Change the ObjExceptionProcessor to write intermediate dumps to a
temporary location until they are confirmed by the
UncaughtExceptionHandler. Because the exception preprocessor uses
heuristics to detect iOS sinkholes, it's possible for an exception to
be identified as fatal, but not actual trigger the uncaught exception
handler. If the processor detects more than one fatal exception, it will
unregister itself and indicate this in the second dump with the key
'MultipleHandledUncaughtNSException'.
This changes also consolidates and simplifies some methods in the
InProcessHandler.
Change-Id: Ifc457e974d25f533b77cfd18b702129fdfb10a75
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3529968
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Also update mini_chromium to f87a38442a9e for python3 changes.
Change-Id: I4ca7aa4cc9dcc97698fc0bc13cfb339421668074
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3542572
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Some sinkholes use objc_exception_throw instead of rethrow, which gives
the preprocessor a second, incorrect, attempt to process the
NSException. This also means if the processor misses the first sinkhole,
on the second attempt the original throwing stack will be missing.
Instead, track the original NSException and ignore any followup calls
to the ObjcExceptionPreprocessor with the same NSException.
Also creates a ExceptionPreprocessorState class to manage the complex
types. This will be used in a followup CL to finalize caught
NSExceptions using the uncaught handler.
Bug: 1300171
Change-Id: I1f9f2c7ee79c7a16585103f04831217979e9332b
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3530246
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>
https://crrev.com/c/3401563 introduced logic to guard the cached
intermediate dump writer from concurrent exceptions with a first
exception wins approach. To prevent the losing exception from returning
immediately and terminating the app before the dump is written, the
losing thread sleeps indefinitely. In the case where the losing
exception is from a call to abort() and the winning exception is a Mach
exception, the process will never terminate because abort() will first
block all signals on all other threads with a sigprocmask. This prevents
the kernel from delivering the signal converted from the Mach exception
and will never terminate. This effectively deadlocks the app.
Instead, unblock all signals corresponding to all Mach exceptions
Crashpad registers for before returning KERN_FAILURE.
Bug: crashpad:391
Change-Id: I96c357e98f09e65e70c67125a45b9b04075c2c06
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3518186
Reviewed-by: Mark Mentovai <mark@chromium.org>
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
in_process_intermediate_dump_handler_test was mixing CPU
architecture (x86_64 vs arm64) and iOS device type (iphoneos vs
iphonesimulator).
Bug: 1306589
Change-Id: Ie43a7f1916d69888e992320d999010071b2575b3
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3527034
Reviewed-by: Rohit Rao <rohitrao@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>
Previous commits[1][2] modified tests to require looking at RawLog
output to validate the success or failure of the iOS InHandlerProcess.
Previously this would use freopen to direct the RawLog to a file.
However, freopen introduces a race where the log file may not be
associated with stderr, and instead may interfere with the
InProcessHandler's cached writer fd. This caused flake with the
intermediate dumps sometimes including stderr logging.
Since the test fixtures only needs to know about the output of RawLog,
instead add a crashpad::internal::SetFileHandleForTesting method to
that swaps out STDERR_FILENO with the test fixture's fd.
[1] https://crrev.com/c/3488826
[2] https://crrev.com/c/3401563
Change-Id: I87b1020db6b896a47bec5a7c916a572c192b884f
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3517773
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Similar to crrev.com/c/3516536, this CL fixes the Windows build
when the UNICODE preprocessor macro is not defined where
code passes Unicode string literals with L"..." to non-Unicode
APIs like RegOpenKeyEx().
This fixes the build by explicitly using RegOpenKeyExW() instead.
Change-Id: I14a827357b9cbd42452e0e5eb13a3430569559a5
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3516538
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
crrev.com/c/3434090 introduced a change that breaks the Windows build
when the `UNICODE` preprocessor macro is not defined, as it passed
a `wchar_t*` to `RegQueryValueEx()`.
This fixes the build by explicitly using `RegQueryValueExW()` instead.
Change-Id: Ic438bd982fdeffba05b4224051242b45e797ebd8
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3516536
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
The IOSSystemDataCollector was previously owned by the iOS CrashHandler
and passed in to the iOS InProcessHandler in each method. Move
ownership to iOS InProcessHandler to simplify.
Change-Id: Ifa41304cb1e3e3825a211e6cce5aa730d0edcc95
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3517965
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Override malloc_default_zone and malloc_default_purgeable_zone
with allocators that exit when called from the signal or Mach exception
threads in XCUITests, to verify the allocator is not used by the
InProcessHandler. Check stderr for error messages to confirm failures.
