diff --git a/CMakeLists.txt b/CMakeLists.txt
index a3e2143078f..bc2c9999dc5 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -224,10 +224,16 @@ if (TARGET_ARCH MATCHES "^arm64" OR TARGET_ARCH MATCHES "^aarch64")
set(AARCH64 1)
set(X64 1)
message(STATUS "Building for AArch64")
+ if (ANDROID)
+ set(ANDROID64 1)
+ endif ()
elseif (TARGET_ARCH MATCHES "^arm")
set(ARM 1) # This means AArch32.
set(X64 OFF)
message(STATUS "Building for ARM")
+ if (ANDROID)
+ set(ANDROID32 1)
+ endif ()
elseif (TARGET_ARCH MATCHES "^riscv64")
set(RISCV64 1)
set(X64 1)
@@ -692,6 +698,7 @@ if (UNIX)
endif (NOT CMAKE_COMPILER_IS_GNUCC)
check_if_linker_is_gnu_gold(LINKER_IS_GNU_GOLD)
+ check_if_linker_is_llvm_lld(LINKER_IS_LLVM_LLD)
# FIXME i#2949: static 32-bit release-build linking with gcc 7.3.1 fails when
# static C++ clients like drmemtrace or drmemtrace_raw2trace are linked in.
@@ -821,6 +828,7 @@ if (UNIX)
# (see i#1034 where this caused problems in the past).
set(BASE_CFLAGS "${BASE_CFLAGS} -funsigned-char")
# Ubuntu defaults to -fstack-protector these days, which depends on app TLS.
+ CHECK_C_COMPILER_FLAG("-no-pie" no_pie_avail)
CHECK_C_COMPILER_FLAG("-fno-stack-protector" no_stack_protector_avail)
if (no_stack_protector_avail)
set(BASE_CFLAGS "${BASE_CFLAGS} -fno-stack-protector")
@@ -875,14 +883,14 @@ if (UNIX)
endif ()
# there's no cmake warning control so we hardcode it
set(WARN "-Wall -Werror -Wwrite-strings -Wvla")
+ CHECK_C_COMPILER_FLAG("-Wno-unused-but-set-variable" nounused_avail)
+ if (nounused_avail)
+ set(WARN "${WARN} -Wno-unused-but-set-variable")
+ endif (nounused_avail)
if (NOT CMAKE_COMPILER_IS_CLANG)
# Old gcc's ignore unknown -W flags, but -Wall -Werror causes clang to
# complain that it doesn't recognize it.
# Actually this is not true: gcc 4.1.2 aborts on unknown -W so we check
- CHECK_C_COMPILER_FLAG("-Wno-unused-but-set-variable" nounused_avail)
- if (nounused_avail)
- set(WARN "${WARN} -Wno-unused-but-set-variable")
- endif (nounused_avail)
# XXX i#3792: DynamoRIO manages '\0' termination and error states itself in
# too many places. In order to activate this warning, this code needs to get
# re-factored for no good reason.
@@ -908,6 +916,21 @@ if (UNIX)
set(WARN "${WARN} -Wno-dangling-pointer")
endif ()
else (NOT CMAKE_COMPILER_IS_CLANG)
+ # Clang emits this warning when a function is used without a defined
+ # prototype, however clang can't find function prototpes in #define macros.
+ # If function prototypes are moved out of #define macros then this warning
+ # can be re-enabled.
+ CHECK_C_COMPILER_FLAG("-Wno-deprecated-non-prototype" nodeprecated_prototype_avail)
+ if (nodeprecated_prototype_avail)
+ set(WARN "${WARN} -Wno-deprecated-non-prototype")
+ endif (nodeprecated_prototype_avail)
+ # Clang emits this warning due to the assembly shared between arm and
+ # aarch64 in os.c not specifying register widths. If the shared code is
+ # split into arm and aarch64 versions, then this warning can be re-enabled.
+ CHECK_C_COMPILER_FLAG("-Wno-asm-operand-widths" noasm_operand_widths)
+ if (noasm_operand_widths)
+ set(WARN "${WARN} -Wno-asm-operand-widths")
+ endif (noasm_operand_widths)
# clang turns off color when it's writing to a pipe, but the user may still
# wish to force color if it eventually goes to a terminal.
option(CLANG_COLOR_DIAGNOSTICS "force colored clang diagnostics" OFF)
diff --git a/api/docs/release.dox b/api/docs/release.dox
index 5359fe8d8d3..5b8faf839ed 100644
--- a/api/docs/release.dox
+++ b/api/docs/release.dox
@@ -134,6 +134,10 @@ changes:
is the encoding of the removed instruction up to a pointer's length. Added
#OFFLINE_FILE_VERSION_RETIRED_INSTRUCTIONS_ONLY to increase the trace version
for drmemtraces with uncompleted instructions removed.
+ - Moved module file read logic into read_module_file() in raw2trace_shared, and
+ removed raw2trace_directory_t::initialize_module_file() since the
+ read_module_file() can be directly used without having to pull in the whole
+ raw2trace_directory_t.
Further non-compatibility-affecting changes include:
- Added X64 Linux support to dr_create_memory_dump(). This API has the same
@@ -143,6 +147,8 @@ Further non-compatibility-affecting changes include:
- Added the AArch64 FPMR register as DR_REG_FPMR.
- Added OPSZ_addr which chooses the appropriate operand size on X86 from 2/4/8 based
on the 32/64 bit mode and the presence or absence of the address size prefix.
+ - Fixed the order of operands for the vpexpandd/vpexpandq opcodes (previously the
+ source and destination operands were reversed).
**************************************************
diff --git a/clients/drcachesim/CMakeLists.txt b/clients/drcachesim/CMakeLists.txt
index a67e5b64504..942e49847b8 100644
--- a/clients/drcachesim/CMakeLists.txt
+++ b/clients/drcachesim/CMakeLists.txt
@@ -161,9 +161,11 @@ add_exported_library(drmemtrace_reuse_distance STATIC tools/reuse_distance.cpp)
add_exported_library(drmemtrace_histogram STATIC tools/histogram.cpp)
add_exported_library(drmemtrace_reuse_time STATIC tools/reuse_time.cpp)
add_exported_library(drmemtrace_basic_counts STATIC tools/basic_counts.cpp)
-add_exported_library(drmemtrace_opcode_mix STATIC tools/opcode_mix.cpp)
+add_exported_library(drmemtrace_opcode_mix STATIC
+ tools/opcode_mix.cpp tracer/raw2trace_shared.cpp)
add_exported_library(drmemtrace_syscall_mix STATIC tools/syscall_mix.cpp)
-add_exported_library(drmemtrace_view STATIC tools/view.cpp)
+add_exported_library(drmemtrace_view STATIC
+ tools/view.cpp tracer/raw2trace_shared.cpp)
add_exported_library(drmemtrace_func_view STATIC tools/func_view.cpp)
add_exported_library(drmemtrace_invariant_checker STATIC tools/invariant_checker.cpp)
add_exported_library(drmemtrace_schedule_stats STATIC tools/schedule_stats.cpp)
@@ -1202,7 +1204,7 @@ if (BUILD_TESTS)
add_executable(tool.drcacheoff.skip_unit_tests tests/skip_unit_tests.cpp)
configure_DynamoRIO_standalone(tool.drcacheoff.skip_unit_tests)
target_link_libraries(tool.drcacheoff.skip_unit_tests drmemtrace_analyzer
- drmemtrace_view drmemtrace_raw2trace test_helpers)
+ drmemtrace_view test_helpers ${zlib_libs})
add_win32_flags(tool.drcacheoff.skip_unit_tests)
use_DynamoRIO_extension(tool.drcacheoff.skip_unit_tests drreg_static)
use_DynamoRIO_extension(tool.drcacheoff.skip_unit_tests drcovlib_static)
diff --git a/clients/drcachesim/tools/opcode_mix.cpp b/clients/drcachesim/tools/opcode_mix.cpp
index 07c712e09d0..6ff25819d31 100644
--- a/clients/drcachesim/tools/opcode_mix.cpp
+++ b/clients/drcachesim/tools/opcode_mix.cpp
@@ -1,5 +1,5 @@
/* **********************************************************
- * Copyright (c) 2017-2023 Google, Inc. All rights reserved.
+ * Copyright (c) 2017-2024 Google, Inc. All rights reserved.
* **********************************************************/
/*
@@ -56,7 +56,7 @@
#include "dr_api.h"
#include "memref.h"
#include "raw2trace.h"
-#include "raw2trace_directory.h"
+#include "raw2trace_shared.h"
#include "reader.h"
#include "trace_entry.h"
#include "utils.h"
@@ -92,13 +92,16 @@ opcode_mix_t::initialize()
return "";
// Legacy trace support where binaries are needed.
// We do not support non-module code for such traces.
