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>
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>
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>
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>
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>
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>
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>
This should prevent accidental "\r\n" line endings from being introduced
again, as happened in 359fc4a1336d, fixed by 31470459b624.
Also includes:
Update mini_chromium to 0512d42698bfb47f2016ac627177c22d22b983d4
0512d42698bf Add .gitattributes, setting “text eol=lf” for all files
Change-Id: Id84c014914fec66632006ed364e2b14b4f1c175c
Reviewed-on: https://chromium-review.googlesource.com/c/crashpad/crashpad/+/1953807
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Robert Sesek <rsesek@chromium.org>
This is a follow-up to c8a016b99d97, following the post-landing
discussion at
https://chromium-review.googlesource.com/c/crashpad/crashpad/+/1393921/5#message-2058541d8c4505d20a990ab7734cd758e437a5f7
base::size, and std::size that will eventually replace it when C++17 is
assured, does not allow the size of non-static data members to be taken
in constant expression context. The remaining uses of ArraySize are in:
minidump/minidump_exception_writer.cc (×1)
minidump/minidump_system_info_writer.cc (×2, also uses base::size)
snapshot/cpu_context.cc (×4, also uses base::size)
util/misc/arraysize_test.cc (×10, of course)
The first of these occurs when initializing a constexpr variable. All
others are in expressions used with static_assert.
Includes:
Update mini_chromium to 737433ebade4d446643c6c07daae02a67e8deccao
f701716d9546 Add Windows ARM64 build target to mini_chromium
87a95a3d6ac2 Remove the arraysize macro
1f7255ead1f7 Placate MSVC in areas of base::size usage
737433ebade4 Add cast
Bug: chromium:837308
Change-Id: I6a5162654461b1bdd9b7b6864d0d71a734bcde19
Reviewed-on: https://chromium-review.googlesource.com/c/1396108
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
This allows clients to use the database to handle uploads themselves,
e.g. on Android, where Crashpad does not yet provide an uploader.
The handler does not launch an upload thread when no url is supplied.
Previously, the handler would move these reports to
completed and record the upload as skipped with kUploadsDisabled.
With this change, these reports would remain pending until pruned,
with no metrics recorded for them in regard to their upload.
Bug: crashpad:30
Change-Id: I4167ab1531634b10e91d03229018ae6aab4103aa
Reviewed-on: https://chromium-review.googlesource.com/1010970
Reviewed-by: Robert Sesek <rsesek@chromium.org>
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
end_to_end_test.py expects g_extra_memory_pointer data to be 32-bit,
so use a fixed-size type.
Change-Id: I5798bc8a895d7e02461671fd31e59dd178f6a6a4
Reviewed-on: https://chromium-review.googlesource.com/957792
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
This caused an error with clang, but not msvc.
At first I thought this might be a discrepancy between the warning
levels used, but it appears msvc was okay with this because ints are
the same size as longs.
Change-Id: I798284fef9aa70b1bfda73308b9babe1779e8f4b
Reviewed-on: https://chromium-review.googlesource.com/941723
Reviewed-by: Mark Mentovai <mark@chromium.org>
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
This change updates CrashReportDatbase::NewReport objects to own the
file handle associated with the new report, now accessible via a
FileWriter. NewReport's destructor closes its file handle and removes
its new report unless disarmed with FinishedWritingCrashReport,
eliminating the need for CallErrorWritingCrashReport.
Bug: crashpad:206
Change-Id: Iccb5bbc0ebadb07a237ff8eb938389afcfeae2a5
Reviewed-on: https://chromium-review.googlesource.com/916941
Commit-Queue: Joshua Peraza <jperaza@chromium.org>
Reviewed-by: Mark Mentovai <mark@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
Change-Id: I4b247d7fae1a212350f8ffcf2bf5ba1fa730f5c1
Reviewed-on: https://chromium-review.googlesource.com/780339
Reviewed-by: Scott Graham <scottmg@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
The design for running all Crashpad unit tests on Chromium’s try- and
buildbots involves pulling all tests into a single monolithic
crashpad_tests executable. Many Crashpad tests base the name of their
child executables or modules on the name of the main test executable.
Since the main test executable will have a different name in the
in-Chromium build, knowledge of the test executable name (referred to as
“module” here) needs to be added to the tests themselves.
This introduces TestPaths::BuildArtifact(), which allows the module name
to be specified. For Crashpad’s standalone build, the module name is
verified against the main test executable’s name.
TestPaths::BuildArtifact() can also locate paths in the alternate 32-bit
output directory for 64-bit Windows tests, taking on the responsibility
for what the new (5e9ed4cb9f69) TestPaths::Output32BitDirectory(), now
obsolete, did.
Bug: chromium:779790
Change-Id: I64c4a2190b6319e487c999812a7cfc512a75a700
Reviewed-on: https://chromium-review.googlesource.com/747536
Reviewed-by: Scott Graham <scottmg@chromium.org>
hanging_program.exe is used by crash_other_program.exe, which is in turn
used by end_to_end_test.py. It hangs by loading loader_lock_dll.dll,
which squats in its entry point function while the loader lock is held.
hanging_program.exe needs to do some work in its Thread1() before the
loader lock is taken (a SetThreadPriority() call), and needs to do some
work in its main thread once the loader lock is held (it needs to signal
crash_other_program.exe that it’s successfully wedged itself).
Previously, proper synchronization was not provided. A 1-second Sleep()
was used to wait for the loader lock to be taken. Thread1() pre-work was
only achieved before the loader lock was taken by sheer luck. Things
didn’t always work out so nicely.
This uses an event handle to provide synchronization. An environment
variable is used to pass the handle to loader_lock_dll.dll, because
there aren’t many better options available. This eliminates both flake
and the unnecessary 1-second delay in hanging_program.exe, and since
this program runs twice during end_to_end_test.py, it improves that
test’s runtime by 2 seconds.
Bug: crashpad:197
Test: end_to_end_test.py
Change-Id: Ib9883215ef96bed7571464cc68e09b6ab6310ae6
Reviewed-on: https://chromium-review.googlesource.com/700076
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Leonard Mosescu <mosescu@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
This is essentially based on a search for “^ *const [^*&]*=[^(]*$”
Change-Id: Id571119d0b9a64c6f387eccd51cea7c9eb530e13
Reviewed-on: https://chromium-review.googlesource.com/585555
Reviewed-by: Leonard Mosescu <mosescu@chromium.org>
This uses “static” at function scope to avoid making local copies, even
in cases where the compiler can’t see that the local copy is
unnecessary. “constexpr” adds additional safety in that it prevents
global state from being initialized from any runtime dependencies, which
would be undesirable.
At namespace scope, “constexpr” is also used where appropriate.
For the most part, this was a mechanical transformation for things
matching '(^| )const [^=]*\['.
Similar transformations could be applied to non-arrays in some cases,
but there’s limited practical impact in most non-array cases relative to
arrays, there are far more use sites, and much more manual intervention
would be required.
Change-Id: I3513b739ee8b0be026f8285475cddc5f9cc81152
Reviewed-on: https://chromium-review.googlesource.com/583997
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Leonard Mosescu <mosescu@chromium.org>
76a67a37b1d0 adds crashpad_handler’s --monitor-self argument, which
results in a second crashpad_handler instance running out of the same
database as the initial crashpad_handler instance that it monitors. The
two handlers start at nearly the same time, and will initially be on
precisely the same schedule for periodic tasks such as scanning for new
reports to upload and pruning the database. This is an unnecessary
duplication of effort.
This adds a new --no-periodic-tasks argument to crashpad_handler. When
the first instance of crashpad_handler starts a second to monitor it, it
will use this argument, which prevents the second instance from
performing these tasks.
When --no-periodic-tasks is in effect, crashpad_handler will still be
able to upload crash reports that it knows about by virtue of having
written them itself, but it will not scan the database for other pending
reports to upload.
Bug: crashpad:143
Test: crashpad_util_test ThreadSafeVector.ThreadSafeVector
Change-Id: I7b249dd7b6d5782448d8071855818f986b98ab5a
Reviewed-on: https://chromium-review.googlesource.com/473827
Reviewed-by: Robert Sesek <rsesek@chromium.org>
This supports the “double handler” or “double handler with low
probability” models from https://crashpad.chromium.org/bug/143.
For crashpad_handler to be become its own client, it needs access to its
own executable path to pass to CrashpadClient::StartHandler(). This was
formerly available in the test-only test::Paths::Executable(). Bring
that function’s implementation to the non-test Paths::Executable() in
util/misc, and rename test::Paths to test::TestPaths to avoid future
confusion.
test::TestPaths must still be used to access TestDataRoot(), which does
not make any sense to non-test code.
test::TestPaths::Executable() is retained for use by tests, which most
likely prefer the fatal semantics of that function. Paths::Executable()
is not fatal because for the purposes of implementing the double
handler, a failure to locate the executable path (which may happen on
some systems in deeply-nested directory hierarchies) shouldn’t cause the
initial crashpad_handler to abort, even if it does prevent a second
crashpad_handler from being started.
