Adding -Wl,--start-group, -Wl,--end-group so that ld.bfd
can resolve library dependencies in case of Android
component build for mipsel.
Bug: 794486
Change-Id: I9ce1867b614aea51382637bc59aa7c7b8c2adbde
Reviewed-on: https://chromium-review.googlesource.com/823905
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Commit-Queue: Robert Sesek <rsesek@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#525039}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 71279763dbdd0d8d3a3cb1fb038587ab752ace47
The delay before VM termination was accidentally moved in
https://chromium-review.googlesource.com/c/chromium/src/+/734128 and so
on fast shutdown the results .json file might not be flushed to the disk
image. This could result in a truncated output.json (see linked bug)
causing bot failures.
Also, unify delay to longer of current values as they're arbitrary
anyway.
TBR=wez@chromium.org
Bug: 796026
Change-Id: Ie8986ec928e37d251571590c9f05b58109df0f1f
Reviewed-on: https://chromium-review.googlesource.com/833757
Reviewed-by: Scott Graham <scottmg@chromium.org>
Commit-Queue: Scott Graham <scottmg@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524966}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: c552dad1797351aa6b54998e2a8fca21d0c07758
We cannot rely on an instrumentation extra to pass the
value of $UBSAN_OPTIONS to Java because some tests, such as
webview_instrumentation_test_apk, do not use "am instrument" to launch
APKs containing native code. The solution is to move the code that
sets $UBSAN_OPTIONS into LibraryLoader and arrange for it to be run
by all clients of LibraryLoader.
There doesn't seem to be a simple way to deduplicate the value of
$UBSAN_OPTIONS between the Java code and the Python code so I've
duplicated it with a comment for now.
Bug: 469376
Change-Id: Iefb225601d5e963efbcd6cb4db8d158bb35afd19
Reviewed-on: https://chromium-review.googlesource.com/826106
Reviewed-by: agrieve <agrieve@chromium.org>
Reviewed-by: Daniel Cheng <dcheng@chromium.org>
Reviewed-by: John Budorick <jbudorick@chromium.org>
Commit-Queue: Peter Collingbourne <pcc@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524845}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: fd089e85bde7973f62edddc417abadc7cf69f95d
This is timing out for me when I use run_chrome_public_test_apk with a filter,
on a nexus 5 device.
There may be a better place to put the timeout, any suggestions welcome :)
Change-Id: Iac12a7843bfa299471ba7b2522de367a82793b5d
Reviewed-on: https://chromium-review.googlesource.com/805136
Reviewed-by: Yoland Yan <yolandyan@chromium.org>
Reviewed-by: John Budorick <jbudorick@chromium.org>
Commit-Queue: Matthew Cary <mattcary@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524720}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 1e934065d2bdb664d1d1042cda9ef8d448bda07f
This patch adds support for clang-based AFDO to Chromium.
AFDO stands for Automatic Feedback Directed Optimization. In short, it
allows you to use the postprocessed output of a special `perf` command
to inform the compiler of hot/cold code, among other things.
This allows users to easily specify a profile to use, and allows for
platforms to specify default profiles. It turns none of this on on its
own.
At the moment, this is very sharp and pointy, since the sample profile
isn't a dependency for anything. Ideally, anything built with AFDO
should depend on the profile. I have a CL to make this easily possible
(CL:827560), but think this is useful enough (given care) to get in now.
(and with clang_use_default_sample_profile set to true). We built
with profiles and failed to `gn gen`, respectively. Patch seemed to be
a nop running on an otherwise "regular" build.
Bug: 794750
Test: Ran the build with and without clang_default_afdo_profile defined
Change-Id: I7b2730d52875f108681232f4d8ce7e82e7570a74
Reviewed-on: https://chromium-review.googlesource.com/828277
Reviewed-by: agrieve <agrieve@chromium.org>
Commit-Queue: George Burgess <gbiv@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524620}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: b822aab9f2aa56d7b34c8b220f37a5debcd3eb33
The freetype unbundle build file is out-of-date, and trying to use it
will result in errors like this:
ERROR Unresolved dependencies.
