Merge pull request #71 from GroundControl-Solutions/non-contiguous-fu…

…nctions

Support non-contiguous functions

.github/workflows/build-and-test.yml

@@ -61,3 +61,24 @@ jobs:
           curl -Lo cvdump.exe https://raw.githubusercontent.com/microsoft/microsoft-pdb/HEAD/cvdump/cvdump.exe &&
           ./cvdump.exe world.pdb >world.cvdump &&
           grep '^S_PUB32: .*, Flags: 00000000, main$' world.cvdump
+      - name: verify split functions
+        shell: bash
+        run: |
+          set -x &&
+          cat >split.c <<-\EOF &&
+          __attribute__((cold,noinline)) void cold() { volatile int i=0; }
+          __attribute__((dllexport)) void split_func(int i)
+          {
+            if (i)
+              cold();
+          }
+          EOF
+
+          gcc -g -O2 -shared -o split-dwarf.dll split.c &&
+          bin/${{env.BUILD_CONFIGURATION}}*/cv2pdb.exe split-dwarf.dll split-cv.dll &&
+
+          ls -l split-dwarf* split-cv* &&
+
+          curl -Lo cvdump.exe https://raw.githubusercontent.com/microsoft/microsoft-pdb/HEAD/cvdump/cvdump.exe &&
+          ./cvdump.exe split-cv.pdb > split-cv.dump &&
+          awk '$2 == "S_GPROC32:" && $NF == "split_func" { scope=gensub(/,/,"",1,$3); } $2 == "S_SEPCODE:" { getline; if ($4 == scope) found=1; } END {exit !found; }' split-cv.dump

src/cv2pdb.h

@@ -171,7 +171,7 @@ class CV2PDB : public LastError
 	bool writeDWARFImage(const TCHAR* opath);
 
 	bool addDWARFSectionContrib(mspdb::Mod* mod, unsigned long pclo, unsigned long pchi);
-	bool addDWARFProc(DWARF_InfoData& id, DIECursor cursor);
+	bool addDWARFProc(DWARF_InfoData& id, const std::vector<RangeEntry> &ranges, DIECursor cursor);
 	int  addDWARFStructure(DWARF_InfoData& id, DIECursor cursor);
 	int  addDWARFFields(DWARF_InfoData& structid, DIECursor cursor, int off);
 	int  addDWARFArray(DWARF_InfoData& arrayid, DIECursor cursor);

src/cv2pdb.vcxproj

@@ -34,7 +34,7 @@
   <Import Project="$(VCTargetsPath)\Microsoft.Cpp.Default.props" />
   <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|Win32'" Label="Configuration">
     <ConfigurationType>Application</ConfigurationType>
-    <PlatformToolset>v120_xp</PlatformToolset>
+    <PlatformToolset>$(DefaultPlatformToolset)</PlatformToolset>
     <CharacterSet>Unicode</CharacterSet>
     <WholeProgramOptimization>true</WholeProgramOptimization>
   </PropertyGroup>
@@ -230,4 +230,4 @@
     </PropertyGroup>
     <Error Condition="!Exists('packages\Microsoft.VisualStudio.Setup.Configuration.Native.1.16.30\build\native\Microsoft.VisualStudio.Setup.Configuration.Native.targets')" Text="$([System.String]::Format('$(ErrorText)', 'packages\Microsoft.VisualStudio.Setup.Configuration.Native.1.16.30\build\native\Microsoft.VisualStudio.Setup.Configuration.Native.targets'))" />
   </Target>
-</Project>
+</Project>

src/dwarf2pdb.cpp

@@ -22,6 +22,58 @@
 #include <string>
 #include <vector>
 
+// Returns all non-empty address ranges specified by id.  The entry point (if applicable) is always the low address of the first range.
+static std::vector<RangeEntry> getRanges(DIECursor cursor, const DWARF_InfoData &id)
+{
+	std::vector<RangeEntry> ret;
+	RangeEntry range;
+	if (id.pclo)
+	{
+		if (id.pchi == 0 || id.pchi == id.pclo)
+			return ret;
+		if (id.pclo < id.pcentry && id.pcentry < id.pchi)
+		{
+			range.pclo = id.pcentry;
+			range.pchi = id.pchi;
+			ret.push_back(range);
+			range.pclo = id.pclo;
+			range.pchi = id.pcentry;
+			ret.push_back(range);
+		}
+		else
+		{
+			range.pclo = id.pclo;
+			range.pchi = id.pchi;
+			ret.push_back(range);
+		}
+	}
+	else if (id.ranges != ~0)
+	{
+		RangeCursor rangeCursor(cursor, id.ranges);
+		while (rangeCursor.readNext(range))
+		{
+			if (range.pclo == range.pchi)
+				continue;
+			if (range.pclo <= id.pcentry && id.pcentry < range.pchi)
+			{
+				std::uint64_t pclo = range.pclo;
+				range.pclo = id.pcentry;
+				ret.insert(ret.begin(), range);
+				if (pclo != id.pcentry)
+				{
+					range.pclo = pclo;
+					range.pchi = id.pcentry;
+					ret.push_back(range);
+				}
+			}
+			else
+			{
+				ret.push_back(range);
+			}
+		}
+	}
+	return ret;
+}
 
 void CV2PDB::checkDWARFTypeAlloc(int size, int add)
 {
@@ -643,10 +695,10 @@ void CV2PDB::appendLexicalBlock(DWARF_InfoData& id, unsigned int proclo)
 	cbUdtSymbols += len;
 }
 
