Regenerating the NOTICE file the other day left me scratching my head at
various "how do they differ?" cases, resolved by this patch.
Test: N/A
Change-Id: I4230bfa1d7ec842a4b9fb8c73dd3b39443d74054
Generate a single NOTICE file rather than one per library. All the
headers come from libc these days anyway.
Test: tools/update_notice.sh
Bug: None
Change-Id: I127da185fdabc2815042e19aea74c507ec108f46
Also updates the license files, since apparently they're out of date.
Test: repo upload --cbr .
Bug: None
Change-Id: Ic8d855a7ee5185c4933da699292868e02ef79531
Poking around with -Wundef exposed a pretty horrifying bug, so we
should probably keep it turned on.
Bug: http://b/31496165
Change-Id: Iada279eec91ded0d59ff3841d5a3ce1a36e34c3a
Add an attribute that tells the preprocessor not to guard a
declaration, primarily for use with symbols that get reexported by
libc++ of the form `namespace std { using ::wctrans; }`.
Bug: http://b/28178111
Change-Id: I08c8751214797e37e8f26e7f7416a19e81c2bb4c
Compile headers with both -D_FILE_OFFSET_BITS=32 and -D_FILE_OFFSET_BITS=64.
Bug: http://b/30170081
Change-Id: I92651e075cc69bdc1a2581f99892c9a7fdcdb35b
Test: python run_tests.py
This was causing redundant guards to be emitted in cases where
declarations were already guarded by another macro.
Bug: http://b/30170081
Change-Id: I9080838fbea5a56225df4b26d0918657fb6c4f21
Test: python run_tests.py
Make sure everything works with relative/absolute paths, and with and
without trailing slashes.
Bug: http://b/30170081
Change-Id: Ia13a8ae2fa44590e835d0f4b006d4c31e9eaea83
Test: python run_tests.py
Move the Android.mk to src/ so that mm in that directory works.
Also, remove the dependency on FORCE_BUILD_LLVM_COMPONENTS.
Change-Id: Iad7f3fc96f5f26f9535141d752ba865b40d5ff7e
Zero initialize the contents of ArchMap, keep track of the symbol name
in Declaration, remove unnecessary copies in loops.
Change-Id: I47b07755846f252b83ffc4c89547a34b2f7ab868
Major refactor to use __attribute__((annotate)) to be able to keep
track of the semantic differences between __INTRODUCED_IN(x) and
__INTRODUCED_IN_X86(x), for use in the upcoming preprocessor.
Bug: http://b/30170081
Change-Id: I6496a8c40ba7f4553de9a2be0bbddcf37c813937
The target was moved by commit 5956b4e, but none of the headers
contained were actually necessary to compile any bionic headers.
Bug: http://b/30170081
Change-Id: I8c1a7464de0992c20af194340c0e36d131a836e7
The android-only kernel headers were moved to their own directory by
commit d6e8b8c. Add a new symlink to match.
Change-Id: I06ce7c3167fbb1e058e4a137fae9375bb5792a46
Some of the error messages emitted by versioner (the ones where it was
invoked incorrectly) go to stderr, which meant that the test runner
ignored them. Merge stdout and stderr, and switch from testing for
exact equality to endswith, because of the compilation errors test.
Change-Id: I0e2c25bcc9dea4c12ea82a6a05b29e561a61a902
Git doesn't track empty directories, so most of the tests would fail on
a fresh checkout because of dependencies/common being missing. Remove
the use of dependencies from all of the non-dependency related tests.
Change-Id: I09cc5765aae1576914c1c5d7dfa3fb666eab4a3f
Search for the header/dependency/platform directories in a hard-coded
path relative to $ANDROID_BUILD_TOP when they're not specified.
Change-Id: I476385cfc0247e3b2009348ec37c1810a0e9a7f7
Add a clang-based tool to inspect header availability attributes and
verify them against the NDK platform definitions.
Bug: http://b/28178111
Change-Id: I1bb1925a620e98cc9606cb5a3360b1224c700bd0
This time with more namespace qualification.