//third_party:freetype_harfbuzz(//build/toolchain/linux/unbundle:default)
needs //third_party/freetype:freetype_source(//build/toolchain/linux/unbundle:default)
This CL changes the unbundled build file to simply toggle the
use_system_freetype flag.
R=thestig@chromium.orgCC=phajdan.jr@chromium.org
Change-Id: If38ccc51098857cd479707879562b23f8f32e85f
Reviewed-on: https://chromium-review.googlesource.com/830974
Reviewed-by: Lei Zhang <thestig@chromium.org>
Commit-Queue: Thomas Anderson <thomasanderson@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524535}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: a7b961302726f6accc1d70682017b6c36127ae7a
Also update the "Notes for other distros" in the build instructions.
All usages of gconf has been removed, and we depend on gsettings, not
dconf directly.
This is the final step in switching from gconf to gsettings.
Bug: 768027
Change-Id: I4430c261ac2c9691838e7dfe9c0d7c6842fa535f
Reviewed-on: https://chromium-review.googlesource.com/831106
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Commit-Queue: Tim Brown <timbrown@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524503}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 36312fccc6b5f9bba0fa427d6b9756daad3b94e6
Due to bugs in the Fall Creators Update SDK we can't build Chromium with
clang with that SDK. However the debuggers package (windbg and friends)
in the "spring" Creators Update SDK can't handle fastlink PDBs reliably.
So, this repackages the previous toolchain with the debuggers package
from the Fall Creators Update SDK.
This makes no difference to the builds, although it does mean that an
updated copy of dbghelp.dll will be copied to the build directories.
I created this package by manually copying over the new debuggers
directory and then running:
python package_from_installed.py --repackage=...
Bug: 773476
Change-Id: I05b37d518e3700e125eee5bbcb2cbf885d01afc5
Reviewed-on: https://chromium-review.googlesource.com/827629
Reviewed-by: Scott Graham <scottmg@chromium.org>
Commit-Queue: Bruce Dawson <brucedawson@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524481}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: ecef9073a76e802f1989dfc87749f4da0f2722ed
Bug: 768027
Change-Id: Idf3e65c743e79b27e189a3ce4cb7f7e675a9fd03
Reviewed-on: https://chromium-review.googlesource.com/830812
Reviewed-by: Thomas Anderson <thomasanderson@chromium.org>
Commit-Queue: Tim Brown <timbrown@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524474}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 2bede458a4744fdf465620afd871a2f3c892266b
Add 'target_defaults' to the list of top level keys ignored by gypi_to_gn.py
so it can process files that contain that.
BUG=795329
Change-Id: I3733a7efb97366827febf87ec1be95ad8723be23
Reviewed-on: https://chromium-review.googlesource.com/830114
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Commit-Queue: Frank Henigman <fjhenigman@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524449}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: f1d7d41d5aec7d6562f94544675f90eb4e915f44
It appears using external process to write to a file we hold an open handle for
isn't 100% reliable (I suspect we need an fsync somewhere or maybe gsutil is
doing 'mv' somewhere and writes to another inode, or something like that).
Reopening the file before reading from it fixes the bug, at least locally for
me.
R=scottmg@chromium.org, kbr@chromium.org
BUG=795330
Change-Id: I498540ec62439c02243b84a5577c4b225ba5e1f3
Reviewed-on: https://chromium-review.googlesource.com/830257
Commit-Queue: Sergey Ulanov <sergeyu@chromium.org>
Reviewed-by: Sergey Ulanov <sergeyu@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524439}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 48fdf63300aae84f8159c22831df8cff24b5d984
This CL switches Linux to use in-tree fontconfig builds. It exposes some leaks
and races detected by LSAN and TSAN. For now, these will be suppressed since
these issues were preexisting, but instrumenting fontconfig exposed the issues.
R=dnicoara@chromium.org,dpranke@chromium.org
BUG=795110,795148
Change-Id: Ia75db4ced6ec78a5f0610af9ebc78a87840b86f7
Reviewed-on: https://chromium-review.googlesource.com/826403
Commit-Queue: Daniel Nicoara <dnicoara@chromium.org>
Reviewed-by: Daniel Nicoara <dnicoara@chromium.org>
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524389}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 4f1a60552748d150783655ee6f7b6f7084e3af95
This reverts commit 8c29b757124e1ceb921ad61d023b6b741eda92d4.
