170f5bc99SReid Kleckner //===-- llvm/lib/CodeGen/AsmPrinter/CodeViewDebug.cpp --*- C++ -*--===//
270f5bc99SReid Kleckner //
370f5bc99SReid Kleckner //                     The LLVM Compiler Infrastructure
470f5bc99SReid Kleckner //
570f5bc99SReid Kleckner // This file is distributed under the University of Illinois Open Source
670f5bc99SReid Kleckner // License. See LICENSE.TXT for details.
770f5bc99SReid Kleckner //
870f5bc99SReid Kleckner //===----------------------------------------------------------------------===//
970f5bc99SReid Kleckner //
1070f5bc99SReid Kleckner // This file contains support for writing Microsoft CodeView debug info.
1170f5bc99SReid Kleckner //
1270f5bc99SReid Kleckner //===----------------------------------------------------------------------===//
1370f5bc99SReid Kleckner 
1470f5bc99SReid Kleckner #include "CodeViewDebug.h"
1570f5bc99SReid Kleckner #include "llvm/DebugInfo/CodeView/CodeView.h"
16*2214ed89SReid Kleckner #include "llvm/DebugInfo/CodeView/Line.h"
1770f5bc99SReid Kleckner #include "llvm/DebugInfo/CodeView/SymbolRecord.h"
1870f5bc99SReid Kleckner #include "llvm/MC/MCExpr.h"
1970f5bc99SReid Kleckner #include "llvm/MC/MCSymbol.h"
2070f5bc99SReid Kleckner #include "llvm/Support/COFF.h"
2170f5bc99SReid Kleckner 
2270f5bc99SReid Kleckner using namespace llvm::codeview;
2370f5bc99SReid Kleckner 
2470f5bc99SReid Kleckner namespace llvm {
2570f5bc99SReid Kleckner 
269533af4fSReid Kleckner StringRef CodeViewDebug::getFullFilepath(const DIFile *File) {
279533af4fSReid Kleckner   std::string &Filepath = FileToFilepathMap[File];
2870f5bc99SReid Kleckner   if (!Filepath.empty())
2970f5bc99SReid Kleckner     return Filepath;
3070f5bc99SReid Kleckner 
319533af4fSReid Kleckner   StringRef Dir = File->getDirectory(), Filename = File->getFilename();
329533af4fSReid Kleckner 
3370f5bc99SReid Kleckner   // Clang emits directory and relative filename info into the IR, but CodeView
3470f5bc99SReid Kleckner   // operates on full paths.  We could change Clang to emit full paths too, but
3570f5bc99SReid Kleckner   // that would increase the IR size and probably not needed for other users.
3670f5bc99SReid Kleckner   // For now, just concatenate and canonicalize the path here.
3770f5bc99SReid Kleckner   if (Filename.find(':') == 1)
3870f5bc99SReid Kleckner     Filepath = Filename;
3970f5bc99SReid Kleckner   else
4070f5bc99SReid Kleckner     Filepath = (Dir + "\\" + Filename).str();
4170f5bc99SReid Kleckner 
4270f5bc99SReid Kleckner   // Canonicalize the path.  We have to do it textually because we may no longer
4370f5bc99SReid Kleckner   // have access the file in the filesystem.
4470f5bc99SReid Kleckner   // First, replace all slashes with backslashes.
4570f5bc99SReid Kleckner   std::replace(Filepath.begin(), Filepath.end(), '/', '\\');
4670f5bc99SReid Kleckner 
4770f5bc99SReid Kleckner   // Remove all "\.\" with "\".
4870f5bc99SReid Kleckner   size_t Cursor = 0;
4970f5bc99SReid Kleckner   while ((Cursor = Filepath.find("\\.\\", Cursor)) != std::string::npos)
5070f5bc99SReid Kleckner     Filepath.erase(Cursor, 2);
5170f5bc99SReid Kleckner 
5270f5bc99SReid Kleckner   // Replace all "\XXX\..\" with "\".  Don't try too hard though as the original
5370f5bc99SReid Kleckner   // path should be well-formatted, e.g. start with a drive letter, etc.
