[clang] [compiler-rt] [XRay] Add support for instrumentation of DSOs on x86_64 (PR #90959)
via llvm-commits
llvm-commits at lists.llvm.org
Fri May 3 04:16:34 PDT 2024
llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT-->
@llvm/pr-subscribers-xray
Author: Sebastian Kreutzer (sebastiankreutzer)
<details>
<summary>Changes</summary>
This PR introduces shared library (DSO) support for XRay based on a revised version of the implementation outlined in [this RFC](https://discourse.llvm.org/t/rfc-upstreaming-dso-instrumentation-support-for-xray/73000).
The feature enables the patching and handling of events from DSOs, supporting both libraries linked at startup or explicitly loaded, e.g. via `dlopen`.
This patch adds the following:
- The `-fxray-enable-shared` flag to enable the feature (turned off by default)
- A small runtime library that is linked into every instrumented DSO, providing position-independent trampolines and code to register with the main XRay runtime
- Changes to the XRay runtime to support management and patching of multiple objects
These changes are fully backward compatible, i.e. running without instrumented DSOs will produce identical traces (in terms of recorded function IDs) to the previous implementation.
Due to my limited ability to test on other architectures, this feature is only implemented and tested with x86_64. Extending support to other architectures is fairly straightforward, requiring only a position-independent implementation of the architecture-specific trampoline implementation (see `compiler-rt/lib/xray/xray_trampoline_x86_64.S` for reference).
This patch does not include any functionality to resolve function IDs from DSOs for the provided logging/tracing modes. These modes still work and will record calls from DSOs, but symbol resolution for these functions in not available. Getting this to work properly requires recording information about the loaded DSOs and should IMO be discussed in a separate RFC, as there are mulitple feasible approaches.
@<!-- -->petrhosek @<!-- -->jplehr
---
Patch is 59.29 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/90959.diff
19 Files Affected:
- (modified) clang/include/clang/Driver/Options.td (+4)
- (modified) clang/include/clang/Driver/XRayArgs.h (+4)
- (modified) clang/lib/Driver/ToolChains/CommonArgs.cpp (+8-4)
- (modified) clang/lib/Driver/XRayArgs.cpp (+7)
- (modified) compiler-rt/cmake/Modules/AllSupportedArchDefs.cmake (+1)
- (modified) compiler-rt/cmake/config-ix.cmake (+4)
- (modified) compiler-rt/include/xray/xray_interface.h (+23)
- (modified) compiler-rt/lib/xray/CMakeLists.txt (+70-3)
- (added) compiler-rt/lib/xray/xray_dso_init.cpp (+62)
- (modified) compiler-rt/lib/xray/xray_init.cpp (+134-24)
- (modified) compiler-rt/lib/xray/xray_interface.cpp (+206-55)
- (modified) compiler-rt/lib/xray/xray_interface_internal.h (+68-15)
- (modified) compiler-rt/lib/xray/xray_trampoline_x86_64.S (+17-7)
- (modified) compiler-rt/lib/xray/xray_x86_64.cpp (+13-10)
- (added) compiler-rt/test/xray/TestCases/Posix/basic-mode-dso.cpp (+47)
- (added) compiler-rt/test/xray/TestCases/Posix/clang-enable-shared.cpp (+14)
- (added) compiler-rt/test/xray/TestCases/Posix/dlopen.cpp (+110)
- (added) compiler-rt/test/xray/TestCases/Posix/patch-premain-dso.cpp (+45)
- (added) compiler-rt/test/xray/TestCases/Posix/patching-unpatching-dso.cpp (+75)