Reason for revert: <INSERT REASONING HERE>
Original change's description:
> Enable frame pointers on arm32.
>
> LLVM fixed the bug where framepointers were not located next to the return
> address in thumb mode. https://bugs.llvm.org/show_bug.cgi?id=18505
>
> With that fix, enabling framepointers on arm32 bring it in line with other
> build configurations and allows for better stack traces.
>
> Bug: 786450
> Change-Id: I01a1c9f82fac22b2bbc15c6f52deed8aca7263ea
> Reviewed-on: https://chromium-review.googlesource.com/826255
> Reviewed-by: Brett Wilson <brettw@chromium.org>
> Commit-Queue: Albert J. Wong <ajwong@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#524127}
TBR=ajwong@chromium.org,brettw@chromium.org
Change-Id: Ie327529394021473cfffd0ee80baecce4bd83e2a
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: 786450
Reviewed-on: https://chromium-review.googlesource.com/828607
Reviewed-by: Albert J. Wong <ajwong@chromium.org>
Commit-Queue: Albert J. Wong <ajwong@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524337}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: f4bb44bdd8d3cee8685dfe08851383bbfba13535
Studying the code in src/build/android/test_runner.py and related
files (like test_instance_factory and gtest_test_instance), there were
no dependencies on this specific argument's name. Change to use the
standard one so that Android works the same as the other platforms.
BUG=794767
Change-Id: I22f3eea9cb9d1029b778b1d4a878d0e74dc6449c
Reviewed-on: https://chromium-review.googlesource.com/826490
Reviewed-by: John Budorick <jbudorick@chromium.org>
Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org>
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524177}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: bb29325c1aedbb530110644fd38430a905af1b96
64-bit builds have been working for a while now.
32-bit chrome/win still cannot build on Mac hosts due to bug 794838.
TBR=brettw
Bug: 774209,794838
Change-Id: I762853e7e5d2b4470642402e8b585dfe8f84b53e
Reviewed-on: https://chromium-review.googlesource.com/826010
Commit-Queue: Nico Weber <thakis@chromium.org>
Reviewed-by: Hans Wennborg <hans@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524129}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: ce96b47065492432a419a12ee14ed2bf3ee59084
LLVM fixed the bug where framepointers were not located next to the return
address in thumb mode. https://bugs.llvm.org/show_bug.cgi?id=18505
With that fix, enabling framepointers on arm32 bring it in line with other
build configurations and allows for better stack traces.
Bug: 786450
Change-Id: I01a1c9f82fac22b2bbc15c6f52deed8aca7263ea
Reviewed-on: https://chromium-review.googlesource.com/826255
Reviewed-by: Brett Wilson <brettw@chromium.org>
Commit-Queue: Albert J. Wong <ajwong@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524127}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 8c29b757124e1ceb921ad61d023b6b741eda92d4
https://chromium-review.googlesource.com/c/chromium/src/+/823219 rolled
clang past the revision including -finstrument-function-entry-bare. This
reduces the instrumentation size and runtime overhead.
Use this new instrumentation when possible.
Size of libchrome.so:
Before: 67MB
After: 55MB
No Instrumentation: 47MB
Bug: 776702
Change-Id: I74bea84e7e6b0fa98c841ce0f2c492b6a7a7e31f
Reviewed-on: https://chromium-review.googlesource.com/826623
Reviewed-by: agrieve <agrieve@chromium.org>
Reviewed-by: Matthew Cary <mattcary@chromium.org>
Commit-Queue: Benoit L <lizeb@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524124}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 1c99c3dbef26ab8b853524b889bfd5dec815a2dc
All usages of gconf have now been removed, so we can remove the build
flag too. After this, I will remove gconf from the instrumented
libraries, sysroots, and install-build-deps.sh.
