2010-02-09 18:08:56 +01:00
|
|
|
// Copyright (c) 2010 Google Inc.
|
2006-09-20 23:16:16 +02:00
|
|
|
// All rights reserved.
|
2006-08-25 23:14:45 +02:00
|
|
|
//
|
2006-09-20 23:16:16 +02:00
|
|
|
// Redistribution and use in source and binary forms, with or without
|
|
|
|
|
// modification, are permitted provided that the following conditions are
|
|
|
|
|
// met:
|
2006-08-25 23:14:45 +02:00
|
|
|
//
|
2006-09-20 23:16:16 +02:00
|
|
|
// * Redistributions of source code must retain the above copyright
|
|
|
|
|
// notice, this list of conditions and the following disclaimer.
|
|
|
|
|
// * Redistributions in binary form must reproduce the above
|
|
|
|
|
// copyright notice, this list of conditions and the following disclaimer
|
|
|
|
|
// in the documentation and/or other materials provided with the
|
|
|
|
|
// distribution.
|
|
|
|
|
// * Neither the name of Google Inc. nor the names of its
|
|
|
|
|
// contributors may be used to endorse or promote products derived from
|
|
|
|
|
// this software without specific prior written permission.
|
2006-08-25 23:14:45 +02:00
|
|
|
//
|
2006-09-20 23:16:16 +02:00
|
|
|
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
|
|
|
|
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
|
|
|
|
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
|
|
|
|
|
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
|
|
|
|
|
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
|
|
|
|
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
|
|
|
|
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
|
|
|
|
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
|
|
|
|
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
|
|
|
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
|
|
|
|
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
2010-10-07 22:31:36 +02:00
|
|
|
//
|
|
|
|
|
// basic_source_line_resolver.cc: BasicSourceLineResolver implementation.
|
|
|
|
|
//
|
|
|
|
|
// See basic_source_line_resolver.h and basic_source_line_resolver_types.h
|
|
|
|
|
// for documentation.
|
|
|
|
|
|
2013-09-25 20:25:13 +02:00
|
|
|
#include <assert.h>
|
2006-08-25 23:14:45 +02:00
|
|
|
#include <stdio.h>
|
|
|
|
|
#include <string.h>
|
2009-02-19 22:26:20 +01:00
|
|
|
#include <sys/types.h>
|
|
|
|
|
#include <sys/stat.h>
|
2006-10-25 23:25:41 +02:00
|
|
|
|
2013-09-25 20:25:13 +02:00
|
|
|
#include <limits>
|
2006-10-23 21:24:58 +02:00
|
|
|
#include <map>
|
2021-07-22 20:13:37 +02:00
|
|
|
#include <memory>
|
2006-08-25 23:14:45 +02:00
|
|
|
#include <utility>
|
2006-10-25 23:25:41 +02:00
|
|
|
#include <vector>
|
|
|
|
|
|
2007-02-14 20:51:05 +01:00
|
|
|
#include "google_breakpad/processor/basic_source_line_resolver.h"
|
2010-10-07 22:31:36 +02:00
|
|
|
#include "processor/basic_source_line_resolver_types.h"
|
|
|
|
|
#include "processor/module_factory.h"
|
|
|
|
|
|
2010-04-09 01:06:23 +02:00
|
|
|
#include "processor/tokenize.h"
|
2006-08-25 23:14:45 +02:00
|
|
|
|
2021-10-20 23:07:55 +02:00
|
|
|
using std::deque;
|
2006-08-30 22:05:05 +02:00
|
|
|
using std::make_pair;
|
2021-10-20 23:07:55 +02:00
|
|
|
using std::map;
|
2021-07-22 20:13:37 +02:00
|
|
|
using std::unique_ptr;
|
2021-10-20 23:07:55 +02:00
|
|
|
using std::vector;
|
2006-08-25 23:14:45 +02:00
|
|
|
|
2007-02-14 20:51:05 +01:00
|
|
|
namespace google_breakpad {
|
2006-08-25 23:14:45 +02:00
|
|
|
|
2013-03-06 15:06:52 +01:00
|
|
|
#ifdef _WIN32
|
2015-08-20 17:09:16 +02:00
|
|
|
#ifdef _MSC_VER
|
2013-03-06 15:06:52 +01:00
|
|
|
#define strtok_r strtok_s
|
2015-08-20 17:09:16 +02:00
|
|
|
#endif
|
2013-03-06 15:06:52 +01:00
|
|
|
#define strtoull _strtoui64
|
|
|
|
|
#endif
|
|
|
|
|
|
2017-11-29 22:29:37 +01:00
|
|
|
namespace {
|
|
|
|
|
|
|
|
|
|
// Utility function to tokenize given the presence of an optional initial
|
|
|
|
|
// field. In this case, optional_field is the expected string for the optional
|
|
|
|
|
// field, and max_tokens is the maximum number of tokens including the optional
|
|
|
|
|
// field. Refer to the documentation for Tokenize for descriptions of the other
|
|
|
|
|
// arguments.
|
2020-06-24 00:55:43 +02:00
|
|
|
bool TokenizeWithOptionalField(char* line,
|
|
|
|
|
const char* optional_field,
|
|
|
|
|
const char* separators,
|
2017-11-29 22:29:37 +01:00
|
|
|
int max_tokens,
|
2020-06-24 00:55:43 +02:00
|
|
|
vector<char*>* tokens) {
|
2017-11-29 22:29:37 +01:00
|
|
|
// First tokenize assuming the optional field is not present. If we then see
|
|
|
|
|
// the optional field, additionally tokenize the last token into two tokens.
|
|
|
|
|
if (!Tokenize(line, separators, max_tokens - 1, tokens)) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (strcmp(tokens->front(), optional_field) == 0) {
|
|
|
|
|
// The optional field is present. Split the last token in two to recover the
|
|
|
|
|
// field prior to the last.
|
|
|
|
|
vector<char*> last_tokens;
|
|
|
|
|
if (!Tokenize(tokens->back(), separators, 2, &last_tokens)) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
// Replace the previous last token with the two new tokens.
|
|
|
|
|
tokens->pop_back();
|
|
|
|
|
tokens->push_back(last_tokens[0]);
|
|
|
|
|
tokens->push_back(last_tokens[1]);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
} // namespace
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
static const char* kWhitespace = " \r\n";
|
2013-07-11 03:36:06 +02:00
|
|
|
static const int kMaxErrorsPrinted = 5;
|
|
|
|
|
static const int kMaxErrorsBeforeBailing = 100;
|
2010-04-09 01:06:23 +02:00
|
|
|
|
2010-10-07 22:31:36 +02:00
|
|
|
BasicSourceLineResolver::BasicSourceLineResolver() :
|
|
|
|
|
SourceLineResolverBase(new BasicModuleFactory) { }
|
2006-08-25 23:14:45 +02:00
|
|
|
|
2013-07-11 03:36:06 +02:00
|
|
|
// static
|
|
|
|
|
void BasicSourceLineResolver::Module::LogParseError(
|
2020-06-24 00:55:43 +02:00
|
|
|
const string& message,
|
2013-07-11 03:36:06 +02:00
|
|
|
int line_number,
|
2020-06-24 00:55:43 +02:00
|
|
|
int* num_errors) {
|
2013-07-11 03:36:06 +02:00
|
|
|
if (++(*num_errors) <= kMaxErrorsPrinted) {
|
|
|
|
|
if (line_number > 0) {
|
|
|
|
|
BPLOG(ERROR) << "Line " << line_number << ": " << message;
|
|
|
|
|
} else {
|
|
|
|
|
BPLOG(ERROR) << message;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
bool BasicSourceLineResolver::Module::LoadMapFromMemory(
|
2020-06-24 00:55:43 +02:00
|
|
|
char* memory_buffer,
|
2013-07-11 03:36:06 +02:00
|
|
|
size_t memory_buffer_size) {
|
2009-02-19 22:26:20 +01:00
|
|
|
linked_ptr<Function> cur_func;
|
|
|
|
|
int line_number = 0;
|
2013-07-11 03:36:06 +02:00
|
|
|
int num_errors = 0;
|
2020-06-24 00:55:43 +02:00
|
|
|
char* save_ptr;
|
2009-04-22 02:55:31 +02:00
|
|
|
|
|
|
|
|
// If the length is 0, we can still pretend we have a symbol file. This is
|
2010-10-07 22:31:36 +02:00
|
|
|
// for scenarios that want to test symbol lookup, but don't necessarily care
|
|
|
|
|
// if certain modules do not have any information, like system libraries.
