On iOS, holding a lock during a slow upload can lead to watchdog kills
if the app is suspended mid-upload. Instead, if the client can obtain
the lock, the database sets a lock-time file attribute and releases the
flock. The file attribute is cleared when the upload is completed. The
lock-time attribute can be used to prevent file access from other
processes, or to discard reports that likely were terminated mid-upload.
Bug:chromium:1342051
Change-Id: Ib878f6ade8eae467ee39acb52288296759c84582
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3739019
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
This adds a runtime exception helper (& test module) for Windows and
plumbing to allow the module to be registered by the crashpad client,
and to trigger the crashpad handler. Embedders can build their own
module to control which exceptions are passed to the handler.
See: go/chrome-windows-runtime-exception-helper for motivation.
When registered (which is the responsibility of the embedding
application), the helper is loaded by WerFault.exe when Windows
Error Reporting receives crashes that are not caught by crashpad's
normal handlers - for instance a control-flow violation when a
module is compiled with /guard:cf.
Registration:
The embedder must arrange for the full path to the helper to
be added in the appropriate Windows Error Reporting\
RuntimeExceptionHelperModules registry key.
Once an embedder's crashpad client is connected to a crashpad
handler (e.g. through SetIpcPipeName()) the embedder calls
RegisterWerModule. Internally, this registration includes handles
used to trigger the crashpad handler, an area reserved to hold an
exception and context, and structures needed by the crashpad handler.
Following a crash:
WerFault.exe handles the crash then validates and loads the helper
module. WER hands the helper module a handle to the crashing target
process and copies of the exception and context for the faulting thread.
The helper then copies out the client's registration data and
duplicates handles to the crashpad handler, then fills back the various structures in the paused client that the crashpad handler will need.
The helper then signals the crashpad handler, which collects a dump then
notifies the helper that it is done.
Support:
WerRegisterExceptionHelperModule has been availble since at least
Windows 7 but WerFault would not pass on the exceptions that crashpad
could not already handle. This changed in Windows 10 20H1 (19041),
which supports HKCU and HKLM registrations, and passes in more types of
crashes. It is harmless to register the module for earlier versions
of Windows as it simply won't be loaded by WerFault.exe.
Tests:
snapshot/win/end_to_end_test.py has been refactored slightly to
group crash generation and output validation in main() by breaking
up RunTests into smaller functions.
As the module works by being loaded in WerFault.exe it is tested
in end_to_end_test.py.
Bug: crashpad:133, 866033, 865632
Change-Id: Id668bd15a510a24c79753e1bb03e9456f41a9780
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3677284
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Alex Gough <ajgo@chromium.org>
This is a reland of 460943dd9a71dc76f68182a8ede766d5543e5341
Original change's description:
> The DoubleForkAndExec() function was taking over 622 milliseconds to run
> on macOS 11 (BigSur) on Intel i5-1038NG7. I did some debugging by adding
> some custom traces and found that the fork() syscall is the bottleneck
> here, i.e., the first fork() takes around 359 milliseconds and the
> nested fork() takes around 263 milliseconds. Replacing the nested fork()
> and exec() with posix_spawn() reduces the time consumption to 257
> milliseconds!
>
> See https://github.com/libuv/libuv/pull/3064 to know why fork() is so
> slow on macOS and why posix_spawn() is a better replacement.
>
> Another point to note is that even base::LaunchProcess() from Chromium
> calls posix_spawnp() on macOS -
> 8f8d82dea0:base/process/launch_mac.cc;l=295-296
The reland isolates the change to non-Android POSIX systems because
posix_spawn and posix_spawnp are available in Android NDK 28, but
Chromium is building with version 23.
Change-Id: If44629f5445bb0e3d0a1d3698b85f047d1cbf04f
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3721655
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
This reverts commit 460943dd9a71dc76f68182a8ede766d5543e5341.
Reason for revert: This fails to compile in Chromium Android.
posix_spawn and posix_spawnp are available in Android NDK 28, but
Chromium is building with version 23.
https://ci.chromium.org/ui/p/chromium/builders/try/android_compile_dbg/1179765/overview
Original change's description:
> posix: Replace DoubleForkAndExec() with ForkAndSpawn()
>
> The DoubleForkAndExec() function was taking over 622 milliseconds to run
> on macOS 11 (BigSur) on Intel i5-1038NG7. I did some debugging by adding
> some custom traces and found that the fork() syscall is the bottleneck
> here, i.e., the first fork() takes around 359 milliseconds and the
> nested fork() takes around 263 milliseconds. Replacing the nested fork()
> and exec() with posix_spawn() reduces the time consumption to 257
> milliseconds!