Bug: 768027
Change-Id: Idf5e8d60da387b3287f875092feba488217ca642
Reviewed-on: https://chromium-review.googlesource.com/826046
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Commit-Queue: Tim Brown <timbrown@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#524120}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: a28f4d062f4c030eeddd085f261988d7958db2ba
1. Updated the max limit on number for entries in link map when
reporting a stack trace from 64 to 512. 64 is not enough, e.g.
component builds of content_unittests load 158 libs.
2. Fixed the code to handle the case when the limit is reached
Previously that code was using sizeof() to get array size.
3. Update regex in runner_common.py to handle the case when pc_offset
part is missing.
Bug: 706592
Change-Id: Iededea88c86206b247948c789043c4f4dfd8a8d9
Reviewed-on: https://chromium-review.googlesource.com/815155
Commit-Queue: Sergey Ulanov <sergeyu@chromium.org>
Reviewed-by: Wez <wez@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#523847}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 449cb549b0e6e5c0863971443c609f08a280ae9e
Add a '--chartjson_result_file' flag to extract ChartJSON test results from
android devices.
R=jbudorick@chromium.org
Bug: webrtc:8566
Change-Id: I050d1f714471a5525e91d36d1e792815cb4578ae
Reviewed-on: https://chromium-review.googlesource.com/819831
Commit-Queue: Edward Lesmes <ehmaldonado@chromium.org>
Reviewed-by: John Budorick <jbudorick@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#523758}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: e627ef8b73fb4abab2094c2c420329c8cc1b8925
We cannot find library dependencies to add to a target's ordered
libraries list by searching the build directory because the build
directory may contain libraries that appear to be dependencies but are
in fact unrelated libraries that share a name with a system library
that the target depends on. For example, we may be building a target
that depends on the system's libGLESv2.so, and the build directory may
contain a library named libGLESv2.so as a result of building ANGLE. The
resulting APK would try to load libGLESv2.so using the crazy linker,
which might fail to load the system's library.
The solution is to use GN's write_runtime_deps feature to export the
runtime dependency list from the build system and create an ordered
libraries list from that.
Bug: 742655
Change-Id: Ifdbf3de6b78b56af22dbfe0d12d80db5a48fe9db
Reviewed-on: https://chromium-review.googlesource.com/821600
Reviewed-by: agrieve <agrieve@chromium.org>
Commit-Queue: Peter Collingbourne <pcc@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#523675}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 66205b643e9c51fe1fffce934e62ff4f8da32063
This reverts commit db213ff641175f136d52a577e21690f7f955f0bd.
Reason for reland: Downstream issue fixed. No changes here.
This CL makes aapt2 perform the final resource processing and packaging for an
apk as well as outputting proguard files.
TBR=agrieve
Bug: 782316
Change-Id: I1b35f49f2e4b4d503cc7ae07dabc18d9014abea5
Reviewed-on: https://chromium-review.googlesource.com/817377
Reviewed-by: agrieve <agrieve@chromium.org>
Commit-Queue: agrieve <agrieve@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#523456}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: f174f4deb83b28f63a30813241c2f279c4a99440
Stripping binaries entirely before shipping them to swarming clients to
be run prevents us from symbolizing stack traces. Since the effect on
binary & bootfs size is not all that great for Release, nor for Debug
component builds, disable stripping entirely for now.
Bug: 792521
Change-Id: Ie9031e26dcd9f77ee17e5bb3648f0ad13839f61f
Reviewed-on: https://chromium-review.googlesource.com/818505
Reviewed-by: Scott Graham <scottmg@chromium.org>
Commit-Queue: Wez <wez@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#523222}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: e1e1c872d3e7c9af677673b493e209b3cb845e8c
Comment [1] explains that libssl1.0.0 can be manually installed on systems that
are expected to have libssl1.0.2. In this case, libssl1.0.0 will appear to be
available in the repos when it is not.
The fix is to check for newer packages first.
[1] https://bugs.chromium.org/p/chromium/issues/detail?id=785818#c13
BUG=785818
R=dpranke@chromium.org
Change-Id: I52a69adc5271b5485f7d09aabd1a01a24ff10352
Reviewed-on: https://chromium-review.googlesource.com/819976
Commit-Queue: Dirk Pranke <dpranke@chromium.org>
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#523195}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 0524b2b7b70eff655189206acfb599fe9b27c40d
Debian builds require this package to be installed in order to build.
