Add line numbers to OProfile. To do this, I added a processDebugLoc()
call to the MachineCodeEmitter interface and made copying the start
line of a function not conditional on whether we're emitting Dwarf
debug information. I'll propagate the processDebugLoc() calls to the
non-X86 targets in a followup patch.
In the long run, it'll probably be better to gather this information
through the DwarfWriter, but the DwarfWriter currently depends on the
AsmPrinter and TargetAsmInfo, and fixing that would be out of the way
for this patch.
There's a bug in OProfile 0.9.4 that makes it ignore line numbers for
addresses above 4G, and a patch fixing it at
http://thread.gmane.org/gmane.linux.oprofile/7634
Sample output:
$ sudo opcontrol --reset; sudo opcontrol --start-daemon; sudo opcontrol --start; `pwd`/Debug/bin/lli fib.bc; sudo opcontrol --stop
Signalling daemon... done
Profiler running.
fib(40) == 165580141
Stopping profiling.
$ opreport -g -d -l `pwd`/Debug/bin/lli|head -60
Overflow stats not available
CPU: Core 2, speed 1998 MHz (estimated)
Counted CPU_CLK_UNHALTED events (Clock cycles when not halted) with a unit mask of 0x00 (Unhalted core cycles) count 100000
vma samples % linenr info image name symbol name
00007f67a30370b0 25489 61.2554 fib.c:24 10946.jo fib_left
00007f67a30370b0 1634 6.4106 fib.c:24
00007f67a30370b1 83 0.3256 fib.c:24
00007f67a30370b9 1997 7.8348 fib.c:24
00007f67a30370c6 2080 8.1604 fib.c:27
00007f67a30370c8 988 3.8762 fib.c:27
00007f67a30370cd 1315 5.1591 fib.c:27
00007f67a30370cf 251 0.9847 fib.c:27
00007f67a30370d3 1191 4.6726 fib.c:27
00007f67a30370d6 975 3.8252 fib.c:27
00007f67a30370db 1010 3.9625 fib.c:27
00007f67a30370dd 242 0.9494 fib.c:27
00007f67a30370e1 2782 10.9145 fib.c:28
00007f67a30370e5 3768 14.7828 fib.c:28
00007f67a30370eb 615 2.4128 (no location information)
00007f67a30370f3 6558 25.7287 (no location information)
00007f67a3037100 15603 37.4973 fib.c:29 10946.jo fib_right
00007f67a3037100 1646 10.5493 fib.c:29
00007f67a3037101 45 0.2884 fib.c:29
00007f67a3037109 2372 15.2022 fib.c:29
00007f67a3037116 2234 14.3178 fib.c:32
00007f67a3037118 612 3.9223 fib.c:32
00007f67a303711d 622 3.9864 fib.c:32
00007f67a303711f 385 2.4675 fib.c:32
00007f67a3037123 404 2.5892 fib.c:32
00007f67a3037126 634 4.0633 fib.c:32
00007f67a303712b 870 5.5759 fib.c:32
00007f67a303712d 62 0.3974 fib.c:32
00007f67a3037131 1848 11.8439 fib.c:33
00007f67a3037135 2840 18.2016 fib.c:33
00007f67a303713a 1 0.0064 fib.c:33
00007f67a303713b 1023 6.5564 (no location information)
00007f67a3037143 5 0.0320 (no location information)
000000000080c1e4 15 0.0360 MachineOperand.h:150 lli llvm::MachineOperand::isReg() const
000000000080c1e4 6 40.0000 MachineOperand.h:150
000000000080c1ec 2 13.3333 MachineOperand.h:150
...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@76102 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/CodeGen/SelectionDAG/SelectionDAGBuild.cpp b/lib/CodeGen/SelectionDAG/SelectionDAGBuild.cpp
index 3c85118..05622f2 100644
--- a/lib/CodeGen/SelectionDAG/SelectionDAGBuild.cpp
+++ b/lib/CodeGen/SelectionDAG/SelectionDAGBuild.cpp
@@ -3918,8 +3918,7 @@
case Intrinsic::dbg_func_start: {
DwarfWriter *DW = DAG.getDwarfWriter();
DbgFuncStartInst &FSI = cast<DbgFuncStartInst>(I);
- if (!isValidDebugInfoIntrinsic(FSI, CodeGenOpt::None) || !DW
- || !DW->ShouldEmitDwarfDebug())
+ if (!isValidDebugInfoIntrinsic(FSI, CodeGenOpt::None))
return 0;
MachineFunction &MF = DAG.getMachineFunction();
@@ -3940,6 +3939,8 @@
// Record the source line.
setCurDebugLoc(ExtractDebugLocation(FSI, MF.getDebugLocInfo()));
+ if (!DW || !DW->ShouldEmitDwarfDebug())
+ return 0;
DebugLocTuple PrevLocTpl = MF.getDebugLocTuple(PrevLoc);
DISubprogram SP(cast<GlobalVariable>(FSI.getSubprogram()));
DICompileUnit CU(PrevLocTpl.CompileUnit);
@@ -3953,7 +3954,9 @@
// This is a beginning of a new function.