>
> See https://github.com/libuv/libuv/pull/3064 to know why fork() is so
> slow on macOS and why posix_spawn() is a better replacement.
>
> Another point to note is that even base::LaunchProcess() from Chromium
> calls posix_spawnp() on macOS -
> 8f8d82dea0:base/process/launch_mac.cc;l=295-296
>
> Change-Id: I25c6ee9629a1ae5d0c32b361b56a1ce0b4b0fd26
> Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3641386
> Reviewed-by: Mark Mentovai <mark@chromium.org>
> Commit-Queue: Mark Mentovai <mark@chromium.org>
Change-Id: I7f6161bc4734c50308438cdde1e193023ee9bfb8
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3719439
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
The DoubleForkAndExec() function was taking over 622 milliseconds to run
on macOS 11 (BigSur) on Intel i5-1038NG7. I did some debugging by adding
some custom traces and found that the fork() syscall is the bottleneck
here, i.e., the first fork() takes around 359 milliseconds and the
nested fork() takes around 263 milliseconds. Replacing the nested fork()
and exec() with posix_spawn() reduces the time consumption to 257
milliseconds!
See https://github.com/libuv/libuv/pull/3064 to know why fork() is so
slow on macOS and why posix_spawn() is a better replacement.
Another point to note is that even base::LaunchProcess() from Chromium
calls posix_spawnp() on macOS -
8f8d82dea0:base/process/launch_mac.cc;l=295-296
Change-Id: I25c6ee9629a1ae5d0c32b361b56a1ce0b4b0fd26
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3641386
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
Because the upload thread uses synchronous upload, calling Stop() on
that thread from the main thread will lock, and trigger a terminate
when transitioning from foreground to background.
Additionally, background assertions now only last 30 seconds, so
shorten the timeout to 20 seconds.
This is a followup to https://crrev.com/c/3517967.
Bug: crashpad:1315441
Change-Id: Ic6886607805667ffce5ecf41716fc63333a341b8
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3577820
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Stop the prune thread and the upload thread when moving to the
inactive/background state. This will reduce the number of 0xdead10cc
system kills from having a file lock during iOS suspend.
Wait to start the prune thread when the application is active.
Otherwise, for iOS prewarmed applications, the prune thread will
regularly start when the application is foregrounded for the first
time when the user intentionally runs the app.
It's still possible for either the prune thread or the upload thread to
have a file lock during iOS suspend, such as when a task started in the
foreground and does not complete in time for suspension. Future work
should include considering BackgroundTasks and/or NSURLSessions, which
can more safely run in the background.
Bug: crashpad: 400
Change-Id: Ic7d4687eb795fe585327f128aa84a5928141f4a9
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3517967
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
iOS applications may be terminated with the exception code 0xdead10cc
when holding on to file locks in the shared container during suspension.
One approach to minimize this is to request additional background
execution time to complete the locking operation (in this case the
CrashReportUpload thread and the PruneIntermediateDumpsAndCrashReports
thread).
Bug: crashpad:400
Change-Id: I4192ae1a92646ea337a09ac071e49761ab2d3860
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3517966
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Crashpad currently has a circular dependency: client->snapshot->client.
The dependency from snapshot -> client only exists to pull in a single
constant for Windows (CrashpadClient::kTriggeredExceptionCode), so this
change breaks the dependency by splitting the constant out into a new
file util/win/exception_codes.h.
Change-Id: I6b74b367df716e097758e63a44c53cb92ea5e04d
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3450763
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Use BUILDFLAG(IS_*) instead of defined(OS_*).
This was generated mostly mechnically by performing the following steps:
- sed -i '' -E -e 's/defined\(OS_/BUILDFLAG(IS_/g' \
-e 's%([ !])OS_([A-Z]+)%\1BUILDFLAG(IS_\2)%g' \
$(git grep -l 'OS_'
'**/*.c' '**/*.cc' '**/*.h' '**/*.m' '**/*.mm')
- sed -i '' -e 's/#ifdef BUILDFLAG(/#if BUILDFLAG(/' \
$(git grep -l '#ifdef BUILDFLAG('
'**/*.c' '**/*.cc' '**/*.h' '**/*.m' '**/*.mm')
- gsed -i -z -E -e \
's%(.*)#include "%\1#include "build/buildflag.h"\n#include "%' \
$(git grep -l 'BUILDFLAG(IS_'
'**/*.c' '**/*.cc' '**/*.h' '**/*.m' '**/*.mm')
- Spot checks to move #include "build/buildflag.h" to the correct parts
of files.
