r348930 - [CUDA][OPENMP][NVPTX]Improve logic of the debug info support.
Alexey Bataev via cfe-commits
cfe-commits at lists.llvm.org
Wed Dec 12 06:52:27 PST 2018
Author: abataev
Date: Wed Dec 12 06:52:27 2018
New Revision: 348930
URL: http://llvm.org/viewvc/llvm-project?rev=348930&view=rev
Log:
[CUDA][OPENMP][NVPTX]Improve logic of the debug info support.
Summary:
Added support for the -gline-directives-only option + fixed logic of the
debug info for CUDA devices. If optimization level is O0, then options
--[no-]cuda-noopt-device-debug do not affect the debug info level. If
the optimization level is >O0, debug info options are used +
--no-cuda-noopt-device-debug is used or no --cuda-noopt-device-debug is
used, the optimization level for the device code is kept and the
emission of the debug directives is used.
If the opt level is > O0, debug info is requested +
--cuda-noopt-device-debug option is used, the optimization is disabled
for the device code + required debug info is emitted.
Reviewers: tra, echristo
Subscribers: aprantl, guansong, JDevlieghere, cfe-commits
Differential Revision: https://reviews.llvm.org/D51554
Modified:
cfe/trunk/include/clang/Driver/ToolChain.h
cfe/trunk/lib/Driver/ToolChains/Clang.cpp
cfe/trunk/lib/Driver/ToolChains/Cuda.cpp
cfe/trunk/lib/Driver/ToolChains/Cuda.h
cfe/trunk/test/Driver/cuda-dwarf-2.cu
cfe/trunk/test/Driver/openmp-offload-gpu.c
Modified: cfe/trunk/include/clang/Driver/ToolChain.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Driver/ToolChain.h?rev=348930&r1=348929&r2=348930&view=diff
==============================================================================
--- cfe/trunk/include/clang/Driver/ToolChain.h (original)
+++ cfe/trunk/include/clang/Driver/ToolChain.h Wed Dec 12 06:52:27 2018
@@ -10,6 +10,7 @@
#ifndef LLVM_CLANG_DRIVER_TOOLCHAIN_H
#define LLVM_CLANG_DRIVER_TOOLCHAIN_H
+#include "clang/Basic/DebugInfoOptions.h"
#include "clang/Basic/LLVM.h"
#include "clang/Basic/Sanitizers.h"
#include "clang/Basic/DebugInfoOptions.h"
@@ -435,6 +436,10 @@ public:
return true;
}
+ /// Adjust debug information kind considering all passed options.
+ virtual void adjustDebugInfoKind(codegenoptions::DebugInfoKind &DebugInfoKind,
+ const llvm::opt::ArgList &Args) const {}
+
/// GetExceptionModel - Return the tool chain exception model.
virtual llvm::ExceptionHandling
GetExceptionModel(const llvm::opt::ArgList &Args) const;
Modified: cfe/trunk/lib/Driver/ToolChains/Clang.cpp
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Driver/ToolChains/Clang.cpp?rev=348930&r1=348929&r2=348930&view=diff
==============================================================================
--- cfe/trunk/lib/Driver/ToolChains/Clang.cpp (original)
+++ cfe/trunk/lib/Driver/ToolChains/Clang.cpp Wed Dec 12 06:52:27 2018
@@ -3233,6 +3233,9 @@ static void RenderDebugOptions(const Too
}
}
+ // Adjust the debug info kind for the given toolchain.
+ TC.adjustDebugInfoKind(DebugInfoKind, Args);
+
RenderDebugEnablingArgs(Args, CmdArgs, DebugInfoKind, DWARFVersion,
DebuggerTuning);
Modified: cfe/trunk/lib/Driver/ToolChains/Cuda.cpp
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Driver/ToolChains/Cuda.cpp?rev=348930&r1=348929&r2=348930&view=diff
==============================================================================
--- cfe/trunk/lib/Driver/ToolChains/Cuda.cpp (original)
+++ cfe/trunk/lib/Driver/ToolChains/Cuda.cpp Wed Dec 12 06:52:27 2018
@@ -278,32 +278,44 @@ void CudaInstallationDetector::print(raw
}
namespace {
- /// Debug info kind.
