Skip to contentSkip to conversation

lib/test_printf.c: fix clang -Wformat warnings

PATCH Superseded Justin Stitt 5 participants 13 messages Jun 29, 2022 – Jul 28, 2022 Library code

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 lib/test_printf.c: fix clang -Wformat warnings

see warnings:

| lib/test_printf.c:157:52: error: format specifies type 'unsigned char'
| but the argument has type 'int' [-Werror,-Wformat] test("0|1|1|128|255",
| "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-
| lib/test_printf.c:158:55: error: format specifies type 'char' but the
| argument has type 'int' [-Werror,-Wformat] test("0|1|1|-128|-1",
| "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
-
| lib/test_printf.c:159:41: error: format specifies type 'unsigned short'
| but the argument has type 'int' [-Werror,-Wformat]
| test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

There's an ongoing movement to eventually enable the -Wformat flag for clang. Previous patches have targeted incorrect usage of format specifiers. In this case, however, the "incorrect" format specifiers are intrinsically part of the test cases. Hence, fixing them would be misaligned with their intended purpose. My proposed fix is to simply disable the warnings so that one day a clean build of the kernel with clang (and -Wformat enabled) would be possible. It would also keep us in the green for alot of the CI bots.

---
 lib/test_printf.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
lib/test_printf.c +10 -0
index 07309c45f327..748591a0c55c 100644
@@ -4,6 +4,12 @@
  */
 
 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+#define DO_PRAGMA(x) _Pragma(#x)
+#define NOWARN(warnoption, ...)
+	    DO_PRAGMA(GCC diagnostic push)
+		    DO_PRAGMA(GCC diagnostic ignored #warnoption)
+			    __VA_ARGS__
+				    DO_PRAGMA(GCC diagnostic pop)
 
 #include <linux/init.h>
 #include <linux/kernel.h>
@@ -154,9 +160,13 @@ test_number(void)
 	test("0x1234abcd  ", "%#-12x", 0x1234abcd);
 	test("  0x1234abcd", "%#12x", 0x1234abcd);
 	test("0|001| 12|+123| 1234|-123|-1234", "%d|%03d|%3d|%+d|% d|%+d|% d", 0, 1, 12, 123, 1234, -123, -1234);
+	/* disable -Wformat for this chunk */
+	NOWARN(-Wformat, 
 	test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
 	test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
 	test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+	)
+	/* end chunk */
 	/*
 	 * POSIX/C99: »The result of converting zero with an explicit
 	 * precision of zero shall be no characters.« Hence the output
Signature · 1 line
2.37.0.rc0.161.g10f37bed90-goog
#1 lore raw
Quoted 14 lines · Justin Stitt

see warnings:

| lib/test_printf.c:157:52: error: format specifies type 'unsigned char'
| but the argument has type 'int' [-Werror,-Wformat] test("0|1|1|128|255",
| "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-
| lib/test_printf.c:158:55: error: format specifies type 'char' but the
| argument has type 'int' [-Werror,-Wformat] test("0|1|1|-128|-1",
| "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
-
| lib/test_printf.c:159:41: error: format specifies type 'unsigned short'
| but the argument has type 'int' [-Werror,-Wformat]
| test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
Quoted 8 lines

There's an ongoing movement to eventually enable the -Wformat flag for clang. Previous patches have targeted incorrect usage of format specifiers. In this case, however, the "incorrect" format specifiers are intrinsically part of the test cases. Hence, fixing them would be misaligned with their intended purpose. My proposed fix is to simply disable the warnings so that one day a clean build of the kernel with clang (and -Wformat enabled) would be possible. It would also keep us in the green for alot of the CI bots.

a lot

...

Quoted 1 line

#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt

+ blank line

lib/test_printf.c line 12

Quoted 6 lines
lib/test_printf.c +6 -0
+#define DO_PRAGMA(x) _Pragma(#x)
+#define NOWARN(warnoption, ...)
+           DO_PRAGMA(GCC diagnostic push)
+                   DO_PRAGMA(GCC diagnostic ignored #warnoption)
+                           __VA_ARGS__
+                                   DO_PRAGMA(GCC diagnostic pop)

Is it deliberately a broken indentation here?