Change-Id: I1bb92e57504d71bbf6c6eaad3571c814e8a6934c
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3488826
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Change signal, uncaught NSExceptions and Mach exception handlers to
prevent re-entrancy with a first-exception-wins approach to prevent
concurrent exceptions from trying to use the same cached intermediate
dump writer. Uses compare-and-swap to either return early for reentrant
signals or to wait indefinitely for anything after the first fatal
exception.
Change the NSException handler generated from the Objective-C exception
preprocessor to not used the cached intermediate dump writer and
not use the same first-exception-wins logic. This is useful because the
Objective-C exception preprocessor is imperfect and may generate
intermediate dumps that are not followed by process termination.
Simplify DumpWithoutCrashing's ownership of its intermediate dump writer
to be thread safe.
Set a handler for SIGPIPE for applications that haven't already
ignored or set a handler for SIGPIPE.
Bug: crashpad:391
Change-Id: Ia8ae61d50be81910fa0af40325300441d9dc01b6
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3401563
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Strings in minidumps are typically NUL-terminated
(https://docs.microsoft.com/en-us/windows/win32/api/minidumpapiset/ns-minidumpapiset-minidump_string).
But the CodeViewRecord types do not use MINDIUMP_STRINGs, and do not
have a separate length field for the pdb_name. Instead, the strings are
always NUL-terminated, with the length derived from the
MINIDUMP_LOCATION_DESCRIPTOR::DataSize field. The writer is correctly
NUL-terminating the debug filename, but ModuleSnapshotMinidump is
off-by-one and including the NUL-terminator.
Change-Id: I8d813b5ef9e9e167dca73a6a938fbbf8dd1580c2
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3482876
Reviewed-by: Mark Mentovai <mark@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>
When trying to update Chromium's copy of crashpad I got this error
message:
Banned functions were used.
third_party\crashpad\crashpad\client\crashpad_client_ios_test.mm:33:
testing::Test should not be used in Objective-C++ code as it does
not drain the autorelease pool at the end of the test. Use
PlatformTest instead.
So, I'm fixing the code as requested.
The change was introduced in crrev.com/c/3418581
Change-Id: I4888febbd41b6365d9bde5ad062565770496243f
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3459403
Reviewed-by: Justin Cohen <justincohen@chromium.org>
Commit-Queue: Bruce Dawson <brucedawson@chromium.org>
When updating Chromium's copy of crashpad I received this very
reasonable warning:
client\ios_handler\exception_processor.h: Includes STL header(s) but does not reference std::
So, this change removes the #include of vector.
Change-Id: I22f05b542fd4e0b582351072a3e3bb4af402b836
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3459402
Reviewed-by: Mark Mentovai <mark@chromium.org>
Reviewed-by: Justin Cohen <justincohen@chromium.org>
Commit-Queue: Bruce Dawson <brucedawson@chromium.org>
kernel32.dll no longer works as a source of truth for Windows versions
because it is not updated with every Windows update. This change grabs
the last two version numbers from the registry, if possible.
This also copies some code cleanup from Chromium (crrev.com/c/3205913).
Bug: chromium:1248324
Change-Id: I9d6745084060f033cd54c56f832aed4ac163e6be
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3434090
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Bruce Dawson <brucedawson@chromium.org>
This patch adds optional support for Arm Pointer Authentication Codes.
X30/LR is not stored to stack at any place and restored for usage.
Therefore only adding PAC flag to .note.gnu.property section.
Change-Id: I9581059dfa1eed88af5a73df15b6a0d299caea13
Bug: crashpad: 1145581
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3440070
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Adenilson Cavalcanti <cavalcantii@chromium.org>
xnu turns hardware faults into Mach exceptions, so the only signal left
to register is SIGABRT, which never starts off as a hardware fault.
Installing a handler for other signals would lead to recording
exceptions twice. As a consequence, Crashpad will not generate
intermediate dumps for anything manually calling raise(SIG*). In
practice, this doesn’t actually happen for crash signals that originate
as hardware faults.
Change-Id: I1be669d10e89b8e8ebcc69cfdf79c1ee20c96f76
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3403042
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
The badAccess test may return different mach exception codes depending
on optimization levels. Simplify by this by turning off optimization
for this particular test.
Also removes testSegv, which was duplicative and not really testing
SIGSEGV.