Tested manually by building clang_x64/android_relocation_packer locally,
as the Android trybot analyze step does not think it needs to run for
this change.
Original description:
> The code was essentially doing 'using Logger::INFO' in the global
> namespace to make its 'LOG(INFO)' macros work. Unfortunately, C++ does
> not allow you to use using decls on classes like this unless you are in
> a derived class. GCC does not accept this code, and Clang was recently
> updated (LLVM r268594) to reject it as well.
>
> This should fix the Chromium Android ASan build with TOT Clang:
> https://build.chromium.org/p/chromium.fyi/builders/ClangToTAndroidASan/TBR=thakis@chromium.org,sgurun@chromium.org
BUG=609543
Review-Url: https://codereview.chromium.org/1952353005
Cr-Commit-Position: refs/heads/master@{#391952}
(cherry picked from commit 5762af8ad13e62957493c3e4314a234ee57a4200)
Change-Id: Ibc6f6023aef028c5029be128ac799dc67fc6683c
Turn messages that are printed on every run of relocation_packer down to
VLOG(1) to clean up ninja build output.
Bug: 24409581
Change-Id: I040aed6a7b4261eefa6f7278fa451180115b0716
On unpacking, reinstate any p_align values that packing reduced to
page size. Ensures a round-trip pack and unpack is bit-equivalent
to the original input.
https://android-review.googlesource.com/#/c/148492/
Extend unit tests to include ia32, x64, and mips32.
Recreate test data for arm32 and arm64. Generate new test data for
ia32, x64, and mips32.
Bug: http://b/20687795
Bug: http://b/18051137
Change-Id: Ifbca8e206ef447297ba4f19272b813702be27a35
DT_MIPS_RLD_MAP2 stores the difference between dynamic
entry address and the address of the _r_debug (used by GDB)
since the dynamic section and target address are on the
different sides of the hole it needs to be adjusted accordingly
Bug: http://b/20658994
Bug: http://b/18051137
Change-Id: I4c5e379e8a4c68f029467df6dc660f81a7013a09
Having p_align > page_size leads to the situation when striping
packed executables results in unnecessary p_vaddr adjustments.
And it also may result (with probability 1 - 1/sizeof(uintptr_t)) in
misaligned segments following .dynstr
Bug: http://b/20629834
Bug: http://b/18051137
Change-Id: I1c5da4911e4409d63cb09f6b6b0a16ef54f6501b
According to runs on /system/lib there using
unsigned leb128 does not save us any additional
space. In order to keep packing as simple as
possible switch to using signed leb128 for
everything.
Bug: http://b/18051137
Change-Id: I1a47cb9eb2175895b3c3f7c13b4c6b1060de86c0
Statically link libc++ to relocation_packer in
order to make it work from prebuilts
Bug: http://b/18051137
Change-Id: I933ed6a0e48780a26b261069eb6a293432824fe7
This is recent addition to bionic linker. The symbol
versioning was not supported before therefore this bug
went unnoticed.
Also normal exit when there is not enought relocations
to pack. This is to enable integration of relocation_packer
to android build system.
Bug: http://b/20139821
Bug: http://b/18051137
Change-Id: Iaf36ae11c8e4b15cf785b6dd1712a3bdcf47cc45
Print is bad and I should feel bad. Use the logging module instead.
Will follow up by adding a persistent log destination instead of just
the console.
Change-Id: I396ff10712f88a03f8d8183b6de29ea273815962
Right now any changes in here would be innocuous because I manually
update bionicbb, but I'd like to check in the various job
configurations. Once I have we don't want anyone to be able to make
the buildbot run any untrusted code.
Change-Id: Ic050859cd5017615f71c75f995ba21bb45407b05
Cleanspecs must not be removed once they have been built. This means
they can't be reverted, or reliably cherry-picked. Just skip any
changes that include them since they make such a mess.
Change-Id: I3df8d81f93651d573485de7a75ecf5c6278c0001
Some of the commentary in relocation packer code is relevant
only to the packing strategy employed by chromium, and no
longer applies here. This change fixes or deletes it.
