All source files in PDBs interpreted as UTF8 (#4111)

include/dxc/DxilContainer/DxilContainer.h

@@ -374,7 +374,7 @@ struct DxilSourceInfo_SourceContentsEntry {
   uint32_t AlignedSizeInBytes;                             // Size of the entry including this header and padding. Aligned to 4-byte boundary.
   uint32_t Flags;                                          // Reserved, must be set to 0.
   uint32_t ContentSizeInBytes;                             // Size of the data following this header, *including* the null terminator
-  // Followed by NameSizeInBytes bytes of the UTF-8-encoded content (including null terminator).
+  // Followed by ContentSizeInBytes bytes of the UTF-8-encoded content (including null terminator).
   // Followed by [0-3] zero bytes to align to a 4-byte boundary.
 };
 

tools/clang/tools/dxcompiler/dxcpdbutils.cpp

@@ -242,7 +242,7 @@ private:
 
   struct Source_File {
     std::wstring Name;
-    CComPtr<IDxcBlob> Content;
+    CComPtr<IDxcBlobEncoding> Content;
   };
 
   CComPtr<IDxcBlob> m_InputBlob;
@@ -365,11 +365,11 @@ private:
           Source_File file;
           file.Name = ToWstring(md_name->getString());
 
-          // File content
-          IFR(hlsl::DxcCreateBlobOnHeapCopy(
-            md_content->getString().data(),
-            md_content->getString().size(),
-            &file.Content));
+          IFR(hlsl::DxcCreateBlob(
+            md_content->getString().data(), md_content->getString().size(),
+            /*bPinned*/false, /*bCopy*/true,
+            /*encodingKnown*/true, CP_UTF8,
+            m_pMalloc, &file.Content));
 
           m_SourceFiles.push_back(std::move(file));
         }
@@ -499,10 +499,11 @@ private:
 
           Source_File source;
           source.Name = ToWstring(source_data.Name);
-          IFR(hlsl::DxcCreateBlobOnHeapCopy(
-            source_data.Content.data(),
-            source_data.Content.size(),
-            &source.Content));
+          IFR(hlsl::DxcCreateBlob(
+            source_data.Content.data(), source_data.Content.size(),
+            /*bPinned*/false, /*bCopy*/true,
+            /*encodingKnown*/true, CP_UTF8,
+            m_pMalloc, &source.Content));
 
           // First file is the main file
           if (i == 0) {

tools/clang/tools/dxcompiler/dxcshadersourceinfo.cpp

@@ -333,7 +333,7 @@ void SourceInfoWriter::Write(llvm::StringRef targetProfile, llvm::StringRef entr
       const size_t entryOffset = m_Buffer.size();
 
       // Write the header
-      hlsl::DxilSourceInfo_SourceNamesEntry entryHeader = {};
+      hlsl::DxilSourceInfo_SourceNamesEntry entryHeader ={};
       entryHeader.NameSizeInBytes = file.Name.size()+1;
       entryHeader.ContentSizeInBytes = file.Content.size()+1;
       entryHeader.AlignedSizeInBytes = PadToFourBytes(sizeof(entryHeader) + file.Name.size() + 1);

tools/clang/unittests/HLSL/CompilerTest.cpp

@@ -1138,12 +1138,16 @@ static void VerifyPdbUtil(dxc::DxcDllSupport &dllSupport,
       CComPtr<IDxcBlobEncoding> pFileContent;
       VERIFY_SUCCEEDED(pPdbUtils->GetSourceName(i, &pFileName));
       VERIFY_SUCCEEDED(pPdbUtils->GetSource(i, &pFileContent));
+
+      CComPtr<IDxcBlobUtf8> pFileContentUtf8;
+      VERIFY_SUCCEEDED(pFileContent.QueryInterface(&pFileContentUtf8));
+      llvm::StringRef FileContentRef(pFileContentUtf8->GetStringPointer(), pFileContentUtf8->GetStringLength());
+
       if (0 == wcscmp(pFileName, pMainFileName)) {
-        VERIFY_IS_TRUE(pFileContent->GetBufferSize() == MainSource.size());
-        VERIFY_IS_TRUE(0 == std::memcmp(pFileContent->GetBufferPointer(), MainSource.data(), MainSource.size()));
+        VERIFY_ARE_EQUAL(FileContentRef, MainSource);
       }
       else {
-        VERIFY_IS_TRUE(0 == std::memcmp(pFileContent->GetBufferPointer(), IncludedFile.data(), IncludedFile.size()));
+        VERIFY_ARE_EQUAL(FileContentRef, IncludedFile);
       }
     }
   }