5470f5bc99SReid Kleckner   Cursor = 0;
5570f5bc99SReid Kleckner   while ((Cursor = Filepath.find("\\..\\", Cursor)) != std::string::npos) {
5670f5bc99SReid Kleckner     // Something's wrong if the path starts with "\..\", abort.
5770f5bc99SReid Kleckner     if (Cursor == 0)
5870f5bc99SReid Kleckner       break;
5970f5bc99SReid Kleckner 
6070f5bc99SReid Kleckner     size_t PrevSlash = Filepath.rfind('\\', Cursor - 1);
6170f5bc99SReid Kleckner     if (PrevSlash == std::string::npos)
6270f5bc99SReid Kleckner       // Something's wrong, abort.
6370f5bc99SReid Kleckner       break;
6470f5bc99SReid Kleckner 
6570f5bc99SReid Kleckner     Filepath.erase(PrevSlash, Cursor + 3 - PrevSlash);
6670f5bc99SReid Kleckner     // The next ".." might be following the one we've just erased.
6770f5bc99SReid Kleckner     Cursor = PrevSlash;
6870f5bc99SReid Kleckner   }
6970f5bc99SReid Kleckner 
7070f5bc99SReid Kleckner   // Remove all duplicate backslashes.
7170f5bc99SReid Kleckner   Cursor = 0;
7270f5bc99SReid Kleckner   while ((Cursor = Filepath.find("\\\\", Cursor)) != std::string::npos)
7370f5bc99SReid Kleckner     Filepath.erase(Cursor, 1);
7470f5bc99SReid Kleckner 
7570f5bc99SReid Kleckner   return Filepath;
7670f5bc99SReid Kleckner }
7770f5bc99SReid Kleckner 
78*2214ed89SReid Kleckner unsigned CodeViewDebug::maybeRecordFile(const DIFile *F) {
79*2214ed89SReid Kleckner   unsigned NextId = FileIdMap.size() + 1;
80*2214ed89SReid Kleckner   auto Insertion = FileIdMap.insert(std::make_pair(F, NextId));
81*2214ed89SReid Kleckner   if (Insertion.second) {
82*2214ed89SReid Kleckner     // We have to compute the full filepath and emit a .cv_file directive.
83*2214ed89SReid Kleckner     StringRef FullPath = getFullFilepath(F);
84*2214ed89SReid Kleckner     NextId = Asm->OutStreamer->EmitCVFileDirective(NextId, FullPath);
85*2214ed89SReid Kleckner     assert(NextId == FileIdMap.size() && ".cv_file directive failed");
86*2214ed89SReid Kleckner   }
87*2214ed89SReid Kleckner   return Insertion.first->second;
88*2214ed89SReid Kleckner }
89*2214ed89SReid Kleckner 
9070f5bc99SReid Kleckner void CodeViewDebug::maybeRecordLocation(DebugLoc DL,
9170f5bc99SReid Kleckner                                         const MachineFunction *MF) {
929533af4fSReid Kleckner   // Skip this instruction if it has the same location as the previous one.
939533af4fSReid Kleckner   if (DL == CurFn->LastLoc)
949533af4fSReid Kleckner     return;
959533af4fSReid Kleckner 
969533af4fSReid Kleckner   const DIScope *Scope = DL.get()->getScope();
9770f5bc99SReid Kleckner   if (!Scope)
9870f5bc99SReid Kleckner     return;
999533af4fSReid Kleckner 
10070f5bc99SReid Kleckner   // Skip this line if it is longer than the maximum we can record.