- std::string error = directory_.initialize_module_file(module_file_path_);
- if (!error.empty())
- return "Failed to initialize directory: " + error;
+ file_t modfile;
+ std::string error = read_module_file(module_file_path_, modfile, modfile_bytes_);
+ if (!error.empty()) {
+ return "Failed to read module file: " + error;
+ }
module_mapper_ =
- module_mapper_t::create(directory_.modfile_bytes_, nullptr, nullptr, nullptr,
- nullptr, knob_verbose_, knob_alt_module_dir_);
+ module_mapper_t::create(modfile_bytes_, nullptr, nullptr, nullptr, nullptr,
+ knob_verbose_, knob_alt_module_dir_);
module_mapper_->get_loaded_modules();
+ dr_close_file(modfile);
error = module_mapper_->get_last_error();
if (!error.empty())
return "Failed to load binaries: " + error;
@@ -110,6 +113,9 @@ opcode_mix_t::~opcode_mix_t()
for (auto &iter : shard_map_) {
delete iter.second;
}
+ if (modfile_bytes_ != nullptr) {
+ delete[] modfile_bytes_;
+ }
}
bool
diff --git a/clients/drcachesim/tools/opcode_mix.h b/clients/drcachesim/tools/opcode_mix.h
index d67fd985855..e07e7997b81 100644
--- a/clients/drcachesim/tools/opcode_mix.h
+++ b/clients/drcachesim/tools/opcode_mix.h
@@ -46,7 +46,6 @@
#include "analysis_tool.h"
#include "memref.h"
#include "raw2trace.h"
-#include "raw2trace_directory.h"
#include "trace_entry.h"
namespace dynamorio {
@@ -185,9 +184,9 @@ class opcode_mix_t : public analysis_tool_t {
std::string module_file_path_;
std::unique_ptr module_mapper_;
std::mutex mapper_mutex_;
- // We reference directory.modfile_bytes throughout operation, so its lifetime
- // must match ours.
- raw2trace_directory_t directory_;
+ // XXX: Perhaps module_mapper_t should be made to own the cleanup of
+ // modfile_bytes_.
+ char *modfile_bytes_ = nullptr;
std::unordered_map shard_map_;
// This mutex is only needed in parallel_shard_init. In all other accesses to
diff --git a/clients/drcachesim/tools/view.cpp b/clients/drcachesim/tools/view.cpp
index 4728c5629d9..264b2797f78 100644
--- a/clients/drcachesim/tools/view.cpp
+++ b/clients/drcachesim/tools/view.cpp
@@ -53,7 +53,7 @@
#include "memref.h"
#include "memtrace_stream.h"
#include "raw2trace.h"
-#include "raw2trace_directory.h"
+#include "raw2trace_shared.h"
#include "trace_entry.h"
#include "utils.h"
@@ -124,9 +124,12 @@ view_t::initialize_stream(memtrace_stream_t *serial_stream)
if (module_file_path_.empty()) {
has_modules_ = false;
} else {
- std::string error = directory_.initialize_module_file(module_file_path_);
+ file_t modfile;
+ std::string error = read_module_file(module_file_path_, modfile, modfile_bytes_);
if (!error.empty())
has_modules_ = false;
+ else
+ dr_close_file(modfile);
}
if (!has_modules_) {
// Continue but omit disassembly to support cases where binaries are
@@ -136,8 +139,8 @@ view_t::initialize_stream(memtrace_stream_t *serial_stream)
// Legacy trace support where binaries are needed.
// We do not support non-module code for such traces.
module_mapper_ =
- module_mapper_t::create(directory_.modfile_bytes_, nullptr, nullptr, nullptr,
- nullptr, knob_verbose_, knob_alt_module_dir_);
+ module_mapper_t::create(modfile_bytes_, nullptr, nullptr, nullptr, nullptr,
+ knob_verbose_, knob_alt_module_dir_);
module_mapper_->get_loaded_modules();
std::string error = module_mapper_->get_last_error();
if (!error.empty())
@@ -662,5 +665,12 @@ view_t::print_results()
return true;
}
+view_t::~view_t()
+{
+ if (modfile_bytes_ != nullptr) {
+ delete[] modfile_bytes_;
+ }
+}
+
} // namespace drmemtrace
} // namespace dynamorio
diff --git a/clients/drcachesim/tools/view.h b/clients/drcachesim/tools/view.h
index 30d5f033e14..4ba6d4102ed 100644
--- a/clients/drcachesim/tools/view.h
+++ b/clients/drcachesim/tools/view.h
@@ -1,5 +1,5 @@
/* **********************************************************
- * Copyright (c) 2018-2023 Google, Inc. All rights reserved.
+ * Copyright (c) 2018-2024 Google, Inc. All rights reserved.
* **********************************************************/
/*
@@ -47,7 +47,6 @@
#include "memref.h"
#include "memtrace_stream.h"
#include "raw2trace.h"
-#include "raw2trace_directory.h"
namespace dynamorio {
namespace drmemtrace {
@@ -61,6 +60,7 @@ class view_t : public analysis_tool_t {
view_t(const std::string &module_file_path, uint64_t skip_refs, uint64_t sim_refs,
const std::string &syntax, unsigned int verbose,
const std::string &alt_module_dir = "");
+ virtual ~view_t();
std::string
initialize_stream(memtrace_stream_t *serial_stream) override;
bool
@@ -136,7 +136,9 @@ class view_t : public analysis_tool_t {
// std::optional here.
std::string module_file_path_;
std::unique_ptr module_mapper_;
- raw2trace_directory_t directory_;
+ // XXX: Perhaps module_mapper_t should be made to own the cleanup of
+ // modfile_bytes_.
+ char *modfile_bytes_ = nullptr;
unsigned int knob_verbose_;
int trace_version_;
diff --git a/clients/drcachesim/tracer/instru.h b/clients/drcachesim/tracer/instru.h
index 7aa6bda6191..bf674b63321 100644
--- a/clients/drcachesim/tracer/instru.h
+++ b/clients/drcachesim/tracer/instru.h
@@ -47,6 +47,7 @@
#include "dr_allocator.h"
#include "dr_api.h"
#include "drvector.h"
+#include "raw2trace_shared.h"
#include "trace_entry.h"
namespace dynamorio {
@@ -54,9 +55,6 @@ namespace drmemtrace {
#define MINSERT instrlist_meta_preinsert
-// Versioning for our drmodtrack custom module fields.
-#define CUSTOM_MODULE_VERSION 1
-
// A std::unordered_set, even using dr_allocator_t, raises transparency risks when
// statically linked on Windows (from lock functions and other non-allocator
// resources). We thus create our own resource-isolated class to track GPR register
@@ -499,11 +497,6 @@ class offline_instru_t : public instru_t {
label_marks_elidable(instr_t *instr, DR_PARAM_OUT int *opnd_index,
DR_PARAM_OUT int *memopnd_index, DR_PARAM_OUT bool *is_write,
DR_PARAM_OUT bool *needs_base);
- static int
- print_module_data_fields(char *dst, size_t max_len, const void *custom_data,
- size_t custom_size,
- int (*user_print_cb)(void *data, char *dst, size_t max_len),
- void *user_cb_data);
private:
struct custom_module_data_t {
diff --git a/clients/drcachesim/tracer/instru_offline.cpp b/clients/drcachesim/tracer/instru_offline.cpp
index bf3e63564e5..a0c3873199f 100644
--- a/clients/drcachesim/tracer/instru_offline.cpp
+++ b/clients/drcachesim/tracer/instru_offline.cpp
@@ -47,6 +47,7 @@
#include "drreg.h"
#include "drutil.h"
#include "drvector.h"
+#include "raw2trace_shared.h"
#include "trace_entry.h"
#include "utils.h"
#include "instru.h"
@@ -186,31 +187,6 @@ offline_instru_t::load_custom_module_data(module_data_t *module, int seg_idx)
return nullptr;
}
-int
-offline_instru_t::print_module_data_fields(
- char *dst, size_t max_len, const void *custom_data, size_t custom_size,
- int (*user_print_cb)(void *data, char *dst, size_t max_len), void *user_cb_data)
-{
- char *cur = dst;
- int len = dr_snprintf(dst, max_len, "v#%d,%zu,", CUSTOM_MODULE_VERSION, custom_size);
- if (len < 0)
- return -1;
- cur += len;
- if (cur - dst + custom_size > max_len)
- return -1;
- if (custom_size > 0) {
- memcpy(cur, custom_data, custom_size);
- cur += custom_size;
- }
- if (user_print_cb != nullptr) {
- int res = (*user_print_cb)(user_cb_data, cur, max_len - (cur - dst));
- if (res == -1)
- return -1;
- cur += res;
- }
- return (int)(cur - dst);
-}
-
int
offline_instru_t::print_custom_module_data(void *data, char *dst, size_t max_len)
{
diff --git a/clients/drcachesim/tracer/raw2trace.cpp b/clients/drcachesim/tracer/raw2trace.cpp
index 397de45050f..a280e5004c0 100644
--- a/clients/drcachesim/tracer/raw2trace.cpp
+++ b/clients/drcachesim/tracer/raw2trace.cpp
@@ -136,74 +136,6 @@ trace_metadata_writer_t::write_timestamp(byte *buffer, uint64 timestamp)
* Module list
*/
-const char *(*module_mapper_t::user_parse_)(const char *src,
- DR_PARAM_OUT void **data) = nullptr;
-void (*module_mapper_t::user_free_)(void *data) = nullptr;
-int (*module_mapper_t::user_print_)(void *data, char *dst, size_t max_len) = nullptr;
-bool module_mapper_t::has_custom_data_global_ = true;
-
-module_mapper_t::module_mapper_t(
- const char *module_map,
- const char *(*parse_cb)(const char *src, DR_PARAM_OUT void **data),
- std::string (*process_cb)(drmodtrack_info_t *info, void *data, void *user_data),
- void *process_cb_user_data, void (*free_cb)(void *data), uint verbosity,
- const std::string &alt_module_dir, file_t encoding_file)
- : modmap_(module_map)
- , cached_user_free_(free_cb)
- , verbosity_(verbosity)
- , alt_module_dir_(alt_module_dir)
- , encoding_file_(encoding_file)
-{
- // We mutate global state because do_module_parsing() uses drmodtrack, which
- // wants global functions. The state isn't needed past do_module_parsing(), so
- // we make sure to reset it afterwards.