Change-Id: I6066fb1f3a39842ec076ec82495f0a44a2e17e93
Reviewed-on: https://chromium-review.googlesource.com/817649
Reviewed-by: Thomas Anderson <thomasanderson@chromium.org>
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Commit-Queue: Conley Owens <cco3@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#523139}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 3aeec4449b15c08e97037cb6caf7eb31ac2c3b2e
This introduces a dedicated template for defining and Annotation
Processor: java_annotation_processor().
Libraries can use them via:
annotation_processor_deps = [":processor_target"]
We now explicitly pass a --processorpath to javac, and always path both
--interface-classpath and --classpath to javac.py. The
interface-classpath is always written to the depfile, but the normal
classpath is passed to javac when one or more annotation processors are
enabled.
Due to the added build_config complexity this brings, android_apk's
build_config is now written by java_library_impl(), and no more
override_build_config support exists.
Bug: 792170
Change-Id: Ieebfa943615b9495010900c0a7be2b1ed9839a98
Reviewed-on: https://chromium-review.googlesource.com/817617
Commit-Queue: agrieve <agrieve@chromium.org>
Reviewed-by: Eric Stevenson <estevenson@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#523109}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: bc0688555c7e9c33ac5cb9030940238591fe9d7a
Rather than create the output directory in apkbuilder.py, add it to the
gn target's output for gn to create it automatically.
Bug: 792534
Change-Id: I3de087be4839e917b4308c38016da3b783668a96
Reviewed-on: https://chromium-review.googlesource.com/815355
Reviewed-by: agrieve <agrieve@chromium.org>
Commit-Queue: Peter Wen <wnwen@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#523103}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 40a260125252553a344e5ba748c232d03285705d
Jumbo builds reduce the total amount of CPU time used for compilation,
but they reduce parallelism. On machines with a modest number of cores
this is pure win, but on massively parallel builds this reduces the
benefits of jumbo. This change tweaks jumbo_file_merge_limit to a much
lower number when doing goma builds. This should also improve the
performance of incremental builds.
Testing was done with various merge limits while building chrome and
webkit_unit_tests and 8 was found to give the best combination of
elapsed-time full build performance and incremental build performance.
As expected the total CPU time spend compiling (and linking) goes up
when using a lower jumbo_file_merge_limit, but the wall clock time drops
(on goma builds). A merge limit of 10 may give slightly faster full
builds, but risks longer incremental builds. The exact number is not
critical but this is definitely in the right range, and much better than
the default of 50.
Testing suggests that jumbo goma builds will now take about 20% less
time than non-jumbo goma builds. goma builds with the old jumbo settings
were somewhere in-between. Results will improve as jumbo grows.
All testing was done on Windows with clang and is assumed to be
sufficiently applicable to other platforms.
Bug: 787983
Change-Id: I4e26341a5b224011d1401efa4563fcf2b53c969c
Reviewed-on: https://chromium-review.googlesource.com/814896
Reviewed-by: Dirk Pranke <dpranke@chromium.org>
Commit-Queue: Bruce Dawson <brucedawson@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#523004}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 396e35d38515d055988c7ad3d3ea43d68769b91f
This CL:
- disables tracing in subprocesses, which was causing occasional hangs
- stops tracing at env teardown time
Bug: 791632
Change-Id: Iaa6cda1f1adcaa1f4cccd7ef147c944e63bdc689
Reviewed-on: https://chromium-review.googlesource.com/816061
Reviewed-by: Benjamin Pastene <bpastene@chromium.org>
Commit-Queue: John Budorick <jbudorick@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#522820}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: ce8da4e667fcf61dbcc9d45e82188249d42c3ba9
This reland prevents the environment variable from being set when
asan is enabled because asan also reads this variable, and we should
not override its defaults.
Bug: 469376, 792905
Change-Id: I5bfaee392ad39a03ed2e5f6d16965dae929733de
Reviewed-on: https://chromium-review.googlesource.com/816598
Reviewed-by: John Budorick <jbudorick@chromium.org>
Commit-Queue: Peter Collingbourne <pcc@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#522802}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 45e1654501f6bd0f407b0ff03dba54f48650f860
*.js and *.css files (used in //extensions) are not dangerous
to have in sources lists so silently let them be.
