[distcc] Re: building gcc for mingw requires a patch
ericw at evcohs.com
Sun Aug 1 04:03:38 GMT 2004
On 31 Jul 2004 at 14:41, Dan Kegel wrote:
> E. Weddington wrote:
> >>Someone recently said that using gcc and distccd on cygwin
> >>worked fine for small apps, but crashed sometimes compiling larger things.
> >>During that discussion, someone asked whether one could just
> >>compile gcc to run on mingw to avoid possible cygwin problems.
> >>For reference, here's a link to the gcc bug report about this being hard;
> >>it contains a link to the latest patch that fixes collect2 so it runs on mingw.
> >> http://gcc.gnu.org/PR14316
> > Also note that for building some targets for host=mingw, an additional fix/patch is needed for
> > fixinc / fixinc.sh. AFAIK, there isn't (yet) a PR for this (I've been meaning to do this for a while).
> > The targets that I know of that require a fix are avr and m68k, but there could very well be
> > others.
> Looking at your previous post,
> I see I'm just repeating stuff you've already said.
> Ah, well, it bears repeating :-)
Yes, and often, until it gets fixed in CVS. :-)
In re-reading that post, the solution that has worked for me is to patch fixinc.sh. In that script is
a list of targets where fixinc is not built. I have put the avr and m68k targets in this list and
everything *seems* to be fine. I don't know if this would be the correct solution, but AFAICT the
avr and m68k don't seem to need fixinc.
> Do you have any sort of URL for the fixing patch you mentioned?
> (http://gcc.gnu.org/ml/gcc/2003-04/msg01295.html looks related,
> but that's for gcc-3.2; 3.3 is supposedly fixed.)
This doesn't seem to be related; this post talks about issues with memcpy and family. The
issue with building fixinc for host=mingw is the POSIX function calls that are in the fixinc
source; this is related to the collect2 issue as there are also POSIX calls in the collect2 source
and hence problems with building for host=mingw for collect2.
More information about the distcc