MF.setDefaultDebugLoc(ExtractDebugLocation(FSI, MF.getDebugLocInfo()));
-
+
+ if (!DW || !DW->ShouldEmitDwarfDebug())
+ return 0;
// llvm.dbg.func_start also defines beginning of function scope.
DW->RecordRegionStart(cast<GlobalVariable>(FSI.getSubprogram()));
return 0;
diff --git a/lib/ExecutionEngine/JIT/JITEmitter.cpp b/lib/ExecutionEngine/JIT/JITEmitter.cpp
index 63ac6f1..1cc16af 100644
--- a/lib/ExecutionEngine/JIT/JITEmitter.cpp
+++ b/lib/ExecutionEngine/JIT/JITEmitter.cpp
@@ -473,7 +473,12 @@
// CurFn - The llvm function being emitted. Only valid during
// finishFunction().
const Function *CurFn;
-
+
+ /// Information about emitted code, which is passed to the
+ /// JITEventListeners. This is reset in startFunction and used in
+ /// finishFunction.
+ JITEvent_EmittedFunctionDetails EmissionDetails;
+
// CurFnStubUses - For a given Function, a vector of stubs that it
// references. This facilitates the JIT detecting that a stub is no
// longer used, so that it may be deallocated.
@@ -488,6 +493,8 @@
// in the JITResolver's ExternalFnToStubMap.
StringMap<void *> ExtFnStubs;
+ DebugLocTuple PrevDLT;
+
public:
JITEmitter(JIT &jit, JITMemoryManager *JMM) : Resolver(jit), CurFn(0) {
MemMgr = JMM ? JMM : JITMemoryManager::CreateDefaultMemManager();
@@ -567,6 +574,8 @@
/// MachineRelocations that reference external functions by name.
const StringMap<void*> &getExternalFnStubs() const { return ExtFnStubs; }
+ virtual void processDebugLoc(DebugLoc DL);
+
virtual void emitLabel(uint64_t LabelID) {
if (LabelLocations.size() <= LabelID)
LabelLocations.resize((LabelID+1)*2);
@@ -676,6 +685,21 @@
FnRefs.insert(CurFn);
}
+void JITEmitter::processDebugLoc(DebugLoc DL) {
+ if (!DL.isUnknown()) {
+ DebugLocTuple CurDLT = EmissionDetails.MF->getDebugLocTuple(DL);
+
+ if (CurDLT.CompileUnit != 0 && PrevDLT != CurDLT) {
+ JITEvent_EmittedFunctionDetails::LineStart NextLine;
+ NextLine.Address = getCurrentPCValue();
+ NextLine.Loc = DL;
+ EmissionDetails.LineStarts.push_back(NextLine);
+ }
+
+ PrevDLT = CurDLT;
+ }
+}
+
static unsigned GetConstantPoolSizeInBytes(MachineConstantPool *MCP,
const TargetData *TD) {
const std::vector<MachineConstantPoolEntry> &Constants = MCP->getConstants();
@@ -918,6 +942,9 @@
TheJIT->updateGlobalMapping(F.getFunction(), CurBufferPtr);
MBBLocations.clear();
+
+ EmissionDetails.MF = &F;
+ EmissionDetails.LineStarts.clear();
}
bool JITEmitter::finishFunction(MachineFunction &F) {
@@ -1028,9 +1055,8 @@
// Invalidate the icache if necessary.
sys::Memory::InvalidateInstructionCache(FnStart, FnEnd-FnStart);
- JITEvent_EmittedFunctionDetails Details;
TheJIT->NotifyFunctionEmitted(*F.getFunction(), FnStart, FnEnd-FnStart,
- Details);
+ EmissionDetails);
DOUT << "JIT: Finished CodeGen of [" << (void*)FnStart
<< "] Function: " << F.getFunction()->getName()
diff --git a/lib/ExecutionEngine/JIT/OProfileJITEventListener.cpp b/lib/ExecutionEngine/JIT/OProfileJITEventListener.cpp
index 6263a5b..888d83e 100644
--- a/lib/ExecutionEngine/JIT/OProfileJITEventListener.cpp
+++ b/lib/ExecutionEngine/JIT/OProfileJITEventListener.cpp
@@ -18,6 +18,8 @@
#define DEBUG_TYPE "oprofile-jit-event-listener"
#include "llvm/Function.h"
+#include "llvm/Analysis/DebugInfo.h"
+#include "llvm/CodeGen/MachineFunction.h"
#include "llvm/ExecutionEngine/JITEventListener.h"
#include "llvm/Support/Debug.h"
#include "llvm/System/Errno.h"
@@ -64,10 +66,47 @@
}
}
+class FilenameCache {
+ // Holds the filename of each CompileUnit, so that we can pass the
+ // pointer into oprofile. These char*s are freed in the destructor.