Change-Id: Idb92731da6a86545ed83c2bbdd200a0b792c579f
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3403040
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
https://crrev.com/c/3399252 fixed a heap overrun in iOS intermediate
dump processing.
This is a follow-up to that change to harden `CrashHandler` against
similar crashes:
1) Ensure the destructor of `ScopedAlternateWriter` is invoked
to restore `InProcessHandler::writer_` state before processing
the intermediate dump (otherwise, a signal raised by the intermediate
dump handler would dereference the empty `std::unique_ptr` in
`InProcessHandler::writer_`).
2) Harden `InProcessHandler` to check if `writer_` is empty before
handling signals or exceptions
Change-Id: I1e63a496395b26681632302e8915b4433897037a
Bug: 391
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3401766
Reviewed-by: Justin Cohen <justincohen@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
Handle multiple simultaneous crashes among threads by having
the first crashing thread set an atomic flag and subsequently crashing
threads check the flag before requesting a dump. If a dump has already
been requested, the threads pause on a futex with a timeout in case the
crashing thread crashes again or otherwise fails to WakeThreads().
The thread_local disabled_for_thread_ is removed and combined with this
flag because accessing thread_locals produces undefined behavior in
signal handlers.
Bug:crashpad:384, chromium:861730
Change-Id: I83bce36e1010d0635ba8aeac937e150c43a4166f
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3403017
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
Both running first chance handlers and checking for disabled signal
handlers should no longer interact with DumpWithoutCrashing().
First-chance-handlers should also run even with disabled crashpad
signal handlers or else those signals would be reported by the next
chained signal handlers as crashes.
Change-Id: I64b3da42c400a1c431c6228d4da181ed56bfda89
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3403413
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Joshua Peraza <jperaza@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>
To avoid unnecessary duplication of potentially large memory regions,
the iOS MemorySnapshot data is owned by the intermediate dump. When
merging two iOS memory snapshots, it's necessary to copy the merged
data into a vector owned by the memory snapshot itself.
Previously the merged snapshot just contained the original data, which
leads to a heap overrun.
Bug: 1288544
Change-Id: I4eb392800141451db188d17c7b9f9ca9bd3bb603
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3399252
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
Mig-generated files contain mig identifiers, which include timestamp and mig build info.
To improve build determinism and goma cachehits we can replace these lines with something stable.
Bug: crashpad:390
Change-Id: Iedb2f6e64428612899587c2ac4d488baf439961f
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3394052
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
Right now there's no clear way to differentiate between Signal and Mach
exceptions. Instead, set EXC_SOFT_SIGNAL as the top level exception for
signal exceptions, moving the signal number to ExceptionInfo() and
and the signal code plus the two previous values into Codes().
Fixed:crashpad:389
Change-Id: Ia57f402b98be2a648febb58b9dee0cb80d9e5954
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3388874
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Justin Cohen <justincohen@chromium.org>
- Formatted util/misc/initialization_state_dcheck.h
- Included build/build_config.h file in
util/stdlib/aligned_allocator_test.cc as it uses an
OS_* macro.
Change-Id: I8fb67f1ae440834d1b60f390658513a341789428
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3390648
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Avi Drissman <avi@chromium.org>
6562d2d0b2a8 Compile as C++17 by using -std=c++17 (and /std:c++17 on
Windows)
Change-Id: I10c209e8147425af77392e7649210ab68a5d27a7
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3371781
Reviewed-by: Joshua Peraza <jperaza@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
From the Fuchsia build:
```
ERROR at //third_party/crashpad/util/BUILD.gn:658:7: Undefined identifier.
defines += [ "CRASHPAD_USE_BORINGSSL" ]
^------
See //src/developer/forensics/crash_reports/BUILD.gn:127:5: which caused the file to be included.
"//third_party/crashpad/util",
^----------------------------
```
Change-Id: I1f563e5bb599b3a7a83ee8211037ee1d7464bd62
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3370891
Reviewed-by: Francois Rousseau <frousseau@google.com>
Commit-Queue: Francois Rousseau <frousseau@google.com>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Fuchsia is currently pulling in lss twice, this rolls crashpad up to
HEAD and redirects the Fuchsia build to use the existing copy in our
dependencies.
Change-Id: Ib698ee72d3d1881d22128bc94a1f6f17185a0755
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/3364805
Reviewed-by: Francois Rousseau <frousseau@google.com>
Commit-Queue: Francois Rousseau <frousseau@google.com>