101*2214ed89SReid Kleckner   LineInfo LI(DL.getLine(), DL.getLine(), /*IsStatement=*/true);
102*2214ed89SReid Kleckner   if (LI.getStartLine() != DL.getLine() || LI.isAlwaysStepInto() ||
103*2214ed89SReid Kleckner       LI.isNeverStepInto())
10470f5bc99SReid Kleckner     return;
10570f5bc99SReid Kleckner 
106*2214ed89SReid Kleckner   ColumnInfo CI(DL.getCol(), /*EndColumn=*/0);
107*2214ed89SReid Kleckner   if (CI.getStartColumn() != DL.getCol())
108*2214ed89SReid Kleckner     return;
10900d9639cSReid Kleckner 
110*2214ed89SReid Kleckner   if (!CurFn->HaveLineInfo)
111*2214ed89SReid Kleckner     CurFn->HaveLineInfo = true;
112*2214ed89SReid Kleckner   unsigned FileId = 0;
113*2214ed89SReid Kleckner   if (CurFn->LastLoc.get() && CurFn->LastLoc->getFile() == DL->getFile())
114*2214ed89SReid Kleckner     FileId = CurFn->LastFileId;
115*2214ed89SReid Kleckner   else
116*2214ed89SReid Kleckner     FileId = CurFn->LastFileId = maybeRecordFile(DL->getFile());
117*2214ed89SReid Kleckner   CurFn->LastLoc = DL;
118*2214ed89SReid Kleckner   Asm->OutStreamer->EmitCVLocDirective(CurFn->FuncId, FileId, DL.getLine(),
119*2214ed89SReid Kleckner                                        DL.getCol(), /*PrologueEnd=*/false,
120*2214ed89SReid Kleckner                                        /*IsStmt=*/false, DL->getFilename());
12170f5bc99SReid Kleckner }
12270f5bc99SReid Kleckner 
12370f5bc99SReid Kleckner CodeViewDebug::CodeViewDebug(AsmPrinter *AP)
12470f5bc99SReid Kleckner     : Asm(nullptr), CurFn(nullptr) {
12570f5bc99SReid Kleckner   MachineModuleInfo *MMI = AP->MMI;
12670f5bc99SReid Kleckner 
12770f5bc99SReid Kleckner   // If module doesn't have named metadata anchors or COFF debug section
12870f5bc99SReid Kleckner   // is not available, skip any debug info related stuff.
12970f5bc99SReid Kleckner   if (!MMI->getModule()->getNamedMetadata("llvm.dbg.cu") ||
13070f5bc99SReid Kleckner       !AP->getObjFileLowering().getCOFFDebugSymbolsSection())
13170f5bc99SReid Kleckner     return;
13270f5bc99SReid Kleckner 
13370f5bc99SReid Kleckner   // Tell MMI that we have debug info.
13470f5bc99SReid Kleckner   MMI->setDebugInfoAvailability(true);
13570f5bc99SReid Kleckner   Asm = AP;
13670f5bc99SReid Kleckner }
13770f5bc99SReid Kleckner 
13870f5bc99SReid Kleckner void CodeViewDebug::endModule() {
13970f5bc99SReid Kleckner   if (FnDebugInfo.empty())
14070f5bc99SReid Kleckner     return;
14170f5bc99SReid Kleckner 
14270f5bc99SReid Kleckner   // FIXME: For functions that are comdat, we should emit separate .debug$S
14370f5bc99SReid Kleckner   // sections that are comdat associative with the main function instead of
14470f5bc99SReid Kleckner   // having one big .debug$S section.
14570f5bc99SReid Kleckner   assert(Asm != nullptr);
14670f5bc99SReid Kleckner   Asm->OutStreamer->SwitchSection(
14770f5bc99SReid Kleckner       Asm->getObjFileLowering().getCOFFDebugSymbolsSection());
14870f5bc99SReid Kleckner   Asm->EmitInt32(COFF::DEBUG_SECTION_MAGIC);
14970f5bc99SReid Kleckner 
15070f5bc99SReid Kleckner   // The COFF .debug$S section consists of several subsections, each starting
15170f5bc99SReid Kleckner   // with a 4-byte control code (e.g. 0xF1, 0xF2, etc) and then a 4-byte length
15270f5bc99SReid Kleckner   // of the payload followed by the payload itself.  The subsections are 4-byte
15370f5bc99SReid Kleckner   // aligned.
15470f5bc99SReid Kleckner 
155*2214ed89SReid Kleckner   // Emit per-function debug information.