-bool CV2PDB::addDWARFProc(DWARF_InfoData& procid, DIECursor cursor)
+bool CV2PDB::addDWARFProc(DWARF_InfoData& procid, const std::vector<RangeEntry> &ranges, DIECursor cursor)
 {
-	unsigned int pclo = procid.pclo - codeSegOff;
-	unsigned int pchi = procid.pchi - codeSegOff;
+	unsigned int pclo = ranges.front().pclo - codeSegOff;
+	unsigned int pchi = ranges.front().pchi - codeSegOff;
 
 	unsigned int len;
 	unsigned int align = 4;
@@ -789,6 +841,26 @@ bool CV2PDB::addDWARFProc(DWARF_InfoData& procid, DIECursor cursor)
 		appendEndArg();
 		appendEnd();
 	}
+
+	for (std::size_t i=1; i<ranges.size(); ++i)
+	{
+		const int sepcode_size = sizeof(codeview_symbol::sepcode_v3);
+		checkUdtSymbolAlloc(sepcode_size);
+		codeview_symbol*cvs = (codeview_symbol*) (udtSymbols + cbUdtSymbols);
+		cvs->sepcode_v3.id = S_SEPCODE_V3;
+		cvs->sepcode_v3.len = sepcode_size - 2;
+		cvs->sepcode_v3.parent = 0;
+		cvs->sepcode_v3.end = 0;
+		cvs->sepcode_v3.length = ranges[i].pchi - ranges[i].pclo;
+		cvs->sepcode_v3.flags = 0;
+		cvs->sepcode_v3.offset = ranges[i].pclo - codeSegOff;
+		cvs->sepcode_v3.parent_offset = pclo;
+		cvs->sepcode_v3.section = img.text.secNo + 1;
+		cvs->sepcode_v3.parent_section = img.text.secNo + 1;
+		cbUdtSymbols += sepcode_size;
+
+		appendEnd();
+	}
 	return true;
 }
 
@@ -1453,42 +1525,19 @@ bool CV2PDB::createTypes()
 			case DW_TAG_subprogram:
 				if (id.name)
 				{
-					if (!id.is_artificial)
+					std::vector<RangeEntry> ranges = getRanges(cursor, id);
+					if (!ranges.empty())
 					{
-						unsigned long entry_point = 0;
-						if (id.pcentry)
-						{
-							entry_point = id.pcentry;
-						}
-						else if (id.pclo)
-						{
-							entry_point = id.pclo;
-						}
-						else if (id.ranges != ~0)
-						{
-							entry_point = ~0;
-							RangeEntry range;
-							RangeCursor rangeCursor(cursor, id.ranges);
-							while (rangeCursor.readNext(range))
-							{
-								entry_point = min(entry_point, range.pclo);
-							}
-
-							if (entry_point == ~0)
-								entry_point = 0;
-						}
-
-						if (entry_point)
+						if (!id.is_artificial)
 						{
+							std::uint64_t entry_point = ranges.front().pclo;
 							if (debug & DbgPdbSyms)
 								fprintf(stderr, "%s:%d: Adding a public: %s at %x\n", __FUNCTION__, __LINE__, id.name, entry_point);
 
 							mod->AddPublic2(id.name, img.text.secNo + 1, entry_point - codeSegOff, 0);
 						}
+						addDWARFProc(id, ranges, cursor.getSubtreeCursor());
 					}
-
-					if (id.pclo && id.pchi)
-						addDWARFProc(id, cursor.getSubtreeCursor());
 				}
 				break;
 

src/mscvpdb.h

@@ -1670,6 +1670,20 @@ union codeview_symbol
 		// return UDT,instance constructor,instance constructor with virtual base
 		unsigned int            unknown2;
 	} funcinfo_32;
+
+	struct
+	{
+		short int               len;
+		short int               id;
+		unsigned int            parent;
+		unsigned int            end;
+		unsigned int            length;
+		unsigned int            flags;
+		unsigned int            offset;
+		unsigned int            parent_offset;
+		short int               section;
+		short int               parent_section;
+	} sepcode_v3;
 };
 
 #define S_COMPILAND_V1  0x0001
@@ -1745,6 +1759,7 @@ union codeview_symbol
 #define S_MSTOOL_V3     0x1116  /* compiler command line options and build information */
 #define S_PUB_FUNC1_V3  0x1125  /* didn't get the difference between the two */
 #define S_PUB_FUNC2_V3  0x1127
+#define S_SEPCODE_V3    0x1132
 #define S_SECTINFO_V3   0x1136
 #define S_SUBSECTINFO_V3 0x1137
 #define S_ENTRYPOINT_V3 0x1138