- sed -i '' -E -e \
's%^(#include "build/buildflag.h")$%#include "build/build_config.h"\n\1%' \
$(grep -L '^#include "build/build_config.h"$'
$(git grep -l 'BUILDFLAG(IS_'
'**/*.c' '**/*.cc' '**/*.h' '**/*.m' '**/*.mm'))
- Add “clang-format off” around tool usage messages.
- git cl format
- Update mini_chromium to 85ba51f98278 (intermediate step).
TESTING ONLY).
- for f in $(git grep -l '^#include "build/buildflag.h"$'
'**/*.c' '**/*.cc' '**/*.h' '**/*.m' '**/*.mm'); do \
grep -v '^#include "build/buildflag.h"$' "${f}" > /tmp/z; \
cp /tmp/z "${f}"; done
- git cl format
- Update mini_chromium to 735143774c5f (intermediate step).
- Update mini_chromium to f41420eb45fa (as checked in).
- Update mini_chromium to 6e2f204b4ae1 (as checked in).
For ease of review and inspection, each of these steps is uploaded as a
new patch set in a review series.
This includes an update of mini_chromium to 6e2f204b4ae1:
f41420eb45fa Use BUILDFLAG for OS checking
6e2f204b4ae1 Include what you use: string_util.h uses build_config.h
Bug: chromium:1234043
Change-Id: Ieef86186f094c64e59b853729737e36982f8cf69
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3400258
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
Following https://crev.com/d3d85ce0b330b11f73f0495b7b99cea0d04d8c63,
the compiler now does not build an unwind library into the ASAN
runtime, nor is one available from the NDK in r23. Restoring Chrome's
standard dependencies (libcxx, libunwind) prevents a link error.
Bug: 1271628
Change-Id: I62d1c066bd7037276d78e2533dd5e4c3cf14f8c7
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3298826
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Richard Townsend <richard.townsend@arm.com>
Only the handler uses util/net. After
8342e6bd613a5b2e44eca1d74288e3115ccef139, the introduction of an
Objective-C class caused Chromium to emit duplicate class defintion
warnings in the component build.
Bug: chromium:1270609
Change-Id: I2770528347aef406bb21a79d295f702498f7b37e
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3290276
Commit-Queue: Robert Sesek <rsesek@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
This implements a per-report retry rate limit (as opposed to per upload
rate limit in ShouldRateLimitUpload). When a report upload ends in a
retry state, an in-memory only timestamp is stored with the next
possible retry time. This timestamp is a backoff from the main thread
work interval, doubling on each attemt. Because this is only stored in
memory, on restart reports in the retry state will always be tried
once, and then fall back into the next backoff. This continues until
5 retry attempts are reached.
Change-Id: Ibde8855a8a9f0743f0b0bd4d5e3de8a45c64bcb6
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3087723
Reviewed-by: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Manage the intermediate minidump generation, and own the crash report
upload thread and database.
Change-Id: I272d790a827cd13f6872e56f4675f366d13719c5
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3087721
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
This change was partially scripted and partially done manually with vim
regex + manually placing the deleted constructors.
The script change looked for destructors in the public: section of a
class, if that existed the deleted constructors would go before the
destructor.
For manual placement I looked for any constructor in the public: section
of the corresponding class. If there wasn't one, then it would ideally
have gone as the first entry except below enums, classes and typedefs.
This may not have been perfect, but is hopefully good enough. Fingers
crossed.
#include "base/macros.h" is removed from files that don't use
ignore_result, which is the only other thing defined in base/macros.h.
Bug: chromium:1010217
Change-Id: I099526255a40b1ac1264904b4ece2f3f503c9418
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3171034
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Peter Boström <pbos@chromium.org>
Standalone Crashpad for Android can now be built with gn.
Change-Id: I0ee7f8e1af8c2bc0edb88e93b345abd7d739f33c
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3034984
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
Remove unneeded base/strings/stringprintf.h includes.
ARCH_CPU_X86_64 macro is used without including build/build_config.h
Missing base/check.h
Change-Id: Ib7864ab7b30ef8fc37649783f7b90b618d0d6a0b
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2920552
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Chromium moved base::size() to base/cxx17_backports.h, so do the same in
mini_chromium and update the users in Crashpad.
Roll mini_chromium to 2f06f83f to make the new base header available.
Bug: chromium:1210983
Change-Id: Ie3dc4c189dcdfcac030b95fe285f94abb29a27bf
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2917779
Commit-Queue: Lei Zhang <thestig@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
Move some common files out of client and handler for iOS usage.