...

lib/test_printf.c test_number(void) · line 167

Quoted 5 lines
lib/test_printf.c +2 -0
+       /* disable -Wformat for this chunk */
+       NOWARN(-Wformat,
        test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
        test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
        test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

Perhaps shift right the lines as well?

Quoted 2 lines
lib/test_printf.c +2 -0
+       )
+       /* end chunk */
Signature · 2 lines
With Best Regards,
Andy Shevchenko
#2 lore raw
Quoted 10 lines · Andy Shevchenko

lib/test_printf.c test_number(void) · line 167

Quoted 6 lines · Justin Stitt
lib/test_printf.c +2 -0
+       /* disable -Wformat for this chunk */
+       NOWARN(-Wformat,
        test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
        test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
        test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

Perhaps shift right the lines as well?

Along these lines, I think it would look nicer to pass a block statement (a group of statements) to the macro rather than use __VA_ARGS__. Here's an example:
https://godbolt.org/z/fsYcGGEMb

You have to be careful with control flow out of blocks like this sometimes, but for these simple localized cases it looks like that should be fine.

As Nathan mentions, you can probably re-use the existing infra in your definition of NOWARN. I do prefer some macro to make it appear that the pragma is scoped to a block statement, rather than multiple lines for the diag push + pop inline.

Signature · 2 lines
Thanks,
~Nick Desaulniers
#3 lore raw
In reply to Justin Stitt

Hi Justin,

Quoted 33 lines · Justin Stitt

see warnings:

| lib/test_printf.c:157:52: error: format specifies type 'unsigned char'
| but the argument has type 'int' [-Werror,-Wformat] test("0|1|1|128|255",
| "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-
| lib/test_printf.c:158:55: error: format specifies type 'char' but the
| argument has type 'int' [-Werror,-Wformat] test("0|1|1|-128|-1",
| "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
-
| lib/test_printf.c:159:41: error: format specifies type 'unsigned short'
| but the argument has type 'int' [-Werror,-Wformat]
| test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

There's an ongoing movement to eventually enable the -Wformat flag for clang. Previous patches have targeted incorrect usage of format specifiers. In this case, however, the "incorrect" format specifiers are intrinsically part of the test cases. Hence, fixing them would be misaligned with their intended purpose. My proposed fix is to simply disable the warnings so that one day a clean build of the kernel with clang (and -Wformat enabled) would be possible. It would also keep us in the green for alot of the CI bots.

---
 lib/test_printf.c | 10 ++++++++++
 1 file changed, 10 insertions(+)
lib/test_printf.c +0 -0
index 07309c45f327..748591a0c55c 100644

lib/test_printf.c line 12

Quoted 9 lines
lib/test_printf.c +6 -0
@@ -4,6 +4,12 @@
  */
 
 #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
+#define DO_PRAGMA(x) _Pragma(#x)
+#define NOWARN(warnoption, ...)
+	    DO_PRAGMA(GCC diagnostic push)
+		    DO_PRAGMA(GCC diagnostic ignored #warnoption)
+			    __VA_ARGS__
+				    DO_PRAGMA(GCC diagnostic pop)

This is basically a duplicate of the __diag infrastructure that exists in include/linux/compiler_types.h and include/linux/compiler-clang.h, I think we should just reuse that here. It eliminates this hunk and the hunk below...

Quoted 6 lines
lib/test_printf.c +0 -0
 #include <linux/init.h>
 #include <linux/kernel.h>
@@ -154,9 +160,13 @@ test_number(void)
 	test("0x1234abcd  ", "%#-12x", 0x1234abcd);
 	test("  0x1234abcd", "%#12x", 0x1234abcd);
 	test("0|001| 12|+123| 1234|-123|-1234", "%d|%03d|%3d|%+d|% d|%+d|% d", 0, 1, 12, 123, 1234, -123, -1234);

lib/test_printf.c test_number(void) · line 172

Quoted 10 lines
lib/test_printf.c +4 -0
+	/* disable -Wformat for this chunk */
+	NOWARN(-Wformat, 
 	test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
 	test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
 	test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+	)
+	/* end chunk */
 	/*
 	 * POSIX/C99: »The result of converting zero with an explicit
 	 * precision of zero shall be no characters.« Hence the output

...becomes a lot simpler in my opinion (feel free to reword the comment however you want).