- DR_ASSERT(user_parse_ == nullptr);
- DR_ASSERT(user_free_ == nullptr);
- DR_ASSERT(user_print_ == nullptr);
-
- user_parse_ = parse_cb;
- user_process_ = process_cb;
- user_process_data_ = process_cb_user_data;
- user_free_ = free_cb;
- // has_custom_data_global_ is potentially mutated in parse_custom_module_data.
- // It is assumed to be set to 'true' initially.
- has_custom_data_global_ = true;
-
- if (modmap_ != nullptr)
- last_error_ = do_module_parsing();
- if (encoding_file_ != INVALID_FILE)
- last_error_ += do_encoding_parsing();
-
- // capture has_custom_data_global_'s value for this instance.
- has_custom_data_ = has_custom_data_global_;
-
- user_parse_ = nullptr;
- user_free_ = nullptr;
-}
-
-module_mapper_t::~module_mapper_t()
-{
- // update user_free_
- user_free_ = cached_user_free_;
- // drmodtrack_offline_exit requires the parameter to be non-null, but we
- // may not have even initialized the modhandle yet.
- if (modhandle_ != nullptr &&
- drmodtrack_offline_exit(modhandle_) != DRCOVLIB_SUCCESS) {
- WARN("Failed to clean up module table data");
- }
- user_free_ = nullptr;
- for (std::vector::iterator mvi = modvec_.begin(); mvi != modvec_.end();
- ++mvi) {
- if (!mvi->is_external && mvi->map_seg_base != NULL && mvi->total_map_size != 0) {
- bool ok = dr_unmap_executable_file(mvi->map_seg_base, mvi->total_map_size);
- if (!ok)
- WARN("Failed to unmap module %s", mvi->path);
- }
- }
- modhandle_ = nullptr;
- modvec_.clear();
-}
-
std::string
raw2trace_t::handle_custom_data(const char *(*parse_cb)(const char *src,
DR_PARAM_OUT void **data),
@@ -218,86 +150,6 @@ raw2trace_t::handle_custom_data(const char *(*parse_cb)(const char *src,
return "";
}
-const char *
-module_mapper_t::parse_custom_module_data(const char *src, DR_PARAM_OUT void **data)
-{
- const char *buf = src;
- const char *skip_comma = strchr(buf, ',');
- // Check the version # to try and handle legacy and newer formats.
- int version = -1;
- if (skip_comma == nullptr || dr_sscanf(buf, "v#%d,", &version) != 1 ||
- version != CUSTOM_MODULE_VERSION) {
- // It's not what we expect. We try to handle legacy formats before bailing.
- static bool warned_once;
- has_custom_data_global_ = false;
- if (!warned_once) { // Race is fine: modtrack parsing is global already.
- WARN("Incorrect module field version %d: attempting to handle legacy format",
- version);
- warned_once = true;
- }
- // First, see if the user_parse_ is happy:
- if (user_parse_ != nullptr) {
- void *user_data;
- buf = (*user_parse_)(buf, &user_data);
- if (buf != nullptr) {
- // Assume legacy format w/ user data but none of our own.
- custom_module_data_t *custom_data = new custom_module_data_t;
- custom_data->user_data = user_data;
- custom_data->contents_size = 0;
- custom_data->contents = nullptr;
- *data = custom_data;
- return buf;
- }
- }
- // Now look for no custom field at all.
- // If the next field looks like a path, we assume it's the old format with
- // no user field and we continue w/o vdso data.
- if (buf[0] == '/' || strstr(buf, "[vdso]") == buf) {
- *data = nullptr;
- return buf;
- }
- // Else, bail.
- WARN("Unable to parse module data: custom field mismatch");
- return nullptr;
- }
- buf = skip_comma + 1;
- skip_comma = strchr(buf, ',');
- size_t size;
- if (skip_comma == nullptr || dr_sscanf(buf, "%zu,", &size) != 1)
- return nullptr; // error
- custom_module_data_t *custom_data = new custom_module_data_t;
- custom_data->contents_size = size;
- buf = skip_comma + 1;
- if (custom_data->contents_size == 0)
- custom_data->contents = nullptr;
- else {
- custom_data->contents = buf;
- buf += custom_data->contents_size;
- }
- if (user_parse_ != nullptr)
- buf = (*user_parse_)(buf, &custom_data->user_data);
- *data = custom_data;
- return buf;
-}
-
-int
-module_mapper_t::print_custom_module_data(void *data, char *dst, size_t max_len)
-{
- custom_module_data_t *custom_data = (custom_module_data_t *)data;
- return offline_instru_t::print_module_data_fields(
- dst, max_len, custom_data->contents, custom_data->contents_size, user_print_,
- custom_data->user_data);
-}
-
-void
-module_mapper_t::free_custom_module_data(void *data)
-{
- custom_module_data_t *custom_data = (custom_module_data_t *)data;
- if (user_free_ != nullptr)
- (*user_free_)(custom_data->user_data);
- delete custom_data;
-}
-
std::string
raw2trace_t::do_module_parsing()
{
@@ -309,76 +161,6 @@ raw2trace_t::do_module_parsing()
return module_mapper_->get_last_error();
}
-std::string
-module_mapper_t::do_module_parsing()
-{
- uint num_mods;
- VPRINT(1, "Reading module file from memory\n");
- if (drmodtrack_add_custom_data(nullptr, nullptr, parse_custom_module_data,
- free_custom_module_data) != DRCOVLIB_SUCCESS) {
- return "Failed to set up custom module parser";
- }
- if (drmodtrack_offline_read(INVALID_FILE, modmap_, NULL, &modhandle_, &num_mods) !=
- DRCOVLIB_SUCCESS)
- return "Failed to parse module file";
- modlist_.resize(num_mods);
- for (uint i = 0; i < num_mods; i++) {
- modlist_[i].struct_size = sizeof(modlist_[i]);
- if (drmodtrack_offline_lookup(modhandle_, i, &modlist_[i]) != DRCOVLIB_SUCCESS)
- return "Failed to query module file";
- if (user_process_ != nullptr) {
- custom_module_data_t *custom = (custom_module_data_t *)modlist_[i].custom;
- std::string error =
- (*user_process_)(&modlist_[i], custom->user_data, user_process_data_);
- if (!error.empty())
- return error;
- }
- }
- return "";
-}
-
-std::string
-module_mapper_t::do_encoding_parsing()
-{
- if (encoding_file_ == INVALID_FILE)
- return "";
- uint64 file_size;
- if (!dr_file_size(encoding_file_, &file_size))
- return "Failed to obtain size of encoding file";
- size_t map_size = (size_t)file_size;
- byte *map_start = reinterpret_cast(
- dr_map_file(encoding_file_, &map_size, 0, NULL, DR_MEMPROT_READ, 0));
- if (map_start == nullptr || map_size < file_size)
- return "Failed to map encoding file";
- byte *map_at = map_start;
- byte *map_end = map_start + file_size;
- uint64_t encoding_file_version = *reinterpret_cast(map_at);
- map_at += sizeof(uint64_t);
- if (encoding_file_version > ENCODING_FILE_VERSION)
- return "Encoding file has invalid version";
- if (encoding_file_version >= ENCODING_FILE_VERSION_HAS_FILE_TYPE) {
- if (map_at + sizeof(uint64_t) > map_end)
- return "Encoding file header is truncated";
- uint64_t encoding_file_type = *reinterpret_cast(map_at);
- map_at += sizeof(uint64_t);
- separate_non_mod_instrs_ =
- TESTANY(ENCODING_FILE_TYPE_SEPARATE_NON_MOD_INSTRS, encoding_file_type);
- }
- uint64_t cumulative_encoding_length = 0;
- while (map_at < map_end) {
- encoding_entry_t *entry = reinterpret_cast(map_at);
- if (entry->length <= sizeof(encoding_entry_t))
- return "Encoding file is corrupted";
- if (map_at + entry->length > map_end)
- return "Encoding file is truncated";
- cum_block_enc_len_to_encoding_id_[cumulative_encoding_length] = entry->id;
- cumulative_encoding_length += (entry->length - sizeof(encoding_entry_t));
- encodings_[entry->id] = entry;
- map_at += entry->length;
- }
- return "";
-}
-
std::string
raw2trace_t::read_and_map_modules()
{
@@ -393,113 +175,6 @@ raw2trace_t::read_and_map_modules()
return module_mapper_->get_last_error();
}
-// Maps each module into the address space.
