A new bug in macOS 11.0db6 20A5364e has broken the {CTL_KERN,
KERN_PROCARGS2} sysctl such that it will not work properly unless
provided with a buffer at least 17 bytes larger than originally
indicated. Work around the bug by providing a buffer a whole 32 bytes
larger.
Bug: crashpad:347, crashpad:355
Test: crashpad_util_test ProcessInfo.{Self,SelfTask,Forked}
Change-Id: I9324a63390875308979a10fefcd4c1c880651aee
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2399646
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
Apple has never exposed the CPU frequency on ARM systems. Report it as 0
on mac-arm64 without attempting to obtain it from the system (which
would log a warning in the process).
This will resolve these harmless warnings produced when Crashpad creates
a snapshot on arm64:
[pid:tid:yyyymmdd,hhmmss.µµµµµµ:WARNING system_snapshot_mac.cc:50] sysctlbyname hw.cpufrequency: No such file or directory (2)
[pid:tid:yyyymmdd,hhmmss.µµµµµµ:WARNING system_snapshot_mac.cc:50] sysctlbyname hw.cpufrequency_max: No such file or directory (2)
Bug: chromium:1103944
Change-Id: Id6217d5b9f756c54f46a6b29742c361e987412f0
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2392076
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
There is no possibility to run 32-bit processes on macOS 10.15 or later.
There is never any possibility to run 32-bit processes on macOS on
arm64.
This transforms ProcessReaderMac::Is64Bit into a compile-time constant
“yes” when building for a system that will never see a 32-bit process.
This is a lightweight way to get much 32-bit support code removed from
optimized compiled output, including all of process_types. In an
optimized build of crashpad_handler for arm64, this is a 3% reduction
from 569kB to 552kB (-17kB).
Change-Id: I8890a170467834b99b017f1aa3dc78f3f33cd13e
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2389010
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
When building for macOS and configured with target_cpu =
"mac_universal", bi-architecture x86_64/arm64 output will be produced.
mac_universal is, so far, a “Crashpad special” that will only work with
mini_chromium and the standalone Crashpad build, and not the in-Chromium
build. It exists to support Keystone, which intends to ship as
x86_64/arm64 universal.
Includes:
Update mini_chromium to e0008f2714a76c7f2a3854fa75774427a886d6b9
e0008f2714a7 mac-arm64: Allow target_cpu = "mac_universal" to create
universal builds
Bug: crashpad:345
Change-Id: I5ff2dce5ffae58186e33757aa94587f8eca20b99
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2387410
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
cl_kernels modules have appeared since OS X 10.10 as MH_BUNDLE modules
with a __TEXT segment, one section of which claims to belong to the __LD
segment. They are produced when OpenCL is asked to compile an OpenCL
kernel for the CPU, but this currently appears impossible on arm64.
The workaround is omitted as it appears to be unnecessary, but the test
still attempts to create an OpenCL kernel for the CPU. If this ever
becomes possible, and the modules are malformed, the test will fail as
an indication that the workaround must be reinstated for arm64.
Bug: crashpad:345
Test: crashpad_snapshot_test ProcessReaderMac.{Self,Child}Modules
Change-Id: Ia3d7163cc9995bb4a33457a77c2a5f0e66f4c1a0
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2386466
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
__builtin_trap uses ud2 on x86_64, producing a SIGILL. On arm64, it uses
brk #1, producing a SIGTRAP. Test expectations must be adjusted
accordingly.
Bug: crashpad:345
Test: crashpad_snapshot_test MachOImageAnnotationsReader.CrashModuleInitialization, crashpad_util_test ExcServerVariants.*,ExceptionPorts.*
Change-Id: I22e75b7b48b8887031b1d95f1cea8a09733daf49
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2386464
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
On x86_64, it’s impossible for a signal handler distinguish between
SIGBUS caused synchronously by a hardware fault and SIGBUS raised
asynchronously by software. This remains true on arm64, and is expanded
to include both SIGILL and SIGSEGV.