Cheers,
Nathan

lib/test_printf.c +3 -0
index 8010de49b6c5..86e088ad9e48 100644
@@ -154,9 +154,12 @@ test_number(void)
 	test("0x1234abcd  ", "%#-12x", 0x1234abcd);
 	test("  0x1234abcd", "%#12x", 0x1234abcd);
 	test("0|001| 12|+123| 1234|-123|-1234", "%d|%03d|%3d|%+d|% d|%+d|% d", 0, 1, 12, 123, 1234, -123, -1234);
+	__diag_push();
+	__diag_ignore(clang, 11, "-Wformat", "These trigger clang's -Wformat and the specifiers should not be changed.");
 	test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
 	test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
 	test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+	__diag_pop();
 	/*
 	 * POSIX/C99: »The result of converting zero with an explicit
 	 * precision of zero shall be no characters.« Hence the output
#4 lore raw

PATCH v2 1/1 lib/test_printf.c: fix clang -Wformat warnings

In reply to Justin Stitt

changes from v1:
* moved NOWARN macro definition to a more appropriate location
* using __diag_ignore_all (thanks Nathan)
* using local scoping for code blocks instead of __VA_ARGS__ (thanks Nick)
* indented affected test cases (thanks Andy)

---
 lib/test_printf.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
lib/test_printf.c +8 -3
index 07309c45f327..1b1755ce9fa7 100644
@@ -30,6 +30,9 @@
 #define PAD_SIZE 16
 #define FILL_CHAR '$'
 
+#define NOWARN(option, comment, block) \
+		__diag_push() __diag_ignore_all(#option, comment) block __diag_pop()
+
 KSTM_MODULE_GLOBALS();
 
 static char *test_buffer __initdata;
@@ -154,9 +157,11 @@ test_number(void)
 	test("0x1234abcd  ", "%#-12x", 0x1234abcd);
 	test("  0x1234abcd", "%#12x", 0x1234abcd);
 	test("0|001| 12|+123| 1234|-123|-1234", "%d|%03d|%3d|%+d|% d|%+d|% d", 0, 1, 12, 123, 1234, -123, -1234);
-	test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-	test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
-	test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+	NOWARN(-Wformat, "Disables clang -Wformat warning", {
+		test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
+		test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
+		test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+	})
 	/*
 	 * POSIX/C99: »The result of converting zero with an explicit
 	 * precision of zero shall be no characters.« Hence the output
Signature · 1 line
2.37.0.rc0.161.g10f37bed90-goog
#5 lore raw
Quoted 11 lines · Justin Stitt

changes from v1:
* moved NOWARN macro definition to a more appropriate location
* using __diag_ignore_all (thanks Nathan)
* using local scoping for code blocks instead of __VA_ARGS__ (thanks Nick)
* indented affected test cases (thanks Andy)

Somehow you replaced the commit message with a changelog. On top of that, I didn't suggest anything important here, so to me it is considered as a credit in the changelog (see previous sentence as well).

Signature · 2 lines
With Best Regards,
Andy Shevchenko
#6 lore raw

PATCH v3 1/1 lib/test_printf.c: fix clang -Wformat warnings

In reply to Justin Stitt

see warnings:
| lib/test_printf.c:157:52: error: format specifies type 'unsigned char' | but the argument has type 'int' [-Werror,-Wformat] test("0|1|1|128|255",
| "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-

| lib/test_printf.c:158:55: error: format specifies type 'char' but the
| argument has type 'int' [-Werror,-Wformat] test("0|1|1|-128|-1",
| "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);

-
| lib/test_printf.c:159:41: error: format specifies type 'unsigned short'
| but the argument has type 'int' [-Werror,-Wformat]
| test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

There's an ongoing movement to eventually enable the -Wformat flag for clang. Previous patches have targeted incorrect usage of format specifiers. In this case, however, the "incorrect" format specifiers are intrinsically part of the test cases. Hence, fixing them would be misaligned with their intended purpose. My proposed fix is to simply disable the warnings so that one day a clean build of the kernel with clang (and -Wformat enabled) would be possible. It would also keep us in the green for alot of the CI bots.

