[llvm-branch-commits] [clang] [llvm] Add format string handling (PR #196568)
Abhina Sree via llvm-branch-commits
llvm-branch-commits at lists.llvm.org
Fri May 22 11:34:54 PDT 2026
https://github.com/abhina-sree updated https://github.com/llvm/llvm-project/pull/196568
>From 719e1d23651b5d6b8252b44a37a0a4fa5256ed62 Mon Sep 17 00:00:00 2001
From: Abhina Sreeskantharajan <Abhina.Sreeskantharajan at ibm.com>
Date: Fri, 8 May 2026 12:19:11 -0400
Subject: [PATCH 1/2] Add format string handling
---
clang/include/clang/AST/FormatString.h | 12 +--
clang/include/clang/Basic/TargetInfo.h | 3 +
clang/include/clang/Lex/TextEncodingConfig.h | 3 +-
clang/lib/AST/FormatString.cpp | 86 ++++++++++---------
clang/lib/AST/FormatStringParsing.h | 36 +++++---
clang/lib/AST/PrintfFormatString.cpp | 89 +++++++++++++-------
clang/lib/AST/ScanfFormatString.cpp | 23 +++--
clang/lib/Basic/TargetInfo.cpp | 3 +
clang/lib/Frontend/CompilerInstance.cpp | 2 +-
clang/lib/Lex/TextEncodingConfig.cpp | 11 ++-
clang/lib/Sema/SemaChecking.cpp | 54 +++++++-----
llvm/include/llvm/Support/TextEncoding.h | 10 +++
llvm/lib/Support/TextEncoding.cpp | 19 +++++
13 files changed, 232 insertions(+), 119 deletions(-)
diff --git a/clang/include/clang/AST/FormatString.h b/clang/include/clang/AST/FormatString.h
index a3382e1a1d007..a24ade2d71ee9 100644
--- a/clang/include/clang/AST/FormatString.h
+++ b/clang/include/clang/AST/FormatString.h
@@ -19,6 +19,7 @@
#define LLVM_CLANG_AST_FORMATSTRING_H
#include "clang/AST/CanonicalType.h"
+#include "llvm/Support/TextEncoding.h"
#include <optional>
namespace clang {
@@ -728,7 +729,8 @@ class FormatStringHandler {
virtual bool HandleInvalidPrintfConversionSpecifier(
const analyze_printf::PrintfSpecifier &FS, const char *startSpecifier,
- unsigned specifierLen) {
+ unsigned specifierLen,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
return true;
}
@@ -744,10 +746,10 @@ class FormatStringHandler {
// Scanf-specific handlers.
- virtual bool
- HandleInvalidScanfConversionSpecifier(const analyze_scanf::ScanfSpecifier &FS,
- const char *startSpecifier,
- unsigned specifierLen) {
+ virtual bool HandleInvalidScanfConversionSpecifier(
+ const analyze_scanf::ScanfSpecifier &FS, const char *startSpecifier,
+ unsigned specifierLen,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
return true;
}
diff --git a/clang/include/clang/Basic/TargetInfo.h b/clang/include/clang/Basic/TargetInfo.h
index 779093f963bc1..41ad751c4b769 100644
--- a/clang/include/clang/Basic/TargetInfo.h
+++ b/clang/include/clang/Basic/TargetInfo.h
@@ -38,6 +38,7 @@
#include "llvm/IR/DerivedTypes.h"
#include "llvm/Support/DataTypes.h"
#include "llvm/Support/Error.h"
+#include "llvm/Support/TextEncoding.h"
#include "llvm/Support/VersionTuple.h"
#include "llvm/TargetParser/Triple.h"
#include <cassert>
@@ -323,6 +324,8 @@ class TargetInfo : public TransferrableTargetInfo,
virtual ~TargetInfo();
+ llvm::TextEncodingConverter *FormatStrConverter;
+
/// Retrieve the target options.
TargetOptions &getTargetOpts() const {
assert(TargetOpts && "Missing target options");
diff --git a/clang/include/clang/Lex/TextEncodingConfig.h b/clang/include/clang/Lex/TextEncodingConfig.h
index 9e1e7649d65c4..97fffd38d28c5 100644
--- a/clang/include/clang/Lex/TextEncodingConfig.h
+++ b/clang/include/clang/Lex/TextEncodingConfig.h
@@ -23,7 +23,8 @@ class TextEncodingConfig {
llvm::TextEncodingConverter *getConverter(ConversionAction Action) const;
static std::error_code
setConvertersFromOptions(TextEncodingConfig &TEC,
- const clang::LangOptions &Opts);
+ const clang::LangOptions &Opts,
+ clang::TargetInfo &TInfo);
llvm::StringRef getExecEncoding() { return ExecEncoding; }
};
diff --git a/clang/lib/AST/FormatString.cpp b/clang/lib/AST/FormatString.cpp
index 7e1ac0de6dcaf..0d449fb5f0904 100644
--- a/clang/lib/AST/FormatString.cpp
+++ b/clang/lib/AST/FormatString.cpp
@@ -33,8 +33,9 @@ FormatStringHandler::~FormatStringHandler() {}
// scanf format strings.
