This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: lto-bootstrap tweeks
On Mon, Jan 10, 2011 at 03:26:58PM +0100, Jan Hubicka wrote:
> Hi,
> > Honza,
> > Could you take a look at bootstrap-lto on x86_64-apple-darwin10?
>
> The PR does not reproduce for me. It would really help if you just explored
> the backtrace from the ICE and tried to work out where we trigger the streaming
> of optimization attribute despite the fact that it is not present in the preprocessed
> testcase...
>
> > I'll reconfirm this with your patch but I suspect we are still blocked
> > by PR46037 from doing an lto-bootstrap. It would be nice if we could
> > get lto-bootstrap working on darwin for gcc 4.6. Also, I suspect that
> > cygwin might also share this problem if -fuse-linker-plugin is dropped
> > since its lto code has common origins with darwin's.
>
> I will give it a try how linker disabled lto bootstrap works on Linux, but doing
> so is quite pointless since all of libbackend.a is dropped.
> You might consider gluing our plugin into apple linker plugin. At first glance these
> seems quite compatible.
>
> Honza
Honza,
Also, can you confirm what the minimal configure flags are? I see that HJ does
Intel64-bootstrap-lto tests with...
configure flags: --enable-clocale=gnu --with-system-zlib --enable-shared --with-demangler-in-ld --with-build-config=bootstrap-lto --with-fpmath=sse --enable-languages=c,c++,fortran,java,lto,objc
However I thought that an explicit --enable-stage1-languages entry for lto was required.
Jack