---
changes from v1 -> v2:
* moved NOWARN macro definition to a more appropriate location
* using __diag_ignore_all (thanks Nathan)
* using local scoping for code blocks instead of __VA_ARGS__ (thanks Nick)
* indented affected test cases (thanks Andy)

changes from v2 -> v3:
* reinserted commit message
* remove Andy's Suggested-by tag
* add issue tracker link

 lib/test_printf.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
lib/test_printf.c +8 -3
index 07309c45f327..1b1755ce9fa7 100644
@@ -30,6 +30,9 @@
 #define PAD_SIZE 16
 #define FILL_CHAR '$'
 
+#define NOWARN(option, comment, block) \
+		__diag_push() __diag_ignore_all(#option, comment) block __diag_pop()
+
 KSTM_MODULE_GLOBALS();
 
 static char *test_buffer __initdata;
@@ -154,9 +157,11 @@ test_number(void)
 	test("0x1234abcd  ", "%#-12x", 0x1234abcd);
 	test("  0x1234abcd", "%#12x", 0x1234abcd);
 	test("0|001| 12|+123| 1234|-123|-1234", "%d|%03d|%3d|%+d|% d|%+d|% d", 0, 1, 12, 123, 1234, -123, -1234);
-	test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-	test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
-	test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+	NOWARN(-Wformat, "Disables clang -Wformat warning", {
+		test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
+		test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
+		test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+	})
 	/*
 	 * POSIX/C99: »The result of converting zero with an explicit
 	 * precision of zero shall be no characters.« Hence the output
Signature · 1 line
2.37.0.rc0.161.g10f37bed90-goog
#7 lore raw
Quoted 49 lines · Justin Stitt

see warnings:
| lib/test_printf.c:157:52: error: format specifies type 'unsigned char' | but the argument has type 'int' [-Werror,-Wformat] test("0|1|1|128|255",
| "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-

| lib/test_printf.c:158:55: error: format specifies type 'char' but the
| argument has type 'int' [-Werror,-Wformat] test("0|1|1|-128|-1",
| "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);

-
| lib/test_printf.c:159:41: error: format specifies type 'unsigned short'
| but the argument has type 'int' [-Werror,-Wformat]
| test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

There's an ongoing movement to eventually enable the -Wformat flag for clang. Previous patches have targeted incorrect usage of format specifiers. In this case, however, the "incorrect" format specifiers are intrinsically part of the test cases. Hence, fixing them would be misaligned with their intended purpose. My proposed fix is to simply disable the warnings so that one day a clean build of the kernel with clang (and -Wformat enabled) would be possible. It would also keep us in the green for alot of the CI bots.

---
changes from v1 -> v2:
* moved NOWARN macro definition to a more appropriate location
* using __diag_ignore_all (thanks Nathan)
* using local scoping for code blocks instead of __VA_ARGS__ (thanks Nick)
* indented affected test cases (thanks Andy)

changes from v2 -> v3:
* reinserted commit message
* remove Andy's Suggested-by tag
* add issue tracker link

 lib/test_printf.c | 11 ++++++++---
 1 file changed, 8 insertions(+), 3 deletions(-)
lib/test_printf.c +0 -0
index 07309c45f327..1b1755ce9fa7 100644

lib/test_printf.c line 34

Quoted 5 lines
lib/test_printf.c +2 -0
@@ -30,6 +30,9 @@
 #define PAD_SIZE 16
 #define FILL_CHAR '$'
 