//===----------------------------------------------------------------------===//
-OptionalAmount clang::analyze_format_string::ParseAmount(const char *&Beg,
- const char *E) {
+OptionalAmount clang::analyze_format_string::ParseAmount(
+ const char *&Beg, const char *E,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
const char *I = Beg;
UpdateOnReturn<const char *> UpdateBeg(Beg, I);
@@ -42,7 +43,7 @@ OptionalAmount clang::analyze_format_string::ParseAmount(const char *&Beg,
bool hasDigits = false;
for (; I != E; ++I) {
- char c = *I;
+ char c = FormatStrConverter.convert(*I);
if (c >= '0' && c <= '9') {
hasDigits = true;
accumulator = (accumulator * 10) + (c - '0');
@@ -60,21 +61,22 @@ OptionalAmount clang::analyze_format_string::ParseAmount(const char *&Beg,
}
OptionalAmount clang::analyze_format_string::ParseNonPositionAmount(
- const char *&Beg, const char *E, unsigned &argIndex) {
- if (*Beg == '*') {
+ const char *&Beg, const char *E, unsigned &argIndex,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
+ if (FormatStrConverter.convert(*Beg) == '*') {
++Beg;
return OptionalAmount(OptionalAmount::Arg, argIndex++, Beg, 0, false);
}
- return ParseAmount(Beg, E);
+ return ParseAmount(Beg, E, FormatStrConverter);
}
OptionalAmount clang::analyze_format_string::ParsePositionAmount(
FormatStringHandler &H, const char *Start, const char *&Beg, const char *E,
- PositionContext p) {
- if (*Beg == '*') {
+ PositionContext p, const llvm::TextEncodingConverter &FormatStrConverter) {
+ if (FormatStrConverter.convert(*Beg) == '*') {
const char *I = Beg + 1;
- const OptionalAmount &Amt = ParseAmount(I, E);
+ const OptionalAmount &Amt = ParseAmount(I, E, FormatStrConverter);
if (Amt.getHowSpecified() == OptionalAmount::NotSpecified) {
H.HandleInvalidPosition(Beg, I - Beg, p);
@@ -89,7 +91,7 @@ OptionalAmount clang::analyze_format_string::ParsePositionAmount(
assert(Amt.getHowSpecified() == OptionalAmount::Constant);
- if (*I == '$') {
+ if (FormatStrConverter.convert(*I) == '$') {
// Handle positional arguments
// Special case: '*0$', since this is an easy mistake.
@@ -109,18 +111,21 @@ OptionalAmount clang::analyze_format_string::ParsePositionAmount(
return OptionalAmount(false);
}
- return ParseAmount(Beg, E);
+ return ParseAmount(Beg, E, FormatStrConverter);
}
bool clang::analyze_format_string::ParseFieldWidth(
FormatStringHandler &H, FormatSpecifier &CS, const char *Start,
- const char *&Beg, const char *E, unsigned *argIndex) {
+ const char *&Beg, const char *E, unsigned *argIndex,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
// FIXME: Support negative field widths.
if (argIndex) {
- CS.setFieldWidth(ParseNonPositionAmount(Beg, E, *argIndex));
+ CS.setFieldWidth(
+ ParseNonPositionAmount(Beg, E, *argIndex, FormatStrConverter));
} else {
const OptionalAmount Amt = ParsePositionAmount(
- H, Start, Beg, E, analyze_format_string::FieldWidthPos);
+ H, Start, Beg, E, analyze_format_string::FieldWidthPos,
+ FormatStrConverter);
if (Amt.isInvalid())
return true;
@@ -129,14 +134,13 @@ bool clang::analyze_format_string::ParseFieldWidth(
return false;
}
-bool clang::analyze_format_string::ParseArgPosition(FormatStringHandler &H,
- FormatSpecifier &FS,
- const char *Start,
- const char *&Beg,
- const char *E) {
+bool clang::analyze_format_string::ParseArgPosition(
+ FormatStringHandler &H, FormatSpecifier &FS, const char *Start,
+ const char *&Beg, const char *E,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
const char *I = Beg;
- const OptionalAmount &Amt = ParseAmount(I, E);
+ const OptionalAmount &Amt = ParseAmount(I, E, FormatStrConverter);
if (I == E) {
// No more characters left?
@@ -144,7 +148,8 @@ bool clang::analyze_format_string::ParseArgPosition(FormatStringHandler &H,
return true;
}
- if (Amt.getHowSpecified() == OptionalAmount::Constant && *(I++) == '$') {
+ if (Amt.getHowSpecified() == OptionalAmount::Constant &&
+ FormatStrConverter.convert(*(I++)) == '$') {
// Warn that positional arguments are non-standard.