156*2214ed89SReid Kleckner   for (auto &P : FnDebugInfo)
157*2214ed89SReid Kleckner     emitDebugInfoForFunction(P.first, P.second);
15870f5bc99SReid Kleckner 
15970f5bc99SReid Kleckner   // This subsection holds a file index to offset in string table table.
16070f5bc99SReid Kleckner   Asm->OutStreamer->AddComment("File index to string table offset subsection");
161*2214ed89SReid Kleckner   Asm->OutStreamer->EmitCVFileChecksumsDirective();
16270f5bc99SReid Kleckner 
16370f5bc99SReid Kleckner   // This subsection holds the string table.
16470f5bc99SReid Kleckner   Asm->OutStreamer->AddComment("String table");
165*2214ed89SReid Kleckner   Asm->OutStreamer->EmitCVStringTableDirective();
16670f5bc99SReid Kleckner 
16770f5bc99SReid Kleckner   clear();
16870f5bc99SReid Kleckner }
16970f5bc99SReid Kleckner 
17070f5bc99SReid Kleckner static void EmitLabelDiff(MCStreamer &Streamer,
17170f5bc99SReid Kleckner                           const MCSymbol *From, const MCSymbol *To,
17270f5bc99SReid Kleckner                           unsigned int Size = 4) {
17370f5bc99SReid Kleckner   MCSymbolRefExpr::VariantKind Variant = MCSymbolRefExpr::VK_None;
17470f5bc99SReid Kleckner   MCContext &Context = Streamer.getContext();
17570f5bc99SReid Kleckner   const MCExpr *FromRef = MCSymbolRefExpr::create(From, Variant, Context),
17670f5bc99SReid Kleckner                *ToRef   = MCSymbolRefExpr::create(To, Variant, Context);
17770f5bc99SReid Kleckner   const MCExpr *AddrDelta =
17870f5bc99SReid Kleckner       MCBinaryExpr::create(MCBinaryExpr::Sub, ToRef, FromRef, Context);
17970f5bc99SReid Kleckner   Streamer.EmitValue(AddrDelta, Size);
18070f5bc99SReid Kleckner }
18170f5bc99SReid Kleckner 
182*2214ed89SReid Kleckner void CodeViewDebug::emitDebugInfoForFunction(const Function *GV,
183*2214ed89SReid Kleckner                                              FunctionInfo &FI) {
18470f5bc99SReid Kleckner   // For each function there is a separate subsection
18570f5bc99SReid Kleckner   // which holds the PC to file:line table.
18670f5bc99SReid Kleckner   const MCSymbol *Fn = Asm->getSymbol(GV);
18770f5bc99SReid Kleckner   assert(Fn);
18870f5bc99SReid Kleckner 
18970f5bc99SReid Kleckner   StringRef FuncName;
19070f5bc99SReid Kleckner   if (auto *SP = getDISubprogram(GV))
19170f5bc99SReid Kleckner     FuncName = SP->getDisplayName();
19270f5bc99SReid Kleckner 
19370f5bc99SReid Kleckner   // If our DISubprogram name is empty, use the mangled name.
19470f5bc99SReid Kleckner   if (FuncName.empty())
19570f5bc99SReid Kleckner     FuncName = GlobalValue::getRealLinkageName(GV->getName());
19670f5bc99SReid Kleckner 
19770f5bc99SReid Kleckner   // Emit a symbol subsection, required by VS2012+ to find function boundaries.