``````````diff
diff --git a/clang/include/clang/Driver/Options.td b/clang/include/clang/Driver/Options.td
index 953f6fc649e621..3e3be5475c0c4d 100644
--- a/clang/include/clang/Driver/Options.td
+++ b/clang/include/clang/Driver/Options.td
@@ -2850,6 +2850,10 @@ def fxray_selected_function_group :
HelpText<"When using -fxray-function-groups, select which group of functions to instrument. Valid range is 0 to fxray-function-groups - 1">,
MarshallingInfoInt<CodeGenOpts<"XRaySelectedFunctionGroup">, "0">;
+def fxray_enable_shared : Flag<["-"], "fxray-enable-shared">, Group<f_Group>, Visibility<[ClangOption, CC1Option]>,
+ HelpText<"Enable shared library instrumentation.">;
+def fno_xray_enable_shared : Flag<["-"], "fno-xray-enable-shared">, Group<f_Group>,
+ Visibility<[ClangOption, CC1Option]>;
defm fine_grained_bitfield_accesses : BoolOption<"f", "fine-grained-bitfield-accesses",
CodeGenOpts<"FineGrainedBitfieldAccesses">, DefaultFalse,
diff --git a/clang/include/clang/Driver/XRayArgs.h b/clang/include/clang/Driver/XRayArgs.h
index bdd3d979547eed..90a21e69586033 100644
--- a/clang/include/clang/Driver/XRayArgs.h
+++ b/clang/include/clang/Driver/XRayArgs.h
@@ -27,6 +27,7 @@ class XRayArgs {
XRayInstrSet InstrumentationBundle;
llvm::opt::Arg *XRayInstrument = nullptr;
bool XRayRT = true;
+ bool XRayEnableShared = false;
public:
/// Parses the XRay arguments from an argument list.
@@ -35,6 +36,9 @@ class XRayArgs {
llvm::opt::ArgStringList &CmdArgs, types::ID InputType) const;
bool needsXRayRt() const { return XRayInstrument && XRayRT; }
+ bool needsXRayDSORt() const {
+ return XRayInstrument && XRayRT && XRayEnableShared;
+ }
llvm::ArrayRef<std::string> modeList() const { return Modes; }
XRayInstrSet instrumentationBundle() const { return InstrumentationBundle; }
};
diff --git a/clang/lib/Driver/ToolChains/CommonArgs.cpp b/clang/lib/Driver/ToolChains/CommonArgs.cpp
index 6796b43a155020..399bf795ce394e 100644
--- a/clang/lib/Driver/ToolChains/CommonArgs.cpp
+++ b/clang/lib/Driver/ToolChains/CommonArgs.cpp
@@ -1520,10 +1520,14 @@ bool tools::addSanitizerRuntimes(const ToolChain &TC, const ArgList &Args,
}
bool tools::addXRayRuntime(const ToolChain&TC, const ArgList &Args, ArgStringList &CmdArgs) {
- if (Args.hasArg(options::OPT_shared))
- return false;
-
- if (TC.getXRayArgs().needsXRayRt()) {
+ if (Args.hasArg(options::OPT_shared)) {
+ if (TC.getXRayArgs().needsXRayDSORt()) {
+ CmdArgs.push_back("-whole-archive");
+ CmdArgs.push_back(TC.getCompilerRTArgString(Args, "xray-dso"));
+ CmdArgs.push_back("-no-whole-archive");
+ return true;
+ }
+ } else if (TC.getXRayArgs().needsXRayRt()) {
CmdArgs.push_back("--whole-archive");
CmdArgs.push_back(TC.getCompilerRTArgString(Args, "xray"));
for (const auto &Mode : TC.getXRayArgs().modeList())
diff --git a/clang/lib/Driver/XRayArgs.cpp b/clang/lib/Driver/XRayArgs.cpp
index 8c5134e2501358..7809cd7ef7c759 100644
--- a/clang/lib/Driver/XRayArgs.cpp
+++ b/clang/lib/Driver/XRayArgs.cpp
@@ -63,6 +63,10 @@ XRayArgs::XRayArgs(const ToolChain &TC, const ArgList &Args) {
<< XRayInstrument->getSpelling() << Triple.str();
}
+ if (Args.hasFlag(options::OPT_fxray_enable_shared,
+ options::OPT_fno_xray_enable_shared, false))
+ XRayEnableShared = true;
+
// Both XRay and -fpatchable-function-entry use
// TargetOpcode::PATCHABLE_FUNCTION_ENTER.
