Carving out -Wformat warning about scoped enums into a subwarning

[ZijunZhaoCCK]

Make it part of -Wformat-pedantic.

Fixes #81647

[llvmbot]

@llvm/pr-subscribers-clang

Author: None (ZijunZhaoCCK)

Changes

Make it part of -Wformat-pedantic.

---

Full diff: https://github.com/llvm/llvm-project/pull/88595.diff

3 Files Affected:

• (modified) clang/include/clang/Basic/DiagnosticSemaKinds.td (+2-2)
• (modified) clang/test/Sema/format-strings-scanf.c (+1-1)
• (modified) clang/test/Sema/format-strings-signedness.c (+3-3)

diff --git a/clang/include/clang/Basic/DiagnosticSemaKinds.td b/clang/include/clang/Basic/DiagnosticSemaKinds.td
index 774d2b53a38252..4ba27d62208da4 100644
--- a/clang/include/clang/Basic/DiagnosticSemaKinds.td
+++ b/clang/include/clang/Basic/DiagnosticSemaKinds.td
@@ -9827,7 +9827,7 @@ def warn_scanf_nonzero_width : Warning<
 def warn_format_conversion_argument_type_mismatch : Warning<
   "format specifies type %0 but the argument has "
   "%select{type|underlying type}2 %1">,
-  InGroup<Format>;
+  InGroup<FormatPedantic>;
 def warn_format_conversion_argument_type_mismatch_pedantic : Extension<
   warn_format_conversion_argument_type_mismatch.Summary>,
   InGroup<FormatPedantic>;
@@ -9840,7 +9840,7 @@ def warn_format_conversion_argument_type_mismatch_confusion : Warning<
 def warn_format_argument_needs_cast : Warning<
   "%select{values of type|enum values with underlying type}2 '%0' should not "
   "be used as format arguments; add an explicit cast to %1 instead">,
-  InGroup<Format>;
+  InGroup<FormatPedantic>;
 def warn_format_argument_needs_cast_pedantic : Warning<
   warn_format_argument_needs_cast.Summary>,
   InGroup<FormatPedantic>, DefaultIgnore;
diff --git a/clang/test/Sema/format-strings-scanf.c b/clang/test/Sema/format-strings-scanf.c
index eb5b8ec36bf7a4..9bdc46bfeebc3b 100644
--- a/clang/test/Sema/format-strings-scanf.c
+++ b/clang/test/Sema/format-strings-scanf.c
@@ -1,7 +1,7 @@
 // RUN: %clang_cc1 -std=c11 -fsyntax-only -verify -Wformat-nonliteral %s
 
 // Test that -Wformat=0 works:
-// RUN: %clang_cc1 -std=c11 -fsyntax-only -Werror -Wformat=0 %s
+// RUN: %clang_cc1 -std=c11 -fsyntax-only -Werror -Wformat=0 -Wno-format-pedantic %s
 
 #include <stdarg.h>
 typedef __SIZE_TYPE__ size_t;
diff --git a/clang/test/Sema/format-strings-signedness.c b/clang/test/Sema/format-strings-signedness.c
index d5a8140d9ef8a0..ccd4d72c2d2635 100644
--- a/clang/test/Sema/format-strings-signedness.c
+++ b/clang/test/Sema/format-strings-signedness.c
@@ -12,8 +12,8 @@
 // RUN: %clang_cc1 -triple=x86_64-pc-linux-gnu -std=c11 -fsyntax-only -Wformat -verify=okay %s
 
 // Verify that -Wformat-signedness with -Wno-format are not reported (gcc compat).
-// RUN: %clang_cc1 -triple=x86_64-pc-linux-gnu -std=c11 -fsyntax-only -Wformat-signedness -Wno-format -verify=okay %s
-// RUN: %clang_cc1 -triple=x86_64-pc-linux-gnu -std=c11 -fsyntax-only -Wno-format -Wformat-signedness -verify=okay %s
+// RUN: %clang_cc1 -triple=x86_64-pc-linux-gnu -std=c11 -fsyntax-only -Wformat-signedness -Wno-format -verify %s
+// RUN: %clang_cc1 -triple=x86_64-pc-linux-gnu -std=c11 -fsyntax-only -Wno-format -Wformat-signedness -verify %s
 // okay-no-diagnostics
 
 int printf(const char *restrict format, ...);
@@ -218,5 +218,5 @@ void test_printf_unsigned_priX16(uint16_t x) {
 void test_suppress(int x)
 {
 #pragma GCC diagnostic ignored "-Wformat"
-    printf("%u", x);
+    printf("%u", x); // expected-warning{{format specifies type 'unsigned int' but the argument has type 'int'}}
 }

[karka228]

The -Wformat-signedness warning was implemented to work the same way as the same warning do in gcc. This patch seems to remove that gcc compatibility. With this patch -Wformat-signedness is now controlled by -Wformat-pedantic instead of -Wformat (as in gcc). I don't know how important it is for clang to be gcc compatible in this way (@AaronBallman what do you think?). At least this comment need to be updated as it no longer match the test below.

[AaronBallman]

Yes, this is intentional, see the discussion at: #81647 (comment)

[karka228]

That discussion is only about -Wformat scoped enums warnings, but this patch currently move all the -Wformat warnings regarding type mismatch into a subwarning, that a lot more warnings than only scoped enums. Is that intended?

[AaronBallman]

Agreed, that seems like an outstanding issue with the PR and should be addressed. Only the scoped enumeration behavior should change.

[AaronBallman]

Please mention that this fixes #81647 in the patch summary so that the issue is automatically closed.

You should also add a release note so that users know about the change (and link to the issue being closed).

[github-actions]

✅ With the latest revision this PR passed the C/C++ code formatter.

[pirama-arumuga-nainar]

Please mention that this fixes #81647 in the patch summary so that the issue is automatically closed.

You should also add a release note so that users know about the change (and link to the issue being closed).

This is also pending. The rest of the changes LGTM.

[AaronBallman]

Please be sure to add a release note in clang/docs/ReleaseNotes.rst. The code changes themselves LGTM, but there are some issues with the tests.

[AaronBallman]

LGTM!