[clang] [analyzer] Fix core.VLASize checker false positive taint reports (PR #68140)

[Daniel Krupp via cfe-commits]

https://github.com/dkrupp created https://github.com/llvm/llvm-project/pull/68140

The checker reported a false positive on this code 

void testTaintedSanitizedVLASize(void) {
  int x;
  scanf("%d", &x);
  if (x<1)
    return;
  int vla[x]; // no-warning
}

After the fix, the checker only emits tainted warning if the vla size is coming from a tainted source and it cannot prove that it is positive.

>From 4b310278d2923ff718d074a7f7c8806ad03c6401 Mon Sep 17 00:00:00 2001
From: Daniel Krupp <daniel.krupp at ericsson.com>
Date: Tue, 3 Oct 2023 19:58:28 +0200
Subject: [PATCH] [analyzer] Fix core.VLASize checker false positive taint
 reports

The checker reported a false positive on this code
void testTaintedSanitizedVLASize(void) {
  int x;
  scanf("%d", &x);
  if (x<1)
    return;
  int vla[x]; // no-warning
}

After the fix, the checker only emits tainted warning if the vla size is
coming from a tainted source and it cannot prove that it is positive.
---
 .../StaticAnalyzer/Checkers/VLASizeChecker.cpp   | 16 ++++++++--------
 clang/test/Analysis/taint-diagnostic-visitor.c   |  4 ++--
 clang/test/Analysis/taint-generic.c              | 11 ++++++++++-
 3 files changed, 20 insertions(+), 11 deletions(-)

diff --git a/clang/lib/StaticAnalyzer/Checkers/VLASizeChecker.cpp b/clang/lib/StaticAnalyzer/Checkers/VLASizeChecker.cpp
index b195d912cadfe9b..46b5f5e10f0e65c 100644
--- a/clang/lib/StaticAnalyzer/Checkers/VLASizeChecker.cpp
+++ b/clang/lib/StaticAnalyzer/Checkers/VLASizeChecker.cpp
@@ -162,12 +162,6 @@ ProgramStateRef VLASizeChecker::checkVLAIndexSize(CheckerContext &C,
   if (SizeV.isUnknown())
     return nullptr;
 
-  // Check if the size is tainted.
-  if (isTainted(State, SizeV)) {
-    reportTaintBug(SizeE, State, C, SizeV);
-    return nullptr;
-  }
-
   // Check if the size is zero.
   DefinedSVal SizeD = SizeV.castAs<DefinedSVal>();
 
@@ -189,10 +183,10 @@ ProgramStateRef VLASizeChecker::checkVLAIndexSize(CheckerContext &C,
   DefinedOrUnknownSVal Zero = SVB.makeZeroVal(SizeTy);
 
   SVal LessThanZeroVal = SVB.evalBinOp(State, BO_LT, SizeD, Zero, SizeTy);
+  ProgramStateRef StatePos, StateNeg;
   if (std::optional<DefinedSVal> LessThanZeroDVal =
           LessThanZeroVal.getAs<DefinedSVal>()) {
     ConstraintManager &CM = C.getConstraintManager();
-    ProgramStateRef StatePos, StateNeg;
 
     std::tie(StateNeg, StatePos) = CM.assumeDual(State, *LessThanZeroDVal);
     if (StateNeg && !StatePos) {
@@ -202,6 +196,12 @@ ProgramStateRef VLASizeChecker::checkVLAIndexSize(CheckerContext &C,
     State = StatePos;
   }
 
+  // Check if the size is tainted.
+  if ((StateNeg || StateZero) && isTainted(State, SizeV)) {
+    reportTaintBug(SizeE, State, C, SizeV);
+    return nullptr;
+  }
+
   return State;
 }
 
@@ -220,7 +220,7 @@ void VLASizeChecker::reportTaintBug(const Expr *SizeE, ProgramStateRef State,
   SmallString<256> buf;
   llvm::raw_svector_ostream os(buf);
   os << "Declared variable-length array (VLA) ";
-  os << "has tainted size";
+  os << "has a tainted (attacker controlled) size, that can be 0 or negative";
 
   auto report = std::make_unique<PathSensitiveBugReport>(*TaintBT, os.str(), N);
   report->addRange(SizeE->getSourceRange());
diff --git a/clang/test/Analysis/taint-diagnostic-visitor.c b/clang/test/Analysis/taint-diagnostic-visitor.c
index 8a7510177f3e444..45369785ed6924e 100644
--- a/clang/test/Analysis/taint-diagnostic-visitor.c
+++ b/clang/test/Analysis/taint-diagnostic-visitor.c
@@ -46,8 +46,8 @@ void taintDiagnosticVLA(void) {
   scanf("%d", &x); // expected-note {{Value assigned to 'x'}}
                    // expected-note at -1 {{Taint originated here}}
                    // expected-note at -2 {{Taint propagated to the 2nd argument}}
-  int vla[x]; // expected-warning {{Declared variable-length array (VLA) has tainted size}}
-              // expected-note at -1 {{Declared variable-length array (VLA) has tainted size}}
+  int vla[x]; // expected-warning {{Declared variable-length array (VLA) has a tainted}}
+              // expected-note at -1 {{Declared variable-length array (VLA) has a tainted}}
 }
 
 
diff --git a/clang/test/Analysis/taint-generic.c b/clang/test/Analysis/taint-generic.c
index c6a01594f15abb7..ae2ae5b23aab3c6 100644
--- a/clang/test/Analysis/taint-generic.c
+++ b/clang/test/Analysis/taint-generic.c
@@ -405,7 +405,16 @@ int testDivByZero(void) {
 void testTaintedVLASize(void) {
   int x;
   scanf("%d", &x);
-  int vla[x]; // expected-warning{{Declared variable-length array (VLA) has tainted size}}
+  int vla[x]; // expected-warning{{Declared variable-length array (VLA) has a tainted (attacker controlled) size, that can be 0 or negative}}
+}
+
+// Tainted-sanitized VLAs.
+void testTaintedSanitizedVLASize(void) {
+  int x;
+  scanf("%d", &x);
+  if (x<1)
+    return;
+  int vla[x]; // no-warning
 }
 
 int testTaintedAllocaMem() {