Bug: crashpad:345
Test: crashpad_util_test Signals.Raise_HandlerReraisesTo*
Change-Id: I181ea35121048dc0c666e2346340e698220ca650
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2386463
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
MacOSXMinorVersion reported just the “y” value for an OS version 10.y.z.
This is no longer sufficient to identify OS versions accurately in macOS
11. A new MacOSVersionNumber function reports the full OS version as
“xxyyzz” for an OS version x.y.z. This is the same format used by
<Availability.h> __MAC_* macros since 10.10.
MacOSXVersion is also renamed to MacOSVersionComponents for
disambiguation and proper modern nomenclature.
Bug: crashpad:347
Test: crashpad_snapshot_test SystemSnapshotMacTest.OSVersion, crashpad_util_test MacUtil.MacOSVersionNumber
Change-Id: I66421954f021c0627095474cb26359970fcd9101
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2386386
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
We're working to decouple ChromeOS and Linux builds of Chrome.
Currently OS_CHROMEOS sets OS_LINUX, so we need to refactor
current OS_LINUX usage to make this explicit.
More information can be found at go/cros_is_linux_os_linux
BUG=chromium:1110266
TEST=manual build
Change-Id: Ie765da1ab6a0bf0286538ae1df3697abaa29aeaa
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2391116
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
4ae896bad0af replaced OS_MACOSX with OS_APPLE and introduced OS_MAC,
disentangled from OS_IOS. This allows !defined(OS_IOS) to be written
more directly as defined(OS_MAC) in cases where OS_APPLE is assured.
Change-Id: I8848503d3318038865dd4c8586a81ce82764af0a
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2384318
Reviewed-by: Justin Cohen <justincohen@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
On macOS 11, system libraries are no longer present on disk as
individual Mach-O files, but are shipped exclusively in the dyld shared
cache. This affects tests that verify the list of modules loaded into a
process. The new _dyld_shared_cache_contains_path function can be used
to determine whether a path exists in the dyld shared cache.
Bug: crashpad:347
Test: crashpad_snapshot_test ProcessReaderMac.{Self,Child}Modules
Change-Id: I19402e45701caeb850bdafd701ccbba4ef9c60bc
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2384322
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
This updates the process_types version of dyld_all_image_infos for macOS
11.0.
Test: crashpad_snapshot_test ProcessTypes.DyldImagesSelf
Bug: crashpad:347
Change-Id: Ieb134f17015070a11c918b26ece316f77a4cc263
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2384317
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
The macOS 11.0 SDK, as of Xcode 12b6 12A8189n, has not updated
<AvailabilityMacros.h> with a MAC_OS_X_VERSION_11_0 or
MAC_OS_X_VERSION_10_16 constant. However, the <Availability.h> interface
has been updated to provide both __MAC_11_0 and __MAC_10_16.
<AvailabilityMacros.h>’s MAC_OS_X_VERSION_MAX_ALLOWED, which is supposed
to identify the SDK version, is broken in the 11.0 SDK in that whenever
the deployment target is set to 10.15 or earlier, the SDK will be
mis-identified through this interface as 10.15. When using the
<Availability.h> equivalent, __MAC_OS_X_VERSION_MAX_ALLOWED, the 11.0
SDK is identified as 10.16 (arguably it should be internally versioned
as 11.0, but at least this interface allows it to be detected
unambiguously.) It’s clear that the <AvailabilityMacros.h> interface
provides no meaningful support for the macOS 11.0 SDK at all, but
<Availability.h> does.
<Availability.h> was introduced in the Mac OS X 10.5 SDK, so there is no
relevant SDK version compatibility problem with this interface.