Code comment change only; no functional effect.
Change-Id: Id229ee1d802bba608be15b79bc75bf90df557dab
Signed-off-by: Simon Baldwin <simonb@google.com>
Removes:
- src/run_length_encoder.h
artifact of the original packing tool, not used here
- test_data/generate_elf_file_unittest_relocs.sh
- test_data/generate_elf_file_unittest_relocs.py
test data generation for chromium/gyp, not usable here
- README.TXT
because it is now almost entirely outdated
Change-Id: Ic4cd372647d9a365dc52833a6cc1cf66f0c95ec9
I still had the service that was relying on this running on top of
local changes, so this hadn't been a problem yet.
Change-Id: I63b45b8c7cf81972dbb7128013c1c777a2342d4c
https://issues.jenkins-ci.org/browse/JENKINS-27256
1.601 broke the ability to get a build's URL before the build had
actually started. The bug is pseudo-fixed, but would require
installing a new plugin and fixing the python jenkinsapi plugin for
the time being to use /queuefix/ rather than /queue/.
Just avoiding logging a URL for now.
Change-Id: Ibf90e5887fc4532dac688082ad7729787897da11
1. One binary for all architectures
2. Generalize (and slightly improve) compression
2.1 works on all relocation types (rela?.dyn section only so far)
2.2 Uses same format to encode ElfW(Rel) as well as ElfW(Rela) tables
Bug: 18051137
Change-Id: I66c95d9076954ca115816fc577d0f5ef274e5e72
Currently, we get a 404 from Gerrit in the event that two projects
have the same Change-Id. We should be able to handle this and actually
cherry-pick each change so we can check multi-project changes, but for
now just skip these changes.
Change-Id: I7bc63208998d58beec83b71b302450d9be3ea026
Guarding based on the Gerrit owner can be circumvented by an arbitrary
user uploading a different patch with a Change-Id that is non-unique,
with the other copy being owned by a Googler.
Change-Id: I5414b679e361d4c38d70bf9c4516c122f668fc49
Begin adding some unit tests. Testing the Jenkins interface will
probably be difficult, but testing that we can properly handle
messages from Gerrit is easy enough.
Change-Id: Id7e8e175e0f064fd10715febe40040a65564c701
Googlers should be able to tell the buildbot to launch builds for an
external contributor's change, but Gerrit is determined to make this
hard for us. The email sent by Gerrit for comments doesn't actually
contain the email address of the commenter in any clear way, only the
name. It does however contain a list of `Gerrit-Reviewer: Name
<email>` entries that could be used for this though.
Change-Id: Iee61f06dcd8f0024f16b535ea9a34765bf01b2e7
Since we often make changes that might break on other architectures,
let the buildbot deal with some of these for us. They can be invoked
by `bionicbb:ARCH`, where `ARCH` is one of:
* arm
* aarch64
* mips
* mips64
* x86
* x86_64
Specifying arm isn't particularly interesting (since the default
target for the buildbot is hammerhead), but there are some differences
in the math instructions available for the default ARM target, so it
could be helpful for testing changes to the compiler-rt builtins.
Change-Id: I94018fd3c30d26fcf405e747fc633cbdd08ff4e5
Apparently Google abandoned the 2 space indent for Python long ago.
Helps to actually read the style guide before trying to adhere to it.
Change-Id: I4feb019f0916f9d8e4f78c0dbeafbe45d8a46bfd
Making `bionicbb` its own package required `PYTHONPATH=..`, which we
don't want. Just use a relative import instead.
Change-Id: I59b9852522118ece89829288a4921902e756e4aa
The only JSON file is the client secret file, which we really don't
want leaking. Same goes for the oauth storage file.
Change-Id: Ie860704574ec37ccfc50ff7c9d59717a7b9b443b
These have been sitting around in a git repo on my machine for a
while. They're now big an important enough that I'd like to both keep
them securely backed up, and also have my changes reviewed.
Change-Id: Ic4545149b4b07f0d57b21cac32aab8553dceb567