H.HandlePosition(Start, I - Start);
@@ -165,16 +170,15 @@ bool clang::analyze_format_string::ParseArgPosition(FormatStringHandler &H,
return false;
}
-bool clang::analyze_format_string::ParseVectorModifier(FormatStringHandler &H,
- FormatSpecifier &FS,
- const char *&I,
- const char *E,
- const LangOptions &LO) {
+bool clang::analyze_format_string::ParseVectorModifier(
+ FormatStringHandler &H, FormatSpecifier &FS, const char *&I, const char *E,
+ const LangOptions &LO,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
if (!LO.OpenCL)
return false;
const char *Start = I;
- if (*I == 'v') {
+ if (FormatStrConverter.convert(*I) == 'v') {
++I;
if (I == E) {
@@ -182,7 +186,7 @@ bool clang::analyze_format_string::ParseVectorModifier(FormatStringHandler &H,
return true;
}
- OptionalAmount NumElts = ParseAmount(I, E);
+ OptionalAmount NumElts = ParseAmount(I, E, FormatStrConverter);
if (NumElts.getHowSpecified() != OptionalAmount::Constant) {
H.HandleIncompleteSpecifier(Start, E - Start);
return true;
@@ -194,22 +198,20 @@ bool clang::analyze_format_string::ParseVectorModifier(FormatStringHandler &H,
return false;
}
-bool clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
- const char *&I,
- const char *E,
- const LangOptions &LO,
- bool IsScanf) {
+bool clang::analyze_format_string::ParseLengthModifier(
+ FormatSpecifier &FS, const char *&I, const char *E, const LangOptions &LO,
+ const llvm::TextEncodingConverter &FormatStrConverter, bool IsScanf) {
LengthModifier::Kind lmKind = LengthModifier::None;
const char *lmPosition = I;
- switch (*I) {
+ switch (FormatStrConverter.convert(*I)) {
default:
return false;
case 'h':
++I;
- if (I != E && *I == 'h') {
+ if (I != E && FormatStrConverter.convert(*I) == 'h') {
++I;
lmKind = LengthModifier::AsChar;
- } else if (I != E && *I == 'l' && LO.OpenCL) {
+ } else if (I != E && FormatStrConverter.convert(*I) == 'l' && LO.OpenCL) {
++I;
lmKind = LengthModifier::AsShortLong;
} else {
@@ -218,7 +220,7 @@ bool clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
break;
case 'l':
++I;
- if (I != E && *I == 'l') {
+ if (I != E && FormatStrConverter.convert(*I) == 'l') {
++I;
lmKind = LengthModifier::AsLongLong;
} else {
@@ -251,7 +253,9 @@ bool clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
// be parsed as the GNU extension 'a' length modifier. If not, this
// will be parsed as a conversion specifier.
++I;
- if (I != E && (*I == 's' || *I == 'S' || *I == '[')) {
+ if (I != E && (FormatStrConverter.convert(*I) == 's' ||
+ FormatStrConverter.convert(*I) == 'S' ||
+ FormatStrConverter.convert(*I) == '[')) {
lmKind = LengthModifier::AsAllocate;
break;
}
@@ -269,7 +273,8 @@ bool clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
// scanf: AsInt64
case 'I':
if (I + 1 != E && I + 2 != E) {
- if (I[1] == '6' && I[2] == '4') {
+ if (FormatStrConverter.convert(I[1]) == '6' &&
+ FormatStrConverter.convert(I[2]) == '4') {
I += 3;
lmKind = LengthModifier::AsInt64;
break;
@@ -277,7 +282,8 @@ bool clang::analyze_format_string::ParseLengthModifier(FormatSpecifier &FS,
if (IsScanf)
return false;
- if (I[1] == '3' && I[2] == '2') {
+ if (FormatStrConverter.convert(I[1]) == '3' &&
+ FormatStrConverter.convert(I[2]) == '2') {
I += 3;
lmKind = LengthModifier::AsInt32;
break;
diff --git a/clang/lib/AST/FormatStringParsing.h b/clang/lib/AST/FormatStringParsing.h
index 401528481a9d6..531bc291e0b5b 100644
--- a/clang/lib/AST/FormatStringParsing.h
+++ b/clang/lib/AST/FormatStringParsing.h
@@ -35,29 +35,43 @@ template <typename T> class UpdateOnReturn {
namespace analyze_format_string {
-OptionalAmount ParseAmount(const char *&Beg, const char *E);
-OptionalAmount ParseNonPositionAmount(const char *&Beg, const char *E,
- unsigned &argIndex);
+OptionalAmount
+ParseAmount(const char *&Beg, const char *E,
+ const llvm::TextEncodingConverter &FormatStrConverter);
-OptionalAmount ParsePositionAmount(FormatStringHandler &H, const char *Start,
- const char *&Beg, const char *E,
- PositionContext p);
+OptionalAmount
+ParseNonPositionAmount(const char *&Beg, const char *E, unsigned &argIndex,
+ const llvm::TextEncodingConverter &FormatStrConverter);
+
+OptionalAmount
+ParsePositionAmount(FormatStringHandler &H, const char *Start, const char *&Beg,
+ const char *E, PositionContext p,
+ const llvm::TextEncodingConverter &FormatStrConverter);
+
+OptionalAmount
+ParsePositionAmount(FormatStringHandler &H, const char *Start, const char *&Beg,
+ const char *E, PositionContext p,
+ const llvm::TextEncodingConverter &FormatStrConverter);
bool ParseFieldWidth(FormatStringHandler &H, FormatSpecifier &CS,
const char *Start, const char *&Beg, const char *E,
- unsigned *argIndex);
+ unsigned *argIndex,
+ const llvm::TextEncodingConverter &FormatStrConverter);
bool ParseArgPosition(FormatStringHandler &H, FormatSpecifier &CS,
- const char *Start, const char *&Beg, const char *E);
+ const char *Start, const char *&Beg, const char *E,
+ const llvm::TextEncodingConverter &FormatStrConverter);
bool ParseVectorModifier(FormatStringHandler &H, FormatSpecifier &FS,
- const char *&Beg, const char *E,
- const LangOptions &LO);
+ const char *&Beg, const char *E, const LangOptions &LO,
+ const llvm::TextEncodingConverter &FormatStrConverter);
/// Returns true if a LengthModifier was parsed and installed in the
/// FormatSpecifier& argument, and false otherwise.
