[PATCH v2 0/6] scripts/Kconfig.toolchain: Document minimum supported versions for dynamic checks

[PATCH v2 0/6] scripts/Kconfig.toolchain: Document minimum supported versions for dynamic checks

[Nathan Chancellor]

Hi all,

Now that we have a fairly unified location for compiler and linker
feature checks in scripts/Kconfig.toolchain, we should make it easy to
clean up the checks as they become obsolete through minimum supported
version bumps by documenting when they were first supported. I somewhat
modeled this after the documentation in compiler_attributes.h but I did
not want it to be as verbose.

In doing this archaeology, I developed a few prerequisite clean up
patches, mostly around removing code that is practically dead through
minimum versions or adjusting options that are only supported in one
compiler. See the individual patches for the details.

---
Changes in v2:
- Fix versions for a few entries (Julian)
- Add patch fixing check for asm goto with tied outputs in riscv code
- Add patch eliminating check for asm goto with tied outputs, as all
  compilers that properly support asm goto with outputs will have
  support for tied outputs
- Add links to tools where feature is supported/changed to make future
  auditing easier
- Adjust comment format to be a little more succinct
- Reword commit messages accordingly
- Link to v1: https://patch.msgid.link/20261110-kconfig-toolchain-updates-v1-0-6d4fb636290e@kernel.org

---
Nathan Chancellor (6):
      scripts/Kconfig.toolchain: Remove CONFIG_CC_HAS_NO_PROFILE_FN_ATTR
      scripts/Kconfig.toolchain: Do not dynamically probe '-fno-addrsig'
      scripts/Kconfig.toolchain: Clean up '-fno-allow-store-data-races' handling
      riscv: checksum: Fix CC_HAS_ASM_GOTO_TIED_OUTPUT check
      Eliminate CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT
      scripts/Kconfig.toolchain: Document minimum supported versions for dynamic checks

 Makefile                       |  5 ---
 arch/riscv/lib/csum.c          |  4 +--
 arch/x86/include/asm/uaccess.h | 82 ++++++++++++++++++++----------------------
 kernel/gcov/Kconfig            |  1 -
 scripts/Kconfig.toolchain      | 80 +++++++++++++++++++++++++++++++++--------
 5 files changed, 107 insertions(+), 65 deletions(-)
---
base-commit: 7f705c0ad1c7cab3d8de99a3a67937bc82fc8481
base-tree: git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
change-id: 20261110-kconfig-toolchain-updates-d8b351370a5f

Best regards,
--  
Cheers,
Nathan

[PATCH v2 1/6] scripts/Kconfig.toolchain: Remove CONFIG_CC_HAS_NO_PROFILE_FN_ATTR

[Nathan Chancellor]

__attribute__((no_profile_instrument_function)) was first supported in
GCC 7 and clang-13, which is higher than both the minimum supported
versions of both compilers. This configuration is always 'y', thus it
can be removed.

Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
 kernel/gcov/Kconfig       | 1 -
 scripts/Kconfig.toolchain | 3 ---
 2 files changed, 4 deletions(-)

diff --git a/kernel/gcov/Kconfig b/kernel/gcov/Kconfig
index 04f4ebdc3cf5..f62de2dea8a3 100644
--- a/kernel/gcov/Kconfig
+++ b/kernel/gcov/Kconfig
@@ -4,7 +4,6 @@ menu "GCOV-based kernel profiling"
 config GCOV_KERNEL
 	bool "Enable gcov-based kernel profiling"
 	depends on DEBUG_FS
-	depends on !ARCH_WANTS_NO_INSTR || CC_HAS_NO_PROFILE_FN_ATTR
 	select CONSTRUCTORS
 	default n
 	help
diff --git a/scripts/Kconfig.toolchain b/scripts/Kconfig.toolchain
index 2acbd40a38f1..666d44cc6ba8 100644
--- a/scripts/Kconfig.toolchain
+++ b/scripts/Kconfig.toolchain
@@ -141,9 +141,6 @@ config CC_HAS_ASSUME
 	# https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106654
 	default y if CC_IS_GCC && GCC_VERSION >= 130100
 
-config CC_HAS_NO_PROFILE_FN_ATTR
-	def_bool $(success,echo '__attribute__((no_profile_instrument_function)) int x();' | $(CC) -x c - -c -o /dev/null -Werror)
-
 # Beware of two problems when using cc-option in Kconfig compared to Kbuild
 #
 # 1. cc-option in Kbuild tests the provided flags with all previously added