Bug: 787836
Change-Id: I8dc2c16505594ed5342cad008dc1ca8b5a499b73
Reviewed-on: https://chromium-review.googlesource.com/814554
Reviewed-by: Bruce Dawson <brucedawson@chromium.org>
Commit-Queue: Daniel Bratell <bratell@opera.com>
Cr-Original-Commit-Position: refs/heads/master@{#522731}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 6e004f6a753bbf3480974058f4c549e93a2babfc
This reverts commit b2754af2ed37b2bf30efac6234fcecbdd941c4bb.
Reason for revert: Caused the step stack_tool_with_logcat_dump to start failing on cronet bots.
Original change's description:
> [Reland] Use llvm symbolizer for stack script.
>
> llvm symbolizer is a more efficient symbolizer than addr2line, objdump,
> etc.
>
> In this cl, I 1)created a wrapper instance to interact with
> llvm symbolizer, 2)made the stack script to use llvm symbolizer instance,
> and then 3)added llvm symbolizer into isolated inputs.
>
> Bug: 774267
> Change-Id: I43d305b4c0c94be614ee4cad35b40f04b3d3ae20
> Reviewed-on: https://chromium-review.googlesource.com/810007
> Reviewed-by: John Budorick <jbudorick@chromium.org>
> Reviewed-by: agrieve <agrieve@chromium.org>
> Reviewed-by: Dirk Pranke <dpranke@chromium.org>
> Commit-Queue: Zhiling Huang <hzl@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#522520}
TBR=dpranke@chromium.org,agrieve@chromium.org,hzl@chromium.org,jbudorick@chromium.org
Change-Id: If4e9f5618f62a5ed2a4d3152af6b18dbc748b1f8
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: 774267
Reviewed-on: https://chromium-review.googlesource.com/816257
Reviewed-by: Peter Collingbourne <pcc@chromium.org>
Commit-Queue: Peter Collingbourne <pcc@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#522711}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 6e6e84820c83c331f13add5fc62b8ee439bd10f0
This allows swarming users to set a gtest filter using
"-e GTEST_FILTER foo".
Bug:
Change-Id: I3b6ec960418d292f47f243cba145334d041bdcca
Reviewed-on: https://chromium-review.googlesource.com/816024
Reviewed-by: John Budorick <jbudorick@chromium.org>
Commit-Queue: Peter Collingbourne <pcc@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#522701}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 5b6975e12c7531505aa35f83110bedd8d48e588b
We were using a deprecated API to determine the existence of a schema,
and we were lazy loading the libgio library in case a distro doesn't
have the library.
All supported distros have libgio and support the new API, hence I fixed
both.
Test: Forced use of GSettings (see patch 1 - I use i3) and watched
chrome: //net-internals#proxy while I changed gsettings
Bug: 768027
Change-Id: Id9a3b157723c0b34e69b5b4eadcce1a7048b8636
Reviewed-on: https://chromium-review.googlesource.com/812471
Commit-Queue: Tim Brown <timbrown@chromium.org>
Reviewed-by: Lei Zhang <thestig@chromium.org>
Reviewed-by: Thomas Anderson <thomasanderson@chromium.org>
Reviewed-by: Eric Roman <eroman@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#522686}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 1c307cc3013b72f716ac0ddd83373818f6d6e1ee
Previously OCSP was broken with RemoteTestServer because OCSP server
port wasn't proxied.
1. Updated testserver.py to accept --ocsp-proxy-port-number argument
which makes it possible to generate correct cerificate for the
proxied connection.
2. Updated chrome_test_server_spawner.py to forward oscp server port.
3. Updated RemtoeTestServer to proxy OCSP connection.
4. Enabled OCSP tests on Fuchsia where they were previously failing.
Bug: 776575, 792300
Change-Id: Iedb1f5b26fc29ad0c73e78f450cd49335a424c18
Reviewed-on: https://chromium-review.googlesource.com/810096
Commit-Queue: Sergey Ulanov <sergeyu@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
Reviewed-by: Eric Roman <eroman@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#522634}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: ee7c8db6b554120fb28385f8759695048f1d8612
This reverts commit c1406d52464e16da16aae9cb189c4f28e6412358.