bool ParseLengthModifier(FormatSpecifier &FS, const char *&Beg, const char *E,
- const LangOptions &LO, bool IsScanf = false);
+ const LangOptions &LO,
+ const llvm::TextEncodingConverter &FormatStrConverter,
+ bool IsScanf = false);
/// Returns true if the invalid specifier in \p SpecifierBegin is a UTF-8
/// string; check that it won't go further than \p FmtStrEnd and write
diff --git a/clang/lib/AST/PrintfFormatString.cpp b/clang/lib/AST/PrintfFormatString.cpp
index 6610a2de9e083..7efcc554ec136 100644
--- a/clang/lib/AST/PrintfFormatString.cpp
+++ b/clang/lib/AST/PrintfFormatString.cpp
@@ -35,14 +35,17 @@ typedef clang::analyze_format_string::SpecifierResult<PrintfSpecifier>
using analyze_format_string::ParseNonPositionAmount;
-static bool ParsePrecision(FormatStringHandler &H, PrintfSpecifier &FS,
- const char *Start, const char *&Beg, const char *E,
- unsigned *argIndex) {
+static bool
+ParsePrecision(FormatStringHandler &H, PrintfSpecifier &FS, const char *Start,
+ const char *&Beg, const char *E, unsigned *argIndex,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
if (argIndex) {
- FS.setPrecision(ParseNonPositionAmount(Beg, E, *argIndex));
+ FS.setPrecision(
+ ParseNonPositionAmount(Beg, E, *argIndex, FormatStrConverter));
} else {
const OptionalAmount Amt = ParsePositionAmount(
- H, Start, Beg, E, analyze_format_string::PrecisionPos);
+ H, Start, Beg, E, analyze_format_string::PrecisionPos,
+ FormatStrConverter);
if (Amt.isInvalid())
return true;
FS.setPrecision(Amt);
@@ -50,11 +53,14 @@ static bool ParsePrecision(FormatStringHandler &H, PrintfSpecifier &FS,
return false;
}
-static bool ParseObjCFlags(FormatStringHandler &H, PrintfSpecifier &FS,
- const char *FlagBeg, const char *E, bool Warn) {
+static bool
+ParseObjCFlags(FormatStringHandler &H, PrintfSpecifier &FS, const char *FlagBeg,
+ const char *E, bool Warn,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
StringRef Flag(FlagBeg, E - FlagBeg);
// Currently there is only one flag.
- if (Flag == "tt") {
+ if (Flag.size() == 2 && FormatStrConverter.convert(FlagBeg[0]) == 't' &&
+ FormatStrConverter.convert(FlagBeg[1]) == 't') {
FS.setHasObjCTechnicalTerm(FlagBeg);
return false;
}
@@ -81,6 +87,8 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
const char *Start = nullptr;
UpdateOnReturn<const char *> UpdateBeg(Beg, I);
+ const llvm::TextEncodingConverter &FormatStrConverter =
+ *Target.FormatStrConverter;
// Look for a '%' character that indicates the start of a format specifier.
for (; I != E; ++I) {
char c = *I;
@@ -89,7 +97,7 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
H.HandleNullChar(I);
return true;
}
- if (c == '%') {
+ if (FormatStrConverter.convert(c) == '%') {
Start = I++; // Record the start of the format specifier.
break;
}
@@ -107,7 +115,7 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
}
PrintfSpecifier FS;
- if (ParseArgPosition(H, FS, Start, I, E))
+ if (ParseArgPosition(H, FS, Start, I, E, FormatStrConverter))
return true;
if (I == E) {
@@ -117,13 +125,17 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
return true;
}
- if (*I == '{') {
+ if (FormatStrConverter.convert(*I) == '{') {
++I;
unsigned char PrivacyFlags = 0;
StringRef MatchedStr;
do {
- StringRef Str(I, E - I);
+ const char *II;
+ std::string S(I, E - I);
+ for (unsigned long i = 0; i < S.length(); ++i)
+ S[i] = FormatStrConverter.convert(S[i]);
+ StringRef Str(S);
std::string Match = "^[[:space:]]*"
"(private|public|sensitive|mask\\.[^[:space:],}]*)"
"[[:space:]]*(,|})";
@@ -132,25 +144,38 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
if (R.match(Str, &Matches)) {
MatchedStr = Matches[1];
+ II = I;
I += Matches[0].size();
+ while (FormatStrConverter.convert(*II) == ' ')
+ ++II;
+
// Set the privacy flag if the privacy annotation in the
// comma-delimited segment is at least as strict as the privacy
// annotations in previous comma-delimited segments.
if (MatchedStr.starts_with("mask")) {
- StringRef MaskType = MatchedStr.substr(sizeof("mask.") - 1);
+ StringRef MaskType(II + sizeof("mask.") - 1,
+ MatchedStr.size() - sizeof("mask.") + 1);
unsigned Size = MaskType.size();
+
if (Warn && (Size == 0 || Size > 8))
H.handleInvalidMaskType(MaskType);
FS.setMaskType(MaskType);
- } else if (MatchedStr == "sensitive")
+ } else if (MatchedStr == "sensitive") {
+ StringRef ProxyMatchedStr(II, sizeof("sensitive") - 1);
+ MatchedStr = ProxyMatchedStr;
PrivacyFlags = clang::analyze_os_log::OSLogBufferItem::IsSensitive;
- else if (PrivacyFlags !=
- clang::analyze_os_log::OSLogBufferItem::IsSensitive &&
- MatchedStr == "private")
+ } else if (PrivacyFlags !=
+ clang::analyze_os_log::OSLogBufferItem::IsSensitive &&
+ MatchedStr == "private") {
+ StringRef ProxyMatchedStr(II, sizeof("private") - 1);
+ MatchedStr = ProxyMatchedStr;
PrivacyFlags = clang::analyze_os_log::OSLogBufferItem::IsPrivate;
- else if (PrivacyFlags == 0 && MatchedStr == "public")
+ } else if (PrivacyFlags == 0 && MatchedStr == "public") {
+ StringRef ProxyMatchedStr(II, sizeof("public") - 1);
+ MatchedStr = ProxyMatchedStr;
PrivacyFlags = clang::analyze_os_log::OSLogBufferItem::IsPublic;
+ }
} else {
size_t CommaOrBracePos =
Str.find_if([](char c) { return c == ',' || c == '}'; });
@@ -165,7 +190,7 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
I += CommaOrBracePos + 1;
}
// Continue until the closing brace is found.