-- 
2.56.0

Re: [PATCH v2 1/6] scripts/Kconfig.toolchain: Remove CONFIG_CC_HAS_NO_PROFILE_FN_ATTR

[Nicolas Schier]

On Wed, Nov 11, 2026 at 05:52:03PM -0700, Nathan Chancellor wrote:
> __attribute__((no_profile_instrument_function)) was first supported in
> GCC 7 and clang-13, which is higher than both the minimum supported
> versions of both compilers. This configuration is always 'y', thus it
> can be removed.
> 
> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> ---
>  kernel/gcov/Kconfig       | 1 -
>  scripts/Kconfig.toolchain | 3 ---
>  2 files changed, 4 deletions(-)

Thanks.

Reviewed-by: Nicolas Schier <nsc@kernel.org>

-- 
Nicolas

[PATCH v2 2/6] scripts/Kconfig.toolchain: Do not dynamically probe '-fno-addrsig'

[Nathan Chancellor]

'-fno-addrsig' avoids generating the LLVM specific .llvm_addrsig
section. This flag is supported by all supported versions of clang, so
convert CONFIG_CC_OPT_NO_ADDRSIG into a simple string symbol that
depends on CC_IS_CLANG to clean up a dynamic probe that is unnecessary.

Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
 scripts/Kconfig.toolchain | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/scripts/Kconfig.toolchain b/scripts/Kconfig.toolchain
index 666d44cc6ba8..570c5a31016c 100644
--- a/scripts/Kconfig.toolchain
+++ b/scripts/Kconfig.toolchain
@@ -164,7 +164,8 @@ config CC_OPT_ZERO_INIT_PADDING_BITS
 	def_string "$(cc-option-str,-fzero-init-padding-bits=all)"
 
 config CC_OPT_NO_ADDRSIG
-	def_string "$(cc-option-str,-fno-addrsig)"
+	# supported in clang, not needed for gcc
+	def_string "-fno-addrsig" if CC_IS_CLANG
 
 config CC_OPT_DIAGNOSTICS_SHOW_CONTEXT
 	def_string "$(cc-option-str,-fdiagnostics-show-context=2)"

-- 
2.56.0

Re: [PATCH v2 2/6] scripts/Kconfig.toolchain: Do not dynamically probe '-fno-addrsig'

[Nicolas Schier]

On Wed, Nov 11, 2026 at 05:52:04PM -0700, Nathan Chancellor wrote:
> '-fno-addrsig' avoids generating the LLVM specific .llvm_addrsig
> section. This flag is supported by all supported versions of clang, so
> convert CONFIG_CC_OPT_NO_ADDRSIG into a simple string symbol that
> depends on CC_IS_CLANG to clean up a dynamic probe that is unnecessary.
> 
> Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> ---
>  scripts/Kconfig.toolchain | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 

Reviewed-by: Nicolas Schier <nsc@kernel.org>

-- 
Nicolas

[PATCH v2 3/6] scripts/Kconfig.toolchain: Clean up '-fno-allow-store-data-races' handling

[Nathan Chancellor]

Commit 454e00cf8ee5 ("kbuild: avoid re-running compiler and linker
probes") added two separate symbols for the "no allow store data races"
GCC flags. These flags serve the same purpose, so combine them using the
second argument of cc-option-str to handle the old value. To avoid
setting the second value of cc-option-str for clang, add a GCC
dependency to the symbol. This will eventually allow the cc-option-str
call to be eliminated when GCC 10 is the minimum version for the kernel.

Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
 Makefile                  | 5 -----
 scripts/Kconfig.toolchain | 7 ++-----
 2 files changed, 2 insertions(+), 10 deletions(-)

diff --git a/Makefile b/Makefile
index d73f2bbd3b93..fb2490faa8d1 100644
--- a/Makefile
+++ b/Makefile
@@ -949,12 +949,7 @@ KBUILD_RUSTFLAGS += -Cdebug-assertions=$(if $(CONFIG_RUST_DEBUG_ASSERTIONS),y,n)
 KBUILD_RUSTFLAGS += -Coverflow-checks=$(if $(CONFIG_RUST_OVERFLOW_CHECKS),y,n)
 
