[llvm] r220996 - R600: Make sure to inline all internal functions
Reid Kleckner
rnk at google.com
Fri Oct 31 15:13:09 PDT 2014
This include from lib/Target/R600:
#include "llvm/Transforms/Utils/Cloning.h"
You probably want to figure something else out.
On Fri, Oct 31, 2014 at 2:39 PM, Duncan P. N. Exon Smith <
dexonsmith at apple.com> wrote:
> llc (among others) won't link for me in cmake, even after r220998.
>
> [862/869] Linking CXX executable bin/llc
> FAILED: : && /usr/bin/clang++ -fPIC -fvisibility-inlines-hidden -Wall -W
> -Wno-unused-parameter -Wwrite-strings -Wmissing-field-initializers
> -pedantic -Wno-long-long -Wcovered-switch-default -std=c++11
> -fcolor-diagnostics -g -Wl,-search_paths_first
> -Wl,-headerpad_max_install_names tools/llc/CMakeFiles/llc.dir/llc.cpp.o
> -o bin/llc lib/libLLVMAArch64CodeGen.a lib/libLLVMAArch64AsmPrinter.a
> lib/libLLVMAArch64AsmParser.a lib/libLLVMAArch64Desc.a
> lib/libLLVMAArch64Info.a lib/libLLVMAArch64Disassembler.a
> lib/libLLVMARMCodeGen.a lib/libLLVMARMAsmPrinter.a
> lib/libLLVMARMAsmParser.a lib/libLLVMARMDesc.a lib/libLLVMARMInfo.a
> lib/libLLVMARMDisassembler.a lib/libLLVMCppBackendCodeGen.a
> lib/libLLVMCppBackendInfo.a lib/libLLVMHexagonCodeGen.a
> lib/libLLVMHexagonDesc.a lib/libLLVMHexagonInfo.a
> lib/libLLVMHexagonDisassembler.a lib/libLLVMMipsCodeGen.a
> lib/libLLVMMipsAsmPrinter.a lib/libLLVMMipsAsmParser.a
> lib/libLLVMMipsDesc.a lib/libLLVMMipsInfo.a lib/libLLVMMipsDisassembler.a
> lib/libLLV!
> MMSP430CodeGen.a lib/libLLVMMSP430AsmPrinter.a lib/libLLVMMSP430Desc.a
> lib/libLLVMMSP430Info.a lib/libLLVMNVPTXCodeGen.a
> lib/libLLVMNVPTXAsmPrinter.a lib/libLLVMNVPTXDesc.a lib/libLLVMNVPTXInfo.a
> lib/libLLVMPowerPCCodeGen.a lib/libLLVMPowerPCAsmPrinter.a
> lib/libLLVMPowerPCAsmParser.a lib/libLLVMPowerPCDesc.a
> lib/libLLVMPowerPCInfo.a lib/libLLVMPowerPCDisassembler.a
> lib/libLLVMR600CodeGen.a lib/libLLVMR600AsmPrinter.a lib/libLLVMR600Desc.a
> lib/libLLVMR600Info.a lib/libLLVMSparcCodeGen.a
> lib/libLLVMSparcAsmPrinter.a lib/libLLVMSparcAsmParser.a
> lib/libLLVMSparcDesc.a lib/libLLVMSparcInfo.a
> lib/libLLVMSparcDisassembler.a lib/libLLVMSystemZCodeGen.a
> lib/libLLVMSystemZAsmPrinter.a lib/libLLVMSystemZAsmParser.a
> lib/libLLVMSystemZDesc.a lib/libLLVMSystemZInfo.a
> lib/libLLVMSystemZDisassembler.a lib/libLLVMX86CodeGen.a
> lib/libLLVMX86AsmPrinter.a lib/libLLVMX86AsmParser.a lib/libLLVMX86Desc.a
> lib/libLLVMX86Info.a lib/libLLVMX86Disassembler.a lib/libLLVMXCoreCodeGen.a
> lib/libLLVMXCoreA!