if (Arg *A = Args.getLastArg(options::OPT_fpatchable_function_entry_EQ))
@@ -177,6 +181,9 @@ void XRayArgs::addArgs(const ToolChain &TC, const ArgList &Args,
Args.addOptOutFlag(CmdArgs, options::OPT_fxray_function_index,
options::OPT_fno_xray_function_index);
+ if (XRayEnableShared)
+ CmdArgs.push_back("-fxray-enable-shared");
+
if (const Arg *A =
Args.getLastArg(options::OPT_fxray_instruction_threshold_EQ)) {
int Value;
diff --git a/compiler-rt/cmake/Modules/AllSupportedArchDefs.cmake b/compiler-rt/cmake/Modules/AllSupportedArchDefs.cmake
index 2fe06273a814c7..ea87edceb1b4a6 100644
--- a/compiler-rt/cmake/Modules/AllSupportedArchDefs.cmake
+++ b/compiler-rt/cmake/Modules/AllSupportedArchDefs.cmake
@@ -84,6 +84,7 @@ else()
set(ALL_XRAY_SUPPORTED_ARCH ${X86_64} ${ARM32} ${ARM64} ${MIPS32} ${MIPS64}
powerpc64le ${HEXAGON} ${LOONGARCH64})
endif()
+set(ALL_XRAY_DSO_SUPPORTED_ARCH ${X86_64})
set(ALL_SHADOWCALLSTACK_SUPPORTED_ARCH ${ARM64})
if (UNIX)
diff --git a/compiler-rt/cmake/config-ix.cmake b/compiler-rt/cmake/config-ix.cmake
index ba740af9e1d60f..1b20015576532c 100644
--- a/compiler-rt/cmake/config-ix.cmake
+++ b/compiler-rt/cmake/config-ix.cmake
@@ -656,6 +656,9 @@ if(APPLE)
list_intersect(XRAY_SUPPORTED_ARCH
ALL_XRAY_SUPPORTED_ARCH
SANITIZER_COMMON_SUPPORTED_ARCH)
+ list_intersect(XRAY_DSO_SUPPORTED_ARCH
+ ALL_XRAY_DSO_SUPPORTED_ARCH
+ SANITIZER_COMMON_SUPPORTED_ARCH)
list_intersect(SHADOWCALLSTACK_SUPPORTED_ARCH
ALL_SHADOWCALLSTACK_SUPPORTED_ARCH
SANITIZER_COMMON_SUPPORTED_ARCH)
@@ -689,6 +692,7 @@ else()
filter_available_targets(CFI_SUPPORTED_ARCH ${ALL_CFI_SUPPORTED_ARCH})
filter_available_targets(SCUDO_STANDALONE_SUPPORTED_ARCH ${ALL_SCUDO_STANDALONE_SUPPORTED_ARCH})
filter_available_targets(XRAY_SUPPORTED_ARCH ${ALL_XRAY_SUPPORTED_ARCH})
+ filter_available_targets(XRAY_DSO_SUPPORTED_ARCH ${ALL_XRAY_DSO_SUPPORTED_ARCH})
filter_available_targets(SHADOWCALLSTACK_SUPPORTED_ARCH
${ALL_SHADOWCALLSTACK_SUPPORTED_ARCH})
filter_available_targets(GWP_ASAN_SUPPORTED_ARCH ${ALL_GWP_ASAN_SUPPORTED_ARCH})
diff --git a/compiler-rt/include/xray/xray_interface.h b/compiler-rt/include/xray/xray_interface.h
index 727431c04e4f73..19c0f6b23175d0 100644
--- a/compiler-rt/include/xray/xray_interface.h
+++ b/compiler-rt/include/xray/xray_interface.h
@@ -97,27 +97,50 @@ enum XRayPatchingStatus {
/// for possible result values.