-// There are several types of mapping entries in the module list:
-// 1) Raw bits directly stored. It is simply pointed at.
-// 2) Extra segments for a module. A single mapping is used for all
-// segments, so extras are ignored.
-// 3) A main segment. The module's file is located by first looking in
-// the alt_module_dir_; if not found, the path present during tracing
-// is searched.
-void
-module_mapper_t::read_and_map_modules()
-{
- if (!last_error_.empty())
- return;
- for (auto it = modlist_.begin(); it != modlist_.end(); ++it) {
- drmodtrack_info_t &info = *it;
- custom_module_data_t *custom_data = (custom_module_data_t *)info.custom;
- if (custom_data != nullptr && custom_data->contents_size > 0) {
- // These raw bytes for vdso is only present for legacy traces; we
- // use encoding entries for new traces.
- // XXX i#2062: Delete this code once we stop supporting legacy traces.
- VPRINT(1, "Using module %d %s stored %zd-byte contents @" PFX "\n",
- (int)modvec_.size(), info.path, custom_data->contents_size,
- custom_data->contents);
- modvec_.push_back(
- module_t(info.path, info.start, (byte *)custom_data->contents, 0,
- custom_data->contents_size, custom_data->contents_size,
- true /*external data*/));
- } else if (strcmp(info.path, "") == 0 ||
- // This should only happen with legacy trace data that's missing
- // the vdso contents.
- (!has_custom_data_ && strcmp(info.path, "[vdso]") == 0)) {
- // We won't be able to decode.
- modvec_.push_back(module_t(info.path, info.start, NULL, 0, 0, 0));
- } else if (info.containing_index != info.index) {
- // For split segments, we assume our mapped layout matches the original.
- byte *seg_map_base = modvec_[info.containing_index].map_seg_base +
- (info.start - modvec_[info.containing_index].orig_seg_base);
- VPRINT(1, "Secondary segment: module %d seg %p-%p = %s\n",
- (int)modvec_.size(), seg_map_base, seg_map_base + info.size,
- info.path);
- // We did not map writable segments. We can't easily detect an internal
- // unmapped writable segment, but for those off the end of our mapping we
- // can avoid pretending there's anything there.
- bool off_end =
- (size_t)(info.start - modvec_[info.containing_index].orig_seg_base) >=
- modvec_[info.containing_index].total_map_size;
- DR_ASSERT(off_end ||
- info.start - modvec_[info.containing_index].orig_seg_base +
- info.size <=
- modvec_[info.containing_index].total_map_size);
- modvec_.push_back(module_t(
- info.path, info.start, off_end ? NULL : seg_map_base,
- off_end ? 0 : info.start - modvec_[info.containing_index].orig_seg_base,
- off_end ? 0 : info.size,
- // 0 total size indicates this is a secondary segment.
- 0));
- } else {
- size_t map_size = 0;
- byte *base_pc = NULL;
- if (!alt_module_dir_.empty()) {
- // First try the specified module dir. It takes precedence to allow
- // overriding the recorded path even when an identical-seeming path
- // exists on the processing machine (e.g., system libraries).
- // XXX: We should add a checksum on UNIX to match Windows and have
- // a sanity check on the library version.
- std::string basename(info.path);
- size_t sep_index = basename.find_last_of(DIRSEP ALT_DIRSEP);
- if (sep_index != std::string::npos)
- basename = std::string(basename, sep_index + 1, std::string::npos);
- std::string new_path = alt_module_dir_ + DIRSEP + basename;
- VPRINT(2, "Trying to map %s\n", new_path.c_str());
- base_pc = dr_map_executable_file(new_path.c_str(),
- DR_MAPEXE_SKIP_WRITABLE, &map_size);
- }
- if (base_pc == NULL) {
- // Try the recorded path.
- VPRINT(2, "Trying to map %s\n", info.path);
- base_pc =
- dr_map_executable_file(info.path, DR_MAPEXE_SKIP_WRITABLE, &map_size);
- }
- if (base_pc == NULL) {
- // We expect to fail to map dynamorio.dll for x64 Windows as it
- // is built /fixed. (We could try to have the map succeed w/o relocs,
- // but we expect to not care enough about code in DR).
- // We also expect to fail for vdso, for which we have encoding entries.
- if (strstr(info.path, "dynamorio") != nullptr ||
- strstr(info.path, "linux-gate") != nullptr ||
- strstr(info.path, "vdso") != nullptr)
- modvec_.push_back(module_t(info.path, info.start, NULL, 0, 0, 0));
- else {
- last_error_ = "Failed to map module " + std::string(info.path);
- return;
- }
- } else {
- VPRINT(1, "Mapped module %d @%p-%p (-%p segment) = %s\n",
- (int)modvec_.size(), base_pc, base_pc + map_size,
- base_pc + info.size, info.path);
- // Be sure to only use the initial segment size to avoid covering
- // another mapping in a segment gap (i#4731).
- modvec_.push_back(
- module_t(info.path, info.start, base_pc, 0, info.size, map_size));
- }
- }
- }
- VPRINT(1, "Successfully read %zu modules\n", modlist_.size());
-}
-
std::string
raw2trace_t::do_module_parsing_and_mapping()
{
@@ -517,68 +192,6 @@ raw2trace_t::find_mapped_trace_address(app_pc trace_address,
return module_mapper_->get_last_error();
}
-// The output range is really a segment and not the whole module.
-app_pc
-module_mapper_t::find_mapped_trace_bounds(app_pc trace_address,
- DR_PARAM_OUT app_pc *module_start,
- DR_PARAM_OUT size_t *module_size)
-{
- if (modvec_.empty()) {
- last_error_ = "Failed to call get_loaded_modules() first";
- return nullptr;
- }
-
- // For simplicity we do a linear search, caching the prior hit.
- if (trace_address >= last_orig_base_ &&
- trace_address < last_orig_base_ + last_map_size_) {
- if (module_start != nullptr)
- *module_start = last_map_base_;
- if (module_size != nullptr)
- *module_size = last_map_size_;
- return trace_address - last_orig_base_ + last_map_base_;
- }
- for (std::vector::iterator mvi = modvec_.begin(); mvi != modvec_.end();
- ++mvi) {
- if (trace_address >= mvi->orig_seg_base &&
- trace_address < mvi->orig_seg_base + mvi->seg_size) {
- app_pc mapped_address =
- trace_address - mvi->orig_seg_base + mvi->map_seg_base;
- last_orig_base_ = mvi->orig_seg_base;
- last_map_size_ = mvi->seg_size;
- last_map_base_ = mvi->map_seg_base;
- if (module_start != nullptr)
- *module_start = last_map_base_;
- if (module_size != nullptr)
- *module_size = last_map_size_;
- return mapped_address;
- }
- }
- last_error_ = "Trace address not found";
- return nullptr;
-}
-
-app_pc
-module_mapper_t::find_mapped_trace_address(app_pc trace_address)
-{
- return find_mapped_trace_bounds(trace_address, nullptr, nullptr);
-}
-
-drcovlib_status_t
-module_mapper_t::write_module_data(char *buf, size_t buf_size,
- int (*print_cb)(void *data, char *dst, size_t max_len),
- DR_PARAM_OUT size_t *wrote)
-{
- user_print_ = print_cb;
- drcovlib_status_t res =
- drmodtrack_add_custom_data(nullptr, print_custom_module_data,
- parse_custom_module_data, free_custom_module_data);
- if (res == DRCOVLIB_SUCCESS) {
- res = drmodtrack_offline_write(modhandle_, buf, buf_size, wrote);
- }
- user_print_ = nullptr;
- return res;
-}
-
/***************************************************************************
* Top-level
*/
diff --git a/clients/drcachesim/tracer/raw2trace.h b/clients/drcachesim/tracer/raw2trace.h
index 794b37cd27d..41662004997 100644
--- a/clients/drcachesim/tracer/raw2trace.h
+++ b/clients/drcachesim/tracer/raw2trace.h
@@ -64,7 +64,6 @@
#include "archive_ostream.h"
#include "dr_api.h"
-#include "drcovlib.h"
#include "drmemtrace.h"
#include "hashtable.h"
#include "instru.h"
@@ -129,32 +128,6 @@ typedef enum {
RAW2TRACE_STAT_MAX,
} raw2trace_statistic_t;
-struct module_t {
- module_t(const char *path, app_pc orig, byte *map, size_t offs, size_t size,
- size_t total_size, bool external = false)
- : path(path)
- , orig_seg_base(orig)
- , map_seg_base(map)
- , seg_offs(offs)
- , seg_size(size)
- , total_map_size(total_size)
- , is_external(external)
- {
- }
- const char *path;
- // We have to handle segments within a module separately, as there can be
- // gaps between them that contain other objects (xref i#4731).
- app_pc orig_seg_base;
- byte *map_seg_base;
- size_t seg_offs;
- size_t seg_size;
- // Despite tracking segments separately, we have a single mapping.
- // The first segment stores that mapping size here; subsequent segments
- // have 0 for this field.
- size_t total_map_size;
- bool is_external; // If true, the data is embedded in drmodtrack custom fields.
-};
-
/**
* instr_summary_t is a compact encapsulation of the information needed by trace
* conversion from decoded instructions.
