Pierre Labastie wrote: > Le 27/03/2013 19:47, Bruce Dubbs a écrit : >> Pierre Labastie wrote: >>> >>> It really looks like it is a tcl error (I cannot think of a reason why >>> gcc would try to compire a *regular* expression). I found this bug report: >>> http://sourceforge.net/tracker/?func=detail&aid=3603557&group_id=10894&atid=110894 >>> It seems to indicate that the limit on memory could be increased at >>> compile time on tcl 8.6. >>> >>> Anyway, I would not count those errors as GCC's. >> Well, yes and no. The error is triggered by the gcc regex. >> >> I looked at the bug and it does say that it is fixed with a compile time >> overridable, but I don't see how to do that. It's not in tcl's >> configure and a search of the source doesn't show anything obvious.
> Downgrading to tcl8.5.13, all the memory errors are gone. > Only the libmudflap "pass41" has an error "execution test", > but with no reference to memeory. I got the head of the development branch for tcl, but that didn't fix it. I may ask on the tcl mailing list. This isn't a critical issue, but somehow I've started to focus on it. I need to dig into dejagnu and expect to try to figure out why these are failing. I have determined that I can isolate the test with, for instance: $ make check-c++ RUNTESTFLAGS="dg.exp=mangle33* --all --debug" The results show up in ./gcc/testsuite/g++/dbg.log For some reason, the above wants to run a subset of mudflap tests too. The problem files seem to compile without error, but the checking shows a problem. I'm going to spend some more time trying to debug these issues. -- Bruce -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