 # Tell gcc to never replace conditional load with a non-conditional one
-ifdef CONFIG_CC_IS_GCC
-# gcc-10 renamed --param=allow-store-data-races=0 to
-# -fno-allow-store-data-races.
-KBUILD_CFLAGS	+= $(CONFIG_CC_OPT_ALLOW_STORE_DATA_RACES_PARAM)
 KBUILD_CFLAGS	+= $(CONFIG_CC_OPT_NO_ALLOW_STORE_DATA_RACES)
-endif
 
 ifdef CONFIG_READABLE_ASM
 # Disable optimizations that make assembler listings hard to read.
diff --git a/scripts/Kconfig.toolchain b/scripts/Kconfig.toolchain
index 570c5a31016c..6520a06a9103 100644
--- a/scripts/Kconfig.toolchain
+++ b/scripts/Kconfig.toolchain
@@ -176,12 +176,9 @@ config CC_OPT_DIAGNOSTICS_SHOW_INLINING_CHAIN
 config CC_OPT_STRICT_FLEX_ARRAYS
 	def_string "$(cc-option-str,-fstrict-flex-arrays=3)"
 
-# gcc-10 renamed --param=allow-store-data-races=0 to -fno-allow-store-data-races
-config CC_OPT_ALLOW_STORE_DATA_RACES_PARAM
-	def_string "$(cc-option-str,--param=allow-store-data-races=0)"
-
 config CC_OPT_NO_ALLOW_STORE_DATA_RACES
-	def_string "$(cc-option-str,-fno-allow-store-data-races)"
+	# gcc-10 renamed --param=allow-store-data-races=0 to -fno-allow-store-data-races
+	def_string "$(cc-option-str,-fno-allow-store-data-races,--param=allow-store-data-races=0)" if CC_IS_GCC
 
 config CC_HAS_MNOP_MCOUNT
 	def_bool $(cc-option,-mnop-mcount)

-- 
2.56.0

Re: [PATCH v2 3/6] scripts/Kconfig.toolchain: Clean up '-fno-allow-store-data-races' handling

[Nicolas Schier]

On Wed, Nov 11, 2026 at 05:52:05PM -0700, Nathan Chancellor wrote:
> Commit 454e00cf8ee5 ("kbuild: avoid re-running compiler and linker
> probes") added two separate symbols for the "no allow store data races"
> GCC flags. These flags serve the same purpose, so combine them using the
> second argument of cc-option-str to handle the old value. To avoid
> setting the second value of cc-option-str for clang, add a GCC
> dependency to the symbol. This will eventually allow the cc-option-str
> call to be eliminated when GCC 10 is the minimum version for the kernel.
> 
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> ---
>  Makefile                  | 5 -----
>  scripts/Kconfig.toolchain | 7 ++-----
>  2 files changed, 2 insertions(+), 10 deletions(-)
> 

Reviewed-by: Nicolas Schier <nsc@kernel.org>

-- 
Nicolas

[PATCH v2 4/6] riscv: checksum: Fix CC_HAS_ASM_GOTO_TIED_OUTPUT check

[Nathan Chancellor]

The defined symbol is CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT, not
CC_HAS_ASM_GOTO_TIED_OUTPUT.

Fixes: 4525462dd0db ("riscv: lib: Check if output in asm goto supported")
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
Cc: Charlie Jenkins <thecharlesjenkins@gmail.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: linux-riscv@lists.infradead.org
---
 arch/riscv/lib/csum.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/riscv/lib/csum.c b/arch/riscv/lib/csum.c
index 5d4379768e5d..c2b8986a0d39 100644
--- a/arch/riscv/lib/csum.c
+++ b/arch/riscv/lib/csum.c
@@ -142,7 +142,7 @@ do_csum_with_alignment(const unsigned char *buff, int len)
 	end = (const unsigned long *)(buff + len);
 	csum = do_csum_common(ptr, end, data);
 
-#ifdef CC_HAS_ASM_GOTO_TIED_OUTPUT
+#ifdef CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT
 	if (IS_ENABLED(CONFIG_RISCV_ISA_ZBB) &&
 	    IS_ENABLED(CONFIG_TOOLCHAIN_HAS_ZBB) &&
 	    riscv_has_extension_likely(RISCV_ISA_EXT_ZBB)) {
@@ -186,7 +186,7 @@ do_csum_with_alignment(const unsigned char *buff, int len)
 		return csum >> 16;
 	}
 