|
2013-07-11 03:36:06 +02:00
|
|
|
if (memory_buffer_size == 0) {
|
2009-04-22 02:55:31 +02:00
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
2013-07-11 03:36:06 +02:00
|
|
|
// Make sure the last character is null terminator.
|
|
|
|
|
size_t last_null_terminator = memory_buffer_size - 1;
|
|
|
|
|
if (memory_buffer[last_null_terminator] != '\0') {
|
|
|
|
|
memory_buffer[last_null_terminator] = '\0';
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Skip any null terminators at the end of the memory buffer, and make sure
|
|
|
|
|
// there are no other null terminators in the middle of the memory buffer.
|
|
|
|
|
bool has_null_terminator_in_the_middle = false;
|
|
|
|
|
while (last_null_terminator > 0 &&
|
|
|
|
|
memory_buffer[last_null_terminator - 1] == '\0') {
|
|
|
|
|
last_null_terminator--;
|
|
|
|
|
}
|
|
|
|
|
for (size_t i = 0; i < last_null_terminator; i++) {
|
|
|
|
|
if (memory_buffer[i] == '\0') {
|
|
|
|
|
memory_buffer[i] = '_';
|
|
|
|
|
has_null_terminator_in_the_middle = true;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (has_null_terminator_in_the_middle) {
|
|
|
|
|
LogParseError(
|
|
|
|
|
"Null terminator is not expected in the middle of the symbol data",
|
|
|
|
|
line_number,
|
|
|
|
|
&num_errors);
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
char* buffer;
|
2010-10-07 22:31:36 +02:00
|
|
|
buffer = strtok_r(memory_buffer, "\r\n", &save_ptr);
|
2006-08-25 23:14:45 +02:00
|
|
|
|
2009-02-19 22:26:20 +01:00
|
|
|
while (buffer != NULL) {
|
2007-05-21 22:09:33 +02:00
|
|
|
++line_number;
|
2009-02-19 22:26:20 +01:00
|
|
|
|
2006-08-25 23:14:45 +02:00
|
|
|
if (strncmp(buffer, "FILE ", 5) == 0) {
|
2007-05-21 22:09:33 +02:00
|
|
|
if (!ParseFile(buffer)) {
|
2013-07-11 03:36:06 +02:00
|
|
|
LogParseError("ParseFile on buffer failed", line_number, &num_errors);
|
2007-05-21 22:09:33 +02:00
|
|
|
}
|
2006-09-28 23:09:37 +02:00
|
|
|
} else if (strncmp(buffer, "STACK ", 6) == 0) {
|
|
|
|
|
if (!ParseStackInfo(buffer)) {
|
2013-07-11 03:36:06 +02:00
|
|
|
LogParseError("ParseStackInfo failed", line_number, &num_errors);
|
2006-09-28 23:09:37 +02:00
|
|
|
}
|
2006-08-25 23:14:45 +02:00
|
|
|
} else if (strncmp(buffer, "FUNC ", 5) == 0) {
|
2007-03-23 00:20:17 +01:00
|
|
|
cur_func.reset(ParseFunction(buffer));
|
|
|
|
|
if (!cur_func.get()) {
|
2013-07-11 03:36:06 +02:00
|
|
|
LogParseError("ParseFunction failed", line_number, &num_errors);
|
|
|
|
|
} else {
|
|
|
|
|
// StoreRange will fail if the function has an invalid address or size.
|
|
|
|
|
// We'll silently ignore this, the function and any corresponding lines
|
|
|
|
|
// will be destroyed when cur_func is released.
|
|
|
|
|
functions_.StoreRange(cur_func->address, cur_func->size, cur_func);
|
2006-10-20 21:50:01 +02:00
|
|
|
}
|
2006-10-20 03:46:38 +02:00
|
|
|
} else if (strncmp(buffer, "PUBLIC ", 7) == 0) {
|
2006-10-20 21:50:01 +02:00
|
|
|
// Clear cur_func: public symbols don't contain line number information.
|
2007-03-23 00:20:17 +01:00
|
|
|
cur_func.reset();
|
2006-10-20 21:50:01 +02:00
|
|
|
|
|
|
|
|
if (!ParsePublicSymbol(buffer)) {
|
2013-07-11 03:36:06 +02:00
|
|
|
LogParseError("ParsePublicSymbol failed", line_number, &num_errors);
|
2006-10-20 21:50:01 +02:00
|
|
|
}
|
2006-11-06 20:34:19 +01:00
|
|
|
} else if (strncmp(buffer, "MODULE ", 7) == 0) {
|
2006-12-12 00:22:54 +01:00
|
|
|
// Ignore these. They're not of any use to BasicSourceLineResolver,
|
|
|
|
|
// which is fed modules by a SymbolSupplier. These lines are present to
|
2006-11-06 20:34:19 +01:00
|
|
|
// aid other tools in properly placing symbol files so that they can
|
|
|
|
|
// be accessed by a SymbolSupplier.
|
|
|
|
|
//
|
|
|
|
|
// MODULE <guid> <age> <filename>
|
2010-10-05 21:39:23 +02:00
|
|
|
} else if (strncmp(buffer, "INFO ", 5) == 0) {
|
|
|
|
|
// Ignore these as well, they're similarly just for housekeeping.