extern XRayPatchingStatus __xray_patch();
+extern XRayPatchingStatus __xray_patch_object(int32_t ObjId);
+
/// Reverses the effect of __xray_patch(). See XRayPatchingStatus for possible
/// result values.
extern XRayPatchingStatus __xray_unpatch();
+extern XRayPatchingStatus __xray_unpatch_object(int32_t ObjId);
+
/// This patches a specific function id. See XRayPatchingStatus for possible
/// result values.
extern XRayPatchingStatus __xray_patch_function(int32_t FuncId);
+extern XRayPatchingStatus __xray_patch_function_in_object(int32_t FuncId,
+ int32_t ObjId);
+
/// This unpatches a specific function id. See XRayPatchingStatus for possible
/// result values.
extern XRayPatchingStatus __xray_unpatch_function(int32_t FuncId);
+extern XRayPatchingStatus __xray_unpatch_function_in_object(int32_t FuncId,
+ int32_t ObjId);
+
/// This function returns the address of the function provided a valid function
/// id. We return 0 if we encounter any error, even if 0 may be a valid function
/// address.
extern uintptr_t __xray_function_address(int32_t FuncId);
+extern uintptr_t __xray_function_address_in_object(int32_t FuncId,
+ int32_t ObjId);
+
/// This function returns the maximum valid function id. Returns 0 if we
/// encounter errors (when there are no instrumented functions, etc.).
extern size_t __xray_max_function_id();
+extern size_t __xray_max_function_id_in_object(int32_t ObjId);
+
+extern size_t __xray_num_objects();
+
+extern int32_t __xray_unpack_function_id(int32_t PackedId);
+
+extern int32_t __xray_unpack_object_id(int32_t PackedId);
+
+extern int32_t __xray_pack_id(int32_t FuncId, int32_t ObjId);
+
/// Initialize the required XRay data structures. This is useful in cases where
/// users want to control precisely when the XRay instrumentation data
/// structures are initialized, for example when the XRay library is built with
diff --git a/compiler-rt/lib/xray/CMakeLists.txt b/compiler-rt/lib/xray/CMakeLists.txt
index cf7b5062aae32d..165e61b6f45762 100644
--- a/compiler-rt/lib/xray/CMakeLists.txt
+++ b/compiler-rt/lib/xray/CMakeLists.txt
@@ -10,6 +10,10 @@ set(XRAY_SOURCES
xray_utils.cpp
)
+set(XRAY_DSO_SOURCES
+ xray_dso_init.cpp
+ )
+
# Implementation files for all XRay modes.
set(XRAY_FDR_MODE_SOURCES
xray_fdr_flags.cpp
@@ -33,6 +37,11 @@ set(x86_64_SOURCES
xray_trampoline_x86_64.S
)
+set(x86_64_DSO_SOURCES
+ xray_trampoline_x86_64.S
+ )
+
+
set(arm_SOURCES
xray_arm.cpp
xray_trampoline_arm.S
@@ -128,10 +137,12 @@ set(XRAY_IMPL_HEADERS
# consumption by tests.
set(XRAY_ALL_SOURCE_FILES
${XRAY_SOURCES}
+ ${XRAY_DSO_SOURCES}
${XRAY_FDR_MODE_SOURCES}
${XRAY_BASIC_MODE_SOURCES}
${XRAY_PROFILING_MODE_SOURCES}
${x86_64_SOURCES}
+ ${x86_64_DSO_SOURCES}
${arm_SOURCES}
${armhf_SOURCES}
${hexagon_SOURCES}
@@ -162,6 +173,9 @@ set(XRAY_CFLAGS
${COMPILER_RT_CXX_CFLAGS})
set(XRAY_COMMON_DEFINITIONS SANITIZER_COMMON_NO_REDEFINE_BUILTINS XRAY_HAS_EXCEPTIONS=1)
+# DSO trampolines need to be compiled with GOT addressing
+set(XRAY_COMMON_DEFINITIONS_DSO ${XRAY_COMMON_DEFINITIONS} XRAY_PIC)