- } while (*(I - 1) == ',');
+ } while (FormatStrConverter.convert(*(I - 1)) == ',');
// Set the privacy flag.
switch (PrivacyFlags) {
@@ -188,7 +213,7 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
// Look for flags (if any).
bool hasMore = true;
for (; I != E; ++I) {
- switch (*I) {
+ switch (FormatStrConverter.convert(*I)) {
default:
hasMore = false;
break;
@@ -225,7 +250,8 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
// Look for the field width (if any).
if (ParseFieldWidth(H, FS, Start, I, E,
- FS.usesPositionalArg() ? nullptr : &argIndex))
+ FS.usesPositionalArg() ? nullptr : &argIndex,
+ FormatStrConverter))
return true;
if (I == E) {
@@ -236,7 +262,7 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
}
// Look for the precision (if any).
- if (*I == '.') {
+ if (FormatStrConverter.convert(*I) == '.') {
++I;
if (I == E) {
if (Warn)
@@ -245,7 +271,8 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
}
if (ParsePrecision(H, FS, Start, I, E,
- FS.usesPositionalArg() ? nullptr : &argIndex))
+ FS.usesPositionalArg() ? nullptr : &argIndex,
+ FormatStrConverter))
return true;
if (I == E) {
@@ -256,11 +283,11 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
}
}
- if (ParseVectorModifier(H, FS, I, E, LO))
+ if (ParseVectorModifier(H, FS, I, E, LO, FormatStrConverter))
return true;
// Look for the length modifier.
- if (ParseLengthModifier(FS, I, E, LO) && I == E) {
+ if (ParseLengthModifier(FS, I, E, LO, FormatStrConverter) && I == E) {
// No more characters left?
if (Warn)
H.HandleIncompleteSpecifier(Start, E - Start);
@@ -274,7 +301,7 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
// enables better recovery, and we don't know if
// these flags are applicable until later.
const char *ObjCModifierFlagsStart = nullptr, *ObjCModifierFlagsEnd = nullptr;
- if (*I == '[') {
+ if (FormatStrConverter.convert(*I) == '[') {
ObjCModifierFlagsStart = I;
++I;
auto flagStart = I;
@@ -286,8 +313,8 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
return true;
}
// Did we find the closing ']'?
- if (*I == ']') {
- if (ParseObjCFlags(H, FS, flagStart, I, Warn))
+ if (FormatStrConverter.convert(*I) == ']') {
+ if (ParseObjCFlags(H, FS, flagStart, I, Warn, FormatStrConverter))
return true;
++I;
break;
@@ -307,7 +334,7 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
// Finally, look for the conversion specifier.
const char *conversionPosition = I++;
ConversionSpecifier::Kind k = ConversionSpecifier::InvalidSpecifier;
- switch (*conversionPosition) {
+ switch (FormatStrConverter.convert(*conversionPosition)) {
default:
break;
// C99: 7.19.6.1 (section 8).
@@ -470,7 +497,8 @@ ParsePrintfSpecifier(FormatStringHandler &H, const char *&Beg, const char *E,
FS.setConversionSpecifier(CS);
}
// Assume the conversion takes one argument.
- return !H.HandleInvalidPrintfConversionSpecifier(FS, Start, Len);
+ return !H.HandleInvalidPrintfConversionSpecifier(FS, Start, Len,
+ FormatStrConverter);
}
return PrintfSpecifierResult(Start, FS);
}
@@ -480,7 +508,6 @@ bool clang::analyze_format_string::ParsePrintfString(
const TargetInfo &Target, bool isFreeBSDKPrintf) {
unsigned argIndex = 0;
-
// Keep looking for a format specifier until we have exhausted the string.
while (I != E) {
const PrintfSpecifierResult &FSR = ParsePrintfSpecifier(
diff --git a/clang/lib/AST/ScanfFormatString.cpp b/clang/lib/AST/ScanfFormatString.cpp
index 90cbbd60bbcf5..c63171844d90d 100644
--- a/clang/lib/AST/ScanfFormatString.cpp
+++ b/clang/lib/AST/ScanfFormatString.cpp
@@ -81,7 +81,8 @@ static ScanfSpecifierResult ParseScanfSpecifier(FormatStringHandler &H,
const char *I = Beg;
const char *Start = nullptr;
UpdateOnReturn<const char *> UpdateBeg(Beg, I);
-
+ const llvm::TextEncodingConverter &FormatStrConverter =
+ *Target.FormatStrConverter;
// Look for a '%' character that indicates the start of a format specifier.
for (; I != E; ++I) {
char c = *I;
@@ -90,7 +91,9 @@ static ScanfSpecifierResult ParseScanfSpecifier(FormatStringHandler &H,
H.HandleNullChar(I);
return true;
}
- if (c == '%') {
+ SmallString<1> ConvertedChar;
+ FormatStrConverter.convert(StringRef(&c, 1), ConvertedChar);
+ if (ConvertedChar[0] == '%') {
Start = I++; // Record the start of the format specifier.
break;
}
@@ -107,7 +110,7 @@ static ScanfSpecifierResult ParseScanfSpecifier(FormatStringHandler &H,
}
ScanfSpecifier FS;
- if (ParseArgPosition(H, FS, Start, I, E))
+ if (ParseArgPosition(H, FS, Start, I, E, FormatStrConverter))
return true;
if (I == E) {
@@ -117,7 +120,7 @@ static ScanfSpecifierResult ParseScanfSpecifier(FormatStringHandler &H,
}
// Look for '*' flag if it is present.
- if (*I == '*') {
+ if (FormatStrConverter.convert(*I) == '*') {
FS.setSuppressAssignment(I);
if (++I == E) {
H.HandleIncompleteSpecifier(Start, E - Start);
@@ -127,7 +130,8 @@ static ScanfSpecifierResult ParseScanfSpecifier(FormatStringHandler &H,
// Look for the field width (if any). Unlike printf, this is either
// a fixed integer or isn't present.
- const OptionalAmount &Amt = clang::analyze_format_string::ParseAmount(I, E);
+ const OptionalAmount &Amt =
+ clang::analyze_format_string::ParseAmount(I, E, FormatStrConverter);
if (Amt.getHowSpecified() != OptionalAmount::NotSpecified) {
assert(Amt.getHowSpecified() == OptionalAmount::Constant);
FS.setFieldWidth(Amt);
@@ -140,7 +144,9 @@ static ScanfSpecifierResult ParseScanfSpecifier(FormatStringHandler &H,
}
// Look for the length modifier.
- if (ParseLengthModifier(FS, I, E, LO, /*IsScanf=*/true) && I == E) {
+ if (ParseLengthModifier(FS, I, E, LO, FormatStrConverter,
+ /*IsScanf=*/true) &&
+ I == E) {
// No more characters left?
H.HandleIncompleteSpecifier(Start, E - Start);
return true;
@@ -155,7 +161,7 @@ static ScanfSpecifierResult ParseScanfSpecifier(FormatStringHandler &H,
// Finally, look for the conversion specifier.
const char *conversionPosition = I++;
ScanfConversionSpecifier::Kind k = ScanfConversionSpecifier::InvalidSpecifier;
- switch (*conversionPosition) {
+ switch (FormatStrConverter.convert(*conversionPosition)) {
default:
break;
case '%':
@@ -262,7 +268,8 @@ static ScanfSpecifierResult ParseScanfSpecifier(FormatStringHandler &H,
FS.setConversionSpecifier(CS);
}
// Assume the conversion takes one argument.
- return !H.HandleInvalidScanfConversionSpecifier(FS, Beg, Len);
+ return !H.HandleInvalidScanfConversionSpecifier(FS, Beg, Len,
+ FormatStrConverter);
}
return ScanfSpecifierResult(Start, FS);
}
diff --git a/clang/lib/Basic/TargetInfo.cpp b/clang/lib/Basic/TargetInfo.cpp
index ad083242d9e3e..83132c2370ed4 100644
--- a/clang/lib/Basic/TargetInfo.cpp
+++ b/clang/lib/Basic/TargetInfo.cpp
@@ -194,6 +194,9 @@ TargetInfo::TargetInfo(const llvm::Triple &T) : Triple(T) {
MaxOpenCLWorkGroupSize = 1024;
MaxBitIntWidth.reset();
+
+ FormatStrConverter = new llvm::TextEncodingConverter(
+ std::move(*llvm::TextEncodingConverter::createNoopConverter()));
}
// Out of line virtual dtor for TargetInfo.
diff --git a/clang/lib/Frontend/CompilerInstance.cpp b/clang/lib/Frontend/CompilerInstance.cpp
index dff396d91f2f1..f4859958e94c2 100644
--- a/clang/lib/Frontend/CompilerInstance.cpp
+++ b/clang/lib/Frontend/CompilerInstance.cpp
@@ -556,7 +556,7 @@ void CompilerInstance::createPreprocessor(TranslationUnitKind TUKind) {
PP->setDependencyDirectivesGetter(*GetDependencyDirectives);
if (auto EC = TextEncodingConfig::setConvertersFromOptions(
- PP->getTextEncodingConfig(), getLangOpts()))
+ PP->getTextEncodingConfig(), getLangOpts(), getTarget()))
PP->getDiagnostics().Report(clang::diag::err_fe_text_encoding_config)
<< PP->getTextEncodingConfig().getExecEncoding();
}
diff --git a/clang/lib/Lex/TextEncodingConfig.cpp b/clang/lib/Lex/TextEncodingConfig.cpp
index b89d5baefcc23..427b75a1c0a8b 100644
--- a/clang/lib/Lex/TextEncodingConfig.cpp
+++ b/clang/lib/Lex/TextEncodingConfig.cpp
@@ -23,7 +23,8 @@ TextEncodingConfig::getConverter(ConversionAction Action) const {
std::error_code
TextEncodingConfig::setConvertersFromOptions(TextEncodingConfig &TEC,
- const clang::LangOptions &Opts) {
+ const clang::LangOptions &Opts,
+ clang::TargetInfo &TInfo) {
using namespace llvm;
const char *UTF8 = "UTF-8";
@@ -41,5 +42,13 @@ TextEncodingConfig::setConvertersFromOptions(TextEncodingConfig &TEC,
new TextEncodingConverter(std::move(*ErrorOrConverter));
else
return ErrorOrConverter.getError();
+
+ ErrorOrConverter = llvm::TextEncodingConverter::create(TEC.SystemEncoding,
+ TEC.InternalEncoding);
+
+ if (ErrorOrConverter)
+ TInfo.FormatStrConverter =
+ new TextEncodingConverter(std::move(*ErrorOrConverter));
+
return std::error_code();
}
diff --git a/clang/lib/Sema/SemaChecking.cpp b/clang/lib/Sema/SemaChecking.cpp
index cc834bbee23c4..33a6659058a61 100644
--- a/clang/lib/Sema/SemaChecking.cpp
+++ b/clang/lib/Sema/SemaChecking.cpp
@@ -104,6 +104,7 @@
#include "llvm/Support/Locale.h"
#include "llvm/Support/MathExtras.h"
#include "llvm/Support/SaveAndRestore.h"