|
|
|
|
|
//
|
|
|
|
|
// INFO CODE_ID <code id> <filename>
|
2021-07-22 20:13:37 +02:00
|
|
|
} else if (strncmp(buffer, "INLINE ", 7) == 0) {
|
|
|
|
|
linked_ptr<Inline> in = ParseInline(buffer);
|
|
|
|
|
if (!in.get())
|
|
|
|
|
LogParseError("ParseInline failed", line_number, &num_errors);
|
|
|
|
|
else
|
|
|
|
|
cur_func->AppendInline(in);
|
|
|
|
|
} else if (strncmp(buffer, "INLINE_ORIGIN ", 14) == 0) {
|
|
|
|
|
if (!ParseInlineOrigin(buffer)) {
|
|
|
|
|
LogParseError("ParseInlineOrigin failed", line_number, &num_errors);
|
|
|
|
|
}
|
2006-08-25 23:14:45 +02:00
|
|
|
} else {
|
2007-03-23 00:20:17 +01:00
|
|
|
if (!cur_func.get()) {
|
2013-07-11 03:36:06 +02:00
|
|
|
LogParseError("Found source line data without a function",
|
|
|
|
|
line_number, &num_errors);
|
|
|
|
|
} else {
|
2020-06-24 00:55:43 +02:00
|
|
|
Line* line = ParseLine(buffer);
|
2013-07-11 03:36:06 +02:00
|
|
|
if (!line) {
|
|
|
|
|
LogParseError("ParseLine failed", line_number, &num_errors);
|
|
|
|
|
} else {
|
|
|
|
|
cur_func->lines.StoreRange(line->address, line->size,
|
|
|
|
|
linked_ptr<Line>(line));
|
|
|
|
|
}
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
2013-07-11 03:36:06 +02:00
|
|
|
}
|
|
|
|
|
if (num_errors > kMaxErrorsBeforeBailing) {
|
|
|
|
|
break;
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
2009-02-19 22:26:20 +01:00
|
|
|
buffer = strtok_r(NULL, "\r\n", &save_ptr);
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
2013-07-11 03:36:06 +02:00
|
|
|
is_corrupt_ = num_errors > 0;
|
2006-08-25 23:14:45 +02:00
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
2021-10-20 23:07:55 +02:00
|
|
|
void BasicSourceLineResolver::Module::ConstructInlineFrames(
|
2021-07-22 20:13:37 +02:00
|
|
|
StackFrame* frame,
|
|
|
|
|
MemAddr address,
|
|
|
|
|
const RangeMap<uint64_t, linked_ptr<Inline>>& inlines,
|
2021-10-20 23:07:55 +02:00
|
|
|
deque<unique_ptr<StackFrame>>* inlined_frames) const {
|
2021-07-22 20:13:37 +02:00
|
|
|
linked_ptr<Inline> in;
|
|
|
|
|
MemAddr inline_base;
|
|
|
|
|
if (!inlines.RetrieveRange(address, &in, &inline_base, nullptr, nullptr))
|
2021-10-20 23:07:55 +02:00
|
|
|
return;
|
2021-08-31 19:50:11 +02:00
|
|
|
auto origin = inline_origins_.find(in->origin_id);
|
|
|
|
|
if (origin == inline_origins_.end())
|
2021-10-20 23:07:55 +02:00
|
|
|
return;
|
2021-07-22 20:13:37 +02:00
|
|
|
|
2021-10-20 23:07:55 +02:00
|
|
|
// Update parent frame's source line and source file.
|
|
|
|
|
frame->source_line = in->call_site_line;
|
|
|
|
|
auto file = files_.find(in->call_site_file_id);
|
|
|
|
|
if (file != files_.end()) {
|
|
|
|
|
frame->source_file_name = file->second;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Create a child frame of `frame`.
|
|
|
|
|
StackFrame child_frame = StackFrame(*frame);
|
|
|
|
|
child_frame.function_name = origin->second->name;
|
2021-07-22 20:13:37 +02:00
|
|
|
// Use the starting adress of the inlined range as inlined function base.
|
2021-10-20 23:07:55 +02:00
|
|
|
child_frame.function_base = child_frame.module->base_address() + inline_base;
|
|
|
|
|
child_frame.trust = StackFrame::FRAME_TRUST_INLINE;
|
|
|
|
|
ConstructInlineFrames(&child_frame, address, in->child_inlines,
|
|
|
|
|
inlined_frames);
|
|
|
|
|
// Add child_frame after ConstructInlineFrames so that the innermost frame is
|
|
|
|
|
// the first frame inside inlined_frames.
|
|
|
|
|
inlined_frames->push_back(
|
|
|
|
|
unique_ptr<StackFrame>(new StackFrame(child_frame)));
|
2021-07-22 20:13:37 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
void BasicSourceLineResolver::Module::LookupAddress(
|
|
|
|
|
StackFrame* frame,
|
2021-10-20 23:07:55 +02:00
|
|
|
deque<unique_ptr<StackFrame>>* inlined_frames) const {
|
2006-12-05 23:52:28 +01:00
|
|
|
MemAddr address = frame->instruction - frame->module->base_address();
|
|
|
|
|
|
2010-01-11 23:31:50 +01:00
|
|
|
// First, look for a FUNC record that covers address. Use
|
|
|
|
|
// RetrieveNearestRange instead of RetrieveRange so that, if there
|
|
|
|
|
// is no such function, we can use the next function to bound the
|
|
|
|
|
// extent of the PUBLIC symbol we find, below. This does mean we
|
|
|
|
|
// need to check that address indeed falls within the function we
|
|
|
|
|
// find; do the range comparison in an overflow-friendly way.
|
2006-09-08 20:03:56 +02:00
|
|
|
linked_ptr<Function> func;
|
2006-10-20 21:50:01 +02:00
|
|
|
linked_ptr<PublicSymbol> public_symbol;
|
|
|
|
|
MemAddr function_base;
|
|
|
|
|
MemAddr function_size;
|
|
|
|
|
MemAddr public_address;
|
2016-06-06 07:41:10 +02:00
|
|
|
if (functions_.RetrieveNearestRange(address, &func, &function_base,
|
|
|
|
|
NULL /* delta */, &function_size) &&
|
Breakpad processor: Fix function and public symbol lookup.
In r480, I botched the change to make the comparisons that decide
whether an address falls within a function's range safe from overflow.
The original code said:
address >= function_base && address < function_base + function_size
which is fine unless the function abuts the end of the address space,
in which case the addition overflows and you get a false negative.
My change subtracted function_size from both sides of the latter
comparison, which is meaning-preserving in true math, and gets you:
address >= function_base && address - function_size < function_base
This not only reads strangely, but also still overflows if
function_size is greater than address. That's rare, but I've added a
case to the unit tests that checks it.
My intent had been to replace the addition which could overflow with a
subtraction that was known not to overflow, namely:
address >= function_base && address - function_base < function_size
This is equivalent to the original in true math, and because of the
first comparison, we know the subtraction won't underflow in MemAddr
math.
The patch includes similar fixes to the public symbol lookup code, and
to FindWindowsFrameInfo, which was the only other function affected by
r480.
a=jimblandy, r=mmentovai
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@503 4c0a9323-5329-0410-9bdc-e9ce6186880e
2010-01-28 06:17:23 +01:00
|
|
|
address >= function_base && address - function_base < function_size) {
|
2006-10-20 21:50:01 +02:00
|
|
|
frame->function_name = func->name;
|
2006-12-05 23:52:28 +01:00
|
|
|
frame->function_base = frame->module->base_address() + function_base;
|
2006-10-20 21:50:01 +02:00
|
|
|
|
|
|
|
|
linked_ptr<Line> line;
|
|
|
|
|
MemAddr line_base;
|
2016-06-06 07:41:10 +02:00
|
|
|
if (func->lines.RetrieveRange(address, &line, &line_base, NULL /* delta */,
|
|
|
|
|
NULL /* size */)) {
|
2006-10-20 21:50:01 +02:00
|
|
|
FileMap::const_iterator it = files_.find(line->source_file_id);
|
|
|
|
|
if (it != files_.end()) {
|
|
|
|
|
frame->source_file_name = files_.find(line->source_file_id)->second;
|
|
|
|
|
}
|
|
|
|
|
frame->source_line = line->line;
|
2006-12-05 23:52:28 +01:00
|
|
|
frame->source_line_base = frame->module->base_address() + line_base;
|
2006-10-20 21:50:01 +02:00
|
|
|
}
|
2021-07-22 20:13:37 +02:00
|
|
|
|
|
|
|
|
// Check if this is inlined function call.