+
# Too many existing bugs, needs cleanup.
append_list_if(COMPILER_RT_HAS_WNO_FORMAT -Wno-format XRAY_CFLAGS)
@@ -201,7 +215,16 @@ if (APPLE)
CFLAGS ${XRAY_CFLAGS}
DEFS ${XRAY_COMMON_DEFINITIONS}
DEPS ${XRAY_DEPS})
+ add_compiler_rt_object_libraries(RTXrayDSO
+ OS ${XRAY_SUPPORTED_OS}
+ ARCHS ${XRAY_DSO_SUPPORTED_ARCH}
+ SOURCES ${XRAY_DSO_SOURCES}
+ ADDITIONAL_HEADERS ${XRAY_IMPL_HEADERS}
+ CFLAGS ${XRAY_CFLAGS}
+ DEFS ${XRAY_COMMON_DEFINITIONS_DSO}
+ DEPS ${XRAY_DEPS})
set(XRAY_RTXRAY_ARCH_LIBS "")
+ set(XRAY_DSO_RTXRAY_ARCH_LIBS "")
foreach(arch ${XRAY_SUPPORTED_ARCH})
if(NOT ${arch} IN_LIST XRAY_SOURCE_ARCHS)
continue()
@@ -215,6 +238,17 @@ if (APPLE)
DEFS ${XRAY_COMMON_DEFINITIONS}
DEPS ${XRAY_DEPS})
list(APPEND XRAY_RTXRAY_ARCH_LIBS RTXray_${arch})
+ if (${arch} IN_LIST XRAY_DSO_SUPPORTED_ARCH)
+ add_compiler_rt_object_libraries(RTXrayDSO_${arch}
+ OS ${XRAY_SUPPORTED_OS}
+ ARCHS ${XRAY_DSO_SUPPORTED_ARCH}
+ SOURCES ${${arch}_DSO_SOURCES}
+ ADDITIONAL_HEADERS ${XRAY_IMPL_HEADERS}
+ CFLAGS ${XRAY_CFLAGS}
+ DEFS ${XRAY_COMMON_DEFINITIONS_DSO}
+ DEPS ${XRAY_DEPS})
+ list(APPEND XRAY_DSO_RTXRAY_ARCH_LIBS RTXrayDSO_${arch})
+ endif()
endforeach()
add_compiler_rt_object_libraries(RTXrayFDR
OS ${XRAY_SUPPORTED_OS}
@@ -252,6 +286,17 @@ if (APPLE)
LINK_FLAGS ${XRAY_LINK_FLAGS} ${WEAK_SYMBOL_LINK_FLAGS}
LINK_LIBS ${XRAY_LINK_LIBS}
PARENT_TARGET xray)
+ add_compiler_rt_runtime(clang_rt.xray-dso
+ STATIC
+ OS ${XRAY_SUPPORTED_OS}
+ ARCHS ${XRAY_DSO_SUPPORTED_ARCH}
+ OBJECT_LIBS RTXrayDSO ${XRAY_DSO_RTXRAY_ARCH_LIBS}
+ CFLAGS ${XRAY_CFLAGS}
+ DEFS ${XRAY_COMMON_DEFINITIONS}
+ LINK_FLAGS ${XRAY_LINK_FLAGS} ${WEAK_SYMBOL_LINK_FLAGS}
+ LINK_LIBS ${XRAY_LINK_LIBS}
+ PARENT_TARGET xray)
+
add_compiler_rt_runtime(clang_rt.xray-fdr
STATIC
OS ${XRAY_SUPPORTED_OS}
@@ -283,6 +328,7 @@ if (APPLE)
LINK_LIBS ${XRAY_LINK_LIBS}
PARENT_TARGET xray)
else() # not Apple
+ message("DSO Supported Archs: ${XRAY_DSO_SUPPORTED_ARCH}")
foreach(arch ${XRAY_SUPPORTED_ARCH})
if(NOT CAN_TARGET_${arch})
continue()
@@ -325,7 +371,7 @@ else() # not Apple
LINK_LIBS ${XRAY_LINK_LIBS}
DEFS ${XRAY_COMMON_DEFINITIONS}
OBJECT_LIBS ${XRAY_COMMON_RUNTIME_OBJECT_LIBS} RTXray
- PARENT_TARGET xray)
+ PARENT_TARGET xray)
# FDR mode runtime archive (addon for clang_rt.xray)
add_compiler_rt_runtime(clang_rt.xray-fdr
STATIC
@@ -346,8 +392,8 @@ else() # not Apple