@@ -398,306 +371,6 @@ struct trace_metadata_writer_t {
write_timestamp(byte *buffer, uint64 timestamp);
};
-/**
- * module_mapper_t maps and unloads application modules, as well as non-module
- * instruction encodings (for raw traces, or if not present in the final trace).
- * Using it assumes a dr_context has already been setup.
- * This class is not thread-safe.
- */
-class module_mapper_t {
-public:
- /**
- * Parses and iterates over the list of modules. This is provided to give the user a
- * method for iterating modules in the presence of the custom field used by drmemtrace
- * that prevents direct use of drmodtrack_offline_read(). Its parsing of the module
- * data will invoke \p parse_cb, which should advance the module data pointer passed
- * in \p src and return it as its return value (or nullptr on error), returning
- * the resulting parsed data in \p data. The \p data pointer will afterwards be
- * passed to both \p process_cb, which can update the module path inside \p info
- * (and return a non-empty string on error), and \b free_cb, which can perform
- * cleanup.
- *
- * The callbacks will only be called during object construction.
- *
- * Additionally parses the non-module instruction encodings file if 'encoding_file'
- * is not nullptr.
- *
- * On success, calls the \p process_cb function for every module in the list.
- * On failure, get_last_error() is non-empty, and indicates the cause.
- */
- static std::unique_ptr
- create(const char *module_map,
- const char *(*parse_cb)(const char *src, DR_PARAM_OUT void **data) = nullptr,
- std::string (*process_cb)(drmodtrack_info_t *info, void *data,
- void *user_data) = nullptr,
- void *process_cb_user_data = nullptr, void (*free_cb)(void *data) = nullptr,
- uint verbosity = 0, const std::string &alt_module_dir = "",
- file_t encoding_file = INVALID_FILE)
- {
- return std::unique_ptr(
- new module_mapper_t(module_map, parse_cb, process_cb, process_cb_user_data,
- free_cb, verbosity, alt_module_dir, encoding_file));
- }
-
- /**
- * All APIs on this type, including constructor, may fail. get_last_error() returns
- * the last error message. The object should be considered unusable if
- * !get_last_error().empty().
- */
- std::string
- get_last_error(void) const
- {
- return last_error_;
- }
-
- /**
- * module_t vector corresponding to the application modules. Lazily loads and caches
- * modules. If the object is invalid, returns an empty vector. The user may check
- * get_last_error() to ensure no error has occurred, or get the applicable error
- * message.
- */
- virtual const std::vector &
- get_loaded_modules()
- {
- if (last_error_.empty() && modvec_.empty())
- read_and_map_modules();
- return modvec_;
- }
-
- app_pc
- get_orig_pc_from_map_pc(app_pc map_pc, uint64 modidx, uint64 modoffs) const
- {
- if (modidx == PC_MODIDX_INVALID) {
- uint64 blockidx = 0;
- uint64 blockoffs = 0;
- convert_modoffs_to_non_mod_block(modoffs, blockidx, blockoffs);
- auto const it = encodings_.find(blockidx);
- if (it == encodings_.end())
- return nullptr;
- encoding_entry_t *entry = it->second;
- return (map_pc - entry->encodings) +
- reinterpret_cast(entry->start_pc);
- } else {
- size_t idx = static_cast(modidx); // Avoid win32 warnings.
- app_pc res = map_pc - modvec_[idx].map_seg_base + modvec_[idx].orig_seg_base;
-#ifdef ARM
- // Match Thumb vs Arm mode by setting LSB.
- if (TESTANY(1, modoffs))
- res = reinterpret_cast(reinterpret_cast(res) | 1);
-#endif
- return res;
- }
- }
-
- app_pc
- get_orig_pc(uint64 modidx, uint64 modoffs) const
- {
- if (modidx == PC_MODIDX_INVALID) {
- uint64 blockidx = 0;
- uint64 blockoffs = 0;
- convert_modoffs_to_non_mod_block(modoffs, blockidx, blockoffs);
- auto const it = encodings_.find(blockidx);
- if (it == encodings_.end())
- return nullptr;
- encoding_entry_t *entry = it->second;
- return reinterpret_cast(entry->start_pc + blockoffs);
- } else {
- size_t idx = static_cast(modidx); // Avoid win32 warnings.
- // Cast to unsigned pointer-sized int first to avoid sign-extending.
- return reinterpret_cast(
- reinterpret_cast(modvec_[idx].orig_seg_base)) +
- (modoffs - modvec_[idx].seg_offs);
- }
- }
-
- app_pc
- get_map_pc(uint64 modidx, uint64 modoffs) const
- {
- if (modidx == PC_MODIDX_INVALID) {
- uint64 blockidx = 0;
- uint64 blockoffs = 0;
- convert_modoffs_to_non_mod_block(modoffs, blockidx, blockoffs);
- auto const it = encodings_.find(blockidx);
- if (it == encodings_.end())
- return nullptr;
- encoding_entry_t *entry = it->second;
- return &entry->encodings[blockoffs];
- } else {
- size_t idx = static_cast(modidx); // Avoid win32 warnings.
- return modvec_[idx].map_seg_base + (modoffs - modvec_[idx].seg_offs);
- }
- }
-
- /**
- * This interface is meant to be used with a final trace rather than a raw
- * trace, using the module log file saved from the raw2trace conversion.
- * After a call to get_loaded_modules(), this routine may be used
- * to convert an instruction program counter in a trace into an address in the
- * current process where the instruction bytes for that instruction are mapped,
- * allowing decoding for obtaining further information than is stored in the trace.
- * This interface only supports code inside modules; generated code is
- * expected to have instruction encodings in the trace itself.
- * Returns the mapped address. Check get_last_error_() if an error occurred.
- */
- app_pc
- find_mapped_trace_address(app_pc trace_address);
-
- /**
- * This is identical to find_mapped_trace_address() but it also returns the
- * bounds of the containing region, allowing the caller to perform its own
- * mapping for any address that is also within those bounds.
- */
- app_pc
- find_mapped_trace_bounds(app_pc trace_address, DR_PARAM_OUT app_pc *module_start,
- DR_PARAM_OUT size_t *module_size);
-
- /**
- * Unload modules loaded with read_and_map_modules(), freeing associated resources.
- */
- virtual ~module_mapper_t();
-
- /**
- * Writes out the module list to \p buf, whose capacity is \p buf_size.
- * The written data includes any modifications made by the \p process_cb
- * passed to create(). Any custom data returned by the \p parse_cb passed to
- * create() is passed to \p print_cb here for serialization. The \p print_cb
- * must return the number of characters printed or -1 on error.
- */
- drcovlib_status_t
- write_module_data(char *buf, size_t buf_size,
- int (*print_cb)(void *data, char *dst, size_t max_len),
- DR_PARAM_OUT size_t *wrote);
-
-protected:
- module_mapper_t(const char *module_map,
- const char *(*parse_cb)(const char *src,
- DR_PARAM_OUT void **data) = nullptr,
- std::string (*process_cb)(drmodtrack_info_t *info, void *data,
- void *user_data) = nullptr,
- void *process_cb_user_data = nullptr,
- void (*free_cb)(void *data) = nullptr, uint verbosity = 0,
- const std::string &alt_module_dir = "",
- file_t encoding_file = INVALID_FILE);
-
- module_mapper_t(const module_mapper_t &) = delete;
- module_mapper_t &
- operator=(const module_mapper_t &) = delete;
-#ifndef WINDOWS
- module_mapper_t(module_mapper_t &&) = delete;
- module_mapper_t &
- operator=(module_mapper_t &&) = delete;
-#endif
- // We store this in drmodtrack_info_t.custom to combine our binary contents
- // data with any user-added module data from drmemtrace_custom_module_data.
- struct custom_module_data_t {
- size_t contents_size;
- const char *contents;
- void *user_data;
- };
-
- void
- convert_modoffs_to_non_mod_block(uint64 modoffs, uint64 &blockidx,
- uint64 &blockoffs) const
- {
- if (!separate_non_mod_instrs_) {
- blockidx = modoffs;
- blockoffs = 0;
- return;
- }
- auto it = cum_block_enc_len_to_encoding_id_.upper_bound(modoffs);
- // Since modoffs >= 0 and the smallest key in cum_block_enc_len_to_encoding_id_ is
- // always zero, `it` should never be the first element of the map.
- DR_ASSERT(it != cum_block_enc_len_to_encoding_id_.begin());
- auto it_prev = it;
- it_prev--;
- DR_ASSERT(it_prev->first <= modoffs &&
- (it == cum_block_enc_len_to_encoding_id_.end() || it->first > modoffs));
- blockidx = it_prev->second;
- blockoffs = modoffs - it_prev->first;
- }
-
- virtual void
- read_and_map_modules(void);
-
- std::string
- do_module_parsing();
-
- std::string
- do_encoding_parsing();
-
- const char *modmap_ = nullptr;
- void *modhandle_ = nullptr;
- std::vector modvec_;
- void (*const cached_user_free_)(void *data) = nullptr;
-
- // Custom module fields that use drmodtrack are global.