|
|
|
|
|
if (inlined_frames) {
|
2021-10-20 23:07:55 +02:00
|
|
|
int source_line = frame->source_line;
|
|
|
|
|
string source_file_name = frame->source_file_name;
|
|
|
|
|
ConstructInlineFrames(frame, address, func->inlines, inlined_frames);
|
|
|
|
|
if (!inlined_frames->empty()) {
|
|
|
|
|
// Update the inner most frame's source line and source file name.
|
|
|
|
|
inlined_frames->front()->source_line = source_line;
|
|
|
|
|
inlined_frames->front()->source_file_name = source_file_name;
|
2021-07-22 20:13:37 +02:00
|
|
|
}
|
|
|
|
|
}
|
2006-10-20 21:50:01 +02:00
|
|
|
} else if (public_symbols_.Retrieve(address,
|
|
|
|
|
&public_symbol, &public_address) &&
|
Breakpad processor: Fix function and public symbol lookup.
In r480, I botched the change to make the comparisons that decide
whether an address falls within a function's range safe from overflow.
The original code said:
address >= function_base && address < function_base + function_size
which is fine unless the function abuts the end of the address space,
in which case the addition overflows and you get a false negative.
My change subtracted function_size from both sides of the latter
comparison, which is meaning-preserving in true math, and gets you:
address >= function_base && address - function_size < function_base
This not only reads strangely, but also still overflows if
function_size is greater than address. That's rare, but I've added a
case to the unit tests that checks it.
My intent had been to replace the addition which could overflow with a
subtraction that was known not to overflow, namely:
address >= function_base && address - function_base < function_size
This is equivalent to the original in true math, and because of the
first comparison, we know the subtraction won't underflow in MemAddr
math.
The patch includes similar fixes to the public symbol lookup code, and
to FindWindowsFrameInfo, which was the only other function affected by
r480.
a=jimblandy, r=mmentovai
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@503 4c0a9323-5329-0410-9bdc-e9ce6186880e
2010-01-28 06:17:23 +01:00
|
|
|
(!func.get() || public_address > function_base)) {
|
2006-10-20 21:50:01 +02:00
|
|
|
frame->function_name = public_symbol->name;
|
2006-12-05 23:52:28 +01:00
|
|
|
frame->function_base = frame->module->base_address() + public_address;
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
2010-01-11 23:31:50 +01:00
|
|
|
}
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
WindowsFrameInfo* BasicSourceLineResolver::Module::FindWindowsFrameInfo(
|
|
|
|
|
const StackFrame* frame) const {
|
2010-01-11 23:31:50 +01:00
|
|
|
MemAddr address = frame->instruction - frame->module->base_address();
|
|
|
|
|
scoped_ptr<WindowsFrameInfo> result(new WindowsFrameInfo());
|
2006-08-25 23:14:45 +02:00
|
|
|
|
2010-04-09 01:06:23 +02:00
|
|
|
// We only know about WindowsFrameInfo::STACK_INFO_FRAME_DATA and
|
|
|
|
|
// WindowsFrameInfo::STACK_INFO_FPO. Prefer them in this order.
|
|
|
|
|
// WindowsFrameInfo::STACK_INFO_FRAME_DATA is the newer type that
|
|
|
|
|
// includes its own program string.
|
|
|
|
|
// WindowsFrameInfo::STACK_INFO_FPO is the older type
|
2010-02-05 19:09:17 +01:00
|
|
|
// corresponding to the FPO_DATA struct. See stackwalker_x86.cc.
|
2010-01-11 23:31:50 +01:00
|
|
|
linked_ptr<WindowsFrameInfo> frame_info;
|
2010-04-09 01:06:23 +02:00
|
|
|
if ((windows_frame_info_[WindowsFrameInfo::STACK_INFO_FRAME_DATA]
|
2010-02-05 19:09:17 +01:00
|
|
|
.RetrieveRange(address, &frame_info))
|
2010-04-09 01:06:23 +02:00
|
|
|
|| (windows_frame_info_[WindowsFrameInfo::STACK_INFO_FPO]
|
2010-02-05 19:09:17 +01:00
|
|
|
.RetrieveRange(address, &frame_info))) {
|
2010-01-11 23:31:50 +01:00
|
|
|
result->CopyFrom(*frame_info.get());
|
|
|
|
|
return result.release();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Even without a relevant STACK line, many functions contain
|
|
|
|
|
// information about how much space their parameters consume on the
|
|
|
|
|
// stack. Use RetrieveNearestRange instead of RetrieveRange, so that
|
|
|
|
|
// we can use the function to bound the extent of the PUBLIC symbol,
|
|
|
|
|
// below. However, this does mean we need to check that ADDRESS
|
|
|
|
|
// falls within the retrieved function's range; do the range
|
|
|
|
|
// comparison in an overflow-friendly way.
|
|
|
|
|
linked_ptr<Function> function;
|
|
|
|
|
MemAddr function_base, function_size;
|
2016-06-06 07:41:10 +02:00
|
|
|
if (functions_.RetrieveNearestRange(address, &function, &function_base,
|
|
|
|
|
NULL /* delta */, &function_size) &&
|
2010-01-11 23:31:50 +01:00
|
|
|
address >= function_base && address - function_base < function_size) {
|
|
|
|
|
result->parameter_size = function->parameter_size;
|
|
|
|
|
result->valid |= WindowsFrameInfo::VALID_PARAMETER_SIZE;
|
|
|
|
|
return result.release();
|
2006-10-20 03:46:38 +02:00
|
|
|
}
|
2006-10-23 21:24:58 +02:00
|
|
|
|
2010-01-11 23:31:50 +01:00
|
|
|
// PUBLIC symbols might have a parameter size. Use the function we
|
|
|
|
|
// found above to limit the range the public symbol covers.
|
|
|
|
|
linked_ptr<PublicSymbol> public_symbol;
|
|
|
|
|
MemAddr public_address;
|
|
|
|
|
if (public_symbols_.Retrieve(address, &public_symbol, &public_address) &&
|
Breakpad processor: Fix function and public symbol lookup.
In r480, I botched the change to make the comparisons that decide
whether an address falls within a function's range safe from overflow.
The original code said:
address >= function_base && address < function_base + function_size
which is fine unless the function abuts the end of the address space,
in which case the addition overflows and you get a false negative.
My change subtracted function_size from both sides of the latter
comparison, which is meaning-preserving in true math, and gets you:
address >= function_base && address - function_size < function_base
This not only reads strangely, but also still overflows if
function_size is greater than address. That's rare, but I've added a
case to the unit tests that checks it.
My intent had been to replace the addition which could overflow with a
subtraction that was known not to overflow, namely:
address >= function_base && address - function_base < function_size
This is equivalent to the original in true math, and because of the
first comparison, we know the subtraction won't underflow in MemAddr
math.