+#define NOWARN(option, comment, block) \
+               __diag_push() __diag_ignore_all(#option, comment) block __diag_pop()

Do you mind putting these 4 on distinct lines with \ at the end of the lines, and terminate the statements with an additional ; if that doesn't produce new warnings about empty statements? Maybe something like:

+#define NOWARN(option, comment, block) \
+ __diag_push(); \
+ __diag_ignore_all(#option, comment); \
+ block \
+ __diag_pop();

(sorry for not formatting that myself, remember to use tabs vs spaces). To me, it seems more readable from a quick glance to have these be distinct lines. You'll find open coded uses of __diag_push and friends add the semicolon, even if not strictly necessary.

lib/test_printf.c test_number(void) · line 160

Quoted 12 lines
lib/test_printf.c +2 -3
+
 KSTM_MODULE_GLOBALS();
 
 static char *test_buffer __initdata;
@@ -154,9 +157,11 @@ test_number(void)
        test("0x1234abcd  ", "%#-12x", 0x1234abcd);
        test("  0x1234abcd", "%#12x", 0x1234abcd);
        test("0|001| 12|+123| 1234|-123|-1234", "%d|%03d|%3d|%+d|% d|%+d|% d", 0, 1, 12, 123, 1234, -123, -1234);
-       test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-       test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
-       test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+       NOWARN(-Wformat, "Disables clang -Wformat warning", {

Technically, this is disabling -Wformat for all compilers. How about a comment string like "Intentionally test narrowing conversion specifiers"?

Quoted 10 lines
lib/test_printf.c +4 -1
+               test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
+               test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
+               test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+       })
        /*
         * POSIX/C99: »The result of converting zero with an explicit
         * precision of zero shall be no characters.« Hence the output
--

2.37.0.rc0.161.g10f37bed90-goog

Signature · 2 lines
Thanks,
~Nick Desaulniers
#8 lore raw

PATCH v4 1/1 lib/test_printf.c: fix clang -Wformat warnings

see warnings:
| lib/test_printf.c:157:52: error: format specifies type 'unsigned char' | but the argument has type 'int' [-Werror,-Wformat] test("0|1|1|128|255",
| "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-

| lib/test_printf.c:158:55: error: format specifies type 'char' but the
| argument has type 'int' [-Werror,-Wformat] test("0|1|1|-128|-1",
| "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);

-
| lib/test_printf.c:159:41: error: format specifies type 'unsigned short'
| but the argument has type 'int' [-Werror,-Wformat]
| test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

There's an ongoing movement to eventually enable the -Wformat flag for clang. Previous patches have targeted incorrect usage of format specifiers. In this case, however, the "incorrect" format specifiers are intrinsically part of the test cases. Hence, fixing them would be misaligned with their intended purpose. My proposed fix is to simply disable the warnings so that one day a clean build of the kernel with clang (and -Wformat enabled) would be possible. It would also keep us in the green for alot of the CI bots.

---
changes from v1 -> v2:
* moved NOWARN macro definition to a more appropriate location
* using __diag_ignore_all (thanks Nathan)
* using local scoping for code blocks instead of __VA_ARGS__ (thanks
* Nick)
* indented affected test cases (thanks Andy)

changes from v2 -> v3:
* reinserted commit message
* remove Andy's Suggested-by tag
* add issue tracker link

changes from v3 -> v4:
* better macro indentation and usage string (thanks Nick)

 lib/test_printf.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)
lib/test_printf.c +11 -3
index 07309c45f327..f78044c1efaa 100644
@@ -30,6 +30,12 @@
 #define PAD_SIZE 16
 #define FILL_CHAR '$'
 
+#define NOWARN(option, comment, block) \
+	__diag_push(); \
+	__diag_ignore_all(#option, comment); \
+	block \
+	__diag_pop();
+
 KSTM_MODULE_GLOBALS();
 
 static char *test_buffer __initdata;
@@ -154,9 +160,11 @@ test_number(void)
 	test("0x1234abcd  ", "%#-12x", 0x1234abcd);
 	test("  0x1234abcd", "%#12x", 0x1234abcd);
 	test("0|001| 12|+123| 1234|-123|-1234", "%d|%03d|%3d|%+d|% d|%+d|% d", 0, 1, 12, 123, 1234, -123, -1234);
-	test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-	test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
-	test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+	NOWARN(-Wformat, "Intentionally test narrowing conversion specifiers.", {
+		test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
+		test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
+		test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+	})
 	/*
 	 * POSIX/C99: »The result of converting zero with an explicit
 	 * precision of zero shall be no characters.« Hence the output
Signature · 1 line
2.37.0.170.g444d1eabd0-goog
Reviewed-by on v4 1/1
#9 lore raw
Quoted 24 lines · Justin Stitt

see warnings:
| lib/test_printf.c:157:52: error: format specifies type 'unsigned char' | but the argument has type 'int' [-Werror,-Wformat] test("0|1|1|128|255",
| "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-

| lib/test_printf.c:158:55: error: format specifies type 'char' but the
| argument has type 'int' [-Werror,-Wformat] test("0|1|1|-128|-1",
| "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);

-
| lib/test_printf.c:159:41: error: format specifies type 'unsigned short'
| but the argument has type 'int' [-Werror,-Wformat]
| test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

There's an ongoing movement to eventually enable the -Wformat flag for clang. Previous patches have targeted incorrect usage of format specifiers. In this case, however, the "incorrect" format specifiers are intrinsically part of the test cases. Hence, fixing them would be misaligned with their intended purpose. My proposed fix is to simply disable the warnings so that one day a clean build of the kernel with clang (and -Wformat enabled) would be possible. It would also keep us in the green for alot of the CI bots.

Quoted 4 lines

Looks good to me:

Best Regards,
Petr

#10 lore raw

Hi Petr,

Quoted 25 lines · Petr Mladek
Quoted 24 lines · Justin Stitt

see warnings:
| lib/test_printf.c:157:52: error: format specifies type 'unsigned char' | but the argument has type 'int' [-Werror,-Wformat] test("0|1|1|128|255",
| "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-

| lib/test_printf.c:158:55: error: format specifies type 'char' but the
| argument has type 'int' [-Werror,-Wformat] test("0|1|1|-128|-1",
| "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);

-
| lib/test_printf.c:159:41: error: format specifies type 'unsigned short'
| but the argument has type 'int' [-Werror,-Wformat]
| test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

There's an ongoing movement to eventually enable the -Wformat flag for clang. Previous patches have targeted incorrect usage of format specifiers. In this case, however, the "incorrect" format specifiers are intrinsically part of the test cases. Hence, fixing them would be misaligned with their intended purpose. My proposed fix is to simply disable the warnings so that one day a clean build of the kernel with clang (and -Wformat enabled) would be possible. It would also keep us in the green for alot of the CI bots.

Quoted 8 lines
Quoted 4 lines

Looks good to me:

Would you be able to take this for 5.20 or should we ask Andrew to pick it up? It seems you two seem to split applying patches to this file and we are trying to get -Wformat enabled for clang in 5.20.

Cheers,
Nathan

#11 lore raw
Quoted 35 lines · Nathan Chancellor

Hi Petr,

Quoted 32 lines · Petr Mladek
Quoted 24 lines · Justin Stitt

see warnings:
| lib/test_printf.c:157:52: error: format specifies type 'unsigned char' | but the argument has type 'int' [-Werror,-Wformat] test("0|1|1|128|255",
| "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-

| lib/test_printf.c:158:55: error: format specifies type 'char' but the
| argument has type 'int' [-Werror,-Wformat] test("0|1|1|-128|-1",
| "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);

-
| lib/test_printf.c:159:41: error: format specifies type 'unsigned short'
| but the argument has type 'int' [-Werror,-Wformat]
| test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

There's an ongoing movement to eventually enable the -Wformat flag for clang. Previous patches have targeted incorrect usage of format specifiers. In this case, however, the "incorrect" format specifiers are intrinsically part of the test cases. Hence, fixing them would be misaligned with their intended purpose. My proposed fix is to simply disable the warnings so that one day a clean build of the kernel with clang (and -Wformat enabled) would be possible. It would also keep us in the green for alot of the CI bots.