-enum DebugInfoKind {
- NoDebug, /// No debug info.
- LineTableOnly, /// Line tables only.
- FullDebug /// Full debug info.
+/// Debug info level for the NVPTX devices. We may need to emit different debug
+/// info level for the host and for the device itselfi. This type controls
+/// emission of the debug info for the devices. It either prohibits disable info
+/// emission completely, or emits debug directives only, or emits same debug
+/// info as for the host.
+enum DeviceDebugInfoLevel {
+ DisableDebugInfo, /// Do not emit debug info for the devices.
+ DebugDirectivesOnly, /// Emit only debug directives.
+ EmitSameDebugInfoAsHost, /// Use the same debug info level just like for the
+ /// host.
};
} // anonymous namespace
-static DebugInfoKind mustEmitDebugInfo(const ArgList &Args) {
- Arg *A = Args.getLastArg(options::OPT_O_Group);
- if (Args.hasFlag(options::OPT_cuda_noopt_device_debug,
- options::OPT_no_cuda_noopt_device_debug,
- !A || A->getOption().matches(options::OPT_O0))) {
- if (const Arg *A = Args.getLastArg(options::OPT_g_Group)) {
- const Option &Opt = A->getOption();
- if (Opt.matches(options::OPT_gN_Group)) {
- if (Opt.matches(options::OPT_g0) || Opt.matches(options::OPT_ggdb0))
- return NoDebug;
- if (Opt.matches(options::OPT_gline_tables_only) ||
- Opt.matches(options::OPT_ggdb1))
- return LineTableOnly;
- }
- return FullDebug;
+/// Define debug info level for the NVPTX devices. If the debug info for both
+/// the host and device are disabled (-g0/-ggdb0 or no debug options at all). If
+/// only debug directives are requested for the both host and device
+/// (-gline-directvies-only), or the debug info only for the device is disabled
+/// (optimization is on and --cuda-noopt-device-debug was not specified), the
+/// debug directves only must be emitted for the device. Otherwise, use the same
+/// debug info level just like for the host (with the limitations of only
+/// supported DWARF2 standard).
+static DeviceDebugInfoLevel mustEmitDebugInfo(const ArgList &Args) {
+ const Arg *A = Args.getLastArg(options::OPT_O_Group);
+ bool IsDebugEnabled = !A || A->getOption().matches(options::OPT_O0) ||
+ Args.hasFlag(options::OPT_cuda_noopt_device_debug,
+ options::OPT_no_cuda_noopt_device_debug,
+ /*Default=*/false);
+ if (const Arg *A = Args.getLastArg(options::OPT_g_Group)) {
+ const Option &Opt = A->getOption();
+ if (Opt.matches(options::OPT_gN_Group)) {
+ if (Opt.matches(options::OPT_g0) || Opt.matches(options::OPT_ggdb0))
+ return DisableDebugInfo;
+ if (Opt.matches(options::OPT_gline_directives_only))
+ return DebugDirectivesOnly;
}
+ return IsDebugEnabled ? EmitSameDebugInfoAsHost : DebugDirectivesOnly;
}
- return NoDebug;
+ return DisableDebugInfo;
}
void NVPTX::Assembler::ConstructJob(Compilation &C, const JobAction &JA,
@@ -337,8 +349,8 @@ void NVPTX::Assembler::ConstructJob(Comp
ArgStringList CmdArgs;
CmdArgs.push_back(TC.getTriple().isArch64Bit() ? "-m64" : "-m32");
- DebugInfoKind DIKind = mustEmitDebugInfo(Args);
- if (DIKind == FullDebug) {
+ DeviceDebugInfoLevel DIKind = mustEmitDebugInfo(Args);
+ if (DIKind == EmitSameDebugInfoAsHost) {
// ptxas does not accept -g option if optimization is enabled, so
// we ignore the compiler's -O* options if we want debug info.
CmdArgs.push_back("-g");
@@ -374,7 +386,7 @@ void NVPTX::Assembler::ConstructJob(Comp
// to no optimizations, but ptxas's default is -O3.