The patch includes similar fixes to the public symbol lookup code, and
to FindWindowsFrameInfo, which was the only other function affected by
r480.
a=jimblandy, r=mmentovai
git-svn-id: http://google-breakpad.googlecode.com/svn/trunk@503 4c0a9323-5329-0410-9bdc-e9ce6186880e
2010-01-28 06:17:23 +01:00
|
|
|
(!function.get() || public_address > function_base)) {
|
2010-01-11 23:31:50 +01:00
|
|
|
result->parameter_size = public_symbol->parameter_size;
|
|
|
|
|
}
|
2010-10-07 22:31:36 +02:00
|
|
|
|
2010-01-11 23:31:50 +01:00
|
|
|
return NULL;
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
CFIFrameInfo* BasicSourceLineResolver::Module::FindCFIFrameInfo(
|
|
|
|
|
const StackFrame* frame) const {
|
2010-03-16 17:28:32 +01:00
|
|
|
MemAddr address = frame->instruction - frame->module->base_address();
|
|
|
|
|
MemAddr initial_base, initial_size;
|
|
|
|
|
string initial_rules;
|
|
|
|
|
|
|
|
|
|
// Find the initial rule whose range covers this address. That
|
|
|
|
|
// provides an initial set of register recovery rules. Then, walk
|
|
|
|
|
// forward from the initial rule's starting address to frame's
|
|
|
|
|
// instruction address, applying delta rules.
|
2016-06-06 07:41:10 +02:00
|
|
|
if (!cfi_initial_rules_.RetrieveRange(address, &initial_rules, &initial_base,
|
|
|
|
|
NULL /* delta */, &initial_size)) {
|
2010-03-16 17:28:32 +01:00
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Create a frame info structure, and populate it with the rules from
|
|
|
|
|
// the STACK CFI INIT record.
|
|
|
|
|
scoped_ptr<CFIFrameInfo> rules(new CFIFrameInfo());
|
|
|
|
|
if (!ParseCFIRuleSet(initial_rules, rules.get()))
|
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
|
|
// Find the first delta rule that falls within the initial rule's range.
|
|
|
|
|
map<MemAddr, string>::const_iterator delta =
|
|
|
|
|
cfi_delta_rules_.lower_bound(initial_base);
|
|
|
|
|
|
|
|
|
|
// Apply delta rules up to and including the frame's address.
|
|
|
|
|
while (delta != cfi_delta_rules_.end() && delta->first <= address) {
|
|
|
|
|
ParseCFIRuleSet(delta->second, rules.get());
|
|
|
|
|
delta++;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return rules.release();
|
|
|
|
|
}
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
bool BasicSourceLineResolver::Module::ParseFile(char* file_line) {
|
2013-09-25 20:25:13 +02:00
|
|
|
long index;
|
2020-06-24 00:55:43 +02:00
|
|
|
char* filename;
|
2013-09-25 20:25:13 +02:00
|
|
|
if (SymbolParseHelper::ParseFile(file_line, &index, &filename)) {
|
|
|
|
|
files_.insert(make_pair(index, string(filename)));
|
|
|
|
|
return true;
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
2013-09-25 20:25:13 +02:00
|
|
|
return false;
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
|
|
|
|
|
2021-07-22 20:13:37 +02:00
|
|
|
bool BasicSourceLineResolver::Module::ParseInlineOrigin(
|
|
|
|
|
char* inline_origin_line) {
|
|
|
|
|
long origin_id;
|
|
|
|
|
char* origin_name;
|
|
|
|
|
if (SymbolParseHelper::ParseInlineOrigin(inline_origin_line, &origin_id,
|
2021-10-20 23:07:55 +02:00
|
|
|
&origin_name)) {
|
|
|
|
|
inline_origins_.insert(make_pair(origin_id, new InlineOrigin(origin_name)));
|
2021-07-22 20:13:37 +02:00
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
linked_ptr<BasicSourceLineResolver::Inline>
|
|
|
|
|
BasicSourceLineResolver::Module::ParseInline(char* inline_line) {
|
|
|
|
|
long inline_nest_level;
|
|
|
|
|
long call_site_line;
|
2021-10-20 23:07:55 +02:00
|
|
|
long call_site_file_id;
|
2021-07-22 20:13:37 +02:00
|
|
|
long origin_id;
|
|
|
|
|
vector<std::pair<MemAddr, MemAddr>> ranges;
|
|
|
|
|
if (SymbolParseHelper::ParseInline(inline_line, &inline_nest_level,
|
2021-10-20 23:07:55 +02:00
|
|
|
&call_site_line, &call_site_file_id,
|
|
|
|
|
&origin_id, &ranges)) {
|
|
|
|
|
return linked_ptr<Inline>(new Inline(inline_nest_level, call_site_line,
|
|
|
|
|
call_site_file_id, origin_id, ranges));
|
2021-07-22 20:13:37 +02:00
|
|
|
}
|
|
|
|
|
return linked_ptr<Inline>();
|
|
|
|
|
}
|
|
|
|
|
|
2006-12-12 00:22:54 +01:00
|
|
|
BasicSourceLineResolver::Function*
|
2020-06-24 00:55:43 +02:00
|
|
|
BasicSourceLineResolver::Module::ParseFunction(char* function_line) {
|
2017-11-29 22:29:37 +01:00
|
|
|
bool is_multiple;
|
2013-09-25 20:25:13 +02:00
|
|
|
uint64_t address;
|
|
|
|
|
uint64_t size;
|
|
|
|
|
long stack_param_size;
|
2020-06-24 00:55:43 +02:00
|
|
|
char* name;
|
2017-11-29 22:29:37 +01:00
|
|
|
if (SymbolParseHelper::ParseFunction(function_line, &is_multiple, &address,
|
|
|
|
|
&size, &stack_param_size, &name)) {
|
|
|
|
|
return new Function(name, address, size, stack_param_size, is_multiple);
|
2006-09-08 20:03:56 +02:00
|
|
|
}
|
2013-09-25 20:25:13 +02:00
|
|
|
return NULL;
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
|
|
|
|
|
2006-12-12 00:22:54 +01:00
|
|
|
BasicSourceLineResolver::Line* BasicSourceLineResolver::Module::ParseLine(
|
2020-06-24 00:55:43 +02:00
|
|
|
char* line_line) {
|
2013-09-25 20:25:13 +02:00
|
|
|
uint64_t address;
|
|
|
|
|
uint64_t size;
|
|
|
|
|
long line_number;
|
|
|
|
|
long source_file;
|
|
|
|
|
|
|
|
|
|
if (SymbolParseHelper::ParseLine(line_line, &address, &size, &line_number,
|
|
|
|
|
&source_file)) {
|
|
|
|
|
return new Line(address, size, source_file, line_number);
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
2013-09-25 20:25:13 +02:00
|
|
|
return NULL;
|
2006-09-28 23:09:37 +02:00
|
|
|
}
|
2006-08-25 23:14:45 +02:00
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
bool BasicSourceLineResolver::Module::ParsePublicSymbol(char* public_line) {
|
2017-11-29 22:29:37 +01:00
|
|
|
bool is_multiple;
|
2013-09-25 20:25:13 +02:00
|
|
|
uint64_t address;
|
|
|
|
|
long stack_param_size;
|
2020-06-24 00:55:43 +02:00
|
|
|
char* name;
|
2013-09-25 20:25:13 +02:00
|
|
|
|
2017-11-29 22:29:37 +01:00
|
|
|
if (SymbolParseHelper::ParsePublicSymbol(public_line, &is_multiple, &address,
|
2013-09-25 20:25:13 +02:00
|
|
|
&stack_param_size, &name)) {
|
|
|
|
|
// A few public symbols show up with an address of 0. This has been seen
|
|
|
|
|
// in the dumped output of ntdll.pdb for symbols such as _CIlog, _CIpow,
|
|
|
|
|
// RtlDescribeChunkLZNT1, and RtlReserveChunkLZNT1. They would conflict
|
|
|
|
|
// with one another if they were allowed into the public_symbols_ map,
|
|
|
|
|
// but since the address is obviously invalid, gracefully accept them
|
|
|
|
|
// as input without putting them into the map.