Quoted 4 lines

Looks good to me:

Quoted 5 lines
Quoted 1 line

Would you be able to take this for 5.20 or should we ask Andrew to pick it up? It seems you two seem to split applying patches to this file and we are trying to get -Wformat enabled for clang in 5.20.

I take most vsprintf-related patches via the printk git tree last few years.

Anyway, I have just committed the patch into printk/linux.git, branch for-5.20.

Best Regards,
Petr

Reviewed-by on v4 1/1
#12 lore raw
In reply to Justin Stitt
Quoted 25 lines · Justin Stitt

see warnings:
| lib/test_printf.c:157:52: error: format specifies type 'unsigned char' | but the argument has type 'int' [-Werror,-Wformat] test("0|1|1|128|255",
| "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-

| lib/test_printf.c:158:55: error: format specifies type 'char' but the
| argument has type 'int' [-Werror,-Wformat] test("0|1|1|-128|-1",
| "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);

-
| lib/test_printf.c:159:41: error: format specifies type 'unsigned short'
| but the argument has type 'int' [-Werror,-Wformat]
| test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);

There's an ongoing movement to eventually enable the -Wformat flag for clang. Previous patches have targeted incorrect usage of format specifiers. In this case, however, the "incorrect" format specifiers are intrinsically part of the test cases. Hence, fixing them would be misaligned with their intended purpose. My proposed fix is to simply disable the warnings so that one day a clean build of the kernel with clang (and -Wformat enabled) would be possible. It would also keep us in the green for alot of the CI bots.

Quoted 4 lines

Thanks for humoring all of our requests. I'm happy with the result.

Quoted 55 lines

---
changes from v1 -> v2:
* moved NOWARN macro definition to a more appropriate location
* using __diag_ignore_all (thanks Nathan)
* using local scoping for code blocks instead of __VA_ARGS__ (thanks
* Nick)
* indented affected test cases (thanks Andy)

changes from v2 -> v3:
* reinserted commit message
* remove Andy's Suggested-by tag
* add issue tracker link

changes from v3 -> v4:
* better macro indentation and usage string (thanks Nick)

 lib/test_printf.c | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)
lib/test_printf.c +11 -4
index 07309c45f327..f78044c1efaa 100644
@@ -30,6 +30,12 @@
 #define PAD_SIZE 16
 #define FILL_CHAR '$'
 
+#define NOWARN(option, comment, block) \
+       __diag_push(); \
+       __diag_ignore_all(#option, comment); \
+       block \
+       __diag_pop();
+
 KSTM_MODULE_GLOBALS();
 
 static char *test_buffer __initdata;
@@ -154,9 +160,11 @@ test_number(void)
        test("0x1234abcd  ", "%#-12x", 0x1234abcd);
        test("  0x1234abcd", "%#12x", 0x1234abcd);
        test("0|001| 12|+123| 1234|-123|-1234", "%d|%03d|%3d|%+d|% d|%+d|% d", 0, 1, 12, 123, 1234, -123, -1234);
-       test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
-       test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
-       test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+       NOWARN(-Wformat, "Intentionally test narrowing conversion specifiers.", {
+               test("0|1|1|128|255", "%hhu|%hhu|%hhu|%hhu|%hhu", 0, 1, 257, 128, -1);
+               test("0|1|1|-128|-1", "%hhd|%hhd|%hhd|%hhd|%hhd", 0, 1, 257, 128, -1);
+               test("2015122420151225", "%ho%ho%#ho", 1037, 5282, -11627);
+       })
        /*
         * POSIX/C99: »The result of converting zero with an explicit
         * precision of zero shall be no characters.« Hence the output
--

2.37.0.170.g444d1eabd0-goog

Signature · 2 lines
Thanks,
~Nick Desaulniers