> smPrinter.a lib/libLLVMXCoreDesc.a lib/libLLVMXCoreInfo.a
> lib/libLLVMXCoreDisassembler.a lib/libLLVMAsmPrinter.a lib/libLLVMCodeGen.a
> lib/libLLVMCore.a lib/libLLVMIRReader.a lib/libLLVMMC.a
> lib/libLLVMScalarOpts.a lib/libLLVMSelectionDAG.a lib/libLLVMSupport.a
> lib/libLLVMTarget.a lib/libLLVMAArch64AsmPrinter.a lib/libLLVMAArch64Info.a
> lib/libLLVMAArch64Utils.a lib/libLLVMARMDesc.a lib/libLLVMARMAsmPrinter.a
> lib/libLLVMARMInfo.a lib/libLLVMHexagonInfo.a lib/libLLVMMipsAsmPrinter.a
> lib/libLLVMMipsInfo.a lib/libLLVMMSP430AsmPrinter.a
> lib/libLLVMNVPTXAsmPrinter.a lib/libLLVMPowerPCAsmPrinter.a
> lib/libLLVMPowerPCInfo.a lib/libLLVMR600AsmPrinter.a
> lib/libLLVMSparcAsmPrinter.a lib/libLLVMSparcInfo.a
> lib/libLLVMSystemZDesc.a lib/libLLVMSystemZAsmPrinter.a
> lib/libLLVMSystemZInfo.a lib/libLLVMX86CodeGen.a lib/libLLVMX86Desc.a
> lib/libLLVMX86AsmPrinter.a lib/libLLVMX86Utils.a lib/libLLVMX86Info.a
> lib/libLLVMXCoreAsmPrinter.a lib/libLLVMAsmPrinter.a
> lib/libLLVMSelectionDAG.a lib/libLLVMCodeGen.a lib/libLLVMScalarOpts.a
> lib/libLLVMInstCombine.a lib/libLLVMP!
> rofileData.a lib/libLLVMObject.a lib/libLLVMMCParser.a
> lib/libLLVMTransformUtils.a lib/libLLVMipa.a lib/libLLVMAnalysis.a
> lib/libLLVMTarget.a lib/libLLVMXCoreInfo.a lib/libLLVMMCDisassembler.a
> lib/libLLVMMC.a lib/libLLVMAsmParser.a lib/libLLVMBitReader.a
> lib/libLLVMCore.a lib/libLLVMSupport.a -lcurses -lpthread -lz && :
> Undefined symbols for architecture x86_64:
> "llvm::createBarrierNoopPass()", referenced from:
> (anonymous namespace)::AMDGPUPassConfig::addIRPasses() in
> libLLVMR600CodeGen.a(AMDGPUTargetMachine.cpp.o)
> "llvm::createAlwaysInlinerPass()", referenced from:
> (anonymous namespace)::AMDGPUPassConfig::addIRPasses() in
> libLLVMR600CodeGen.a(AMDGPUTargetMachine.cpp.o)
> ld: symbol(s) not found for architecture x86_64
> clang: error: linker command failed with exit code 1 (use -v to see
> invocation)
>
> Can you have a look?
>
> > On 2014-Oct-31, at 13:52, Tom Stellard <thomas.stellard at amd.com> wrote:
> >
> > Author: tstellar
> > Date: Fri Oct 31 15:52:02 2014
> > New Revision: 220996
> >
> > URL: http://llvm.org/viewvc/llvm-project?rev=220996&view=rev
> > Log:
> > R600: Make sure to inline all internal functions
> >
> > Function calls aren't supported yet.
> >
> > Added:
> > llvm/trunk/lib/Target/R600/AMDGPUAlwaysInlinePass.cpp
> > llvm/trunk/test/CodeGen/R600/inline-calls.ll
> > Modified:
> > llvm/trunk/lib/Target/R600/AMDGPU.h
> > llvm/trunk/lib/Target/R600/AMDGPUTargetMachine.cpp
> > llvm/trunk/test/CodeGen/R600/call.ll
> >
> > Modified: llvm/trunk/lib/Target/R600/AMDGPU.h
> > URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/R600/AMDGPU.h?rev=220996&r1=220995&r2=220996&view=diff
> >
> ==============================================================================
> > --- llvm/trunk/lib/Target/R600/AMDGPU.h (original)
> > +++ llvm/trunk/lib/Target/R600/AMDGPU.h Fri Oct 31 15:52:02 2014
> > @@ -57,6 +57,7 @@ extern char &SILoadStoreOptimizerID;
> > FunctionPass *createAMDGPUPromoteAlloca(const AMDGPUSubtarget &ST);
> > Pass *createAMDGPUStructurizeCFGPass();
> > FunctionPass *createAMDGPUISelDag(TargetMachine &tm);
> > +ModulePass *createAMDGPUAlwaysInlinePass();
> >
> > /// \brief Creates an AMDGPU-specific Target Transformation Info pass.