|
|
|
|
|
if (address == 0) {
|
|
|
|
|
return true;
|
|
|
|
|
}
|
2006-10-20 21:50:01 +02:00
|
|
|
|
2013-09-25 20:25:13 +02:00
|
|
|
linked_ptr<PublicSymbol> symbol(new PublicSymbol(name, address,
|
2017-11-29 22:29:37 +01:00
|
|
|
stack_param_size,
|
|
|
|
|
is_multiple));
|
2013-09-25 20:25:13 +02:00
|
|
|
return public_symbols_.Store(address, symbol);
|
2006-10-20 21:50:01 +02:00
|
|
|
}
|
2013-09-25 20:25:13 +02:00
|
|
|
return false;
|
2006-10-20 21:50:01 +02:00
|
|
|
}
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
bool BasicSourceLineResolver::Module::ParseStackInfo(char* stack_info_line) {
|
2006-09-28 23:09:37 +02:00
|
|
|
// Skip "STACK " prefix.
|
|
|
|
|
stack_info_line += 6;
|
2006-08-25 23:14:45 +02:00
|
|
|
|
2010-02-05 19:10:51 +01:00
|
|
|
// Find the token indicating what sort of stack frame walking
|
|
|
|
|
// information this is.
|
|
|
|
|
while (*stack_info_line == ' ')
|
|
|
|
|
stack_info_line++;
|
2020-06-24 00:55:43 +02:00
|
|
|
const char* platform = stack_info_line;
|
2010-04-09 01:06:23 +02:00
|
|
|
while (!strchr(kWhitespace, *stack_info_line))
|
2010-02-05 19:10:51 +01:00
|
|
|
stack_info_line++;
|
|
|
|
|
*stack_info_line++ = '\0';
|
|
|
|
|
|
|
|
|
|
// MSVC stack frame info.
|
2010-04-09 01:06:23 +02:00
|
|
|
if (strcmp(platform, "WIN") == 0) {
|
2010-05-14 00:21:00 +02:00
|
|
|
int type = 0;
|
2013-03-06 15:04:42 +01:00
|
|
|
uint64_t rva, code_size;
|
2010-04-09 01:06:23 +02:00
|
|
|
linked_ptr<WindowsFrameInfo>
|
|
|
|
|
stack_frame_info(WindowsFrameInfo::ParseFromString(stack_info_line,
|
|
|
|
|
type,
|
|
|
|
|
rva,
|
|
|
|
|
code_size));
|
|
|
|
|
if (stack_frame_info == NULL)
|
|
|
|
|
return false;
|
|
|
|
|
|
|
|
|
|
// TODO(mmentovai): I wanted to use StoreRange's return value as this
|
|
|
|
|
// method's return value, but MSVC infrequently outputs stack info that
|
|
|
|
|
// violates the containment rules. This happens with a section of code
|
|
|
|
|
// in strncpy_s in test_app.cc (testdata/minidump2). There, problem looks
|
|
|
|
|
// like this:
|
|
|
|
|
// STACK WIN 4 4242 1a a 0 ... (STACK WIN 4 base size prolog 0 ...)
|
|
|
|
|
// STACK WIN 4 4243 2e 9 0 ...
|
|
|
|
|
// ContainedRangeMap treats these two blocks as conflicting. In reality,
|
|
|
|
|
// when the prolog lengths are taken into account, the actual code of
|
|
|
|
|
// these blocks doesn't conflict. However, we can't take the prolog lengths
|
|
|
|
|
// into account directly here because we'd wind up with a different set
|
|
|
|
|
// of range conflicts when MSVC outputs stack info like this:
|
|
|
|
|
// STACK WIN 4 1040 73 33 0 ...
|
|
|
|
|
// STACK WIN 4 105a 59 19 0 ...
|
|
|
|
|
// because in both of these entries, the beginning of the code after the
|
|
|
|
|
// prolog is at 0x1073, and the last byte of contained code is at 0x10b2.
|
|
|
|
|
// Perhaps we could get away with storing ranges by rva + prolog_size
|
|
|
|
|
// if ContainedRangeMap were modified to allow replacement of
|
|
|
|
|
// already-stored values.
|
|
|
|
|
|
|
|
|
|
windows_frame_info_[type].StoreRange(rva, code_size, stack_frame_info);
|
|
|
|
|
return true;
|
|
|
|
|
} else if (strcmp(platform, "CFI") == 0) {
|
|
|
|
|
// DWARF CFI stack frame info
|
2010-03-16 17:28:32 +01:00
|
|
|
return ParseCFIFrameInfo(stack_info_line);
|
2006-10-20 03:46:38 +02:00
|
|
|
} else {
|
2010-04-09 01:06:23 +02:00
|
|
|
// Something unrecognized.
|
|
|
|
|
return false;
|
|
|
|
|
}
|
2006-08-25 23:14:45 +02:00
|
|
|
}
|
|
|
|
|
|
2010-03-16 17:28:32 +01:00
|
|
|
bool BasicSourceLineResolver::Module::ParseCFIFrameInfo(
|
2020-06-24 00:55:43 +02:00
|
|
|
char* stack_info_line) {
|
|
|
|
|
char* cursor;
|
2010-03-16 17:28:32 +01:00
|
|
|
|
|
|
|
|
// Is this an INIT record or a delta record?
|
2020-06-24 00:55:43 +02:00
|
|
|
char* init_or_address = strtok_r(stack_info_line, " \r\n", &cursor);
|
2010-03-16 17:28:32 +01:00
|
|
|
if (!init_or_address)
|
|
|
|
|
return false;
|
|
|
|
|
|
|
|
|
|
if (strcmp(init_or_address, "INIT") == 0) {
|
|
|
|
|
// This record has the form "STACK INIT <address> <size> <rules...>".
|
2020-06-24 00:55:43 +02:00
|
|
|
char* address_field = strtok_r(NULL, " \r\n", &cursor);
|
2010-03-16 17:28:32 +01:00
|
|
|
if (!address_field) return false;
|
2010-10-07 22:31:36 +02:00
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
char* size_field = strtok_r(NULL, " \r\n", &cursor);
|
2010-03-16 17:28:32 +01:00
|
|
|
if (!size_field) return false;
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
char* initial_rules = strtok_r(NULL, "\r\n", &cursor);
|
2010-03-16 17:28:32 +01:00
|
|
|
if (!initial_rules) return false;
|
|
|
|
|
|
|
|
|
|
MemAddr address = strtoul(address_field, NULL, 16);
|
|
|
|
|
MemAddr size = strtoul(size_field, NULL, 16);
|
|
|
|
|
cfi_initial_rules_.StoreRange(address, size, initial_rules);
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// This record has the form "STACK <address> <rules...>".