Reason for revert: The builds are failing on linking of some fuzzers: https://logs.chromium.org/v/?s=chromium%2Fbb%2Fchromium.fyi%2FLibfuzzer_Upload_Linux_ASan%2F7088%2F%2B%2Frecipes%2Fsteps%2Fcompile%2F0%2Fstdout
Original change's description:
> Migrate to -fsanitize=fuzzer-no-link when use_fuzzing_engine=true.
>
> -fsanitize=fuzzer and -fsanitize=fuzzer-no-link are two compilation
> flags that enable coverage instrumentation needed for libFuzzer.
>
> The instrumentation has more stuff under the hood compared to
> -fsanitize=trace-pc-guard. Also, it can be changing over time
> without a need to update GN flags again and again (e.g. move from
> edge to trace-pc-guard or something like that).
>
> Bug: 764514
> Change-Id: I48ef328dee49a9620a1b44bd5cd920f116e1bc1b
> Reviewed-on: https://chromium-review.googlesource.com/802395
> Commit-Queue: Max Moroz <mmoroz@chromium.org>
> Reviewed-by: Oliver Chang <ochang@chromium.org>
> Reviewed-by: Dirk Pranke <dpranke@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#522505}
TBR=kcc@chromium.org,inferno@chromium.org,dpranke@chromium.org,ochang@chromium.org,mmoroz@chromium.org
Change-Id: I86f316ba7b13c6cb015a3245e3860047ab4859ea
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: 764514
Reviewed-on: https://chromium-review.googlesource.com/815460
Reviewed-by: Max Moroz <mmoroz@chromium.org>
Commit-Queue: Max Moroz <mmoroz@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#522615}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 8181cec6f208b34465eaacb64e247a1fcce9bcd6
This reverts commit d47b98b28e50b303181dc5e4122deda25cce6fbb.
Reason for revert:
Suspected to cause test failures on cronet bots.
Original change's description:
> Set $UBSAN_OPTIONS during executable tests and disable UBSan's signal handlers.
>
> The signal handlers interfere with the breakpad and sandbox tests, so
> we need to disable them there. We will also want to use the same set of
> UBSan options during executable tests as we do during instrumentation
> tests, so move the $UBSAN_OPTIONS value into the test harness, and
> pass it into the instrumentation test runner using a string extra.
>
> Bug: 469376
> Change-Id: I76ffdd4a8221a54a23e257ac4d2321b8f28d2f42
> Reviewed-on: https://chromium-review.googlesource.com/812982
> Reviewed-by: John Budorick <jbudorick@chromium.org>
> Commit-Queue: Peter Collingbourne <pcc@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#522393}
TBR=pcc@chromium.org,jbudorick@chromium.org
Change-Id: Icb1a52726e7ec971d7d3a158fd91a5c052008269
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: 469376, 792905
Reviewed-on: https://chromium-review.googlesource.com/815545
Reviewed-by: Peter Collingbourne <pcc@chromium.org>
Commit-Queue: Peter Collingbourne <pcc@chromium.org>
Cr-Original-Commit-Position: refs/heads/master@{#522589}
Cr-Mirrored-From: https://chromium.googlesource.com/chromium/src
Cr-Mirrored-Commit: 547f308ad62690009af97c05f760861be6bca55e
Gordana Cmiljanovic
Scott Graham
Peter Collingbourne
Matthew Cary
George Burgess IV
Tom Anderson
Tim Brown
Bruce Dawson
Frank Henigman
Vadim Shtayura
Albert J. Wong
Sergey Ulanov
agrieve
Mohamed Heikal
Kenneth Russell
Nico Weber
Benoit Lize
Eric Stevenson
Benjamin Pastene
Edward Lemur
Mohamed Heikal
Wez
Conley Owens
Peter Wen
Yaron Friedman
John Budorick
Daniel Bratell
Max Moroz
Justin Cohen