> > ImmutablePass *
> >
> > Added: llvm/trunk/lib/Target/R600/AMDGPUAlwaysInlinePass.cpp
> > URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/R600/AMDGPUAlwaysInlinePass.cpp?rev=220996&view=auto
> >
> ==============================================================================
> > --- llvm/trunk/lib/Target/R600/AMDGPUAlwaysInlinePass.cpp (added)
> > +++ llvm/trunk/lib/Target/R600/AMDGPUAlwaysInlinePass.cpp Fri Oct 31
> 15:52:02 2014
> > @@ -0,0 +1,66 @@
> > +//===-- AMDGPUAlwaysInlinePass.cpp - Promote Allocas
> ----------------------===//
> > +//
> > +// The LLVM Compiler Infrastructure
> > +//
> > +// This file is distributed under the University of Illinois Open Source
> > +// License. See LICENSE.TXT for details.
> > +//
> >
> +//===----------------------------------------------------------------------===//
> > +//
> > +/// \file
> > +/// This pass marks all internal functions as always_inline and creates
> > +/// duplicates of all other functions a marks the duplicates as
> always_inline.
> > +//
> >
> +//===----------------------------------------------------------------------===//
> > +
> > +#include "AMDGPU.h"
> > +#include "llvm/IR/Module.h"
> > +#include "llvm/Transforms/Utils/Cloning.h"
> > +
> > +using namespace llvm;
> > +
> > +namespace {
> > +
> > +class AMDGPUAlwaysInline : public ModulePass {
> > +
> > + static char ID;
> > +
> > +public:
> > + AMDGPUAlwaysInline() : ModulePass(ID) { }
> > + bool runOnModule(Module &M) override;
> > + const char *getPassName() const override { return "AMDGPU Always
> Inline Pass"; }
> > +};
> > +
> > +} // End anonymous namespace
> > +
> > +char AMDGPUAlwaysInline::ID = 0;
> > +
> > +bool AMDGPUAlwaysInline::runOnModule(Module &M) {
> > +
> > + std::vector<Function*> FuncsToClone;
> > + for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I) {
> > + Function &F = *I;
> > + if (!F.hasLocalLinkage() && !F.isDeclaration() && !F.use_empty())
> > + FuncsToClone.push_back(&F);
> > + }
> > +
> > + for (Function *F : FuncsToClone) {
> > + ValueToValueMapTy VMap;
> > + Function *NewFunc = CloneFunction(F, VMap, false);
> > + NewFunc->setLinkage(GlobalValue::InternalLinkage);
> > + F->getParent()->getFunctionList().push_back(NewFunc);
> > + F->replaceAllUsesWith(NewFunc);
> > + }
> > +
> > + for (Module::iterator I = M.begin(), E = M.end(); I != E; ++I) {
> > + Function &F = *I;
> > + if (F.hasLocalLinkage()) {
> > + F.addFnAttr(Attribute::AlwaysInline);
> > + }
> > + }
> > + return false;
> > +}
> > +
> > +ModulePass *llvm::createAMDGPUAlwaysInlinePass() {
> > + return new AMDGPUAlwaysInline();
> > +}
> >
> > Modified: llvm/trunk/lib/Target/R600/AMDGPUTargetMachine.cpp
> > URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/lib/Target/R600/AMDGPUTargetMachine.cpp?rev=220996&r1=220995&r2=220996&view=diff
> >
> ==============================================================================
> > --- llvm/trunk/lib/Target/R600/AMDGPUTargetMachine.cpp (original)
> > +++ llvm/trunk/lib/Target/R600/AMDGPUTargetMachine.cpp Fri Oct 31
> 15:52:02 2014
> > @@ -80,6 +80,7 @@ public:
> > return nullptr;
> > }
> >
> > + void addIRPasses() override;
> > void addCodeGenPrepare() override;
> > bool addPreISel() override;
> > bool addInstSelector() override;
> > @@ -106,6 +107,19 @@ void AMDGPUTargetMachine::addAnalysisPas
> > PM.add(createAMDGPUTargetTransformInfoPass(this));
> > }
> >
> > +void AMDGPUPassConfig::addIRPasses() {
> > + // Function calls are not supported, so make sure we inline
> everything.