Key differences between these interfaces for the purposes used by
Crashpad:
- <AvailabilityMacros.h> → <Availability.h>
- MAC_OS_X_VERSION_MIN_REQUIRED (DT) → __MAC_OS_X_VERSION_MIN_REQUIRED
- MAC_OS_X_VERSION_MAX_ALLOWED (SDK) → __MAC_OS_X_VERSION_MAX_ALLOWED
- MAC_OS_X_VERSION_x_y → __MAC_x_y
- <Availability.h> __MAC_OS_X_VERSION_* SDK/DT macros are only
available when targeting macOS, while <AvailabilityMacros.h>
MAC_OS_X_VERSION_* SDK/DT macros are available on all Apple platforms,
which may be a source of confusion. (<Availability.h> __MAC_* macros
do remain available on all Apple platforms.)
This change was made mostly mechanically by:
sed -i '' -Ee 's/<AvailabilityMacros.h>/<Availability.h>/g' \
$(git grep -E -l '<AvailabilityMacros.h>' |
grep -v AvailabilityMacros.h)
sed -i '' -Ee 's/(MAC_OS_X_VERSION_(MIN_REQUIRED|MAX_ALLOWED))/__\1/g' \
$(git grep -E -l 'MAC_OS_X_VERSION_(MIN_REQUIRED|MAX_ALLOWED)' |
grep -v AvailabilityMacros.h)
sed -i '' -Ee 's/(MAC_OS_X_VERSION_(10_[0-9]+))/__MAC_\2/g' \
$(git grep -E -l 'MAC_OS_X_VERSION_(10_[0-9]+)' |
grep -v AvailabilityMacros.h)
Bug: crashpad:347
Change-Id: Ibdcd7a6215a82f7060b7b67d98691f88454085fc
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2382421
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
This removes the rotted functionality for running tests on Fuchsia. It
had previously been broken by other platform changes. Other tools are
from previous SDKs being removed too; this approach is no longer going
to work.
The preferred way is to connect via SSH to the device, however, that
requires using the femu.sh from the SDK, which in turn requires `sudo`
to create a network device, so it won't directly work on bots anyway.
I started trying to update the to use femu.sh, fserve.sh, & fpublish.sh,
but that requires building a .far, which uses GN templates which
differ from the in-tree versions, and don't seem (?) to support
packaging resources into the package.
So, for now because it was confusing people (see linked bug) give up and
delete the code for the time being.
Bug: fuchsia:54031
Change-Id: Iac7af80094b150d11e71474cba4bd93eb8e80639
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2354160
Commit-Queue: Scott Graham <scottmg@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
Semaphore is implemented with a mutex and condition variable.
When a semaphore is used in TimerTest, Run() acquires a lock
to guard its CV but may be interrupted by the signal handler
before releasing the lock which results in a hang when
Signal(), called from the signal handler, also needs to
acquire the lock.
Bug: chromium:1050178
Change-Id: I0d7ef65bd159ee2de9258ba24537f9470f4c621d
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2318554
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
We'll indirect through this group while we rename crashpad{_,-}tests.
Bug: fuchsia:22679
Change-Id: Ic4ca73ecbd6d3e2776a7b7ee8bbbd5012459be52
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2315003
Reviewed-by: Scott Graham <scottmg@chromium.org>
Commit-Queue: Scott Graham <scottmg@chromium.org>
Commit-Queue: Tamir Duberstein <tamird@google.com>
Update buildtools to 9e121212d42be62a7cce38072f925f8398d11e49
6b2a302b65ba Roll GN from 7d7e8dee..cd3869be
6b951673d176 Roll GN from cd3869be..b6203d18
00f6f7c80e3c Update chromium's DEPS-file presubmits with new 'Str' type
eb3987ec709b Roll GN from b6203d18..d585128c
9e121212d42b Roll GN from d585128c..3028c6a4
Update mini_chromium to 85e41f6b6cb5dd31521c4a06b15c25d3345a6194
85e41f6b6cb5 apple: Use `frameworks` instead of `libs` in GN targets
Change-Id: I1fd5cc4d48bdcc2ec762c6edf570fa8c7e55dc88
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2311421
Commit-Queue: Scott Graham <scottmg@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
Disable the stack-overrun test since ASAN will detect the
stack corruption and kill the process.