- static const char *(*user_parse_)(const char *src, DR_PARAM_OUT void **data);
- static void (*user_free_)(void *data);
- static int (*user_print_)(void *data, char *dst, size_t max_len);
- static const char *
- parse_custom_module_data(const char *src, DR_PARAM_OUT void **data);
- static int
- print_custom_module_data(void *data, char *dst, size_t max_len);
- static void
- free_custom_module_data(void *data);
- static bool has_custom_data_global_;
-
- bool has_custom_data_ = false;
-
- // We store module info for do_module_parsing.
- std::vector modlist_;
- std::string (*user_process_)(drmodtrack_info_t *info, void *data,
- void *user_data) = nullptr;
- void *user_process_data_ = nullptr;
- app_pc last_orig_base_ = 0;
- size_t last_map_size_ = 0;
- byte *last_map_base_ = nullptr;
- bool separate_non_mod_instrs_ = false;
- std::map cum_block_enc_len_to_encoding_id_;
-
- uint verbosity_ = 0;
- std::string alt_module_dir_;
- std::string last_error_;
-
- file_t encoding_file_ = INVALID_FILE;
- std::unordered_map encodings_;
-};
-
-/**
- * Subclasses module_mapper_t and replaces the module loading with a buffer
- * of encoded instr_t. Useful for tests where we want to mock the module
- * files with an in-memory buffer of instrs.
- */
-class test_module_mapper_t : public module_mapper_t {
-public:
- test_module_mapper_t(instrlist_t *instrs, void *drcontext)
- : module_mapper_t(nullptr)
- {
- // We encode for 1-based addresses for simpler tests with low values while
- // avoiding null pointer manipulation complaints (xref i#6196).
- byte *pc = instrlist_encode_to_copy(
- drcontext, instrs, decode_buf_,
- reinterpret_cast(static_cast(4)), nullptr, true);
- DR_ASSERT(pc != nullptr);
- DR_ASSERT(pc - decode_buf_ < MAX_DECODE_SIZE);
- // Clear do_module_parsing error; we can't cleanly make virtual b/c it's
- // called from the constructor.
- last_error_ = "";
- }
-
-protected:
- void
- read_and_map_modules() override
- {
- modvec_.push_back(module_t("fake_exe", 0, decode_buf_, 0, MAX_DECODE_SIZE,
- MAX_DECODE_SIZE, true));
- }
-
-private:
- static const int MAX_DECODE_SIZE = 1024;
- byte decode_buf_[MAX_DECODE_SIZE];
-};
-
/**
* Header of raw trace.
*/
diff --git a/clients/drcachesim/tracer/raw2trace_directory.cpp b/clients/drcachesim/tracer/raw2trace_directory.cpp
index bddb8c7fa82..fb9ec8b6b7b 100644
--- a/clients/drcachesim/tracer/raw2trace_directory.cpp
+++ b/clients/drcachesim/tracer/raw2trace_directory.cpp
@@ -1,5 +1,5 @@
/* **********************************************************
- * Copyright (c) 2017-2023 Google, Inc. All rights reserved.
+ * Copyright (c) 2017-2024 Google, Inc. All rights reserved.
* **********************************************************/
/*
@@ -43,6 +43,7 @@
# include
#endif
+#include
#include
#include
#include
@@ -55,6 +56,7 @@
#include "directory_iterator.h"
#include "dr_api.h" // Must be after windows.h.
#include "raw2trace_directory.h" // Includes dr_api.h which must be after windows.h.
+#include "raw2trace_shared.h"
#include "reader.h"
#include "raw2trace.h"
#include "record_file_reader.h"
@@ -432,22 +434,6 @@ raw2trace_directory_t::open_cpu_schedule_file()
#endif
}
-std::string
-raw2trace_directory_t::read_module_file(const std::string &modfilename)
-{
- modfile_ = dr_open_file(modfilename.c_str(), DR_FILE_READ);
- if (modfile_ == INVALID_FILE)
- return "Failed to open module file " + modfilename;
- uint64 modfile_size;
- if (!dr_file_size(modfile_, &modfile_size))
- return "Failed to get module file size: " + modfilename;
- size_t modfile_size_ = (size_t)modfile_size;
- modfile_bytes_ = new char[modfile_size_];
- if (dr_read_file(modfile_, modfile_bytes_, modfile_size_) < (ssize_t)modfile_size_)
- return "Didn't read whole module file " + modfilename;
- return "";
-}
-
bool
raw2trace_directory_t::is_window_subdir(const std::string &dir)
{
@@ -552,7 +538,7 @@ raw2trace_directory_t::initialize(const std::string &indir, const std::string &o
}
std::string modfilename =
modfile_dir + std::string(DIRSEP) + DRMEMTRACE_MODULE_LIST_FILENAME;
- std::string err = read_module_file(modfilename);
+ std::string err = read_module_file(modfilename, modfile_, modfile_bytes_);
if (!err.empty())
return err;
@@ -603,12 +589,6 @@ raw2trace_directory_t::initialize(const std::string &indir, const std::string &o
return open_thread_files();
}
-std::string
-raw2trace_directory_t::initialize_module_file(const std::string &module_file_path)
-{
- return read_module_file(module_file_path);
-}
-
std::string
raw2trace_directory_t::initialize_funclist_file(
const std::string &funclist_file_path,
diff --git a/clients/drcachesim/tracer/raw2trace_directory.h b/clients/drcachesim/tracer/raw2trace_directory.h
index 1d490c6c4c7..c8de4f30928 100644
--- a/clients/drcachesim/tracer/raw2trace_directory.h
+++ b/clients/drcachesim/tracer/raw2trace_directory.h
@@ -1,5 +1,5 @@
/* **********************************************************
- * Copyright (c) 2017-2023 Google, Inc. All rights reserved.
+ * Copyright (c) 2017-2024 Google, Inc. All rights reserved.
* **********************************************************/
/*
@@ -78,11 +78,6 @@ class raw2trace_directory_t {
initialize(const std::string &indir, const std::string &outdir,
const std::string &compress = DEFAULT_TRACE_COMPRESSION_TYPE,
const std::string &syscall_template_file = "");
- // Use this instead of initialize() to only fill in modfile_bytes, for
- // constructing a module_mapper_t. Returns "" on success or an error message on
- // failure.
- std::string
- initialize_module_file(const std::string &module_file_path);
// Use this instead of initialize() to only read the funcion map file.
// Returns "" on success or an error message on failure.
// On success, pushes the parsed entries from the file into "entries".
@@ -115,8 +110,6 @@ class raw2trace_directory_t {
std::string
trace_suffix();
std::string
- read_module_file(const std::string &modfilename);
- std::string
open_thread_files();
std::string
open_thread_log_file(const char *basename);
diff --git a/clients/drcachesim/tracer/raw2trace_shared.cpp b/clients/drcachesim/tracer/raw2trace_shared.cpp
index ae6ffa7e3ed..ca2b17e82b2 100644
--- a/clients/drcachesim/tracer/raw2trace_shared.cpp
+++ b/clients/drcachesim/tracer/raw2trace_shared.cpp
@@ -1,5 +1,5 @@
/* **********************************************************
- * Copyright (c) 2016-2023 Google, Inc. All rights reserved.
+ * Copyright (c) 2016-2024 Google, Inc. All rights reserved.
* **********************************************************/
/*
@@ -33,6 +33,7 @@
#include "raw2trace_shared.h"
#include
+#include
#include
#include
#include
@@ -44,6 +45,28 @@
namespace dynamorio {
namespace drmemtrace {
+#define WARN(msg, ...) \
+ do { \
+ fprintf(stderr, "WARNING: " msg "\n", ##__VA_ARGS__); \
+ fflush(stderr); \
+ } while (0)
+
+#define VPRINT_HEADER() \
+ do { \
+ fprintf(stderr, "[drmemtrace]: "); \
+ } while (0)
+
+// We fflush for Windows cygwin where stderr is not flushed.
+#undef VPRINT
+#define VPRINT(level, ...) \
+ do { \
+ if (this->verbosity_ >= (level)) { \
+ VPRINT_HEADER(); \
+ fprintf(stderr, __VA_ARGS__); \
+ fflush(stderr); \
+ } \
+ } while (0)
+
bool
trace_metadata_reader_t::is_thread_start(const offline_entry_t *entry,
DR_PARAM_OUT std::string *error,
@@ -143,5 +166,442 @@ drmemtrace_get_timestamp_from_offline_trace(const void *trace, size_t trace_size
return DRMEMTRACE_SUCCESS;
}
+std::string
+read_module_file(const std::string &modfilename, file_t &modfile, char *&modfile_bytes)
+{
+ modfile = dr_open_file(modfilename.c_str(), DR_FILE_READ);
+ if (modfile == INVALID_FILE)
+ return "Failed to open module file " + modfilename;
+ uint64 modfile_size;
+ if (!dr_file_size(modfile, &modfile_size)) {
+ dr_close_file(modfile);
+ modfile = INVALID_FILE;
+ return "Failed to get module file size: " + modfilename;
+ }
+ size_t modfile_size_ = (size_t)modfile_size;
+ modfile_bytes = new char[modfile_size_];
+ if (dr_read_file(modfile, modfile_bytes, modfile_size_) < (ssize_t)modfile_size_) {
+ dr_close_file(modfile);
+ delete[] modfile_bytes;
+ modfile = INVALID_FILE;
+ modfile_bytes = nullptr;
+ return "Didn't read whole module file " + modfilename;
+ }
+ return "";
+}
+
+// The output range is really a segment and not the whole module.