-#endif /* CC_HAS_ASM_GOTO_TIED_OUTPUT */
+#endif /* CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT */
 #ifndef CONFIG_32BIT
 	csum += ror64(csum, 32);
 	csum >>= 32;

-- 
2.56.0

Re: [PATCH v2 4/6] riscv: checksum: Fix CC_HAS_ASM_GOTO_TIED_OUTPUT check

[Nicolas Schier]

On Wed, Nov 11, 2026 at 05:52:06PM -0700, Nathan Chancellor wrote:
> The defined symbol is CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT, not
> CC_HAS_ASM_GOTO_TIED_OUTPUT.
> 
> Fixes: 4525462dd0db ("riscv: lib: Check if output in asm goto supported")
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> ---
> Cc: Charlie Jenkins <thecharlesjenkins@gmail.com>
> Cc: Paul Walmsley <pjw@kernel.org>
> Cc: Palmer Dabbelt <palmer@dabbelt.com>
> Cc: linux-riscv@lists.infradead.org
> ---
>  arch/riscv/lib/csum.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 

Thanks for catching!

Reviewed-by: Nicolas Schier <nsc@kernel.org>

Since 4525462dd0db (v6.7) still supported clang-11: shall we add 'Cc: stable'?

-- 
Nicolas

Re: [PATCH v2 4/6] riscv: checksum: Fix CC_HAS_ASM_GOTO_TIED_OUTPUT check

[Nathan Chancellor]

On Thu, Nov 12, 2026 at 12:26:32PM +0100, Nicolas Schier wrote:
> On Wed, Nov 11, 2026 at 05:52:06PM -0700, Nathan Chancellor wrote:
> > The defined symbol is CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT, not
> > CC_HAS_ASM_GOTO_TIED_OUTPUT.
> > 
> > Fixes: 4525462dd0db ("riscv: lib: Check if output in asm goto supported")
> > Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> > ---
> > Cc: Charlie Jenkins <thecharlesjenkins@gmail.com>
> > Cc: Paul Walmsley <pjw@kernel.org>
> > Cc: Palmer Dabbelt <palmer@dabbelt.com>
> > Cc: linux-riscv@lists.infradead.org
> > ---
> >  arch/riscv/lib/csum.c | 4 ++--
> >  1 file changed, 2 insertions(+), 2 deletions(-)
> > 
> 
> Thanks for catching!
> 
> Reviewed-by: Nicolas Schier <nsc@kernel.org>
> 
> Since 4525462dd0db (v6.7) still supported clang-11: shall we add 'Cc: stable'?

I was a little on the fence about the 'Cc: stable'. On the one hand, it
is definitely a bug and it would totally qualify for stable. On the
other hand, nobody has noticed for almost three years and it is
exercising a new code path. I'll leave that up to you when you apply it
(or the RISC-V people can give their input as well).

-- 
Cheers,
Nathan

[PATCH v2 5/6] Eliminate CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT

[Nathan Chancellor]

Commit 1aa0e8b144b6 ("Kconfig: Add option for asm goto w/ tied outputs
to workaround clang-13 bug") added CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT to
avoid a bug in clang's asm goto with outputs support when a tied output
was used. The fix in clang was to match GCC's documented behavior.

Now that the minimum supported version of clang has long been bumped
past the version where the fix landed (clang-14), any compiler that
satisfies the dependencies of CONFIG_CC_HAS_ASM_GOTO_OUTPUT will support
tied outputs properly. Eliminate CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT in
favor of just checking for CONFIG_CC_HAS_ASM_GOTO_OUTPUT directly.

Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
Cc: Linus Torvalds <torvalds@linuxfoundation.org>
Cc: Thomas Gleixner <tglx@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: x86@kernel.org
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Charlie Jenkins <thecharlesjenkins@gmail.com>
Cc: Paul Walmsley <pjw@kernel.org>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: linux-riscv@lists.infradead.org
---
 arch/riscv/lib/csum.c          |  4 +--
 arch/x86/include/asm/uaccess.h | 82 ++++++++++++++++++++----------------------
 scripts/Kconfig.toolchain      |  5 ---
 3 files changed, 41 insertions(+), 50 deletions(-)

diff --git a/arch/riscv/lib/csum.c b/arch/riscv/lib/csum.c
index c2b8986a0d39..cd461a7b61a5 100644
--- a/arch/riscv/lib/csum.c
+++ b/arch/riscv/lib/csum.c
@@ -142,7 +142,7 @@ do_csum_with_alignment(const unsigned char *buff, int len)
 	end = (const unsigned long *)(buff + len);
 	csum = do_csum_common(ptr, end, data);
 
