov_reset): New functions.
> > > (__gcov_dump): Ditto.
> > > (gcov_dump_complete): New global variable.
> > > (__gcov_flush): Outline functionality now in gcov_clear.
> > > * gcc/gcov-io.h (__gcov_reset, __gcov_dump): Declare.
> > > * libgcc/Makefile.in (L_gcov_reset, L_gcov_dump): Define.
> > > * gcc/doc/gcov.texi: Add note on using __gcov_reset and __gcov_dump.
> >
> > It seems OK now, though gcov_clear will end up being in both gcov_reset and
> > gcov_flush,
> > but I suppose it is short enough to make this issue mood
> > (otherwise we could make it hidden exported from gcov_flush psection)
> >
> > Honza
>
> The same issue is going to exist with gcov_exit and gcov_dump/gcov_flush, and
> with gcov_dump_complete. I agree that it seems cleaner to export them hidden
> so I have done that. Bootstrapped and re-testing is in progress. Ok for trunk
> assuming regression tests pass? New patch is below.
OK,
thanks
Honza