+#include "llvm/Support/TextEncoding.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/TargetParser/RISCVTargetParser.h"
#include "llvm/TargetParser/Triple.h"
@@ -7860,10 +7861,10 @@ class CheckFormatHandler : public analyze_format_string::FormatStringHandler {
ArrayRef<FixItHint> Fixit = {});
protected:
- bool HandleInvalidConversionSpecifier(unsigned argIndex, SourceLocation Loc,
- const char *startSpec,
- unsigned specifierLen,
- const char *csStart, unsigned csLen);
+ bool HandleInvalidConversionSpecifier(
+ unsigned argIndex, SourceLocation Loc, const char *startSpec,
+ unsigned specifierLen, const char *csStart, unsigned csLen,
+ const llvm::TextEncodingConverter &FormatStrConverter);
void HandlePositionalNonpositionalArgs(SourceLocation Loc,
const char *startSpec,
@@ -8093,7 +8094,8 @@ void UncoveredArgHandler::Diagnose(Sema &S, bool IsFunctionCall,
bool CheckFormatHandler::HandleInvalidConversionSpecifier(
unsigned argIndex, SourceLocation Loc, const char *startSpec,
- unsigned specifierLen, const char *csStart, unsigned csLen) {
+ unsigned specifierLen, const char *csStart, unsigned csLen,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
bool keepGoing = true;
if (argIndex < NumDataArgs) {
// Consider the argument coverered, even though the specifier doesn't
@@ -8108,7 +8110,13 @@ bool CheckFormatHandler::HandleInvalidConversionSpecifier(
keepGoing = false;
}
- StringRef Specifier(csStart, csLen);
+ // The csStart points to a character that has already been converted to the
+ // exec charset, so we have to reverse the conversion to allow diagnostic
+ // message to match an expected value when using -verify option,
+ std::string RS(csStart, csLen);
+ for (unsigned int i = 0; i < RS.size(); ++i)
+ RS[i] = FormatStrConverter.convert(RS[i]);
+ StringRef Specifier(RS);
// If the specifier in non-printable, it could be the first byte of a UTF-8
// sequence. In that case, print the UTF-8 code point. If not, print the byte
@@ -8262,7 +8270,8 @@ class CheckPrintfHandler : public CheckFormatHandler {
bool HandleInvalidPrintfConversionSpecifier(
const analyze_printf::PrintfSpecifier &FS, const char *startSpecifier,
- unsigned specifierLen) override;
+ unsigned specifierLen,
+ const llvm::TextEncodingConverter &FormatStrConverter) override;
void handleInvalidMaskType(StringRef MaskType) override;
@@ -8402,13 +8411,14 @@ class DecomposePrintfHandler : public CheckPrintfHandler {
bool CheckPrintfHandler::HandleInvalidPrintfConversionSpecifier(
const analyze_printf::PrintfSpecifier &FS, const char *startSpecifier,
- unsigned specifierLen) {
+ unsigned specifierLen,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
const analyze_printf::PrintfConversionSpecifier &CS =
FS.getConversionSpecifier();
return HandleInvalidConversionSpecifier(
FS.getArgIndex(), getLocationOfByte(CS.getStart()), startSpecifier,
- specifierLen, CS.getStart(), CS.getLength());
+ specifierLen, CS.getStart(), CS.getLength(), FormatStrConverter);
}
void CheckPrintfHandler::handleInvalidMaskType(StringRef MaskType) {
@@ -8916,15 +8926,15 @@ bool CheckPrintfHandler::HandlePrintfSpecifier(
// Check for using an Objective-C specific conversion specifier
// in a non-ObjC literal.
if (!allowsObjCArg() && CS.isObjCArg()) {
- return HandleInvalidPrintfConversionSpecifier(FS, startSpecifier,
- specifierLen);
+ return HandleInvalidPrintfConversionSpecifier(
+ FS, startSpecifier, specifierLen, *Target.FormatStrConverter);
}
// %P can only be used with os_log.
if (FSType != FormatStringType::OSLog &&
CS.getKind() == ConversionSpecifier::PArg) {
- return HandleInvalidPrintfConversionSpecifier(FS, startSpecifier,
- specifierLen);
+ return HandleInvalidPrintfConversionSpecifier(
+ FS, startSpecifier, specifierLen, *Target.FormatStrConverter);
}
// %n is not allowed with os_log.
@@ -8943,8 +8953,8 @@ bool CheckPrintfHandler::HandlePrintfSpecifier(
(CS.getKind() == ConversionSpecifier::PArg ||
CS.getKind() == ConversionSpecifier::sArg ||
CS.getKind() == ConversionSpecifier::ObjCObjArg)) {
- return HandleInvalidPrintfConversionSpecifier(FS, startSpecifier,
- specifierLen);
+ return HandleInvalidPrintfConversionSpecifier(
+ FS, startSpecifier, specifierLen, *Target.FormatStrConverter);
}
// Check for use of public/private annotation outside of os_log().
