This is the mail archive of the crossgcc@sourceware.org mailing list for the crossgcc project.
See the CrossGCC FAQ for lots more information.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
Other format: | [Raw text] |
On Tue, Oct 19, 2010 at 12:28 PM, Arnaud Lacombe <lacombar@gmail.com> wrote: > I'll try to have a look to the GCC default behavior. CT-ng's "I know > more than everybody else what should be done" looks definitively "odd" > [to stay polite]. Indeed. I am confused about what gcc's default behavior should be as well. > A. > > On Tue, Oct 19, 2010 at 3:19 PM, Anthony Foiani > <anthony.foiani@gmail.com> wrote: >> Greetings, Bryan -- >> >> On Tue, Oct 19, 2010 at 10:52 AM, Bryan Hundven <bryanhundven@gmail.com> wrote: >>> I got the same problem after rebuilding. This is related to the "c++ >>> includes not getting searched for correctly?" thread. >> >> It might be related to it, but as I pointed out, I can't compile that >> as a regular C program, either. ÂSo it's not just C++ includes. ÂMaybe >> we should start calling it the "GCC includes problem?" Â:) >> >>> As Arnaud has said, the "--with-gxx-include-dir" option is not correct >>> and does not solve the problem. >>> >>> As I have mentioned, I think that creating and removing the >>> ${CT_SYSROOT_DIR}/include symlink is also wrong and suggestive that we >>> are forced to create this because we are making a mistake somewhere >>> else (binutils? gcc_static/shared_core? libc-startfiles?) and we >>> should fix that problem first. I believe that if we can solve that >>> problem, the c++ headers issue will go away. >> >> Given that someone (Arnaud?) already traced through the build log and >> saw the eglibc "fenv.h" get overwritten by the GCC version, it seems >> most likely to be somewhere in the GCC portion of things, presumably >> when the final compiler is installed. >> >> I know I got confused when I was looking at the issue before: he's >> correct that the host include files need not (and, arguably, should >> not) be present in the target sys-root, but I know we do need to have >> the host libraries (libgcc, libstdc++) in the target sys-root. So the ${CT_PREFIX_DIR}/${CT_TARGET}/include is created by the final g++, and we want these headers in the sysroot? The host include files are the ones installed by the "host" gcc_core_{static,shared} toolchains, right? >>> I will focus on this issue today, and will have some updates later. >> >> Thanks for looking into this. ÂI'll continue to poke at it as well. >> >> Regards, >> Tony >> > Thank you, -Bryan -- For unsubscribe information see http://sourceware.org/lists.html#faq
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |