https://github.com/juanvazquez created
https://github.com/llvm/llvm-project/pull/214967
This patch extends the unsafe buffer usage warning to cover std::string_view
constructors that take a pointer and size, similar to the existing check for
std::span.
The warning message has been updated to be generic ('container construction'
instead of 'span construction') and existing tests have been updated to match.
Reapplys #180471 by @pjalwadi adding a fix for the crash that lead to the
revert in #185692 by @nico.
Fixes #166644
>From af04ac1a10c4465e924cc748c87a703901fca7ae Mon Sep 17 00:00:00 2001
From: Prajwal P J <[email protected]>
Date: Sat, 8 Aug 2026 13:41:19 +0200
Subject: [PATCH 1/2] Reapply "[Clang][UnsafeBufferUsage] Warn about two-arg
string_view constructors. (#180471)" (#185692)
This reverts commit ffdf216cdada5d3caa844854c40290f1c61f4bfd.
---
clang/docs/ReleaseNotes.md | 2 +
.../Analysis/Analyses/UnsafeBufferUsage.h | 4 +
.../Analyses/UnsafeBufferUsageGadgets.def | 1 +
clang/include/clang/Basic/DiagnosticGroups.td | 3 +-
.../clang/Basic/DiagnosticSemaKinds.td | 6 +-
clang/lib/Analysis/UnsafeBufferUsage.cpp | 131 ++++++++++++++++++
clang/lib/Sema/AnalysisBasedWarnings.cpp | 36 ++++-
.../warn-unsafe-buffer-usage-string-view.cpp | 44 ++++++
8 files changed, 222 insertions(+), 5 deletions(-)
create mode 100644 clang/test/SemaCXX/warn-unsafe-buffer-usage-string-view.cpp
diff --git a/clang/docs/ReleaseNotes.md b/clang/docs/ReleaseNotes.md
index 04aaa038930d1..6f1ea7b970fb7 100644
--- a/clang/docs/ReleaseNotes.md
+++ b/clang/docs/ReleaseNotes.md
@@ -362,6 +362,8 @@ features cannot lower the translation-unit ABI level;
- Clang now attempts to print enumerator names rather than C-style cast
expressions
in more diagnostics.
+- `-Wunsafe-buffer-usage` now warns about unsafe two-parameter constructors of
+ `std::string_view` (pointer and size), consistent with the existing warning
for `std::span`.
### Improvements to Clang's time-trace
diff --git a/clang/include/clang/Analysis/Analyses/UnsafeBufferUsage.h
b/clang/include/clang/Analysis/Analyses/UnsafeBufferUsage.h
index 65b79cfc1c2ff..05a580bed5c39 100644
--- a/clang/include/clang/Analysis/Analyses/UnsafeBufferUsage.h
+++ b/clang/include/clang/Analysis/Analyses/UnsafeBufferUsage.h
@@ -129,6 +129,10 @@ class UnsafeBufferUsageHandler {
bool IsRelatedToDecl,
ASTContext &Ctx) = 0;
+ virtual void handleUnsafeOperationInStringView(const Stmt *Operation,
+ bool IsRelatedToDecl,
+ ASTContext &Ctx) = 0;
+
/// Invoked when a fix is suggested against a variable. This function groups
/// all variables that must be fixed together (i.e their types must be
changed
/// to the same target type to prevent type mismatches) into a single fixit.
diff --git a/clang/include/clang/Analysis/Analyses/UnsafeBufferUsageGadgets.def
b/clang/include/clang/Analysis/Analyses/UnsafeBufferUsageGadgets.def
index 129ce95c1c0e0..7ce3c5f0fc7c5 100644
--- a/clang/include/clang/Analysis/Analyses/UnsafeBufferUsageGadgets.def
+++ b/clang/include/clang/Analysis/Analyses/UnsafeBufferUsageGadgets.def
@@ -42,6 +42,7 @@ WARNING_OPTIONAL_GADGET(ArraySubscript)
WARNING_OPTIONAL_GADGET(UnsafeLibcFunctionCall)
WARNING_OPTIONAL_GADGET(UnsafeFormatAttributedFunctionCall)
WARNING_OPTIONAL_GADGET(SpanTwoParamConstructor) // Uses of `std::span(arg0,
arg1)`
+WARNING_OPTIONAL_GADGET(StringViewTwoParamConstructor)
FIXABLE_GADGET(ULCArraySubscript) // `DRE[any]` in an Unspecified
Lvalue Context
FIXABLE_GADGET(DerefSimplePtrArithFixable)
FIXABLE_GADGET(PointerDereference)
diff --git a/clang/include/clang/Basic/DiagnosticGroups.td
b/clang/include/clang/Basic/DiagnosticGroups.td
index 3f6f88cb46bca..caf365ddca84a 100644
--- a/clang/include/clang/Basic/DiagnosticGroups.td
+++ b/clang/include/clang/Basic/DiagnosticGroups.td
@@ -1982,7 +1982,8 @@ def UnsafeBufferUsageInLibcCall :
DiagGroup<"unsafe-buffer-usage-in-libc-call">;
def UnsafeBufferUsageInUniquePtrArrayAccess :
DiagGroup<"unsafe-buffer-usage-in-unique-ptr-array-access">;
def UnsafeBufferUsageInStaticSizedArray :
DiagGroup<"unsafe-buffer-usage-in-static-sized-array">;
def UnsafeBufferUsageInFormatAttrCall :
DiagGroup<"unsafe-buffer-usage-in-format-attr-call">;
-def UnsafeBufferUsage : DiagGroup<"unsafe-buffer-usage",
[UnsafeBufferUsageInContainer, UnsafeBufferUsageInLibcCall,
UnsafeBufferUsageInUniquePtrArrayAccess, UnsafeBufferUsageInStaticSizedArray,
UnsafeBufferUsageInFormatAttrCall]>;
+def UnsafeBufferUsageInStringView :
DiagGroup<"unsafe-buffer-usage-in-string-view">;
+def UnsafeBufferUsage : DiagGroup<"unsafe-buffer-usage",
[UnsafeBufferUsageInContainer, UnsafeBufferUsageInLibcCall,
UnsafeBufferUsageInUniquePtrArrayAccess, UnsafeBufferUsageInStaticSizedArray,
UnsafeBufferUsageInFormatAttrCall, UnsafeBufferUsageInStringView]>;
// Warnings and notes InstallAPI verification.
def InstallAPIViolation : DiagGroup<"installapi-violation">;
diff --git a/clang/include/clang/Basic/DiagnosticSemaKinds.td
b/clang/include/clang/Basic/DiagnosticSemaKinds.td
index b57ff6c197d1d..368a4b689b302 100644
--- a/clang/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/clang/include/clang/Basic/DiagnosticSemaKinds.td
@@ -13909,8 +13909,12 @@ def note_unsafe_buffer_variable_fixit_together : Note<
def note_safe_buffer_usage_suggestions_disabled : Note<
"pass -fsafe-buffer-usage-suggestions to receive code hardening
suggestions">;
def warn_unsafe_buffer_usage_in_container : Warning<
- "the two-parameter std::span construction is unsafe as it can introduce
mismatch between buffer size and the bound information">,
+ "the two-parameter %0 construction is unsafe as it can introduce mismatch "
+ "between buffer size and the bound information">,
InGroup<UnsafeBufferUsageInContainer>, DefaultIgnore;
+def warn_unsafe_buffer_usage_in_string_view : Warning<
+ "the two-parameter std::string_view construction is unsafe">,
+ InGroup<UnsafeBufferUsageInStringView>, DefaultIgnore;
def warn_unsafe_buffer_usage_unique_ptr_array_access : Warning<"direct access
using operator[] on std::unique_ptr<T[]> is unsafe due to lack of bounds
checking">,
InGroup<UnsafeBufferUsageInUniquePtrArrayAccess>, DefaultIgnore;
def warn_unsafe_buffer_usage_in_static_sized_array : Warning<"direct access on
T[N] is unsafe due to the lack of bounds checking">,
diff --git a/clang/lib/Analysis/UnsafeBufferUsage.cpp
b/clang/lib/Analysis/UnsafeBufferUsage.cpp
index e80dfc82e60ba..41efb5a20054c 100644
--- a/clang/lib/Analysis/UnsafeBufferUsage.cpp
+++ b/clang/lib/Analysis/UnsafeBufferUsage.cpp
@@ -701,6 +701,71 @@ static bool isSafeSpanTwoParamConstruct(const
CXXConstructExpr &Node,
return isPtrBufferSafe(Arg0, Arg1, Ctx);
}
+static bool isSafeStringViewTwoParamConstruct(const CXXConstructExpr &Node,
+ ASTContext &Ctx) {
+ const Expr *Arg0 = Node.getArg(0)->IgnoreParenImpCasts();
+ const Expr *Arg1 = Node.getArg(1)->IgnoreParenImpCasts();
+
+ // Pattern 1: String Literals
+ if (const auto *SL = dyn_cast<StringLiteral>(Arg0)) {
+ if (auto ArgSize = Arg1->getIntegerConstantExpr(Ctx)) {
+ if (llvm::APSInt::compareValues(
+ llvm::APSInt::getUnsigned(SL->getLength()), *ArgSize) >= 0)
+ return true;
+ return false; // Explicitly unsafe if size > length
+ }
+ }
+
+ // Pattern 2: Constant Arrays
+ if (const auto *CAT = Ctx.getAsConstantArrayType(Arg0->getType())) {
+ if (auto ArgSize = Arg1->getIntegerConstantExpr(Ctx)) {
+ if (llvm::APSInt::compareValues(llvm::APSInt(CAT->getSize(), true),
+ *ArgSize) >= 0)
+ return true;
+ return false; // Explicitly unsafe if size > ArraySize
+ }
+ }
+
+ // Pattern 3: Zero length
+ if (auto Val = Arg1->getIntegerConstantExpr(Ctx)) {
+ if (Val->isZero())
+ return true;
+ }
+
+ // Pattern 4: string_view(it, it) - Only safe if it's .begin() and .end() of
+ // the SAME object
+ auto GetContainerObj = [](const Expr *E) -> const Expr * {
+ E = E->IgnoreParenImpCasts();
+ if (const auto *MCE = dyn_cast<CXXMemberCallExpr>(E)) {
+ const auto *MD = MCE->getMethodDecl();
+ if (MD && (MD->getName() == "begin" || MD->getName() == "end"))
+ return MCE->getImplicitObjectArgument()->IgnoreParenImpCasts();
+ }
+ return nullptr;
+ };
+
+ const Expr *Obj0 = GetContainerObj(Arg0);
+ const Expr *Obj1 = GetContainerObj(Arg1);
+
+ if (Obj0 && Obj1) {
+ const auto *DRE0 = dyn_cast<DeclRefExpr>(Obj0);
+ const auto *DRE1 = dyn_cast<DeclRefExpr>(Obj1);
+
+ // If both are references to variables, they MUST point to the same
+ // declaration.
+ if (DRE0 && DRE1) {
+ if (DRE0->getDecl()->getCanonicalDecl() ==
+ DRE1->getDecl()->getCanonicalDecl())
+ return true;
+ }
+
+ // If they aren't both DeclRefExprs or don't match, we DO NOT return true.
+ // This ensures v1.begin(), v2.end() triggers a warning.
+ }
+
+ return false; // Default to unsafe
+}
+
static bool isSafeArraySubscript(const ArraySubscriptExpr &Node,
const ASTContext &Ctx,
const bool IgnoreStaticSizedArrays) {
@@ -1804,6 +1869,70 @@ class SpanTwoParamConstructorGadget : public
WarningGadget {
SmallVector<const Expr *, 1> getUnsafePtrs() const override { return {}; }
};
+class StringViewTwoParamConstructorGadget : public WarningGadget {
+ static constexpr const char *const StringViewTwoParamConstructorTag =
+ "stringViewTwoParamConstructor";
+ const CXXConstructExpr *Ctor; // the string_view constructor expression
+
+public:
+ StringViewTwoParamConstructorGadget(const MatchResult &Result)
+ : WarningGadget(Kind::StringViewTwoParamConstructor),
+ Ctor(Result.getNodeAs<CXXConstructExpr>(
+ StringViewTwoParamConstructorTag)) {}
+
+ static bool classof(const Gadget *G) {
+ return G->getKind() == Kind::StringViewTwoParamConstructor;
+ }
+
+ static bool matches(const Stmt *S, ASTContext &Ctx, MatchResult &Result) {
+ const auto *CE = dyn_cast<CXXConstructExpr>(S);
+ if (!CE)
+ return false;
+ const auto *CDecl = CE->getConstructor();
+ const auto *CRecordDecl = CDecl->getParent();
+
+ // MATCH: std::basic_string_view
+ bool IsStringView =
+ CRecordDecl->isInStdNamespace() &&
+ CDecl->getDeclName().getAsString() == "basic_string_view" &&
+ CE->getNumArgs() == 2;
+
+ if (!IsStringView || isSafeStringViewTwoParamConstruct(*CE, Ctx))
+ return false;
+
+ Result.addNode(StringViewTwoParamConstructorTag,
DynTypedNode::create(*CE));
+ return true;
+ }
+
+ static bool matches(const Stmt *S, ASTContext &Ctx,
+ const UnsafeBufferUsageHandler *Handler,
+ MatchResult &Result) {
+ if (ignoreUnsafeBufferInContainer(*S, Handler))
+ return false;
+ return matches(S, Ctx, Result);
+ }
+
+ void handleUnsafeOperation(UnsafeBufferUsageHandler &Handler,
+ bool IsRelatedToDecl,
+ ASTContext &Ctx) const override {
+ Handler.handleUnsafeOperationInStringView(Ctor, IsRelatedToDecl, Ctx);
+ }
+
+ SourceLocation getSourceLoc() const override { return Ctor->getBeginLoc(); }
+
+ DeclUseList getClaimedVarUseSites() const override {
+ // If the constructor call is of the form `std::string_view{var, n}`, `var`
+ // is considered an unsafe variable.
+ if (auto *DRE = dyn_cast<DeclRefExpr>(Ctor->getArg(0))) {
+ if (isa<VarDecl>(DRE->getDecl()))
+ return {DRE};
+ }
+ return {};
+ }
+
+ SmallVector<const Expr *, 1> getUnsafePtrs() const override { return {}; }
+};
+
/// A pointer initialization expression of the form:
/// \code
/// int *p = q;
@@ -4703,6 +4832,8 @@ bool clang::matchUnsafePointers(const DynTypedNode &N,
ASTContext &Ctx,
const Expr *UnsafeArg = nullptr) override {}
void handleUnsafeOperationInContainer(const Stmt *, bool,
ASTContext &) override {}
+ void handleUnsafeOperationInStringView(const Stmt *, bool,
+ ASTContext &) override {}
void handleUnsafeVariableGroup(const VarDecl *,
const VariableGroupsManager &, FixItList &&,
const Decl *,
diff --git a/clang/lib/Sema/AnalysisBasedWarnings.cpp
b/clang/lib/Sema/AnalysisBasedWarnings.cpp
index 9960806d5d09b..d0500a6defd64 100644
--- a/clang/lib/Sema/AnalysisBasedWarnings.cpp
+++ b/clang/lib/Sema/AnalysisBasedWarnings.cpp
@@ -2591,12 +2591,19 @@ class UnsafeBufferUsageReporter : public
UnsafeBufferUsageHandler {
SourceRange Range;
unsigned MsgParam = 0;
- // This function only handles SpanTwoParamConstructorGadget so far, which
- // always gives a CXXConstructExpr.
const auto *CtorExpr = cast<CXXConstructExpr>(Operation);
Loc = CtorExpr->getLocation();
+ Range = CtorExpr->getSourceRange();
+
+ std::string ContainerName = "std::span";
+ if (auto *TD = CtorExpr->getConstructor()->getParent()) {
+ // This will provide "std::span" if it's in the std namespace
+ ContainerName = TD->getQualifiedNameAsString();
+ }
+
+ // FIX: Pass the container name to fill the %0 parameter
+ S.Diag(Loc, diag::warn_unsafe_buffer_usage_in_container) << ContainerName;
- S.Diag(Loc, diag::warn_unsafe_buffer_usage_in_container);
if (IsRelatedToDecl) {
assert(!SuggestSuggestions &&
"Variables blamed for unsafe buffer usage without suggestions!");
@@ -2604,6 +2611,29 @@ class UnsafeBufferUsageReporter : public
UnsafeBufferUsageHandler {
}
}
+ void handleUnsafeOperationInStringView(const Stmt *Operation,
+ bool IsRelatedToDecl,
+ ASTContext &Ctx) override {
+ // Extracting location: prioritize the specific location of the constructor
+ SourceLocation Loc = Operation->getBeginLoc();
+ SourceRange Range = Operation->getSourceRange();
+
+ if (const auto *CtorExpr = dyn_cast<CXXConstructExpr>(Operation)) {
+ Loc = CtorExpr->getLocation();
+ }
+
+ // 1. Emit the primary warning for string_view
+ S.Diag(Loc, diag::warn_unsafe_buffer_usage_in_container)
+ << "std::string_view";
+
+ // 2. If a specific variable is 'blamed', emit the note
+ if (IsRelatedToDecl) {
+ // MsgParam 0 is "unsafe operation"
+ // Range helps the IDE underline the whole expression
+ S.Diag(Loc, diag::note_unsafe_buffer_operation) << 0 << Range;
+ }
+ }
+
void handleUnsafeVariableGroup(const VarDecl *Variable,
const VariableGroupsManager &VarGrpMgr,
FixItList &&Fixes, const Decl *D,
diff --git a/clang/test/SemaCXX/warn-unsafe-buffer-usage-string-view.cpp
b/clang/test/SemaCXX/warn-unsafe-buffer-usage-string-view.cpp
new file mode 100644
index 0000000000000..f63986b3f2554
--- /dev/null
+++ b/clang/test/SemaCXX/warn-unsafe-buffer-usage-string-view.cpp
@@ -0,0 +1,44 @@
+// RUN: %clang_cc1 -std=c++20 -Wunsafe-buffer-usage-in-container -verify %s
+
+namespace std {
+ typedef __SIZE_TYPE__ size_t;
+ template <typename T> class basic_string_view {
+ public:
+ basic_string_view(const T *, size_t);
+ template <typename It> basic_string_view(It, It);
+ };
+ typedef basic_string_view<char> string_view;
+ typedef basic_string_view<wchar_t> wstring_view;
+ template <typename T> class vector { public: T* begin(); T* end(); };
+}
+
+typedef std::size_t size_t;
+
+void test_final_coverage() {
+ std::vector<char> v1, v2;
+
+ // 1. Iterator Pairs
+ std::string_view it_ok(v1.begin(), v1.end()); // no-warning
+ // expected-warning@+1 {{the two-parameter std::string_view construction is
unsafe}}
+ std::string_view it_bad(v1.begin(), v2.end());
+
+ // 2. Character Types
+ std::string_view s1("hi", 2); // no-warning
+ // expected-warning@+1 {{the two-parameter std::string_view construction is
unsafe}}
+ std::string_view s2("hi", 3);
+
+ std::wstring_view w1(L"hi", 2); // no-warning
+ // expected-warning@+1 {{the two-parameter std::string_view construction is
unsafe}}
+ std::wstring_view w2(L"hi", 3);
+
+ // 3. Arrays
+ char arr[5];
+ std::string_view a1(arr, 5); // no-warning
+ // expected-warning@+1 {{the two-parameter std::string_view construction is
unsafe}}
+ std::string_view a2(arr, 6);
+
+ // 4. Dynamic/Unknown
+ extern size_t get_size();
+ // expected-warning@+1 {{the two-parameter std::string_view construction is
unsafe}}
+ std::string_view d1("hi", get_size());
+}
>From 5b9e111df2b5129176669ef8cfb65c4352d3556d Mon Sep 17 00:00:00 2001
From: Juan Vazquez <[email protected]>
Date: Sat, 8 Aug 2026 16:22:23 +0200
Subject: [PATCH 2/2] Fix assertion failure on iterator to ptr implicit
conversion
Fixes a crash introduced in PR #180471 where GetContainerObj triggered an
isIdentifier() assertion when attempting to read the name of a C++ operator
resulting from implicit iterator conversions.
---
clang/lib/Analysis/UnsafeBufferUsage.cpp | 5 +++--
.../warn-unsafe-buffer-usage-string-view.cpp | 18 ++++++++++++++++++
2 files changed, 21 insertions(+), 2 deletions(-)
diff --git a/clang/lib/Analysis/UnsafeBufferUsage.cpp
b/clang/lib/Analysis/UnsafeBufferUsage.cpp
index 41efb5a20054c..64b524de829b5 100644
--- a/clang/lib/Analysis/UnsafeBufferUsage.cpp
+++ b/clang/lib/Analysis/UnsafeBufferUsage.cpp
@@ -738,8 +738,9 @@ static bool isSafeStringViewTwoParamConstruct(const
CXXConstructExpr &Node,
E = E->IgnoreParenImpCasts();
if (const auto *MCE = dyn_cast<CXXMemberCallExpr>(E)) {
const auto *MD = MCE->getMethodDecl();
- if (MD && (MD->getName() == "begin" || MD->getName() == "end"))
- return MCE->getImplicitObjectArgument()->IgnoreParenImpCasts();
+ if (MD && MD->getIdentifier())
+ if (MD->getName() == "begin" || MD->getName() == "end")
+ return MCE->getImplicitObjectArgument()->IgnoreParenImpCasts();
}
return nullptr;
};
diff --git a/clang/test/SemaCXX/warn-unsafe-buffer-usage-string-view.cpp
b/clang/test/SemaCXX/warn-unsafe-buffer-usage-string-view.cpp
index f63986b3f2554..fa5d4ab84b009 100644
--- a/clang/test/SemaCXX/warn-unsafe-buffer-usage-string-view.cpp
+++ b/clang/test/SemaCXX/warn-unsafe-buffer-usage-string-view.cpp
@@ -42,3 +42,21 @@ void test_final_coverage() {
// expected-warning@+1 {{the two-parameter std::string_view construction is
unsafe}}
std::string_view d1("hi", get_size());
}
+
+template <typename T>
+struct vector_iterator {
+ operator T*() const;
+};
+
+struct my_vector {
+ vector_iterator <char> begin();
+ vector_iterator <char> end();
+ int length();
+};
+
+void test_iterators_with_implicit_conversion() {
+ my_vector v;
+ // expected-warning@+1 {{the two-parameter std::string_view construction is
unsafe}}
+ std::string_view sv(v.begin(), v.length());
+ std::string_view sv2(v.begin(), v.end()); // no-warning
+}
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits