[llvm-branch-commits] [clang] Convert to exec-charset inside getPredefinedStringLiteralFromCache (PR #196569)
Abhina Sree via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Wed Oct 7 07:57:54 PDT 2026
https://github.com/abhina-sree updated https://github.com/llvm/llvm-project/pull/196569
>From 8b99097451e212505cc469b46703af3d94d891aa Mon Sep 17 00:00:00 2001
From: Abhina Sreeskantharajan <Abhina.Sreeskantharajan at ibm.com>
Date: Fri, 8 May 2026 12:20:45 -0400
Subject: [PATCH 1/2] convert to exec-charset inside
getPredefinedStringLiteralFromCache, test __builtin_FILE()
---
clang/include/clang/Basic/TargetInfo.h | 2 ++
clang/lib/AST/ASTContext.cpp | 10 ++++++++++
clang/lib/Basic/TargetInfo.cpp | 3 +++
clang/lib/Lex/TextEncoding.cpp | 3 ++-
clang/test/CodeGen/systemz-charset.cpp | 4 +++-
5 files changed, 20 insertions(+), 2 deletions(-)
diff --git a/clang/include/clang/Basic/TargetInfo.h b/clang/include/clang/Basic/TargetInfo.h
index 6311b6b567a5e..81b1a056a4c4a 100644
--- a/clang/include/clang/Basic/TargetInfo.h
+++ b/clang/include/clang/Basic/TargetInfo.h
@@ -329,6 +329,8 @@ class TargetInfo : public TransferrableTargetInfo,
virtual ~TargetInfo();
+ llvm::TextEncodingConverter *ExecStrConverter;
+
/// Retrieve the target options.
TargetOptions &getTargetOpts() const {
assert(TargetOpts && "Missing target options");
diff --git a/clang/lib/AST/ASTContext.cpp b/clang/lib/AST/ASTContext.cpp
index 35a2a0b131a3f..067a16b46d0ea 100644
--- a/clang/lib/AST/ASTContext.cpp
+++ b/clang/lib/AST/ASTContext.cpp
@@ -13897,6 +13897,16 @@ ASTContext::getPredefinedStringLiteralFromCache(StringRef Key) const {
*this, Key, StringLiteralKind::Ordinary,
/*Pascal*/ false, getStringLiteralArrayType(CharTy, Key.size()),
SourceLocation());
+
+ llvm::TextEncodingConverter *Converter = getTargetInfo().ExecStrConverter;
+ if (Converter) {
+ SmallString<128> Converted;
+ Converter->convert(Result->getString(), Converted);
+ Result = StringLiteral::Create(
+ *this, Converted, StringLiteralKind::Ordinary, /*Pascal*/ false,
+ getStringLiteralArrayType(CharTy, Converted.size()), SourceLocation());
+ }
+
return Result;
}
diff --git a/clang/lib/Basic/TargetInfo.cpp b/clang/lib/Basic/TargetInfo.cpp
index 1bb0026d8422e..9cc2a737fafbf 100644
--- a/clang/lib/Basic/TargetInfo.cpp
+++ b/clang/lib/Basic/TargetInfo.cpp
@@ -198,6 +198,9 @@ TargetInfo::TargetInfo(const llvm::Triple &T) : Triple(T) {
MaxOpenCLWorkGroupSize = 1024;
MaxBitIntWidth.reset();
+
+ ExecStrConverter = new llvm::TextEncodingConverter(
+ std::move(*llvm::TextEncodingConverter::createNoopConverter()));
}
// Out of line virtual dtor for TargetInfo.
diff --git a/clang/lib/Lex/TextEncoding.cpp b/clang/lib/Lex/TextEncoding.cpp
index c988a0123e4ef..90def3a1c4e31 100644
--- a/clang/lib/Lex/TextEncoding.cpp
+++ b/clang/lib/Lex/TextEncoding.cpp
@@ -69,7 +69,8 @@ TextEncoding::setConvertersFromOptions(TextEncoding &TE,
if (ErrorOrConverter)
TE.FromIBM1047Converter =
std::make_unique<TextEncodingConverter>(std::move(*ErrorOrConverter));
- else
+ TInfo.ExecStrConverter = TEC.ToLiteralEncodingConverter.get();
+ } else
return ErrorOrConverter.getError();
return std::error_code();
diff --git a/clang/test/CodeGen/systemz-charset.cpp b/clang/test/CodeGen/systemz-charset.cpp
index b6b79842468c1..7cb100f05206b 100644
--- a/clang/test/CodeGen/systemz-charset.cpp
+++ b/clang/test/CodeGen/systemz-charset.cpp
@@ -73,7 +73,6 @@ const char32_t *UnicodeUCNString32 = U"\u00E2\u00AC\U000000DF";
//CHECK: [4 x i32] [i32 226, i32 172, i32 223, i32 0]
//CHECK-UTF8: [4 x i32] [i32 226, i32 172, i32 223, i32 0]
-
struct string_view {
int S;
const char* D;
@@ -100,3 +99,6 @@ void function()
// CHECK-UTF8: asm{{.*}}|foo
}
+const char *file = __builtin_FILE();
+//CHECK: {{.*}}\A2\A8\A2\A3\85\94\A9`\83\88\81\99\A2\85\A3K\83\97\97\00"
+//CHECK-UTF8: {{.*}}systemz-charset.cpp\00"
>From d16d9b80f6d49ae66c4c28adaf92ebc053ca653f Mon Sep 17 00:00:00 2001
From: Abhina Sreeskantharajan <Abhina.Sreeskantharajan at ibm.com>
Date: Fri, 22 May 2026 08:51:37 -0400
Subject: [PATCH 2/2] Convert the key before cache lookup to prevent encoding
differences
---
clang/lib/AST/ASTContext.cpp | 18 +++++++++---------
clang/lib/Lex/TextEncoding.cpp | 5 +++--
2 files changed, 12 insertions(+), 11 deletions(-)
diff --git a/clang/lib/AST/ASTContext.cpp b/clang/lib/AST/ASTContext.cpp
index 067a16b46d0ea..ba902065a8ba6 100644
--- a/clang/lib/AST/ASTContext.cpp
+++ b/clang/lib/AST/ASTContext.cpp
@@ -13891,6 +13891,15 @@ QualType ASTContext::getStringLiteralArrayType(QualType EltTy,
StringLiteral *
ASTContext::getPredefinedStringLiteralFromCache(StringRef Key) const {
+ // Apply encoding conversion to the key before cache lookup to ensure
+ // proper deduplication when the same source location is used multiple times
+ SmallString<128> ConvertedKey;
+ llvm::TextEncodingConverter *Converter = getTargetInfo().ExecStrConverter;
+ if (Converter) {
+ Converter->convert(Key, ConvertedKey);
+ Key = ConvertedKey;
+ }
+
StringLiteral *&Result = StringLiteralCache[Key];
if (!Result)
Result = StringLiteral::Create(
@@ -13898,15 +13907,6 @@ ASTContext::getPredefinedStringLiteralFromCache(StringRef Key) const {
/*Pascal*/ false, getStringLiteralArrayType(CharTy, Key.size()),
SourceLocation());
- llvm::TextEncodingConverter *Converter = getTargetInfo().ExecStrConverter;
- if (Converter) {
- SmallString<128> Converted;
- Converter->convert(Result->getString(), Converted);
- Result = StringLiteral::Create(
- *this, Converted, StringLiteralKind::Ordinary, /*Pascal*/ false,
- getStringLiteralArrayType(CharTy, Converted.size()), SourceLocation());
- }
-
return Result;
}
diff --git a/clang/lib/Lex/TextEncoding.cpp b/clang/lib/Lex/TextEncoding.cpp
index 90def3a1c4e31..86b7cb1fbca58 100644
--- a/clang/lib/Lex/TextEncoding.cpp
+++ b/clang/lib/Lex/TextEncoding.cpp
@@ -47,6 +47,8 @@ TextEncoding::setConvertersFromOptions(TextEncoding &TE,
return ErrorOrLiteralConverter.getError();
}
+ TInfo.ExecStrConverter = TE.ToLiteralEncodingConverter.get();
+
if (TInfo.getDefaultOrdinaryLiteralEncoding() == UTF8)
return std::error_code();
@@ -69,8 +71,7 @@ TextEncoding::setConvertersFromOptions(TextEncoding &TE,
if (ErrorOrConverter)
TE.FromIBM1047Converter =
std::make_unique<TextEncodingConverter>(std::move(*ErrorOrConverter));
- TInfo.ExecStrConverter = TEC.ToLiteralEncodingConverter.get();
- } else
+ else
return ErrorOrConverter.getError();
return std::error_code();
More information about the llvm-branch-commits
mailing list