DEFS ${XRAY_COMMON_DEFINITIONS}
OBJECT_LIBS RTXrayBASIC
PARENT_TARGET xray)
- # Profiler Mode runtime
- add_compiler_rt_runtime(clang_rt.xray-profiling
+ # Profiler Mode runtime
+ add_compiler_rt_runtime(clang_rt.xray-profiling
STATIC
ARCHS ${arch}
CFLAGS ${XRAY_CFLAGS}
@@ -356,6 +402,27 @@ else() # not Apple
DEFS ${XRAY_COMMON_DEFINITIONS}
OBJECT_LIBS RTXrayPROFILING
PARENT_TARGET xray)
+
+ if (${arch} IN_LIST XRAY_DSO_SUPPORTED_ARCH)
+ # TODO: Only implemented for X86 at the moment
+ add_compiler_rt_object_libraries(RTXrayDSO
+ ARCHS ${arch}
+ SOURCES ${XRAY_DSO_SOURCES} ${${arch}_DSO_SOURCES}
+ ADDITIONAL_HEADERS ${XRAY_IMPL_HEADERS}
+ CFLAGS ${XRAY_CFLAGS}
+ DEFS ${XRAY_COMMON_DEFINITIONS_DSO}
+ DEPS ${XRAY_DEPS})
+ # DSO runtime archive
+ add_compiler_rt_runtime(clang_rt.xray-dso
+ STATIC
+ ARCHS ${arch}
+ CFLAGS ${XRAY_CFLAGS}
+ LINK_FLAGS ${XRAY_LINK_FLAGS}
+ LINK_LIBS ${XRAY_LINK_LIBS}
+ DEFS ${XRAY_COMMON_DEFINITIONS}
+ OBJECT_LIBS RTXrayDSO
+ PARENT_TARGET xray)
+ endif()
endforeach()
endif() # not Apple
diff --git a/compiler-rt/lib/xray/xray_dso_init.cpp b/compiler-rt/lib/xray/xray_dso_init.cpp
new file mode 100644
index 00000000000000..ad5f91f3f2448f
--- /dev/null
+++ b/compiler-rt/lib/xray/xray_dso_init.cpp
@@ -0,0 +1,62 @@
+//===-- xray_init.cpp -------------------------------------------*- C++ -*-===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+// This file is a part of XRay, a dynamic runtime instrumentation system.
+//
+// XRay initialisation logic for DSOs.
+//===----------------------------------------------------------------------===//
+
+#include "sanitizer_common/sanitizer_atomic.h"
+#include "xray_defs.h"
+#include "xray_flags.h"
+#include "xray_interface_internal.h"
+
+using namespace __sanitizer;
+
+extern "C" {
+extern const XRaySledEntry __start_xray_instr_map[] __attribute__((weak))
+__attribute__((visibility("hidden")));
+extern const XRaySledEntry __stop_xray_instr_map[] __attribute__((weak))
+__attribute__((visibility("hidden")));
+extern const XRayFunctionSledIndex __start_xray_fn_idx[] __attribute__((weak))
+__attribute__((visibility("hidden")));
+extern const XRayFunctionSledIndex __stop_xray_fn_idx[] __attribute__((weak))
+__attribute__((visibility("hidden")));
+
+#if SANITIZER_MAC
+// HACK: This is a temporary workaround to make XRay build on
+// Darwin, but it will probably not work at runtime.