CmdArgs.push_back("-O0");
}
- if (DIKind == LineTableOnly)
+ if (DIKind == DebugDirectivesOnly)
CmdArgs.push_back("-lineinfo");
// Pass -v to ptxas if it was passed to the driver.
@@ -445,7 +457,7 @@ void NVPTX::Linker::ConstructJob(Compila
CmdArgs.push_back(TC.getTriple().isArch64Bit() ? "-64" : "-32");
CmdArgs.push_back(Args.MakeArgString("--create"));
CmdArgs.push_back(Args.MakeArgString(Output.getFilename()));
- if (mustEmitDebugInfo(Args) == FullDebug)
+ if (mustEmitDebugInfo(Args) == EmitSameDebugInfoAsHost)
CmdArgs.push_back("-g");
for (const auto& II : Inputs) {
@@ -498,7 +510,7 @@ void NVPTX::OpenMPLinker::ConstructJob(C
CmdArgs.push_back(Output.getFilename());
} else
assert(Output.isNothing() && "Invalid output.");
- if (mustEmitDebugInfo(Args) == FullDebug)
+ if (mustEmitDebugInfo(Args) == EmitSameDebugInfoAsHost)
CmdArgs.push_back("-g");
if (Args.hasArg(options::OPT_v))
@@ -704,6 +716,21 @@ bool CudaToolChain::supportsDebugInfoOpt
O.matches(options::OPT_gcolumn_info);
}
+void CudaToolChain::adjustDebugInfoKind(
+ codegenoptions::DebugInfoKind &DebugInfoKind, const ArgList &Args) const {
+ switch (mustEmitDebugInfo(Args)) {
+ case DisableDebugInfo:
+ DebugInfoKind = codegenoptions::NoDebugInfo;
+ break;
+ case DebugDirectivesOnly:
+ DebugInfoKind = codegenoptions::DebugDirectivesOnly;
+ break;
+ case EmitSameDebugInfoAsHost:
+ // Use same debug info level as the host.
+ break;
+ }
+}
+
void CudaToolChain::AddCudaIncludeArgs(const ArgList &DriverArgs,
ArgStringList &CC1Args) const {
// Check our CUDA version if we're going to include the CUDA headers.
Modified: cfe/trunk/lib/Driver/ToolChains/Cuda.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Driver/ToolChains/Cuda.h?rev=348930&r1=348929&r2=348930&view=diff
==============================================================================
--- cfe/trunk/lib/Driver/ToolChains/Cuda.h (original)
+++ cfe/trunk/lib/Driver/ToolChains/Cuda.h Wed Dec 12 06:52:27 2018
@@ -159,6 +159,8 @@ public:
bool isPICDefaultForced() const override { return false; }
bool SupportsProfiling() const override { return false; }
bool supportsDebugInfoOption(const llvm::opt::Arg *A) const override;
+ void adjustDebugInfoKind(codegenoptions::DebugInfoKind &DebugInfoKind,
+ const llvm::opt::ArgList &Args) const override;
bool IsMathErrnoDefault() const override { return false; }
void AddCudaIncludeArgs(const llvm::opt::ArgList &DriverArgs,
Modified: cfe/trunk/test/Driver/cuda-dwarf-2.cu
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/Driver/cuda-dwarf-2.cu?rev=348930&r1=348929&r2=348930&view=diff
==============================================================================
--- cfe/trunk/test/Driver/cuda-dwarf-2.cu (original)
+++ cfe/trunk/test/Driver/cuda-dwarf-2.cu Wed Dec 12 06:52:27 2018
@@ -1,25 +1,28 @@
// REQUIRES: clang-driver
//
-// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -g -O0 --no-cuda-noopt-device-debug 2>&1 | \
-// RUN: FileCheck %s -check-prefix NO_DEBUG
+// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -g -O1 --no-cuda-noopt-device-debug 2>&1 | \
+// RUN: FileCheck %s -check-prefix DEBUG_DIRECTIVES
// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -g -O3 2>&1 | \
-// RUN: FileCheck %s -check-prefix NO_DEBUG
+// RUN: FileCheck %s -check-prefix DEBUG_DIRECTIVES
// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -g -O3 --no-cuda-noopt-device-debug 2>&1 | \
-// RUN: FileCheck %s -check-prefix NO_DEBUG
+// RUN: FileCheck %s -check-prefix DEBUG_DIRECTIVES
// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -g0 2>&1 | \
// RUN: FileCheck %s -check-prefix NO_DEBUG
// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -ggdb0 -O3 --cuda-noopt-device-debug 2>&1 | \
// RUN: FileCheck %s -check-prefix NO_DEBUG
-// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -ggdb1 2>&1 | \
-// RUN: FileCheck %s -check-prefix NO_DEBUG -check-prefix LINE_TABLE
-// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -gline-tables-only -O2 --cuda-noopt-device-debug 2>&1 | \
-// RUN: FileCheck %s -check-prefix NO_DEBUG -check-prefix LINE_TABLE
+// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -gline-directives-only -O2 --cuda-noopt-device-debug 2>&1 | \
+// RUN: FileCheck %s -check-prefix DEBUG_DIRECTIVES
// NO_DEBUG-NOT: warning: debug
-// LINE_TABLE-NOT: warning: debug
+// DEBUG_DIRECTIVES-NOT: warning: debug
+// NO_DEBUG: "-fcuda-is-device"
+// NO_DEBUG-NOT: "-debug-info-kind=
// NO_DEBUG: ptxas
// NO_DEBUG-NOT: "-g"
-// LINE_TABLE: "-lineinfo"
+// DEBUG_DIRECTIVES: "-fcuda-is-device"
+// DEBUG_DIRECTIVES-SAME: "-debug-info-kind=line-directives-only"
+// DEBUG_DIRECTIVES: ptxas
+// DEBUG_DIRECTIVES-SAME: "-lineinfo"
// NO_DEBUG: fatbinary
// NO_DEBUG-NOT: "-g"
@@ -27,6 +30,8 @@
// RUN: FileCheck %s -check-prefix HAS_DEBUG
// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -g -O0 --cuda-noopt-device-debug 2>&1 | \
// RUN: FileCheck %s -check-prefix HAS_DEBUG
+// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -g -O0 --no-cuda-noopt-device-debug 2>&1 | \
+// RUN: FileCheck %s -check-prefix HAS_DEBUG
// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -g -O3 --cuda-noopt-device-debug 2>&1 | \
// RUN: FileCheck %s -check-prefix HAS_DEBUG
// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -g2 2>&1 | \
@@ -37,9 +42,14 @@
// RUN: FileCheck %s -check-prefix HAS_DEBUG
// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -ggdb3 -O3 --cuda-noopt-device-debug 2>&1 | \
// RUN: FileCheck %s -check-prefix HAS_DEBUG
+// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -ggdb1 2>&1 | \
+// RUN: FileCheck %s -check-prefix HAS_DEBUG
+// RUN: %clang -### -target x86_64-linux-gnu -c --cuda-gpu-arch=sm_20 %s -gline-tables-only -O2 --cuda-noopt-device-debug 2>&1 | \
+// RUN: FileCheck %s -check-prefix HAS_DEBUG
// HAS_DEBUG-NOT: warning: debug
// HAS_DEBUG: "-fcuda-is-device"
+// HAS_DEBUG-SAME: "-debug-info-kind={{limited|line-tables-only}}"
// HAS_DEBUG-SAME: "-dwarf-version=2"
// HAS_DEBUG: ptxas
// HAS_DEBUG-SAME: "-g"
Modified: cfe/trunk/test/Driver/openmp-offload-gpu.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/Driver/openmp-offload-gpu.c?rev=348930&r1=348929&r2=348930&view=diff
==============================================================================
--- cfe/trunk/test/Driver/openmp-offload-gpu.c (original)
+++ cfe/trunk/test/Driver/openmp-offload-gpu.c Wed Dec 12 06:52:27 2018
@@ -190,29 +190,35 @@
// CHK-BCLIB-WARN: No library 'libomptarget-nvptx-sm_20.bc' found in the default clang lib directory or in LIBRARY_PATH. Expect degraded performance due to no inlining of runtime functions on target devices.