@@ -9602,10 +9612,10 @@ class CheckScanfHandler : public CheckFormatHandler {
const char *startSpecifier,
unsigned specifierLen) override;
- bool
- HandleInvalidScanfConversionSpecifier(const analyze_scanf::ScanfSpecifier &FS,
- const char *startSpecifier,
- unsigned specifierLen) override;
+ bool HandleInvalidScanfConversionSpecifier(
+ const analyze_scanf::ScanfSpecifier &FS, const char *startSpecifier,
+ unsigned specifierLen,
+ const llvm::TextEncodingConverter &FormatStrConverter) override;
void HandleIncompleteScanList(const char *start, const char *end) override;
};
@@ -9621,13 +9631,15 @@ void CheckScanfHandler::HandleIncompleteScanList(const char *start,
bool CheckScanfHandler::HandleInvalidScanfConversionSpecifier(
const analyze_scanf::ScanfSpecifier &FS, const char *startSpecifier,
- unsigned specifierLen) {
+ unsigned specifierLen,
+ const llvm::TextEncodingConverter &FormatStrConverter) {
+
const analyze_scanf::ScanfConversionSpecifier &CS =
FS.getConversionSpecifier();
return HandleInvalidConversionSpecifier(
FS.getArgIndex(), getLocationOfByte(CS.getStart()), startSpecifier,
- specifierLen, CS.getStart(), CS.getLength());
+ specifierLen, CS.getStart(), CS.getLength(), FormatStrConverter);
}
bool CheckScanfHandler::HandleScanfSpecifier(
diff --git a/llvm/include/llvm/Support/TextEncoding.h b/llvm/include/llvm/Support/TextEncoding.h
index 8a304910aa5dd..8f5a6122ede45 100644
--- a/llvm/include/llvm/Support/TextEncoding.h
+++ b/llvm/include/llvm/Support/TextEncoding.h
@@ -105,6 +105,8 @@ class TextEncodingConverter {
LLVM_ABI static ErrorOr<TextEncodingConverter> create(StringRef From,
StringRef To);
+ LLVM_ABI static ErrorOr<TextEncodingConverter> createNoopConverter();
+
TextEncodingConverter(const TextEncodingConverter &) = delete;
TextEncodingConverter &operator=(const TextEncodingConverter &) = delete;
@@ -135,6 +137,14 @@ class TextEncodingConverter {
return std::string(Result);
return EC;
}
+
+ char convert(char SingleChar) const {
+ SmallString<1> Result;
+ auto EC = Converter->convert(StringRef(&SingleChar, 1), Result);
+ if (!EC)
+ return Result[0];
+ return '\0';
+ }
};
} // namespace llvm
diff --git a/llvm/lib/Support/TextEncoding.cpp b/llvm/lib/Support/TextEncoding.cpp
index d36f02c1300b9..5c1d9696686a2 100644
--- a/llvm/lib/Support/TextEncoding.cpp
+++ b/llvm/lib/Support/TextEncoding.cpp
@@ -356,3 +356,22 @@ ErrorOr<TextEncodingConverter> TextEncodingConverter::create(StringRef From,
return std::make_error_code(std::errc::invalid_argument);
#endif
}
+
+class TextEncodingConverterNoop final
+ : public details::TextEncodingConverterImplBase {
+
+public:
+ TextEncodingConverterNoop() {}
+
+ std::error_code convertString(StringRef Source,
+ SmallVectorImpl<char> &Result) override {
+ Result.assign(Source.begin(), Source.end());
+ return std::error_code();
+ }
+
+ void reset() override {}
+};
+
+ErrorOr<TextEncodingConverter> TextEncodingConverter::createNoopConverter() {
+ return TextEncodingConverter(std::make_unique<TextEncodingConverterNoop>());
+}
>From 971993e62b780e6614d1814048271e9a7d1be563 Mon Sep 17 00:00:00 2001
From: Abhina Sreeskantharajan <Abhina.Sreeskantharajan at ibm.com>
Date: Wed, 13 May 2026 15:10:35 -0400
Subject: [PATCH 2/2] fix CI
---
clang/include/clang/Lex/TextEncodingConfig.h | 1 +
clang/lib/Lex/TextEncodingConfig.cpp | 4 ++--
2 files changed, 3 insertions(+), 2 deletions(-)
diff --git a/clang/include/clang/Lex/TextEncodingConfig.h b/clang/include/clang/Lex/TextEncodingConfig.h
index 97fffd38d28c5..3cc51cc9588e3 100644
--- a/clang/include/clang/Lex/TextEncodingConfig.h
+++ b/clang/include/clang/Lex/TextEncodingConfig.h
@@ -10,6 +10,7 @@
#define LLVM_CLANG_LEX_TEXTENCODINGCONFIG_H
#include "clang/Basic/LangOptions.h"
+#include "clang/Basic/TargetInfo.h"
#include "llvm/ADT/StringRef.h"
#include "llvm/Support/TextEncoding.h"
diff --git a/clang/lib/Lex/TextEncodingConfig.cpp b/clang/lib/Lex/TextEncodingConfig.cpp
index 427b75a1c0a8b..9cafdec6b9655 100644
--- a/clang/lib/Lex/TextEncodingConfig.cpp
+++ b/clang/lib/Lex/TextEncodingConfig.cpp
@@ -43,8 +43,8 @@ TextEncodingConfig::setConvertersFromOptions(TextEncodingConfig &TEC,
else
return ErrorOrConverter.getError();
- ErrorOrConverter = llvm::TextEncodingConverter::create(TEC.SystemEncoding,
- TEC.InternalEncoding);
+ ErrorOrConverter = llvm::TextEncodingConverter::create(
+ TInfo.getDefaultNarrowTextEncoding(), UTF8);
if (ErrorOrConverter)
TInfo.FormatStrConverter =
More information about the llvm-branch-commits
mailing list