-#ifdef CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT
+#ifdef CONFIG_CC_HAS_ASM_GOTO_OUTPUT
 	if (IS_ENABLED(CONFIG_RISCV_ISA_ZBB) &&
 	    IS_ENABLED(CONFIG_TOOLCHAIN_HAS_ZBB) &&
 	    riscv_has_extension_likely(RISCV_ISA_EXT_ZBB)) {
@@ -186,7 +186,7 @@ do_csum_with_alignment(const unsigned char *buff, int len)
 		return csum >> 16;
 	}
 
-#endif /* CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT */
+#endif /* CONFIG_CC_HAS_ASM_GOTO_OUTPUT */
 #ifndef CONFIG_32BIT
 	csum += ror64(csum, 32);
 	csum >>= 32;
diff --git a/arch/x86/include/asm/uaccess.h b/arch/x86/include/asm/uaccess.h
index f4700ad6b5ec..64b648e59df0 100644
--- a/arch/x86/include/asm/uaccess.h
+++ b/arch/x86/include/asm/uaccess.h
@@ -248,6 +248,44 @@ do {									\
 		     : [umem] "m" (__m(addr))				\
 		     : : label)
 
+#define __try_cmpxchg_user_asm(itype, ltype, _ptr, _pold, _new, label)	({ \
+	bool success;							\
+	__typeof__(_ptr) _old = (__typeof__(_ptr))(_pold);		\
+	__typeof__(*(_ptr)) __old = *_old;				\
+	__typeof__(*(_ptr)) __new = (_new);				\
+	asm_goto_output("\n"						\
+		     "1: " LOCK_PREFIX "cmpxchg"itype" %[new], %[ptr]\n"\
+		     _ASM_EXTABLE_UA(1b, %l[label])			\
+		     : "=@ccz" (success),				\
+		       [ptr] "+m" (*_ptr),				\
+		       [old] "+a" (__old)				\
+		     : [new] ltype (__new)				\
+		     : "memory"						\
+		     : label);						\
+	if (unlikely(!success))						\
+		*_old = __old;						\
+	likely(success);					})
+
+#ifdef CONFIG_X86_32
+#define __try_cmpxchg64_user_asm(_ptr, _pold, _new, label)	({	\
+	bool success;							\
+	__typeof__(_ptr) _old = (__typeof__(_ptr))(_pold);		\
+	__typeof__(*(_ptr)) __old = *_old;				\
+	__typeof__(*(_ptr)) __new = (_new);				\
+	asm_goto_output("\n"						\
+		     "1: " LOCK_PREFIX "cmpxchg8b %[ptr]\n"		\
+		     _ASM_EXTABLE_UA(1b, %l[label])			\
+		     : "=@ccz" (success),				\
+		       "+A" (__old),					\
+		       [ptr] "+m" (*_ptr)				\
+		     : "b" ((u32)__new),				\
+		       "c" ((u32)((u64)__new >> 32))			\
+		     : "memory"						\
+		     : label);						\
+	if (unlikely(!success))						\
+		*_old = __old;						\
+	likely(success);					})
+#endif // CONFIG_X86_32
 #else // !CONFIG_CC_HAS_ASM_GOTO_OUTPUT
 
 #ifdef CONFIG_X86_32
@@ -313,48 +351,6 @@ do {									\
 		     : [umem] "m" (__m(addr)),				\
 		       "0" (err))
 
-#endif // CONFIG_CC_HAS_ASM_GOTO_OUTPUT
-
-#ifdef CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT
-#define __try_cmpxchg_user_asm(itype, ltype, _ptr, _pold, _new, label)	({ \
-	bool success;							\
-	__typeof__(_ptr) _old = (__typeof__(_ptr))(_pold);		\
-	__typeof__(*(_ptr)) __old = *_old;				\
-	__typeof__(*(_ptr)) __new = (_new);				\
-	asm_goto_output("\n"						\
-		     "1: " LOCK_PREFIX "cmpxchg"itype" %[new], %[ptr]\n"\
-		     _ASM_EXTABLE_UA(1b, %l[label])			\
-		     : "=@ccz" (success),				\
-		       [ptr] "+m" (*_ptr),				\
-		       [old] "+a" (__old)				\
-		     : [new] ltype (__new)				\
-		     : "memory"						\
-		     : label);						\
-	if (unlikely(!success))						\
-		*_old = __old;						\
-	likely(success);					})
-
-#ifdef CONFIG_X86_32
-#define __try_cmpxchg64_user_asm(_ptr, _pold, _new, label)	({	\
-	bool success;							\
-	__typeof__(_ptr) _old = (__typeof__(_ptr))(_pold);		\
-	__typeof__(*(_ptr)) __old = *_old;				\
-	__typeof__(*(_ptr)) __new = (_new);				\
-	asm_goto_output("\n"						\
-		     "1: " LOCK_PREFIX "cmpxchg8b %[ptr]\n"		\
-		     _ASM_EXTABLE_UA(1b, %l[label])			\
-		     : "=@ccz" (success),				\
-		       "+A" (__old),					\
-		       [ptr] "+m" (*_ptr)				\
-		     : "b" ((u32)__new),				\
-		       "c" ((u32)((u64)__new >> 32))			\
-		     : "memory"						\
-		     : label);						\
-	if (unlikely(!success))						\
-		*_old = __old;						\
-	likely(success);					})
-#endif // CONFIG_X86_32
-#else  // !CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT
 #define __try_cmpxchg_user_asm(itype, ltype, _ptr, _pold, _new, label)	({ \
 	int __err = 0;							\
 	bool success;							\
@@ -410,7 +406,7 @@ do {									\
 		*_old = __old;						\
 	likely(__result);					})
 #endif // CONFIG_X86_32
-#endif // CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT
+#endif // CONFIG_CC_HAS_ASM_GOTO_OUTPUT
 
 /* FIXME: this hack is definitely wrong -AK */
 struct __large_struct { unsigned long buf[100]; };
diff --git a/scripts/Kconfig.toolchain b/scripts/Kconfig.toolchain
index 6520a06a9103..9454ef5294d5 100644
--- a/scripts/Kconfig.toolchain
+++ b/scripts/Kconfig.toolchain
@@ -120,11 +120,6 @@ config CC_HAS_ASM_GOTO_OUTPUT
 	depends on !GCC_ASM_GOTO_OUTPUT_BROKEN
 	depends on $(success,echo 'int foo(int x) { asm goto ("": "=r"(x) ::: bar); return x; bar: return 0; }' | $(CC) -x c - -c -o /dev/null)
 
-config CC_HAS_ASM_GOTO_TIED_OUTPUT
-	depends on CC_HAS_ASM_GOTO_OUTPUT
-	# Detect buggy gcc and clang, fixed in gcc-11 clang-14.
-	def_bool $(success,echo 'int foo(int *x) { asm goto (".long (%l[bar]) - .": "+m"(*x) ::: bar); return *x; bar: return 0; }' | $CC -x c - -c -o /dev/null)
-
 config TOOLS_SUPPORT_RELR
 	def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
 

-- 
2.56.0

Re: [PATCH v2 5/6] Eliminate CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT

[Nicolas Schier]

On Wed, Nov 11, 2026 at 05:52:07PM -0700, Nathan Chancellor wrote:
> Commit 1aa0e8b144b6 ("Kconfig: Add option for asm goto w/ tied outputs
> to workaround clang-13 bug") added CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT to
> avoid a bug in clang's asm goto with outputs support when a tied output
> was used. The fix in clang was to match GCC's documented behavior.
> 
> Now that the minimum supported version of clang has long been bumped
> past the version where the fix landed (clang-14), any compiler that
> satisfies the dependencies of CONFIG_CC_HAS_ASM_GOTO_OUTPUT will support
> tied outputs properly. Eliminate CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT in
> favor of just checking for CONFIG_CC_HAS_ASM_GOTO_OUTPUT directly.
> 
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> ---