/// Check that debug info is emitted in dwarf-2
-// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O0 --no-cuda-noopt-device-debug 2>&1 \
-// RUN: | FileCheck -check-prefix=NO_DEBUG %s
+// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O1 --no-cuda-noopt-device-debug 2>&1 \
+// RUN: | FileCheck -check-prefix=DEBUG_DIRECTIVES %s
// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O3 2>&1 \
-// RUN: | FileCheck -check-prefix=NO_DEBUG %s
+// RUN: | FileCheck -check-prefix=DEBUG_DIRECTIVES %s
// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O3 --no-cuda-noopt-device-debug 2>&1 \
-// RUN: | FileCheck -check-prefix=NO_DEBUG %s
+// RUN: | FileCheck -check-prefix=DEBUG_DIRECTIVES %s
// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g0 2>&1 \
// RUN: | FileCheck -check-prefix=NO_DEBUG %s
// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -ggdb0 -O3 --cuda-noopt-device-debug 2>&1 \
// RUN: | FileCheck -check-prefix=NO_DEBUG %s
-// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -gline-tables-only 2>&1 \
-// RUN: | FileCheck -check-prefix=NO_DEBUG -check-prefix=LINE_TABLE %s
-// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -ggdb1 -O2 --cuda-noopt-device-debug 2>&1 \
-// RUN: | FileCheck -check-prefix=NO_DEBUG -check-prefix=LINE_TABLE %s
+// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -gline-directives-only 2>&1 \
+// RUN: | FileCheck -check-prefix=DEBUG_DIRECTIVES %s
-// LINE_TABLE-NOT: warning: debug
+// DEBUG_DIRECTIVES-NOT: warning: debug
// NO_DEBUG-NOT: warning: debug
+// NO_DEBUG: "-fopenmp-is-device"
+// NO_DEBUG-NOT: "-debug-info-kind=
// NO_DEBUG: ptxas
-// LINE_TABLE: "-lineinfo"
+// DEBUG_DIRECTIVES: "-triple" "nvptx64-nvidia-cuda"
+// DEBUG_DIRECTIVES-SAME: "-debug-info-kind=line-directives-only"
+// DEBUG_DIRECTIVES-SAME: "-fopenmp-is-device"
+// DEBUG_DIRECTIVES: ptxas
+// DEBUG_DIRECTIVES: "-lineinfo"
// NO_DEBUG-NOT: "-g"
// NO_DEBUG: nvlink
// NO_DEBUG-NOT: "-g"
+// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O0 --no-cuda-noopt-device-debug 2>&1 \
+// RUN: | FileCheck -check-prefix=HAS_DEBUG %s
// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g 2>&1 \
// RUN: | FileCheck -check-prefix=HAS_DEBUG %s
// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -g -O0 --cuda-noopt-device-debug 2>&1 \
@@ -227,9 +233,14 @@
// RUN: | FileCheck -check-prefix=HAS_DEBUG %s
// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -ggdb3 -O2 --cuda-noopt-device-debug 2>&1 \
// RUN: | FileCheck -check-prefix=HAS_DEBUG %s
+// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -gline-tables-only 2>&1 \
+// RUN: | FileCheck -check-prefix=HAS_DEBUG %s
+// RUN: %clang -### -no-canonical-prefixes -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target -march=sm_60 %s -ggdb1 -O2 --cuda-noopt-device-debug 2>&1 \
+// RUN: | FileCheck -check-prefix=HAS_DEBUG %s
// HAS_DEBUG-NOT: warning: debug
// HAS_DEBUG: "-triple" "nvptx64-nvidia-cuda"
+// HAS_DEBUG-SAME: "-debug-info-kind={{limited|line-tables-only}}"
// HAS_DEBUG-SAME: "-dwarf-version=2"
// HAS_DEBUG-SAME: "-fopenmp-is-device"
// HAS_DEBUG: ptxas
More information about the cfe-commits
mailing list