+extern const XRaySledEntry __start_xray_instr_map[] = {};
+extern const XRaySledEntry __stop_xray_instr_map[] = {};
+extern const XRayFunctionSledIndex __start_xray_fn_idx[] = {};
+extern const XRayFunctionSledIndex __stop_xray_fn_idx[] = {};
+#endif
+}
+
+// Handler functions to call in the patched entry/exit sled.
+extern atomic_uintptr_t XRayPatchedFunction;
+extern atomic_uintptr_t XRayArgLogger;
+extern atomic_uintptr_t XRayPatchedCustomEvent;
+extern atomic_uintptr_t XRayPatchedTypedEvent;
+
+static int __xray_object_id{-1};
+
+// Note: .preinit_array initialization does not work for DSOs
+__attribute__((constructor(0))) static void
+__xray_init_dso() XRAY_NEVER_INSTRUMENT {
+ // Register sleds in main XRay runtime.
+ __xray_object_id =
+ __xray_register_dso(__start_xray_instr_map, __stop_xray_instr_map,
+ __start_xray_fn_idx, __stop_xray_fn_idx, {});
+}
+
+__attribute__((destructor(0))) static void
+__xray_finalize_dso() XRAY_NEVER_INSTRUMENT {
+ // Inform the main runtime that this DSO is no longer used.
+ __xray_deregister_dso(__xray_object_id);
+}
diff --git a/compiler-rt/lib/xray/xray_init.cpp b/compiler-rt/lib/xray/xray_init.cpp
index f22a31b95686d0..c3570d97701eee 100644
--- a/compiler-rt/lib/xray/xray_init.cpp
+++ b/compiler-rt/lib/xray/xray_init.cpp
@@ -16,6 +16,8 @@
#include <unistd.h>
#include "sanitizer_common/sanitizer_common.h"
+#include "xray/xray_interface.h"
+#include "xray_allocator.h"
#include "xray_defs.h"
#include "xray_flags.h"
#include "xray_interface_internal.h"
@@ -28,7 +30,7 @@ extern const XRayFunctionSledIndex __start_xray_fn_idx[] __attribute__((weak));
extern const XRayFunctionSledIndex __stop_xray_fn_idx[] __attribute__((weak));
#if SANITIZER_APPLE
-// HACK: This is a temporary workaround to make XRay build on
+// HACK: This is a temporary workaround to make XRay build on
// Darwin, but it will probably not work at runtime.
const XRaySledEntry __start_xray_instr_map[] = {};
extern const XRaySledEntry __stop_xray_instr_map[] = {};
@@ -50,7 +52,11 @@ atomic_uint8_t XRayInitialized{0};
// This should always be updated before XRayInitialized is updated.
SpinMutex XRayInstrMapMutex;
-XRaySledMap XRayInstrMap;
+// XRaySledMap XRayInstrMap;
+// Contains maps for the main executable as well as DSOs.
+// std::vector<XRaySledMap> XRayInstrMaps;
+XRaySledMap *XRayInstrMaps;
+atomic_uint32_t XRayNumObjects;
// Global flag to determine whether the flags have been initialized.
atomic_uint8_t XRayFlagsInitialized{0};
@@ -58,6 +64,60 @@ atomic_uint8_t XRayFlagsInitialized{0};
// A mutex to allow only one thread to initialize the XRay data structures.
