Age | Commit message (Collapse) | Author |
|
Change-Id: I991306d6c9df62b04fa60f20774bc54d7e4af979
|
|
Change-Id: I1ad2848148a1df11b1885193643b5ee1e422b918
|
|
Change-Id: Ic58a37cc6e0b9191dbd320a6b1f164ba50bb3f02
|
|
Change-Id: Ieaa6b4e094d0112402a7d3a72dcb8b660cddc23d
|
|
The king is dead, long live the king.
Change-Id: Ie1b9b97f8fd98d870e7c6e0799bef8327683f0d1
Link: https://android.googlesource.com/platform/prebuilts/clang/host/linux-x86/+/master/GCC_4_9_DEPRECATION.md
Bug: 147240075
Test: N/A
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
|
|
Change-Id: I413988e460414739ddb0b902f50703a34c6d4beb
|
|
ca9212a5c3
Change-Id: I744691899014f66669bdf38aaf7676bb2cc83f9e
|
|
Change-Id: I917466515f9a545d04f7b1d19208e25e0dc74e97
|
|
Change-Id: I27ce746729728e95c05ef4fea75dcc7cdb54f5d5
|
|
As mentioned in the commit where this was added, this doesn't seem to be
needed any more. It is also actively preventing us from removing the code in
the Android LLVM build script that copies bionic's stdatomic.h into clang's
include directory, since clang's stdatomic.h will try to use the system's
stdatomic.h (i.e. this symlink in host builds) if available, and bionic's
stdatomic.h depends on bits/stdatomic.h in C, which doesn't have its own
symlink from here, resulting in C builds that include stdatomic.h being broken
(which is a relatively good sign that nothing is actually depending on this
symlink being here).
Bug: 147765083
Change-Id: I4f704e423de12ea2927d6dc95c97384cb7cfd07d
|
|
Change-Id: I142eaae42a080e7740d37ee6f8ae2e94a97d684d
|
|
51b055fd85 am: f9884bc921 am: 35b4d5b5d7
Change-Id: I8bc7281ea8c75fa5691a9cb38ae6a97071853b37
|
|
51b055fd85 am: f9884bc921
Change-Id: Id33594c9555a626085216d94e2258f2915fa7374
|
|
am: 51b055fd85
Change-Id: I2f26e76d48844555617a7859ab4d05e3e3f00d65
|
|
This reverts commit 4416b2071855d71b6ee06d71540acac1a1d66672.
This re-renames the gcc binary to cause explicit breakage. It may be reverted by buildcop.
Reason for revert: unless the mingw builds also required this, I don't think this was the source of any of the reported failures.
Bug: 147240075
Change-Id: Ib82436bd003150c6a54041ac61399fcbbceb6eeb
|
|
Change-Id: I01471ed0d7bcae9447f613b433799ccf814c1df1
|
|
cb338fb333 am: 960a33dfa3
Change-Id: I2102825e34c4d245cc3f8ab1606d26a5dc5192d7
|
|
cb338fb333
Change-Id: Ifebf3f0ddb1d05f41353304c3976f9ce86c0acef
|
|
am: 4416b20718
Change-Id: Ib606a9b8b92bfbfde3df8ec11b3f6b18b06cde97
|
|
This reverts commit faf5c0c0ba267a56ee16a88bc8983ec9668229e6.
Reason for revert: BUG: 147258869
Change-Id: Ie6ae101dfea020815c54140a3325154d742cc2ed
|
|
am: 4dfff5c29f
Change-Id: I5e8d14f805ebcfd51a1f6d5d2517cd4792468368
|
|
Change-Id: I299b13bae77e5cf864d3fb4366ebb993c34c38e3
|
|
am: faf5c0c0ba
Change-Id: I3fbc00aec5c2ab1cde263fadf69598115487bbb7
|
|
As per:
https://android.googlesource.com/platform/prebuilts/clang/host/linux-x86/+/master/GCC_4_9_DEPRECATION.md
We're about to remove gcc-4.9; move the binaries to see if we can spot
any build breakages in pre or post submit testing.
Build cops may revert this change if necessary.
Please report any unexpected build breakages to: android-llvm@googlegroups.com
Bug: 147240075
Change-Id: Ic28f409ffe7e526ed0eb1b0d55b9162f69d508c3
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
|
|
am: 10b3b0d2fd
Change-Id: I2dc765bed3479df981eef26e8c4e3494be90d9c1
|
|
am: 38a84132fa
Change-Id: If978195103a7894e0ee7e6353239bf27bd22be44
|
|
am: 33581f7b65
Change-Id: I0435e4e98154915e04a563ba0afb8e230a5a4e56
|
|
This CL checks in glibc 2.17. This was built using the newly-included
build-with-previous-gcc.sh script. build-*-multilib-toolchain.sh is more
or less the same as it was; it learned to copy its wrapper into the new
toolchain root, and now knows how to apply things in sysroot-patches.
For a full diff between old and new, look here (sorry external people;
please use `diff` manually):
https://paste.googleplex.com/4796800347144192
In the 2.15 upgrade CL (I355e28b922101700719c0cc27f588782f96979d4), the
commit message notes that it also:
- added a symbolic link to stdatomic.h
- copied libxcb.so.1{,.1.0} from precise's /usr/lib/* to the prebuilts
glibc dir.
Despite no builds breaking without it, I added the first bullet as a
sysroot-patch.
For the second bullet, it seems like a really bad idea to try to copy
libxcb.so from a modernish Debian into an essentially 5 year old
sysroot. cs/ for `libxcb package:^android$ -file:^prebuilts/
-file:quipper/testdata -file:^external/` gives me a single result.
Namely, a directory in vendor/ apparently provides its own libxcb.so.1.
(And a few places in external/ reference it in documentation? Hard to
tell if we ever actually need it, though)
So, unless we can find something it breaks, I'm dropping the libxcb bits
from this (note that the sysroot *does* still contain a libxcb on its
own, like the previous one does.)
Manual actions:
- Copied over CAPS root files from the glibc-2.15 dir not already
present in this directory (MODULE_LICENSE_GPL, OWNERS, NOTICE, etc.)
Bug: 111358231
Test: A few:
- checkbuilds of taimen-eng, hikey960-eng, marlin-eng,
sdk_gphone_x86-userdebug
- `make -j80 dist TARGET_PRODUCT=cf_x86_phone ndk_translation_all`
- `emulator -no-display` on sdk_gphone_x86-userdebug booted and was
`adb`able
- built the emulator branch
(I8936e4a3bddfd500dadb8034f0b272eb1c463978); tests pass. Using the
emulator on a newly-built sdk_gphone_x86-userdebug image boots and
is `adb`able
- `make -j80 -k PRODUCT-sdk_x86-sdk dist` builds
Change-Id: Ieeed444801160e3d288bbad7d871067e81411118
|
|
|