Age | Commit message (Collapse) | Author | Files |
|
|
|
|
|
and '86581f7de9bc72fe5e21778b4442cc0e59e96fbf'
|
|
pleasure.
|
|
|
|
|
|
|
|
from poky
|
|
|
|
|
|
|
|
|
|
common .inc file
|
|
binutils configure (from poky)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
binutils-cross: ditto
|
|
|
|
*mumbles something about unifying recipes*
|
|
|
|
|
|
|
|
see email thread at http://lists.openembedded.org/pipermail/openembedded-devel/2006-October/000523.html for details
|
|
OSX claims that stopc might end up uninitialized. Initialize
it to '0' to and hope it works out. This fixes the compile and
I have an ARM EABI toolchain on OSX that could work...
|
|
|
|
|
|
|
|
remove binutils 2.15.* except 2.15.94
|
|
from the changelog:
1. Update from binutils 2006 0427.
2. Fix an objcopy regression (PR 2593).
3. Reduce ar memory usage (PR 2467).
4. Allow application specific ELF sections (PR 2537).
5. Fix an i386 TLS linker bug (PR 2513).
6. Speed up ia64 linker by 1300X in some cases (PR 2442).
7. Check illegal immediate register operand in i386 assembler (PR
2533).
8. Fix a strings bug (PR 2584).
9. Better handle corrupted ELF files (PR 2257).
10. Fix a MIPS linker bug (PR 2267).
|
|
|
|
be of use on other versions but I've not tested that.
|
|
|
|
The GNU EABI override for ARM
was wrong. Simply remove it. If we
change the DEFAULT_PREFERENCE for glibc2.4
we can update again.
|
|
Make older binutils snapshot the default
for familiar distribution in the dev
tree to avoid further errors when newer
versions gets added.
packages/binutils/binutils:
Lower the default preference on the binutils
CVS snapshot. While it can be used with
glibc 2.4 with EABI and without it, glibc
2.3.5 is broken with this version
|
|
|
|
non-arm-linuxeabi builds
|
|
|
|
|
|
- the same bug exists in earlier versions and is probably fixed by the
same patch, objcopy (==strip) would identify an ro file and abort the
overwrite (smart_rename), but the caller in objcopy fails to check the
error code and so the error is not reported
|
|
x.x.x+cvsYYYYMMDD format and some includes some minor whitespace cleanup.
|
|
|
|
|
|
|
|
This patch fixes problems with the ARM thumb specific code generation. It
also adds debugging to bfd/elf32-arm.c to detect ref counting problems -
one error message can fire in theory in the absence of thumb support if
a union refcount field is used to refcount when it is set to something else.
The error should be ignorable, in that the code should not have changed in
effect.
|
|
2.16 now includes the patches which ensure that HOST_CC_ARCH is passed to
the build, other than that this commit should only affect binutils for
linux-uclibc targets. The patches which enable recognition of
linux-uclibc*' as well as 'linux-gnu*' have integrated - in fact there are
more places now where the change is made, which may improve some things.
The apparently spurious 'autoreconf' of certain subdirectories in the build
tree has been commented out (it did nothing because there is no configure.in
or .am in the build directory).
|
|
packages in a build. The change fixes problems in the following packages:
binutils{,-cross}_2.15.94.0.1
HOST_CC_ARCH was not being passed to HOST_CC. The problem does
not seem to be present in binutils* 2.16, however this version
is still used for uClibc builds.
gcc{,-cross,-cross-initial}_3.4.4
HOST/TARGET_CC_ARCH was not used in the appropriate places, plus
this was not sufficient - FLAGS_FOR_TARGET (a gcc make variable)
is now preloaded with the target CC_ARCH values, which causes all
the builds (gcc, g++, gcj, fortran) to get the correct values.
gcc OTHER
*NOT* fixed - other versions of gcc will still need the _CC_ARCH
fix in the same lines (CC_FOR_TARGET, GCC_FOR_TARGET, CXX_FOR_TARGET),
however so long as these builds use gcc3-*.inc the FLAGS_FOR_TARGET
fixx will be there.
uclibc{,-initial}_0.9.27
TARGET_CC_ARCH added to the do_compile. For the do_stage step the
build actually compiles target code (make utils), but this will not
accept a CC with whitespace. The TARGET_CC_ARCH flags have therefore
been added to CFLAGS (a bit ugly, but it works).
These changes have been tested for both cross and target builds, the changes
should be fine on self-hosted (i.e. target-hosted) builds, if anyone actually
does these, but note that the gcc build specifically tests for build==host and
does *not* do certain _FOR_TARGET settings in this case (this was the case
before this change too) - if you are seeing strange settings on self-hosted
gcc builds (before and after this change) check the do_compile environment
variable settings.
|