19870f5bc99SReid Kleckner   MCSymbol *SymbolsBegin = Asm->MMI->getContext().createTempSymbol(),
19970f5bc99SReid Kleckner            *SymbolsEnd = Asm->MMI->getContext().createTempSymbol();
20070f5bc99SReid Kleckner   Asm->OutStreamer->AddComment("Symbol subsection for " + Twine(FuncName));
20170f5bc99SReid Kleckner   Asm->EmitInt32(unsigned(ModuleSubstreamKind::Symbols));
20270f5bc99SReid Kleckner   EmitLabelDiff(*Asm->OutStreamer, SymbolsBegin, SymbolsEnd);
20370f5bc99SReid Kleckner   Asm->OutStreamer->EmitLabel(SymbolsBegin);
20470f5bc99SReid Kleckner   {
20570f5bc99SReid Kleckner     MCSymbol *ProcSegmentBegin = Asm->MMI->getContext().createTempSymbol(),
20670f5bc99SReid Kleckner              *ProcSegmentEnd = Asm->MMI->getContext().createTempSymbol();
20770f5bc99SReid Kleckner     EmitLabelDiff(*Asm->OutStreamer, ProcSegmentBegin, ProcSegmentEnd, 2);
20870f5bc99SReid Kleckner     Asm->OutStreamer->EmitLabel(ProcSegmentBegin);
20970f5bc99SReid Kleckner 
21070f5bc99SReid Kleckner     Asm->EmitInt16(unsigned(SymbolRecordKind::S_GPROC32_ID));
21170f5bc99SReid Kleckner 
21270f5bc99SReid Kleckner     // Some bytes of this segment don't seem to be required for basic debugging,
21370f5bc99SReid Kleckner     // so just fill them with zeroes.
21470f5bc99SReid Kleckner     Asm->OutStreamer->EmitFill(12, 0);
21570f5bc99SReid Kleckner     // This is the important bit that tells the debugger where the function
21670f5bc99SReid Kleckner     // code is located and what's its size:
21770f5bc99SReid Kleckner     EmitLabelDiff(*Asm->OutStreamer, Fn, FI.End);
21870f5bc99SReid Kleckner     Asm->OutStreamer->EmitFill(12, 0);
21970f5bc99SReid Kleckner     Asm->OutStreamer->EmitCOFFSecRel32(Fn);
22070f5bc99SReid Kleckner     Asm->OutStreamer->EmitCOFFSectionIndex(Fn);
22170f5bc99SReid Kleckner     Asm->EmitInt8(0);
22270f5bc99SReid Kleckner     // Emit the function display name as a null-terminated string.
22370f5bc99SReid Kleckner     Asm->OutStreamer->EmitBytes(FuncName);
22470f5bc99SReid Kleckner     Asm->EmitInt8(0);
22570f5bc99SReid Kleckner     Asm->OutStreamer->EmitLabel(ProcSegmentEnd);
22670f5bc99SReid Kleckner 
22770f5bc99SReid Kleckner     // We're done with this function.
22870f5bc99SReid Kleckner     Asm->EmitInt16(0x0002);
22970f5bc99SReid Kleckner     Asm->EmitInt16(unsigned(SymbolRecordKind::S_PROC_ID_END));
23070f5bc99SReid Kleckner   }
23170f5bc99SReid Kleckner   Asm->OutStreamer->EmitLabel(SymbolsEnd);
23270f5bc99SReid Kleckner   // Every subsection must be aligned to a 4-byte boundary.
23370f5bc99SReid Kleckner   Asm->OutStreamer->EmitFill((-FuncName.size()) % 4, 0);
23470f5bc99SReid Kleckner 
235*2214ed89SReid Kleckner   // We have an assembler directive that takes care of the whole line table.
236*2214ed89SReid Kleckner   Asm->OutStreamer->EmitCVLinetableDirective(FI.FuncId, Fn, FI.End);
23770f5bc99SReid Kleckner }
23870f5bc99SReid Kleckner 
23970f5bc99SReid Kleckner void CodeViewDebug::beginFunction(const MachineFunction *MF) {
24070f5bc99SReid Kleckner   assert(!CurFn && "Can't process two functions at once!");
24170f5bc99SReid Kleckner 
24270f5bc99SReid Kleckner   if (!Asm || !Asm->MMI->hasDebugInfo())
24370f5bc99SReid Kleckner     return;
24470f5bc99SReid Kleckner 
24570f5bc99SReid Kleckner   const Function *GV = MF->getFunction();
24670f5bc99SReid Kleckner   assert(FnDebugInfo.count(GV) == false);
24770f5bc99SReid Kleckner   CurFn = &FnDebugInfo[GV];
248*2214ed89SReid Kleckner   CurFn->FuncId = NextFuncId++;
24970f5bc99SReid Kleckner 
25070f5bc99SReid Kleckner   // Find the end of the function prolog.