SpinMutex XRayInitMutex;
+int32_t
+__xray_register_sleds(const XRaySledEntry *SledsBegin,
+ const XRaySledEntry *SledsEnd,
+ const XRayFunctionSledIndex *FnIndexBegin,
+ const XRayFunctionSledIndex *FnIndexEnd, bool FromDSO,
+ XRayTrampolines Trampolines) XRAY_NEVER_INSTRUMENT {
+ if (!SledsBegin || !SledsEnd) {
+ return -1;
+ }
+ XRaySledMap SledMap;
+ SledMap.FromDSO = FromDSO;
+ SledMap.Loaded = true;
+ SledMap.Trampolines = Trampolines;
+ SledMap.Sleds = SledsBegin;
+ SledMap.Entries = SledsEnd - SledsBegin;
+ if (FnIndexBegin != nullptr) {
+ SledMap.SledsIndex = FnIndexBegin;
+ SledMap.Functions = FnIndexEnd - FnIndexBegin;
+ } else {
+ size_t CountFunctions = 0;
+ uint64_t LastFnAddr = 0;
+
+ for (std::size_t I = 0; I < SledMap.Entries; I++) {
+ const auto &Sled = SledMap.Sleds[I];
+ const auto Function = Sled.function();
+ if (Function != LastFnAddr) {
+ CountFunctions++;
+ LastFnAddr = Function;
+ }
+ }
+
+ SledMap.Functions = CountFunctions;
+ }
+ if (SledMap.Functions >= XRayMaxFunctions) {
+ Report("Too many functions! Maximum is %ld\n", XRayMaxFunctions);
+ return -1;
+ }
+
+ if (Verbosity()) {
+ Report("Registering %d new functions!\n", SledMap.Functions);
+ }
+
+ {
+ SpinMutexLock Guard(&XRayInstrMapMutex);
+ auto Idx = atomic_fetch_add(&XRayNumObjects, 1, memory_order_acq_rel);
+ if (Idx >= XRayMaxObjects) {
+ Report("Too many objects registered! Maximum is %ld\n", XRayMaxObjects);
+ return -1;
+ }
+ XRayInstrMaps[Idx] = std::move(SledMap);
+ return Idx;
+ }
+}
+
// __xray_init() will do the actual loading of the current process' memory map
// and then proceed to look for the .xray_instr_map section/segment.
void __xray_init() XRAY_NEVER_INSTRUMENT {
@@ -80,29 +140,14 @@ void __xray_init() XRAY_NEVER_INSTRUMENT {
return;
}
- {
- SpinMutexLock Guard(&XRayInstrMapMutex);
- XRayInstrMap.Sleds = __start_xray_instr_map;
- XRayInstrMap.Entries = __stop_xray_instr_map - __start_xray_instr_map;
- if (__start_xray_fn_idx != nullptr) {
- XRayInstrMap.SledsIndex = __start_xray_fn_idx;
- XRayInstrMap.Functions = __stop_xray_fn_idx - __start_xray_fn_idx;
- } else {
- size_t CountFunctions = 0;
- uint64_t LastFnAddr = 0;
-
- for (std::size_t I = 0; I < XRayInstrMap.Entries; I++) {
- const auto &Sled = XRayInstrMap.Sleds[I];
- const auto Function = Sled.function();
- if (Function != LastFnAddr) {
- CountFunctions++;
- LastFnAddr = Function;
- }
- }
+ atomic_store(&XRayNumObjects, 0, memory_order_release);
+
+ // Pre-allocation takes up approx. 5kB for XRayMaxObjects=64.
+ XRayInstrMaps = allocateBuffer<XRaySledMap>(XRayMaxObjects);
+
+ __xray_register_sleds(__start_xray_instr_map, __stop_xray_instr_map,
+ __start_xray_fn_idx, __stop_xray_fn_idx, false, {});
- XRayInstrMap.Functions = CountFunctions;
- }
- }
atomic_...
[truncated]
``````````
</details>
https://github.com/llvm/llvm-project/pull/90959
More information about the llvm-commits
mailing list