Bug: crashpad: 31
Change-Id: I1933eaaa7580a81017c52b77dfb636a8fa31ee78
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2851059
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
LogOutputStreamTest.{WriteAbort,FlushAbort} are flaky because the logcat
is sometimes overloaded earlier than expected causing FlushAbort to fail
during Write() or either test to fail to write the abort message.
This change updates LogOutputStream to detect logcat overloads (EAGAIN)
and make one attempt at writing the abort message, even if the output
cap hasn't been reached.
This change also updates LogOutputStream's interface to defer log writes
to a Delegate. In tests, the Delegate implements a mock log and in
production, writes to Android's logcat.
I've removed VerifyGuards because LogOutputStream no longer writes
guards if Write() has never been called and the guards are tested in
other tests.
Change-Id: Icad83524aaf573c3e082469f1de095b6ca2c4839
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2439641
Reviewed-by: Mark Mentovai <mark@chromium.org>
Depends on https://chromium-review.googlesource.com/c/chromium/mini_chromium/+/2424890
Although logging to files is not yet supported by mini_chromium, it is
the default behavior for OS_WIN in chromium. This change should
cause crashpad to log via OutputDebugString() on Windows, instead of
debug.log files. Future work (crbug.com/crashpad/26) should arrange for
logs to be uploaded with reports, embedded in associated minidumps or as
file attachments.
Bug: chromium:711159
Change-Id: I0f9004f7de94dd29d555cc7d23c48a63da6b4bba
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2425108
Reviewed-by: Mark Mentovai <mark@chromium.org>
Includes DEPS roll of mini_chromium:
f0bd14b Pull build_config.h source set into separate build file
65fb5c9 Update path to win_helper after moving to build/config
Change-Id: Ic9f5c68e2cebd8bf86492766684bdb422da1aa9e
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2426989
Reviewed-by: Mark Mentovai <mark@chromium.org>
This change prepares crashpad for the upcoming switch of base::string16
to std::u16string on all platforms. It does so by replacing Windows-only
instances of base::string16 with std::wstring, and using appropriate
string utility functions.
Bug: chromium:911896
Change-Id: Ibb0b8a4e4dc7fae1d24d18823f8dbb6da31f8239
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2332402
Commit-Queue: Jan Wilken Dörrie <jdoerrie@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
CFI attempts to verify that the dynamic type of a function object
matches the static type of the function pointer used to call it.
https://clang.llvm.org/docs/ControlFlowIntegrity.html#indirect-function-call-checking
However, the analyzer does not have enough information to check
cross-dso calls. In these instances, CFI crashes upon calling the
function with an error like:
pthread_create_linux.cc:60:16: runtime error:
control flow integrity check for type
'int (unsigned long *, const pthread_attr_t *, void *(*)(void *), void *)'
failed during indirect function call
(/lib/x86_64-linux-gnu/libpthread.so.0+0x9200):
note: (unknown) defined here pthread_create_linux.cc:60:16:
note: check failed in crashpad_handler,
destination function located in /lib/x86_64-linux-gnu/libpthread.so.0
Change-Id: Ib29dabfe714f2ee9cc06a5d17e6899ff81a06df4
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2339332
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
Reviewed-by: Mark Mentovai <mark@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>
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>
Implemented the AddAttachment(), InitializeAttachments(), CleanDatabase() functions
on Windows.
Added attachment=FILE_NAME option to the handler, and
"attachments" argument for Windows and Linux to StartHandler function.
On crash it will create the corresponding attachments in the database
and copy content of the specified files to the database.
Bug: b/157144387
Change-Id: Ia238de39028e07112a7b971b5b7d5e71a5864f53
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2248099
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
If the file just needs the CHECK/CHECK_OP/NOTREACHED
macros, use the appropriate header for that instead.
Or if logging.h is not needed at all, remove it.
This is both a nice cleanup (logging.h is a big header,
and including it unnecessarily has compile-time costs),
and part of the final step towards making logging.h no
longer include check.h and the others.
Bug: chromium:1031540
Change-Id: Ia46806bd95fe498bcf3cf6d2c13ffa4081678043
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2255361
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Hans Wennborg <hans@chromium.org>
Add direct includes for things provided transitively by logging.h
(or by other headers including logging.h).
This is in preparation for cleaning up unnecessary includes of
logging.h in header files (so if something depends on logging.h,
it needs include it explicitly), and for when Chromium's logging.h
no longer includes check.h, check_op.h, and notreached.h.
DEPS is also updated to roll mini_chromium to ae14a14ab4 which
includes these new header files.
Bug: chromium:1031540
Change-Id: I36f646d0a93854989dc602d0dc7139dd7a7b8621
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/2250251
Commit-Queue: Hans Wennborg <hans@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
