Inspector::ReadMessages as was done before r627. The "hello" message contains
the parameter count and is referenced while the message reader loops through
parameter messages. Prior to r627, both messages were named |message|, which
was confusing, probably caused a compiler warning, and apparently provided the
motivation to share them. This caused the crash inspector to fail to properly
collect the parameters. The common failure mode (although others are possible)
was for the inspector to attempt tor read more parameter messages than were
available, resulting in an IPC timeout and inspector death. No crash report
would be written, and the application expecting its crash to be inspected
would time out waiting for a response from the inspector and then _exit. This
is effectively a failure to properly handle crashes.
The inner message is reintroduced, and named parameter_message for
disambiguation.
BUG=chromium:49821
TEST=Crashes catchable by the Mac Breakpad framework
Review URL: http://breakpad.appspot.com/123002
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@628 4c0a9323-5329-0410-9bdc-e9ce6186880e
This patch avoids allocating many copies of identical strings appearing in
debugging information. Without this patch, running dump_syms on Mozilla's
libxul.so (with 173MiB of debugging information) has a peak resident set of
around 450MiB. With this patch, the peak is around 365MiB.
a=jimblandy, r=mark
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@626 4c0a9323-5329-0410-9bdc-e9ce6186880e
At present, the Linux symbol dumper maps the ELF file into memory to
examine the debugging information it contains, but then also calls
google_breakpad::FileID::ElfFileIdentifier, which maps the ELF file into
memory again. Some of our object files are large; Mozilla's libxul.so is
1.1GiB. Trying to map such files twice can interfere with tools like
valgrind that map themselves into high addresses (in an attempt to stay out
of the way of ordinary programs).
The FileID class has another method, ElfFileIdentifierFromMappedFile, that
operates on an already-loaded image of the file; use that instead.
a=jimblandy, r=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@625 4c0a9323-5329-0410-9bdc-e9ce6186880e
This patch avoids unnecessary use of the <cfoo> headers in files that don't
actually use the identifiers they declare in the std:: namespace.
It also changes some files to better conform with the "Names and Order of
Includes" rules in the Google C++ Style Guide.
A=jimb R=mark
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@619 4c0a9323-5329-0410-9bdc-e9ce6186880e
I came across a bunch of comments Neal had made on issue 55011 that I
hadn't addressed. This patch takes care of them.
A=jimb R=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@618 4c0a9323-5329-0410-9bdc-e9ce6186880e
The dwarf_cu_to_module_unittest and bytereader_unittest test executables
include object files from which they use no code.
A=jimb R=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@617 4c0a9323-5329-0410-9bdc-e9ce6186880e
This patch adds all the appropriate symbol dumper unit tests to the Mac
XCode dump_syms project. This allows us to test this code on a 64-bit
platform.
A=jimb R=mark
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@616 4c0a9323-5329-0410-9bdc-e9ce6186880e
The subclasses of CallFrameInfo::Rule store the rule currently in
force for recovering a register or computing the canonical frame
address. Their sole responsibility is to accurately convey rules from
the parser, which creates them, to a CallFrameInfo::Handler member
function, which consumes them. So, the types of their data members
should match those of the corresponding arguments of the corresponding
Handler member function.
CallFrameInfo::OffsetRule and CallFrameInfo::ValOffsetRule use an
'int' to store the rule's offset value, but
CallFrameInfo::Handler::OffsetRule and ...::ValOffsetRule expect a
'long'. On ABIs where 'long' is larger than 'int', this can cause
values to be truncated or sign-extended unexpectedly.
This patch changes those members to 'long'.
Fortunately, offsets appearing in real DWARF call frame information
never even come close to the limits of a 32-bit int, so this bug is
unlikely to cause any practical problems.
A=jimb R=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@615 4c0a9323-5329-0410-9bdc-e9ce6186880e
This patch rewrites the Mac symbol dumper to use the same set of classes
the Linux dumper does for reading debugging information from various
sources, consolidating them into a single table, and writing that out as a
Breakpad symbol file.
In the process, it also adds support for dumping DWARF call frame
information and .eh_frame exception-handling information as Breakpad 'STACK
CFI' records. This allows the Breakpad processor to generate stack traces
from code compiled with -fomit-frame-pointer.
The patch also replaces the DumpSymbols Objective C++ class with
google_breakpad::DumpSymbols, a plain C++ class. The code still uses some
Objective C++ to use the Foundation facilities for dealing with file names
in a file-system-independent fashion, and for examining the contents of
.dSYM bundles.
Since the code has been entirely rewritten, I have changed the author
lines.
A=jimb R=mark
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@614 4c0a9323-5329-0410-9bdc-e9ce6186880e
The #inclusions of <elf.h> and <link.h> were inherited from older code, but
the current code doesn't need anything from them, so they should be
removed.
A=jimb R=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@613 4c0a9323-5329-0410-9bdc-e9ce6186880e
For some reason, Mac OS X places DWARF debugging information in sections
whose names begin with "__", rather than the names beginning with "." given
in the DWARF spec. This patch changes google_breakpad::DwarfCUToModule to
look for line number information under both names.
A=jimb R=mark
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@612 4c0a9323-5329-0410-9bdc-e9ce6186880e
Instead of using bzero in main, use constructors to initialize the
Options structure.
Use C++ bool, not Objective-C BOOL.
Use a const NXArchInfo * to represent the architecture name, so that we can
use the NXGetLocalArchInfo, NXGetArchInfoFromName, etc. to handle things.
Delete the 'uuidStr' member; it is unused.
Leave Options::srcPath as an NSString, so that we can continue to use the
filesystem path abstraction methods provided by the Foundation framework.
A=jimb R=mark
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@611 4c0a9323-5329-0410-9bdc-e9ce6186880e
This patch adds files defining new classes in the google_breakpad::Mach_O
namespace for parsing fat binaries and Mach-O files. These are used in the
new dumper to handle STABS debugging information, DWARF call frame
information, and .eh_frame exception handling stack walking information.
These new classes are independent of endianness and word size, and
therefore can be used on binaries of all the relevant architectures: x86,
x86_64, ppc, and ARM.
The patch adds a complete set of unit tests for the new classes.
A=jimb R=mark (http://breakpad.appspot.com/93001/show, http://breakpad.appspot.com/115001/show)
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@610 4c0a9323-5329-0410-9bdc-e9ce6186880e
Created the exception_handler_test that test the generation of dump and the dumps themselves.
Moved all dump analysis code from minidump to its right class DumpAnalysis. The class is used by both minidump_test and exception_handler_test. The tests are way simpler that way (ie. no handling of HANDLE).
minidump_test now uses the minidump_generator class instead of using Win32. It works well and pass all tests.
exception_handler now passes both the exception and assertion infos to the client to generate the dump. If one is NULL it's going to be handled correctly.
crash_generation_client can now RequestDump with both exception and assertion info.
minidump_generator returns both the mini and full dump string pointers, and output both (or either) depending on which was generated.
All original interfaces and method signature are still there, but call the new functions if possible.
Review URL: http://codereview.chromium.org/1994015
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@596 4c0a9323-5329-0410-9bdc-e9ce6186880e
I had to remove the dependency from base (was using FilePath and ScopedHandle, replaced them by standard std::wstring and HANDLE). Also removed the logging and the main from the original files.
This will serve as a base for testing breakpad's dump generation. It is kept like this for easier tracking.
Review URL: http://codereview.chromium.org/1964006
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@592 4c0a9323-5329-0410-9bdc-e9ce6186880e
The XCode project file has become encrufted with duplicate Executable
entres and some strange settings. This patch deletes and recreates various
entries to make things neat again. It should have no effect on the
project's visible behavior.
a=jimblandy, r=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@591 4c0a9323-5329-0410-9bdc-e9ce6186880e
to TestAssembler::Section.
This patch helps the TestAssembler classes generate Mach-O object files for
use as test input.
This patch adds a new AppendCString overloading to TestAssembler::Section
for emitting null-terminated strings in fixed-length buffers, where the
string is truncated and the terminating null character omitted if the
string is too large for the buffer.
The patch includes unit tests for the new AppendCString overloading. It
also provides some for the existing overloading, which had been neglected.
a=jimblandy, r=mark
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@590 4c0a9323-5329-0410-9bdc-e9ce6186880e
Breakpad's Macintosh symbol dumper uses deprecated functions for
dealing with mixed-endianness code. This patch provides an overloaded
function, ByteSwap, that automatically chooses the OSSwap* functions
from <libkern/OSByteOrder.h> appropriate for its argument's
size.
This patch does *not* address warnings in src/common/mac/dump_syms.mm,
because that code is about to be replaced entirely; there's no reason to
bother reviewing a big, detailed patch against it.
a=jimblandy, r=mark
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@589 4c0a9323-5329-0410-9bdc-e9ce6186880e
It's possible to imagine an implementation of google_breakpad::Module in
which calling SetLoadAddress at different times as the Module is populated
would produce different output. For the Mac dumper, we'd like to depend on
its current behavior --- that the load address is subtracted off only when
writing the symbol file, and can be set at any time prior to that.
This patch makes that promise part of Module's contract, and adjusts the
test suite to verify that that promise is met.
a=jimblandy, r=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@588 4c0a9323-5329-0410-9bdc-e9ce6186880e
This patch addresses two differences between Linux and Macintosh OS X STABS
data:
- StabsReader assumes that the STABS entries follow the conventions for
storing STABS data in object file sections (that is, .stabs and
.stabstr), rather than in the object files's linker symbol table. On Mac
OS X, STABS entries live in the Mach-O file's LC_SYMTAB load command,
along with all the other linker symbols; they are not grouped into units
by N_UNDF entries.
This patch adds a boolean argument to the StabsReader constructor
indicating whether the parser should treat N_UNDF entries as unit
boundaries; this argument should be true on Linux, and false on Mac. The
patch changes src/common/linux/dump_symbols.cc to pass this new argument.
- Mac OS X STABS place SLINE (line number) records immediately before the
FUN record for the function to which they belong, and the values of such
records are absolute, not relative to the function start.
This patch extends the parser to queue up such records and report them to
the handler when we do see the FUN record. The meaning of
StabsHandler::Line remains unchanged; existing handlers do not need to be
adjusted.
This patch also adds unit tests for the new parser behaviors.
a=jimblandy, r=mark
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@587 4c0a9323-5329-0410-9bdc-e9ce6186880e
This patch factors out some of the common code in the StabsReader unit
tests into a fixture class. Pretty mechanical.
a=jimblandy, r=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@586 4c0a9323-5329-0410-9bdc-e9ce6186880e
An N_FUN stabs with no name is an explicit end-of-function marker, whose
value is the size of the function. This patch changes the stabs reader to
recognize these and use them to compute the function's ending address,
instead of treating them as functions with no names and mysterious
addresses. It also adds appropriate unit tests.
a=jimblandy, r=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@585 4c0a9323-5329-0410-9bdc-e9ce6186880e
All the other classes which receive debugging data from some sort of parser
and use it to populate a Module have names ending in "ToModule":
DwarfCUToModule, DwarfCFIToModule. Also, DumpStabsHandler doesn't actually
dump anything.
This patch renames the DumpStabsHandler class to StabsToModule, which is
more consistent and descriptive.
a=jimblandy, r=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@584 4c0a9323-5329-0410-9bdc-e9ce6186880e
StabsReader simply applies a reinterpret_cast to treat the stab entry data
as an array of 'struct nlist' structures, making the parser specific on the
host endianness, word size, and alignment rules. On Mac OS X, a single fat
binary file may contain object files of different ABIs, of which the user
chooses one at run time.
This patch changes the parser to read the data using the google_breakpad::
ByteCursor class, which can handle different endiannesses and word sizes.
The StabsReader constructor now takes arguments indicating the endianness
of the data and the size of each entry's value field. The patch changes
src/common/linux/dump_symbols.cc to pass the new argument.
This patch changes the StabsReader unit tests to use the google_breakpad::
TestAssembler classes to generate test data, rather than reading it from a
file. This makes it easy to generate test data in various endiannesses and
word sizes. It also adds tests for the new parser behaviors.
a=jimblandy, r=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@583 4c0a9323-5329-0410-9bdc-e9ce6186880e
The ByteBuffer and ByteCursor classes are utility classes for reading
binary files, handling endianness and word size issues in a portable way.
a=jimblandy, r=thestig
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@582 4c0a9323-5329-0410-9bdc-e9ce6186880e
