[ovs-dev] [PATCH 02/22] deal with platforms where backtrace() is in a different library than libc.

Ben Pfaff blp at nicira.com
Tue Apr 23 18:05:07 UTC 2013


On Tue, Apr 23, 2013 at 09:34:13AM -0700, Ben Pfaff wrote:
> On Tue, Apr 23, 2013 at 01:30:22AM +0000, YAMAMOTO Takashi wrote:
> > hi,
> > 
> > > On Mon, Apr 22, 2013 at 10:20:00PM +0900, YAMAMOTO Takashi wrote:
> > >> execinfo for NetBSD and ubacktrace for uClibc.
> > >> i don't know if the latter is relevant to Open vSwitch, though.
> > >> 
> > >> Signed-off-by: YAMAMOTO Takashi <yamt at mwd.biglobe.ne.jp>
> > > 
> > > I'm pretty sure this change is going to break (or at least reduce
> > > functionality of) lib/timeval.c on GNU/Linux because it checks for
> > > HAVE_EXECINFO_H and I think that this change removes that check.
> > 
> > thanks for reviewing.
> > 
> > it seems HAVE_EXECINFO_H is provided by another execinfo.h check in
> > configure.ac.  (why having the check in two places?)
> > 
> > my change seems to break the "Defines HAVE_BACKTRACE" part, though.
> > 
> > is there any reason why you check HAVE_EXECINFO_H in lib/timeval.c
> > and HAVE_BACKTRACE in lib/backtrace.c?
> 
> I didn't realized that we had the check in two places.  I'll take a
> closer look soon.

I applied this patch to master.



More information about the dev mailing list