|
2020-06-24 00:55:43 +02:00
|
|
|
char* address_field = init_or_address;
|
|
|
|
|
char* delta_rules = strtok_r(NULL, "\r\n", &cursor);
|
2010-03-16 17:28:32 +01:00
|
|
|
if (!delta_rules) return false;
|
|
|
|
|
MemAddr address = strtoul(address_field, NULL, 16);
|
|
|
|
|
cfi_delta_rules_[address] = delta_rules;
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
2021-07-22 20:13:37 +02:00
|
|
|
bool BasicSourceLineResolver::Function::AppendInline(linked_ptr<Inline> in) {
|
|
|
|
|
// This happends if in's parent wasn't added due to a malformed INLINE record.
|
|
|
|
|
if (in->inline_nest_level > last_added_inline_nest_level + 1)
|
|
|
|
|
return false;
|
|
|
|
|
RangeMap<MemAddr, linked_ptr<Inline>>* current_inlines = &this->inlines;
|
|
|
|
|
auto iter = recent_inlines.find(in->inline_nest_level - 1);
|
|
|
|
|
if (iter != recent_inlines.end())
|
|
|
|
|
current_inlines = &iter->second->child_inlines;
|
|
|
|
|
else
|
|
|
|
|
assert(in->inline_nest_level == 0);
|
|
|
|
|
|
|
|
|
|
last_added_inline_nest_level = in->inline_nest_level;
|
|
|
|
|
recent_inlines[last_added_inline_nest_level] = in;
|
|
|
|
|
|
|
|
|
|
// Store all ranges into current level of inlines.
|
|
|
|
|
for (auto range : in->inline_ranges)
|
|
|
|
|
current_inlines->StoreRange(range.first, range.second, in);
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
2013-09-25 20:25:13 +02:00
|
|
|
// static
|
2020-06-24 00:55:43 +02:00
|
|
|
bool SymbolParseHelper::ParseFile(char* file_line, long* index,
|
|
|
|
|
char** filename) {
|
2013-09-25 20:25:13 +02:00
|
|
|
// FILE <id> <filename>
|
|
|
|
|
assert(strncmp(file_line, "FILE ", 5) == 0);
|
|
|
|
|
file_line += 5; // skip prefix
|
|
|
|
|
|
|
|
|
|
vector<char*> tokens;
|
|
|
|
|
if (!Tokenize(file_line, kWhitespace, 2, &tokens)) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
char* after_number;
|
2013-09-25 20:25:13 +02:00
|
|
|
*index = strtol(tokens[0], &after_number, 10);
|
|
|
|
|
if (!IsValidAfterNumber(after_number) || *index < 0 ||
|
|
|
|
|
*index == std::numeric_limits<long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
*filename = tokens[1];
|
2016-06-10 15:51:14 +02:00
|
|
|
if (!*filename) {
|
2013-09-25 20:25:13 +02:00
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
2021-07-22 20:13:37 +02:00
|
|
|
// static
|
|
|
|
|
bool SymbolParseHelper::ParseInlineOrigin(char* inline_origin_line,
|
|
|
|
|
long* origin_id,
|
|
|
|
|
char** name) {
|
2021-10-20 23:07:55 +02:00
|
|
|
// INLINE_ORIGIN <origin_id> <name>
|
2021-07-22 20:13:37 +02:00
|
|
|
assert(strncmp(inline_origin_line, "INLINE_ORIGIN ", 14) == 0);
|
|
|
|
|
inline_origin_line += 14; // skip prefix
|
|
|
|
|
vector<char*> tokens;
|
2021-10-20 23:07:55 +02:00
|
|
|
if (!Tokenize(inline_origin_line, kWhitespace, 2, &tokens)) {
|
2021-07-22 20:13:37 +02:00
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
char* after_number;
|
|
|
|
|
*origin_id = strtol(tokens[0], &after_number, 10);
|
|
|
|
|
if (!IsValidAfterNumber(after_number) || *origin_id < 0 ||
|
|
|
|
|
*origin_id == std::numeric_limits<long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
2021-10-20 23:07:55 +02:00
|
|
|
*name = tokens[1];
|
2021-07-22 20:13:37 +02:00
|
|
|
if (!*name) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// static
|
|
|
|
|
bool SymbolParseHelper::ParseInline(
|
|
|
|
|
char* inline_line,
|
|
|
|
|
long* inline_nest_level,
|
|
|
|
|
long* call_site_line,
|
2021-10-20 23:07:55 +02:00
|
|
|
long* call_site_file_id,
|
2021-07-22 20:13:37 +02:00
|
|
|
long* origin_id,
|
|
|
|
|
vector<std::pair<MemAddr, MemAddr>>* ranges) {
|
2021-10-20 23:07:55 +02:00
|
|
|
// INLINE <inline_nest_level> <call_site_line> <call_site_file_id> <origin_id>
|
|
|
|
|
// [<address> <size>]+
|
2021-07-22 20:13:37 +02:00
|
|
|
assert(strncmp(inline_line, "INLINE ", 7) == 0);
|
|
|
|
|
inline_line += 7; // skip prefix
|
|
|
|
|
|
|
|
|
|
vector<char*> tokens;
|
|
|
|
|
Tokenize(inline_line, kWhitespace, std::numeric_limits<int>::max(), &tokens);
|
|
|
|
|
|
2021-10-20 23:07:55 +02:00
|
|
|
// The length of the vector should be at least 6 and an even number.
|
|
|
|
|
if (tokens.size() < 6 || tokens.size() % 2 != 0)
|
2021-07-22 20:13:37 +02:00
|
|
|
return false;
|
|
|
|
|
|
|
|
|
|
char* after_number;
|
|
|
|
|
*inline_nest_level = strtol(tokens[0], &after_number, 10);
|
|
|
|
|
if (!IsValidAfterNumber(after_number) || *inline_nest_level < 0 ||
|
|
|
|
|
*inline_nest_level == std::numeric_limits<long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
*call_site_line = strtol(tokens[1], &after_number, 10);
|
|
|
|
|
if (!IsValidAfterNumber(after_number) || *call_site_line < 0 ||
|
|
|
|
|
*call_site_line == std::numeric_limits<long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
2021-10-20 23:07:55 +02:00
|
|
|
*call_site_file_id = strtol(tokens[2], &after_number, 10);
|
|
|
|
|
// If the file id is -1, it might be an artificial function that doesn't have
|
|
|
|
|
// file id. So, we consider -1 as a valid special case.