Also allocate extra alt-stack space when ASAN is enabled.
Change-Id: Id8711d2aaa31c90c78b7feceec2877c18dc9be00
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2303591
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
This prevents errors creating the attachment if a previous test
failed to delete it for some reason.
Change-Id: I2622d86e08987ad3ff201da84dc9ae210acf776d
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2302722
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
If the process' threads can't be read, then the cpu context object won't
be able to be initialized.
Previously, the process snapshot always assumed that the context would
be filled out, as there was no error returned, which could result in
later checks failing.
Return an error from the exception snapshot's initialization so that
process snapshot can correctly handle failure to initialize.
Bug: fuchsia:55837
Change-Id: Ia3fecef1230a19dfa23401b0339c6a94370c6baf
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2296039
Commit-Queue: Scott Graham <scottmg@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
hw.cputype is just CPU_TYPE_ARM64, which isn’t terribly useful.
hw.cpufamily is somewhat better as it uses <mach/machine.h> CPUFAMILY_*
values, which distinguish between different CPU generations.
CPUFAMILY_ARM_VORTEX_TEMPEST identifies A12, for example. (The fun cores
are Vortex and the boring cores are Tempest.)
Bug: crashpad:345
Change-Id: I88be4fa0b305b2fa15bd24358f63dc7d72192b27
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2289041
Reviewed-by: Nico Weber <thakis@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
This #include was omitted from 0c3f50c8e14a.
Bug: crashpad:345
Change-Id: Ic3392bfc611ba847e50743ef7129263bf69c81e2
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2288393
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
This gets all tests building. They don’t all pass, and there aren’t any
guarantees that anything else works yet, either.
This is mostly a lot of CPU context shuffling.
Bug: crashpad:345
Change-Id: I684017a5816f44917392964d7fb6d08083770b38
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2285962
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
This gets all production code for Chrome building, excluding tests.
There aren’t any guarantees that anything works yet.
This is mostly a lot of CPU context shuffling.
In contrast to macOS on x86, there’s no need to support 32-bit arm on
macOS, because this new platform is 64-bit-only from its inception.
Bug: crashpad:345
Change-Id: I187239b6a969005a3458af7fe30c44147a57f95f
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2285961
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
On Android, the compat library built compat/android/sys/mman.cc, which
provides a fallback for mmap, and compat/linux/sys/mman.cc, which
provides a fallback for memfd_create. This can result in two object
files colliding in the library, which is flagged as a “gn analyze”
error:
> ERROR at //crashpad/compat/BUILD.gn:62:5: Duplicate object file
> static_library(target_name) {
> ^----------------------------
> The target //crashpad/compat:compat
> generates two object files with the same name:
> obj/crashpad/compat/compat/mman.o
>
> It could be you accidentally have a file listed twice in the
> sources. Or, depending on how your toolchain maps sources to
> object files, two source files with the same name in different
> directories could map to the same object file.
>
> In the latter case, either rename one of the files or move one of
> the sources to a separate source_set to avoid them both being in
> the same target.
The files are renamed to avoid this collision. The associated headers
cannot be renamed because they need to shadow the SDK’s copies. There is
no “gn analyze” conflict reported for headers with the same name.
Change-Id: Ia49ef5ff8375673395597e96555f72f7c69e3564
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2285965
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Chromium relies on a global gn sources assignment filter, even though
it’s kind of discouraged. Crashpad doesn’t use this, instead selecting
which files to build entirely with the logic in its .gn files. In some
cases, Chromium’s filters prevent _linux files from building on Android,
and _mac files from building on iOS, even though Crashpad’s build has
explicitly requested these inclusions. To overcome this problem,
clear the gn sources_assignment_filter in all of Crashpad’s own .gn
files when building in Chromium.
Change-Id: Iab0af29f4e5aff4be8eec2ee12b2e3f991c0a86d
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2285959
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Rohit Rao <rohitrao@chromium.org>