> > + addPass(createAMDGPUAlwaysInlinePass());
> > + addPass(createAlwaysInlinerPass());
> > + // We need to add the barrier noop pass, otherwise adding the function
> > + // inlining pass will cause all of the PassConfigs passes to be run
> > + // one function at a time, which means if we have a nodule with two
> > + // functions, then we will generate code for the first function
> > + // without ever running any passes on the second.
> > + addPass(createBarrierNoopPass());
> > + TargetPassConfig::addIRPasses();
> > +}
> > +
> > void AMDGPUPassConfig::addCodeGenPrepare() {
> > const AMDGPUSubtarget &ST = TM->getSubtarget<AMDGPUSubtarget>();
> > if (ST.isPromoteAllocaEnabled()) {
> >
> > Modified: llvm/trunk/test/CodeGen/R600/call.ll
> > URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/R600/call.ll?rev=220996&r1=220995&r2=220996&view=diff
> >
> ==============================================================================
> > --- llvm/trunk/test/CodeGen/R600/call.ll (original)
> > +++ llvm/trunk/test/CodeGen/R600/call.ll Fri Oct 31 15:52:02 2014
> > @@ -1,7 +1,7 @@
> > ; RUN: not llc -march=r600 -mcpu=SI -verify-machineinstrs< %s 2>&1 |
> FileCheck %s
> > ; RUN: not llc -march=r600 -mcpu=cypress < %s 2>&1 | FileCheck %s
> >
> > -; CHECK: error: unsupported call to function defined_function in
> test_call
> > +; CHECK: error: unsupported call to function external_function in
> test_call_external
> >
> >
> > declare i32 @external_function(i32) nounwind
> >
> > Added: llvm/trunk/test/CodeGen/R600/inline-calls.ll
> > URL:
> http://llvm.org/viewvc/llvm-project/llvm/trunk/test/CodeGen/R600/inline-calls.ll?rev=220996&view=auto
> >
> ==============================================================================
> > --- llvm/trunk/test/CodeGen/R600/inline-calls.ll (added)
> > +++ llvm/trunk/test/CodeGen/R600/inline-calls.ll Fri Oct 31 15:52:02 2014
> > @@ -0,0 +1,24 @@
> > +; RUN: llc -march=r600 -mcpu=SI -verify-machineinstrs < %s | FileCheck
> %s
> > +; RUN: llc -march=r600 -mcpu=redwood -verify-machineinstrs < %s |
> FileCheck %s
> > +
> > +; CHECK-NOT: {{^}}func:
> > +define internal fastcc i32 @func(i32 %a) {
> > +entry:
> > + %tmp0 = add i32 %a, 1
> > + ret i32 %tmp0
> > +}
> > +
> > +; CHECK: {{^}}kernel:
> > +define void @kernel(i32 addrspace(1)* %out) {
> > +entry:
> > + %tmp0 = call i32 @func(i32 1)
> > + store i32 %tmp0, i32 addrspace(1)* %out
> > + ret void
> > +}
> > +
> > +; CHECK: {{^}}kernel2:
> > +define void @kernel2(i32 addrspace(1)* %out) {
> > +entry:
> > + call void @kernel(i32 addrspace(1)* %out)
> > + ret void
> > +}
> >
> >
> > _______________________________________________
> > llvm-commits mailing list
> > llvm-commits at cs.uiuc.edu
> > http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
>
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20141031/31d8560e/attachment.html>
More information about the llvm-commits
mailing list