|
|
|
|
|
if (!IsValidAfterNumber(after_number) || *call_site_file_id < -1 ||
|
|
|
|
|
*call_site_file_id == std::numeric_limits<long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
*origin_id = strtol(tokens[3], &after_number, 10);
|
2021-07-22 20:13:37 +02:00
|
|
|
if (!IsValidAfterNumber(after_number) || *origin_id < 0 ||
|
|
|
|
|
*origin_id == std::numeric_limits<long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
2021-10-20 23:07:55 +02:00
|
|
|
for (size_t i = 4; i < tokens.size();) {
|
2021-07-22 20:13:37 +02:00
|
|
|
MemAddr address = strtoull(tokens[i++], &after_number, 16);
|
|
|
|
|
if (!IsValidAfterNumber(after_number) ||
|
|
|
|
|
address == std::numeric_limits<unsigned long long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
MemAddr size = strtoull(tokens[i++], &after_number, 16);
|
|
|
|
|
if (!IsValidAfterNumber(after_number) ||
|
|
|
|
|
size == std::numeric_limits<unsigned long long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
ranges->push_back({address, size});
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
2013-09-25 20:25:13 +02:00
|
|
|
// static
|
2020-06-24 00:55:43 +02:00
|
|
|
bool SymbolParseHelper::ParseFunction(char* function_line, bool* is_multiple,
|
|
|
|
|
uint64_t* address, uint64_t* size,
|
|
|
|
|
long* stack_param_size, char** name) {
|
2017-11-29 22:29:37 +01:00
|
|
|
// FUNC [<multiple>] <address> <size> <stack_param_size> <name>
|
2013-09-25 20:25:13 +02:00
|
|
|
assert(strncmp(function_line, "FUNC ", 5) == 0);
|
|
|
|
|
function_line += 5; // skip prefix
|
|
|
|
|
|
|
|
|
|
vector<char*> tokens;
|
2017-11-29 22:29:37 +01:00
|
|
|
if (!TokenizeWithOptionalField(function_line, "m", kWhitespace, 5, &tokens)) {
|
2013-09-25 20:25:13 +02:00
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
2017-11-29 22:29:37 +01:00
|
|
|
*is_multiple = strcmp(tokens[0], "m") == 0;
|
|
|
|
|
int next_token = *is_multiple ? 1 : 0;
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
char* after_number;
|
2017-11-29 22:29:37 +01:00
|
|
|
*address = strtoull(tokens[next_token++], &after_number, 16);
|
2013-09-25 20:25:13 +02:00
|
|
|
if (!IsValidAfterNumber(after_number) ||
|
|
|
|
|
*address == std::numeric_limits<unsigned long long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
2017-11-29 22:29:37 +01:00
|
|
|
*size = strtoull(tokens[next_token++], &after_number, 16);
|
2013-09-25 20:25:13 +02:00
|
|
|
if (!IsValidAfterNumber(after_number) ||
|
|
|
|
|
*size == std::numeric_limits<unsigned long long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
2017-11-29 22:29:37 +01:00
|
|
|
*stack_param_size = strtol(tokens[next_token++], &after_number, 16);
|
2013-09-25 20:25:13 +02:00
|
|
|
if (!IsValidAfterNumber(after_number) ||
|
|
|
|
|
*stack_param_size == std::numeric_limits<long>::max() ||
|
|
|
|
|
*stack_param_size < 0) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
2017-11-29 22:29:37 +01:00
|
|
|
*name = tokens[next_token++];
|
2013-09-25 20:25:13 +02:00
|
|
|
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// static
|
2020-06-24 00:55:43 +02:00
|
|
|
bool SymbolParseHelper::ParseLine(char* line_line, uint64_t* address,
|
|
|
|
|
uint64_t* size, long* line_number,
|
|
|
|
|
long* source_file) {
|
2013-09-25 20:25:13 +02:00
|
|
|
// <address> <size> <line number> <source file id>
|
|
|
|
|
vector<char*> tokens;
|
|
|
|
|
if (!Tokenize(line_line, kWhitespace, 4, &tokens)) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
char* after_number;
|
2013-09-25 20:25:13 +02:00
|
|
|
*address = strtoull(tokens[0], &after_number, 16);
|
|
|
|
|
if (!IsValidAfterNumber(after_number) ||
|
|
|
|
|
*address == std::numeric_limits<unsigned long long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
*size = strtoull(tokens[1], &after_number, 16);
|
|
|
|
|
if (!IsValidAfterNumber(after_number) ||
|
|
|
|
|
*size == std::numeric_limits<unsigned long long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
*line_number = strtol(tokens[2], &after_number, 10);
|
|
|
|
|
if (!IsValidAfterNumber(after_number) ||
|
|
|
|
|
*line_number == std::numeric_limits<long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
*source_file = strtol(tokens[3], &after_number, 10);
|
|
|
|
|
if (!IsValidAfterNumber(after_number) || *source_file < 0 ||
|
|
|
|
|
*source_file == std::numeric_limits<long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Valid line numbers normally start from 1, however there are functions that
|
|
|
|
|
// are associated with a source file but not associated with any line number
|
|
|
|
|
// (block helper function) and for such functions the symbol file contains 0
|
|
|
|
|
// for the line numbers. Hence, 0 should be treated as a valid line number.
|
|
|
|
|
// For more information on block helper functions, please, take a look at:
|
|
|
|
|
// http://clang.llvm.org/docs/Block-ABI-Apple.html
|
|
|
|
|
if (*line_number < 0) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// static
|
2020-06-24 00:55:43 +02:00
|
|
|
bool SymbolParseHelper::ParsePublicSymbol(char* public_line, bool* is_multiple,
|
|
|
|
|
uint64_t* address,
|
|
|
|
|
long* stack_param_size,
|
|
|
|
|
char** name) {
|
2017-11-29 22:29:37 +01:00
|
|
|
// PUBLIC [<multiple>] <address> <stack_param_size> <name>
|
2013-09-25 20:25:13 +02:00
|
|
|
assert(strncmp(public_line, "PUBLIC ", 7) == 0);
|
|
|
|
|
public_line += 7; // skip prefix
|
|
|
|
|
|
|
|
|
|
vector<char*> tokens;
|
2017-11-29 22:29:37 +01:00
|
|
|
if (!TokenizeWithOptionalField(public_line, "m", kWhitespace, 4, &tokens)) {
|
2013-09-25 20:25:13 +02:00
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
2017-11-29 22:29:37 +01:00
|
|
|
*is_multiple = strcmp(tokens[0], "m") == 0;
|
|
|
|
|
int next_token = *is_multiple ? 1 : 0;
|
|
|
|
|
|
2020-06-24 00:55:43 +02:00
|
|
|
char* after_number;
|
2017-11-29 22:29:37 +01:00
|
|
|
*address = strtoull(tokens[next_token++], &after_number, 16);
|
2013-09-25 20:25:13 +02:00
|
|
|
if (!IsValidAfterNumber(after_number) ||
|
|
|
|
|
*address == std::numeric_limits<unsigned long long>::max()) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
2017-11-29 22:29:37 +01:00
|
|
|
*stack_param_size = strtol(tokens[next_token++], &after_number, 16);
|
2013-09-25 20:25:13 +02:00
|
|
|
if (!IsValidAfterNumber(after_number) ||
|
|
|
|
|
*stack_param_size == std::numeric_limits<long>::max() ||
|
|
|
|
|
*stack_param_size < 0) {
|
|
|
|
|
return false;
|
|
|
|
|
}
|
2017-11-29 22:29:37 +01:00
|
|
|
*name = tokens[next_token++];
|
2013-09-25 20:25:13 +02:00
|
|
|
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// static
|
2020-06-24 00:55:43 +02:00
|
|
|
bool SymbolParseHelper::IsValidAfterNumber(char* after_number) {
|
2013-09-25 20:25:13 +02:00
|
|
|
if (after_number != NULL && strchr(kWhitespace, *after_number) != NULL) {
|
|
|
|
|
return true;
|
|
|
|
|
}
|
|
|
|
|
return false;
|
|
|
|
|
}
|
|
|
|
|
|
2007-02-14 20:51:05 +01:00
|
|
|
} // namespace google_breakpad
|