Age | Commit message (Collapse) | Author | Files |
|
|
|
* an image compiled from scratch boots straight into X, although the avahi seggault is still there
|
|
have been 'CPP'. I have tested build for gcc-cross-4.2.2 recipe.
|
|
|
|
|
|
* complimentary to 4b9bf269fccb2d8469b9d1ccfe6d73ac060a6b29
* "upstream: n/a" should really have a more verbose explanation
is it a hack, is upstream being a d0rk, did you try but failed?
|
|
* closes 3607 3732
|
|
violation and makes sure that gcc-cross-sdk will build as well.
* For gcc4.3 we need to look at the gmp/mpfr again as they will be required for
everything and not just gfortran
* Apply a patch to gcc_4.2.2 but only if we want to build gcc. We want to be able
to use arm-angstrom-linux-gnueabi-gfortran and the GF="$(GFORTRAN)" would have set
the fortran to 'gfortran' and we ended up with a build error. Some more experiments
with GFORTRAN_FOR_TARGET are needed.
|
|
|
|
f951 was not executable because libmpfr and libgmp which were build by
OpenEmbedded were not found. They resist in the staging lib directory and
we do not have the LD_LIBRARY_PATH set.
--enable-rpath for gcc was not an option, as we would have difficulties to
say which path would be checked (we install to cross and not staging).
Attempt to set an rpath when linking to mpfr/gmp. We pass --with-mpfr to the
gcc-cross, patch the configure to add an -Wl,-rpath, to gmplibs, this gets
set to GMPLIBS, which gets replaced to HOST_GMPLIBS in the Makefile inside the
builddir, which will be used when linking the f951 utility.
|
|
Move eglibc to revision 4355 which has -Os support.
|
|
|