release/22.x: [Sema] Fix ICE due to incorrect _Bool handling in format string checking (#174684)

[llvmbot]

Backport 15365d3

Requested by: @zyn0217

[llvmbot]

@cor3ntin What do you think about merging this PR to the release branch?

[llvmbot]

@llvm/pr-subscribers-clang

Author: None (llvmbot)

Changes

Backport 15365d3

Requested by: @zyn0217

---

Full diff: https://github.com/llvm/llvm-project/pull/178133.diff

2 Files Affected:

• (modified) clang/lib/AST/FormatString.cpp (+1-1)
• (modified) clang/test/Sema/format-strings.c (+32)

diff --git a/clang/lib/AST/FormatString.cpp b/clang/lib/AST/FormatString.cpp
index d4cb89b43ae87..36c5f57671631 100644
--- a/clang/lib/AST/FormatString.cpp
+++ b/clang/lib/AST/FormatString.cpp
@@ -371,7 +371,7 @@ static clang::analyze_format_string::ArgType::MatchKind
 matchesSizeTPtrdiffT(ASTContext &C, QualType T, QualType E) {
   using MatchKind = clang::analyze_format_string::ArgType::MatchKind;
 
-  if (!T->isIntegerType())
+  if (!T->isIntegerType() || T->isBooleanType())
     return MatchKind::NoMatch;
 
   if (C.hasSameType(T, E))
diff --git a/clang/test/Sema/format-strings.c b/clang/test/Sema/format-strings.c
index 5280549adc3d7..cdac83fc0bb91 100644
--- a/clang/test/Sema/format-strings.c
+++ b/clang/test/Sema/format-strings.c
@@ -983,3 +983,35 @@ void test_promotion(void) {
   // pointers
   printf("%s", i); // expected-warning{{format specifies type 'char *' but the argument has type 'int'}}
 }
+
+void test_bool(_Bool b, _Bool* bp)
+{
+  printf("%zu", b); // expected-warning-re{{format specifies type 'size_t' (aka '{{.+}}') but the argument has type '_Bool'}}
+  printf("%td", b); // expected-warning-re{{format specifies type 'ptrdiff_t' (aka '{{.+}}') but the argument has type '_Bool'}}
+  printf("%jd", b); // expected-warning-re{{format specifies type 'intmax_t' (aka '{{.+}}') but the argument has type '_Bool'}}
+  printf("%lld", b); // expected-warning{{format specifies type 'long long' but the argument has type '_Bool'}}
+  printf("%ld", b); // expected-warning{{format specifies type 'long' but the argument has type '_Bool'}}
+  printf("%d", b); // promoted from _Bool to int
+  printf("%hhd", b); // promoted from _Bool to int
+  printf("%hd", b); // promoted from _Bool to int
+#if !defined(__Fuchsia__) && !defined(__ANDROID__) //'%n' specifier not supported on this platform
+  // The n conversion specifier only supports signed types
+  printf("%zn", bp); // expected-warning-re{{format specifies type 'signed size_t *' (aka '{{.+}}') but the argument has type '_Bool *'}}
+  printf("%jn", bp); // expected-warning-re{{format specifies type 'intmax_t *' (aka '{{.+}}') but the argument has type '_Bool *'}}
+  printf("%lln", bp); // expected-warning{{format specifies type 'long long *' but the argument has type '_Bool *'}}
+  printf("%ln", bp); // expected-warning{{format specifies type 'long *' but the argument has type '_Bool *'}}
+  printf("%n", bp); // expected-warning{{format specifies type 'int *' but the argument has type '_Bool *'}}
+  printf("%hhn", bp); // expected-warning{{format specifies type 'signed char *' but the argument has type '_Bool *'}}
+  printf("%hn", bp); // belong to -Wformat-type-confusion
+#endif
+  printf("%c", b); // expected-warning{{using '%c' format specifier, but argument has boolean value}}
+  printf("%s", b); // expected-warning{{format specifies type 'char *' but the argument has type '_Bool'}}
+  printf("%d", b); // promoted from _Bool to int
+  printf("%o", b); // promoted from _Bool to int
+  printf("%x", b); // promoted from _Bool to int
+  printf("%u", b); // promoted from _Bool to int
+  printf("%f", b); // expected-warning{{format specifies type 'double' but the argument has type '_Bool'}}
+  printf("%e", b); // expected-warning{{format specifies type 'double' but the argument has type '_Bool'}}
+  printf("%a", b); // expected-warning{{format specifies type 'double' but the argument has type '_Bool'}}
+  printf("%g", b); // expected-warning{{format specifies type 'double' but the argument has type '_Bool'}}
+}

[dyung]

Note that this change alone is causing a lit test failure on 32-bit arm buildbots. See #174684 for details.

[c-rhodes]

test was fixed in #178450

[cor3ntin]

This pr should include the changes in #178450 then

[c-rhodes]

@zyn0217 any plans to fix this test? This has been open over a week now and rc3 is in less than a week

[zyn0217]

Sorry I missed this. I'll replace it with a manual backport PR

[zyn0217]

New PR in #179637