+app_pc
+module_mapper_t::find_mapped_trace_bounds(app_pc trace_address,
+ DR_PARAM_OUT app_pc *module_start,
+ DR_PARAM_OUT size_t *module_size)
+{
+ if (modvec_.empty()) {
+ last_error_ = "Failed to call get_loaded_modules() first";
+ return nullptr;
+ }
+
+ // For simplicity we do a linear search, caching the prior hit.
+ if (trace_address >= last_orig_base_ &&
+ trace_address < last_orig_base_ + last_map_size_) {
+ if (module_start != nullptr)
+ *module_start = last_map_base_;
+ if (module_size != nullptr)
+ *module_size = last_map_size_;
+ return trace_address - last_orig_base_ + last_map_base_;
+ }
+ for (std::vector::iterator mvi = modvec_.begin(); mvi != modvec_.end();
+ ++mvi) {
+ if (trace_address >= mvi->orig_seg_base &&
+ trace_address < mvi->orig_seg_base + mvi->seg_size) {
+ app_pc mapped_address =
+ trace_address - mvi->orig_seg_base + mvi->map_seg_base;
+ last_orig_base_ = mvi->orig_seg_base;
+ last_map_size_ = mvi->seg_size;
+ last_map_base_ = mvi->map_seg_base;
+ if (module_start != nullptr)
+ *module_start = last_map_base_;
+ if (module_size != nullptr)
+ *module_size = last_map_size_;
+ return mapped_address;
+ }
+ }
+ last_error_ = "Trace address not found";
+ return nullptr;
+}
+
+app_pc
+module_mapper_t::find_mapped_trace_address(app_pc trace_address)
+{
+ return find_mapped_trace_bounds(trace_address, nullptr, nullptr);
+}
+
+drcovlib_status_t
+module_mapper_t::write_module_data(char *buf, size_t buf_size,
+ int (*print_cb)(void *data, char *dst, size_t max_len),
+ DR_PARAM_OUT size_t *wrote)
+{
+ user_print_ = print_cb;
+ drcovlib_status_t res =
+ drmodtrack_add_custom_data(nullptr, print_custom_module_data,
+ parse_custom_module_data, free_custom_module_data);
+ if (res == DRCOVLIB_SUCCESS) {
+ res = drmodtrack_offline_write(modhandle_, buf, buf_size, wrote);
+ }
+ user_print_ = nullptr;
+ return res;
+}
+
+// Maps each module into the address space.
+// There are several types of mapping entries in the module list:
+// 1) Raw bits directly stored. It is simply pointed at.
+// 2) Extra segments for a module. A single mapping is used for all
+// segments, so extras are ignored.
+// 3) A main segment. The module's file is located by first looking in
+// the alt_module_dir_; if not found, the path present during tracing
+// is searched.
+void
+module_mapper_t::read_and_map_modules()
+{
+ if (!last_error_.empty())
+ return;
+ for (auto it = modlist_.begin(); it != modlist_.end(); ++it) {
+ drmodtrack_info_t &info = *it;
+ custom_module_data_t *custom_data = (custom_module_data_t *)info.custom;
+ if (custom_data != nullptr && custom_data->contents_size > 0) {
+ // These raw bytes for vdso is only present for legacy traces; we
+ // use encoding entries for new traces.
+ // XXX i#2062: Delete this code once we stop supporting legacy traces.
+ VPRINT(1, "Using module %d %s stored %zd-byte contents @" PFX "\n",
+ (int)modvec_.size(), info.path, custom_data->contents_size,
+ custom_data->contents);
+ modvec_.push_back(
+ module_t(info.path, info.start, (byte *)custom_data->contents, 0,
+ custom_data->contents_size, custom_data->contents_size,
+ true /*external data*/));
+ } else if (strcmp(info.path, "") == 0 ||
+ // This should only happen with legacy trace data that's missing
+ // the vdso contents.
+ (!has_custom_data_ && strcmp(info.path, "[vdso]") == 0)) {
+ // We won't be able to decode.
+ modvec_.push_back(module_t(info.path, info.start, NULL, 0, 0, 0));
+ } else if (info.containing_index != info.index) {
+ // For split segments, we assume our mapped layout matches the original.
+ byte *seg_map_base = modvec_[info.containing_index].map_seg_base +
+ (info.start - modvec_[info.containing_index].orig_seg_base);
+ VPRINT(1, "Secondary segment: module %d seg %p-%p = %s\n",
+ (int)modvec_.size(), seg_map_base, seg_map_base + info.size,
+ info.path);
+ // We did not map writable segments. We can't easily detect an internal
+ // unmapped writable segment, but for those off the end of our mapping we
+ // can avoid pretending there's anything there.
+ bool off_end =
+ (size_t)(info.start - modvec_[info.containing_index].orig_seg_base) >=
+ modvec_[info.containing_index].total_map_size;
+ DR_ASSERT(off_end ||
+ info.start - modvec_[info.containing_index].orig_seg_base +
+ info.size <=
+ modvec_[info.containing_index].total_map_size);
+ modvec_.push_back(module_t(
+ info.path, info.start, off_end ? NULL : seg_map_base,
+ off_end ? 0 : info.start - modvec_[info.containing_index].orig_seg_base,
+ off_end ? 0 : info.size,
+ // 0 total size indicates this is a secondary segment.
+ 0));
+ } else {
+ size_t map_size = 0;
+ byte *base_pc = NULL;
+ if (!alt_module_dir_.empty()) {
+ // First try the specified module dir. It takes precedence to allow
+ // overriding the recorded path even when an identical-seeming path
+ // exists on the processing machine (e.g., system libraries).
+ // XXX: We should add a checksum on UNIX to match Windows and have
+ // a sanity check on the library version.
+ std::string basename(info.path);
+ size_t sep_index = basename.find_last_of(DIRSEP ALT_DIRSEP);
+ if (sep_index != std::string::npos)
+ basename = std::string(basename, sep_index + 1, std::string::npos);
+ std::string new_path = alt_module_dir_ + DIRSEP + basename;
+ VPRINT(2, "Trying to map %s\n", new_path.c_str());
+ base_pc = dr_map_executable_file(new_path.c_str(),
+ DR_MAPEXE_SKIP_WRITABLE, &map_size);
+ }
+ if (base_pc == NULL) {
+ // Try the recorded path.
+ VPRINT(2, "Trying to map %s\n", info.path);
+ base_pc =
+ dr_map_executable_file(info.path, DR_MAPEXE_SKIP_WRITABLE, &map_size);
+ }
+ if (base_pc == NULL) {
+ // We expect to fail to map dynamorio.dll for x64 Windows as it
+ // is built /fixed. (We could try to have the map succeed w/o relocs,
+ // but we expect to not care enough about code in DR).
+ // We also expect to fail for vdso, for which we have encoding entries.
+ if (strstr(info.path, "dynamorio") != nullptr ||
+ strstr(info.path, "linux-gate") != nullptr ||
+ strstr(info.path, "vdso") != nullptr)
+ modvec_.push_back(module_t(info.path, info.start, NULL, 0, 0, 0));
+ else {
+ last_error_ = "Failed to map module " + std::string(info.path);
+ return;
+ }
+ } else {
+ VPRINT(1, "Mapped module %d @%p-%p (-%p segment) = %s\n",
+ (int)modvec_.size(), base_pc, base_pc + map_size,
+ base_pc + info.size, info.path);
+ // Be sure to only use the initial segment size to avoid covering
+ // another mapping in a segment gap (i#4731).
+ modvec_.push_back(
+ module_t(info.path, info.start, base_pc, 0, info.size, map_size));
+ }
+ }
+ }
+ VPRINT(1, "Successfully read %zu modules\n", modlist_.size());
+}
+
+std::string
+module_mapper_t::do_module_parsing()
+{
+ uint num_mods;
+ VPRINT(1, "Reading module file from memory\n");
+ if (drmodtrack_add_custom_data(nullptr, nullptr, parse_custom_module_data,
+ free_custom_module_data) != DRCOVLIB_SUCCESS) {
+ return "Failed to set up custom module parser";
+ }
+ if (drmodtrack_offline_read(INVALID_FILE, modmap_, NULL, &modhandle_, &num_mods) !=
+ DRCOVLIB_SUCCESS)
+ return "Failed to parse module file";
+ modlist_.resize(num_mods);
+ for (uint i = 0; i < num_mods; i++) {
+ modlist_[i].struct_size = sizeof(modlist_[i]);
+ if (drmodtrack_offline_lookup(modhandle_, i, &modlist_[i]) != DRCOVLIB_SUCCESS)
+ return "Failed to query module file";
+ if (user_process_ != nullptr) {
+ custom_module_data_t *custom = (custom_module_data_t *)modlist_[i].custom;
+ std::string error =
+ (*user_process_)(&modlist_[i], custom->user_data, user_process_data_);
+ if (!error.empty())
+ return error;
+ }
+ }
+ return "";
+}
+
+std::string
+module_mapper_t::do_encoding_parsing()
+{
+ if (encoding_file_ == INVALID_FILE)
+ return "";
+ uint64 file_size;
+ if (!dr_file_size(encoding_file_, &file_size))
+ return "Failed to obtain size of encoding file";
+ size_t map_size = (size_t)file_size;
+ byte *map_start = reinterpret_cast(
+ dr_map_file(encoding_file_, &map_size, 0, NULL, DR_MEMPROT_READ, 0));
+ if (map_start == nullptr || map_size < file_size)
+ return "Failed to map encoding file";
+ byte *map_at = map_start;
+ byte *map_end = map_start + file_size;
+ uint64_t encoding_file_version = *reinterpret_cast(map_at);
+ map_at += sizeof(uint64_t);
+ if (encoding_file_version > ENCODING_FILE_VERSION)
+ return "Encoding file has invalid version";
+ if (encoding_file_version >= ENCODING_FILE_VERSION_HAS_FILE_TYPE) {
+ if (map_at + sizeof(uint64_t) > map_end)
+ return "Encoding file header is truncated";
+ uint64_t encoding_file_type = *reinterpret_cast(map_at);
+ map_at += sizeof(uint64_t);
+ separate_non_mod_instrs_ =
+ TESTANY(ENCODING_FILE_TYPE_SEPARATE_NON_MOD_INSTRS, encoding_file_type);
+ }
+ uint64_t cumulative_encoding_length = 0;
+ while (map_at < map_end) {
+ encoding_entry_t *entry = reinterpret_cast(map_at);
+ if (entry->length <= sizeof(encoding_entry_t))
+ return "Encoding file is corrupted";
+ if (map_at + entry->length > map_end)
+ return "Encoding file is truncated";
+ cum_block_enc_len_to_encoding_id_[cumulative_encoding_length] = entry->id;
+ cumulative_encoding_length += (entry->length - sizeof(encoding_entry_t));
+ encodings_[entry->id] = entry;
+ map_at += entry->length;
+ }
+ return "";
+}
+
+const char *
+module_mapper_t::parse_custom_module_data(const char *src, DR_PARAM_OUT void **data)
+{
+ const char *buf = src;
+ const char *skip_comma = strchr(buf, ',');
+ // Check the version # to try and handle legacy and newer formats.
+ int version = -1;
+ if (skip_comma == nullptr || dr_sscanf(buf, "v#%d,", &version) != 1 ||
+ version != CUSTOM_MODULE_VERSION) {
+ // It's not what we expect. We try to handle legacy formats before bailing.
+ static bool warned_once;
+ has_custom_data_global_ = false;
+ if (!warned_once) { // Race is fine: modtrack parsing is global already.
+ WARN("Incorrect module field version %d: attempting to handle legacy format",
+ version);
+ warned_once = true;
+ }
+ // First, see if the user_parse_ is happy:
+ if (user_parse_ != nullptr) {
+ void *user_data;
+ buf = (*user_parse_)(buf, &user_data);
+ if (buf != nullptr) {
+ // Assume legacy format w/ user data but none of our own.
+ custom_module_data_t *custom_data = new custom_module_data_t;
+ custom_data->user_data = user_data;
+ custom_data->contents_size = 0;
+ custom_data->contents = nullptr;
+ *data = custom_data;
+ return buf;
+ }
+ }
+ // Now look for no custom field at all.
+ // If the next field looks like a path, we assume it's the old format with
+ // no user field and we continue w/o vdso data.
+ if (buf[0] == '/' || strstr(buf, "[vdso]") == buf) {
+ *data = nullptr;
+ return buf;
+ }
+ // Else, bail.
+ WARN("Unable to parse module data: custom field mismatch");
+ return nullptr;
+ }
+ buf = skip_comma + 1;
+ skip_comma = strchr(buf, ',');
+ size_t size;
+ if (skip_comma == nullptr || dr_sscanf(buf, "%zu,", &size) != 1)
+ return nullptr; // error
+ custom_module_data_t *custom_data = new custom_module_data_t;
+ custom_data->contents_size = size;
+ buf = skip_comma + 1;
+ if (custom_data->contents_size == 0)
+ custom_data->contents = nullptr;
+ else {
+ custom_data->contents = buf;
+ buf += custom_data->contents_size;
+ }
+ if (user_parse_ != nullptr)
+ buf = (*user_parse_)(buf, &custom_data->user_data);
+ *data = custom_data;
+ return buf;
+}
+
+int
+module_mapper_t::print_custom_module_data(void *data, char *dst, size_t max_len)
+{
+ custom_module_data_t *custom_data = (custom_module_data_t *)data;
+ return print_module_data_fields(dst, max_len, custom_data->contents,
+ custom_data->contents_size, user_print_,
+ custom_data->user_data);
+}
+
+void
+module_mapper_t::free_custom_module_data(void *data)
+{
+ custom_module_data_t *custom_data = (custom_module_data_t *)data;
+ if (user_free_ != nullptr)
+ (*user_free_)(custom_data->user_data);
+ delete custom_data;
+}
+
+const char *(*module_mapper_t::user_parse_)(const char *src,
+ DR_PARAM_OUT void **data) = nullptr;
+void (*module_mapper_t::user_free_)(void *data) = nullptr;
+int (*module_mapper_t::user_print_)(void *data, char *dst, size_t max_len) = nullptr;
+bool module_mapper_t::has_custom_data_global_ = true;
+
+module_mapper_t::module_mapper_t(
+ const char *module_map,
+ const char *(*parse_cb)(const char *src, DR_PARAM_OUT void **data),
+ std::string (*process_cb)(drmodtrack_info_t *info, void *data, void *user_data),
+ void *process_cb_user_data, void (*free_cb)(void *data), uint verbosity,
+ const std::string &alt_module_dir, file_t encoding_file)
+ : modmap_(module_map)
+ , cached_user_free_(free_cb)
+ , verbosity_(verbosity)
+ , alt_module_dir_(alt_module_dir)
+ , encoding_file_(encoding_file)
+{
+ // We mutate global state because do_module_parsing() uses drmodtrack, which
+ // wants global functions. The state isn't needed past do_module_parsing(), so
+ // we make sure to reset it afterwards.
+ DR_ASSERT(user_parse_ == nullptr);
+ DR_ASSERT(user_free_ == nullptr);
+ DR_ASSERT(user_print_ == nullptr);
+
+ user_parse_ = parse_cb;
+ user_process_ = process_cb;
+ user_process_data_ = process_cb_user_data;
+ user_free_ = free_cb;
+ // has_custom_data_global_ is potentially mutated in parse_custom_module_data.
+ // It is assumed to be set to 'true' initially.
+ has_custom_data_global_ = true;
+
+ if (modmap_ != nullptr)
+ last_error_ = do_module_parsing();
+ if (encoding_file_ != INVALID_FILE)
+ last_error_ += do_encoding_parsing();
+
+ // capture has_custom_data_global_'s value for this instance.
+ has_custom_data_ = has_custom_data_global_;
+
+ user_parse_ = nullptr;
+ user_free_ = nullptr;
+}
+
+module_mapper_t::~module_mapper_t()
+{
+ // update user_free_
+ user_free_ = cached_user_free_;
+ // drmodtrack_offline_exit requires the parameter to be non-null, but we
+ // may not have even initialized the modhandle yet.
+ if (modhandle_ != nullptr &&
+ drmodtrack_offline_exit(modhandle_) != DRCOVLIB_SUCCESS) {
+ WARN("Failed to clean up module table data");
+ }
+ user_free_ = nullptr;
+ for (std::vector::iterator mvi = modvec_.begin(); mvi != modvec_.end();
+ ++mvi) {
+ if (!mvi->is_external && mvi->map_seg_base != NULL && mvi->total_map_size != 0) {
+ bool ok = dr_unmap_executable_file(mvi->map_seg_base, mvi->total_map_size);
+ if (!ok)
+ WARN("Failed to unmap module %s", mvi->path);
+ }
+ }
+ modhandle_ = nullptr;
+ modvec_.clear();
+}
+
+int
+print_module_data_fields(char *dst, size_t max_len, const void *custom_data,
+ size_t custom_size,
+ int (*user_print_cb)(void *data, char *dst, size_t max_len),
+ void *user_cb_data)
+{
+ char *cur = dst;
+ int len = dr_snprintf(dst, max_len, "v#%d,%zu,", CUSTOM_MODULE_VERSION, custom_size);
+ if (len < 0)
+ return -1;
+ cur += len;
+ if (cur - dst + custom_size > max_len)
+ return -1;
+ if (custom_size > 0) {
+ memcpy(cur, custom_data, custom_size);
+ cur += custom_size;
+ }
+ if (user_print_cb != nullptr) {
+ int res = (*user_print_cb)(user_cb_data, cur, max_len - (cur - dst));
+ if (res == -1)
+ return -1;
+ cur += res;
+ }
+ return (int)(cur - dst);
+}
+
} // namespace drmemtrace
} // namespace dynamorio
diff --git a/clients/drcachesim/tracer/raw2trace_shared.h b/clients/drcachesim/tracer/raw2trace_shared.h
index 1fa67a1e3f1..47952f9e6f5 100644
--- a/clients/drcachesim/tracer/raw2trace_shared.h
+++ b/clients/drcachesim/tracer/raw2trace_shared.h
@@ -42,8 +42,10 @@
*/
#include
+#include