[...]
> diff --git a/scripts/Kconfig.toolchain b/scripts/Kconfig.toolchain
> index 6520a06a9103..9454ef5294d5 100644
> --- a/scripts/Kconfig.toolchain
> +++ b/scripts/Kconfig.toolchain
> @@ -120,11 +120,6 @@ config CC_HAS_ASM_GOTO_OUTPUT
>  	depends on !GCC_ASM_GOTO_OUTPUT_BROKEN
>  	depends on $(success,echo 'int foo(int x) { asm goto ("": "=r"(x) ::: bar); return x; bar: return 0; }' | $(CC) -x c - -c -o /dev/null)
>  
> -config CC_HAS_ASM_GOTO_TIED_OUTPUT
> -	depends on CC_HAS_ASM_GOTO_OUTPUT
> -	# Detect buggy gcc and clang, fixed in gcc-11 clang-14.
> -	def_bool $(success,echo 'int foo(int *x) { asm goto (".long (%l[bar]) - .": "+m"(*x) ::: bar); return *x; bar: return 0; }' | $CC -x c - -c -o /dev/null)

The comments reads as if gcc 8 to 10 are affected.  Is the comment
somehow misleading, or should we still keep this until gcc-10 is the
minimum gcc version?

-- 
Nicolas

Re: [PATCH v2 5/6] Eliminate CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT

[Nathan Chancellor]

On Thu, Nov 12, 2026 at 12:34:50PM +0100, Nicolas Schier wrote:
> On Wed, Nov 11, 2026 at 05:52:07PM -0700, Nathan Chancellor wrote:
> > -config CC_HAS_ASM_GOTO_TIED_OUTPUT
> > -	depends on CC_HAS_ASM_GOTO_OUTPUT
> > -	# Detect buggy gcc and clang, fixed in gcc-11 clang-14.
> > -	def_bool $(success,echo 'int foo(int *x) { asm goto (".long (%l[bar]) - .": "+m"(*x) ::: bar); return *x; bar: return 0; }' | $CC -x c - -c -o /dev/null)
> 
> The comments reads as if gcc 8 to 10 are affected.  Is the comment
> somehow misleading, or should we still keep this until gcc-10 is the
> minimum gcc version?

It is a little misleading. GCC 11 was the first version that supported
asm goto with outputs, so this program will not compile with GCC 8 to 10
as a result:

  <stdin>: In function 'foo':
  <stdin>:1:51: error: expected ':' before string constant

The bug that it is trying to trigger was fixed during the GCC 11
development cycle between the initial asm goto with outputs commit and
the final release. As a result, for GCC, if it supports asm goto with
outputs, it supports tied outputs in the expected manner.

This test was mostly to catch buggy versions of clang, as asm goto with
outputs was first shipped in clang-11 but using tied outputs breaks the
build until they were fixed to be compatible with GCC in clang-14:

  <stdin>:1:29: error: invalid operand in inline asm: '.long (${1:l}) - .'
  int foo(int *x) { asm goto (".long (%l[bar]) - .": "+m"(*x) ::: bar); return *x; bar: return 0; }
                              ^
  <stdin>:1:29: error: unknown token in expression in '.long' directive
  <inline asm>:1:9: note: instantiated into assembly here
          .long () - .
                 ^
  2 errors generated.

So if the compiler results in CONFIG_CC_HAS_ASM_GOTO_OUTPUT=y, we know
that CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT=y will be present as well.

-- 
Cheers,
Nathan

Re: [PATCH v2 5/6] Eliminate CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT

[Nicolas Schier]

On Thu, Nov 12, 2026 at 10:47:53AM -0700, Nathan Chancellor wrote:
> On Thu, Nov 12, 2026 at 12:34:50PM +0100, Nicolas Schier wrote:
> > On Wed, Nov 11, 2026 at 05:52:07PM -0700, Nathan Chancellor wrote:
> > > -config CC_HAS_ASM_GOTO_TIED_OUTPUT
> > > -	depends on CC_HAS_ASM_GOTO_OUTPUT
> > > -	# Detect buggy gcc and clang, fixed in gcc-11 clang-14.
> > > -	def_bool $(success,echo 'int foo(int *x) { asm goto (".long (%l[bar]) - .": "+m"(*x) ::: bar); return *x; bar: return 0; }' | $CC -x c - -c -o /dev/null)
> > 
> > The comments reads as if gcc 8 to 10 are affected.  Is the comment
> > somehow misleading, or should we still keep this until gcc-10 is the
> > minimum gcc version?
> 
> It is a little misleading. GCC 11 was the first version that supported
> asm goto with outputs, so this program will not compile with GCC 8 to 10
> as a result:
> 
>   <stdin>: In function 'foo':
>   <stdin>:1:51: error: expected ':' before string constant
> 
> The bug that it is trying to trigger was fixed during the GCC 11
> development cycle between the initial asm goto with outputs commit and
> the final release. As a result, for GCC, if it supports asm goto with
> outputs, it supports tied outputs in the expected manner.
> 
> This test was mostly to catch buggy versions of clang, as asm goto with
> outputs was first shipped in clang-11 but using tied outputs breaks the
> build until they were fixed to be compatible with GCC in clang-14:
> 
>   <stdin>:1:29: error: invalid operand in inline asm: '.long (${1:l}) - .'
>   int foo(int *x) { asm goto (".long (%l[bar]) - .": "+m"(*x) ::: bar); return *x; bar: return 0; }
>                               ^
>   <stdin>:1:29: error: unknown token in expression in '.long' directive
>   <inline asm>:1:9: note: instantiated into assembly here
>           .long () - .
>                  ^
>   2 errors generated.
> 
> So if the compiler results in CONFIG_CC_HAS_ASM_GOTO_OUTPUT=y, we know
> that CONFIG_CC_HAS_ASM_GOTO_TIED_OUTPUT=y will be present as well.