25170f5bc99SReid Kleckner   // FIXME: is there a simpler a way to do this? Can we just search
25270f5bc99SReid Kleckner   // for the first instruction of the function, not the last of the prolog?
25370f5bc99SReid Kleckner   DebugLoc PrologEndLoc;
25470f5bc99SReid Kleckner   bool EmptyPrologue = true;
25570f5bc99SReid Kleckner   for (const auto &MBB : *MF) {
25670f5bc99SReid Kleckner     if (PrologEndLoc)
25770f5bc99SReid Kleckner       break;
25870f5bc99SReid Kleckner     for (const auto &MI : MBB) {
25970f5bc99SReid Kleckner       if (MI.isDebugValue())
26070f5bc99SReid Kleckner         continue;
26170f5bc99SReid Kleckner 
26270f5bc99SReid Kleckner       // First known non-DBG_VALUE and non-frame setup location marks
26370f5bc99SReid Kleckner       // the beginning of the function body.
26470f5bc99SReid Kleckner       // FIXME: do we need the first subcondition?
26570f5bc99SReid Kleckner       if (!MI.getFlag(MachineInstr::FrameSetup) && MI.getDebugLoc()) {
26670f5bc99SReid Kleckner         PrologEndLoc = MI.getDebugLoc();
26770f5bc99SReid Kleckner         break;
26870f5bc99SReid Kleckner       }
26970f5bc99SReid Kleckner       EmptyPrologue = false;
27070f5bc99SReid Kleckner     }
27170f5bc99SReid Kleckner   }
27270f5bc99SReid Kleckner   // Record beginning of function if we have a non-empty prologue.
27370f5bc99SReid Kleckner   if (PrologEndLoc && !EmptyPrologue) {
27470f5bc99SReid Kleckner     DebugLoc FnStartDL = PrologEndLoc.getFnDebugLoc();
27570f5bc99SReid Kleckner     maybeRecordLocation(FnStartDL, MF);
27670f5bc99SReid Kleckner   }
27770f5bc99SReid Kleckner }
27870f5bc99SReid Kleckner 
27970f5bc99SReid Kleckner void CodeViewDebug::endFunction(const MachineFunction *MF) {
28070f5bc99SReid Kleckner   if (!Asm || !CurFn)  // We haven't created any debug info for this function.
28170f5bc99SReid Kleckner     return;
28270f5bc99SReid Kleckner 
28370f5bc99SReid Kleckner   const Function *GV = MF->getFunction();
28470f5bc99SReid Kleckner   assert(FnDebugInfo.count(GV));
28570f5bc99SReid Kleckner   assert(CurFn == &FnDebugInfo[GV]);
28670f5bc99SReid Kleckner 
287*2214ed89SReid Kleckner   // Don't emit anything if we don't have any line tables.
288*2214ed89SReid Kleckner   if (!CurFn->HaveLineInfo) {
28970f5bc99SReid Kleckner     FnDebugInfo.erase(GV);
29070f5bc99SReid Kleckner   } else {
29170f5bc99SReid Kleckner     CurFn->End = Asm->getFunctionEnd();
29270f5bc99SReid Kleckner   }
29370f5bc99SReid Kleckner   CurFn = nullptr;
29470f5bc99SReid Kleckner }
29570f5bc99SReid Kleckner 
29670f5bc99SReid Kleckner void CodeViewDebug::beginInstruction(const MachineInstr *MI) {
29770f5bc99SReid Kleckner   // Ignore DBG_VALUE locations and function prologue.
29870f5bc99SReid Kleckner   if (!Asm || MI->isDebugValue() || MI->getFlag(MachineInstr::FrameSetup))
29970f5bc99SReid Kleckner     return;
30070f5bc99SReid Kleckner   DebugLoc DL = MI->getDebugLoc();
30170f5bc99SReid Kleckner   if (DL == PrevInstLoc || !DL)
30270f5bc99SReid Kleckner     return;
30370f5bc99SReid Kleckner   maybeRecordLocation(DL, Asm->MF);
30470f5bc99SReid Kleckner }
30570f5bc99SReid Kleckner }
306