Bug: crashpad:143
Test: crashpad_util_test Paths.*, crashpad_test_test TestPaths.*
Change-Id: I9f75bf61839ce51e33c9f7c0d7031cebead6a156
Reviewed-on: https://chromium-review.googlesource.com/466346
Reviewed-by: Scott Graham <scottmg@chromium.org>
Commit-Queue: Mark Mentovai <mark@chromium.org>
This is like 270490ff79df, but for things run by end_to_end_test.py, and
things run for it by crash_other_program.exe.
Change-Id: Iabf3c762c50f41eb61ab31f714c646364196e745
Reviewed-on: https://chromium-review.googlesource.com/458822
Commit-Queue: Mark Mentovai <mark@chromium.org>
Reviewed-by: Scott Graham <scottmg@chromium.org>
ReadFile() attempted to continue reading after a short read. In most
cases, this is fine. However, ReadFile() would keep trying to fill a
partially-filled buffer until experiencing a 0-length read(), signaling
end-of-file. For certain weird file descriptors like terminal input, EOF
is an ephemeral condition, and attempting to read beyond EOF doesn’t
actually return 0 (EOF) provided that they remain open, it will block
waiting for more input. Consequently, ReadFile() and anything based on
ReadFile() had an undocumented and quirky interface, which was that any
short read that it returned (not an underlying short read) actually
indicated EOF.
This facet of ReadFile() was unexpected, so it’s being removed. The new
behavior is that ReadFile() will return an underlying short read. The
behavior of FileReaderInterface::Read() is updated in accordance with
this change.
Upon experiencing a short read, the caller can determine the best
action. Most callers were already prepared for this behavior. Outside of
util/file, only crashpad_database_util properly implemented EOF
detection according to previous semantics, and adapting it to new
semantics is trivial.
Callers who require an exact-length read can use the new
ReadFileExactly(), or the newly renamed LoggingReadFileExactly() or
CheckedReadFileExactly(). These functions will retry following a short
read. The renamed functions were previously called LoggingReadFile() and
CheckedReadFile(), but those names implied that they were simply
wrapping ReadFile(), which is not the case. They wrapped ReadFile() and
further, insisted on a full read. Since ReadFile()’s semantics are now
changing but these functions’ are not, they’re now even more distinct
from ReadFile(), and must be renamed to avoid confusion.
Test: *
Change-Id: I06b77e0d6ad8719bd2eb67dab93a8740542dd908
Reviewed-on: https://chromium-review.googlesource.com/456676
Reviewed-by: Robert Sesek <rsesek@chromium.org>
I haven't been able to reproduce this locally, but we see errors in
crash dumps where the unloaded module list consists of a number of
modules with invalid names and implausible addresses. My assumption is
that RTL_UNLOAD_EVENT_TRACE isn't correct for some OS levels. Instead of
trying to finesse and test that, use RtlGetUnloadEventTraceEx() instead
of RtlGetUnloadEventTrace(), which returns an element size. (This
function is Vista+ which is why it wasn't used the first time around.)
R=mark@chromium.org
BUG=chromium:620175
Change-Id: I4d7080a03623276f9c1c038d6e7329af70e4a64c
Reviewed-on: https://chromium-review.googlesource.com/421564
Reviewed-by: Mark Mentovai <mark@chromium.org>
The Windows 10 loader starts a few extra threads before main(). In a few
of the test cases, the tests were relying on thread ordering (generally,
the test thread being at index #1). Instead, use other signals to find
the correct thread to verify.
R=mark@chromium.org
Change-Id: Icb1f5a8fdf3a0ea6d82ab65960dbcb650965f269
Reviewed-on: https://chromium-review.googlesource.com/407073
Reviewed-by: Mark Mentovai <mark@chromium.org>
Previously, StartHandler() launched the handler process, then connected
over a pipe to register for crash handling. Instead, the initial client
can create and inherit handles to the handler and pass those handle
values and other data (addresses, etc.) on the command line.
This should improve startup time as there's no need to synchronize with
the process at startup, and allows avoiding a call to CreateProcess()
directly in StartHandler(), which is important for registration for
crash reporting from DllMain().
Incidentally adds new utility functions for string/number conversion and
string splitting.
Note: API change; UseHandler() is removed for all platforms.
BUG=chromium:567850,chromium:656800
Change-Id: I1602724183cb107f805f109674c53e95841b24fd
Reviewed-on: https://chromium-review.googlesource.com/400015
Reviewed-by: Mark Mentovai <mark@chromium.org>
Includes mini_chromium DEPS roll for:
88e0a3e Add stub of sparse_histogram.h
R=mark@chromium.org
BUG=crashpad:100
Change-Id: I4c541a33be0f7f47e972af638d4765bd06682acf
Reviewed-on: https://chromium-review.googlesource.com/386385
Reviewed-by: Mark Mentovai <mark@chromium.org>