+ DenseMap<GlobalVariable*, char*> Filenames;
+ // Used as the scratch space in DICompileUnit::getFilename().
+ std::string TempFilename;
+
+ public:
+ const char* getFilename(GlobalVariable *CompileUnit) {
+ char *&Filename = Filenames[CompileUnit];
+ if (Filename == NULL) {
+ DICompileUnit CU(CompileUnit);
+ Filename = strdup(CU.getFilename(TempFilename).c_str());
+ }
+ return Filename;
+ }
+ ~FilenameCache() {
+ for (DenseMap<GlobalVariable*, char*>::iterator
+ I = Filenames.begin(), E = Filenames.end(); I != E;++I) {
+ free(I->second);
+ }
+ }
+};
+
+static debug_line_info LineStartToOProfileFormat(
+ const MachineFunction &MF, FilenameCache &Filenames,
+ uintptr_t Address, DebugLoc Loc) {
+ debug_line_info Result;
+ Result.vma = Address;
+ const DebugLocTuple& tuple = MF.getDebugLocTuple(Loc);
+ Result.lineno = tuple.Line;
+ Result.filename = Filenames.getFilename(tuple.CompileUnit);
+ DOUT << "Mapping " << reinterpret_cast<void*>(Result.vma) << " to "
+ << Result.filename << ":" << Result.lineno << "\n";
+ return Result;
+}
+
// Adds the just-emitted function to the symbol table.
void OProfileJITEventListener::NotifyFunctionEmitted(
const Function &F, void *FnStart, size_t FnSize,
- const EmittedFunctionDetails &) {
+ const EmittedFunctionDetails &Details) {
const char *const FnName = F.getNameStart();
assert(FnName != 0 && FnStart != 0 && "Bad symbol to add");
if (op_write_native_code(Agent, FnName,
@@ -75,6 +114,35 @@
FnStart, FnSize) == -1) {
DOUT << "Failed to tell OProfile about native function " << FnName
<< " at [" << FnStart << "-" << ((char*)FnStart + FnSize) << "]\n";
+ return;
+ }
+
+ // Now we convert the line number information from the address/DebugLoc format
+ // in Details to the address/filename/lineno format that OProfile expects.
+ // OProfile 0.9.4 (and maybe later versions) has a bug that causes it to
+ // ignore line numbers for addresses above 4G.
+ FilenameCache Filenames;
+ std::vector<debug_line_info> LineInfo;
+ LineInfo.reserve(1 + Details.LineStarts.size());
+ if (!Details.MF->getDefaultDebugLoc().isUnknown()) {
+ LineInfo.push_back(LineStartToOProfileFormat(
+ *Details.MF, Filenames,
+ reinterpret_cast<uintptr_t>(FnStart),
+ Details.MF->getDefaultDebugLoc()));
+ }
+ for (std::vector<EmittedFunctionDetails::LineStart>::const_iterator
+ I = Details.LineStarts.begin(), E = Details.LineStarts.end();
+ I != E; ++I) {
+ LineInfo.push_back(LineStartToOProfileFormat(
+ *Details.MF, Filenames, I->Address, I->Loc));
+ }
+ if (!LineInfo.empty()) {
+ if (op_write_debug_line_info(Agent, FnStart,
+ LineInfo.size(), &*LineInfo.begin()) == -1) {
+ DOUT << "Failed to tell OProfile about line numbers for native function "
+ << FnName << " at [" << FnStart << "-" << ((char*)FnStart + FnSize)
+ << "]\n";
+ }
}
}
diff --git a/lib/Target/X86/X86CodeEmitter.cpp b/lib/Target/X86/X86CodeEmitter.cpp
index 0831449..825c51c 100644
--- a/lib/Target/X86/X86CodeEmitter.cpp
+++ b/lib/Target/X86/X86CodeEmitter.cpp
@@ -465,6 +465,8 @@
const TargetInstrDesc *Desc) {
DOUT << MI;
+ MCE.processDebugLoc(MI.getDebugLoc());
+
unsigned Opcode = Desc->Opcode;
// Emit the lock opcode prefix as needed.
@@ -816,4 +818,3 @@
llvm_unreachable(0);
}
}
-