ah ok, thanks for clarification!

Reviewed-by: Nicolas Schier <nsc@kernel.org>

-- 
Nicolas

[PATCH v2 6/6] scripts/Kconfig.toolchain: Document minimum supported versions for dynamic checks

[Nathan Chancellor]

To make it easier to clean up unnecessary dynamic checks when the
minimum supported version of a toolchain component is updated, add
comments documenting when features were first supported. Request that
future additions document this as well.

Signed-off-by: Nathan Chancellor <nathan@kernel.org>
---
 scripts/Kconfig.toolchain | 64 ++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 63 insertions(+), 1 deletion(-)

diff --git a/scripts/Kconfig.toolchain b/scripts/Kconfig.toolchain
index 9454ef5294d5..608da5023504 100644
--- a/scripts/Kconfig.toolchain
+++ b/scripts/Kconfig.toolchain
@@ -1,4 +1,9 @@
 # SPDX-License-Identifier: GPL-2.0-only
+# When adding a new dynamic check (such as {cc,ld}-option{,-str} or testing a
+# small program), please add a comment documenting the first version of the
+# tool that supported the feature being tested, ideally with a link to the tool
+# change as well. For cc-option{,-str}, note the "beware" comment below.
+
 config CC_VERSION_TEXT
 	string
 	default "$(CC_VERSION_TEXT)"
@@ -118,12 +123,16 @@ config GCC_ASM_GOTO_OUTPUT_BROKEN
 config CC_HAS_ASM_GOTO_OUTPUT
 	def_bool y
 	depends on !GCC_ASM_GOTO_OUTPUT_BROKEN
+	# gcc-11: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=e3b3b59683c1e7d31a9d313dd97394abebf644be
+	# clang-11: https://github.com/llvm/llvm-project/commit/50cac248773c3a023e8f6ceb9938bdd5e9f15da2
 	depends on $(success,echo 'int foo(int x) { asm goto ("": "=r"(x) ::: bar); return x; bar: return 0; }' | $(CC) -x c - -c -o /dev/null)
 
 config TOOLS_SUPPORT_RELR
 	def_bool $(success,env "CC=$(CC)" "LD=$(LD)" "NM=$(NM)" "OBJCOPY=$(OBJCOPY)" $(srctree)/scripts/tools-support-relr.sh)
 
 config CC_HAS_ASM_INLINE
+	# gcc-9: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=5b76e75f760f3e4787851366359d5d50f989877c
+	# clang-11: https://github.com/llvm/llvm-project/commit/246398ece7115b57a02dbe7876d86ae8e72406ef
 	def_bool $(success,echo 'void foo(void) { asm inline (""); }' | $(CC) -x c - -c -o /dev/null)
 
 config CC_HAS_ASSUME
@@ -156,6 +165,8 @@ config CC_HAS_ASSUME
 #    pitfall.
 
 config CC_OPT_ZERO_INIT_PADDING_BITS
+	# gcc-15: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=0547dbb725b6d8e878a79e28a2e171eafcfbc1aa
+	# clang-20 does this by default without a flag: https://github.com/llvm/llvm-project/commit/7a086e1b2dc05f54afae3591614feede727601fa
 	def_string "$(cc-option-str,-fzero-init-padding-bits=all)"
 
 config CC_OPT_NO_ADDRSIG
@@ -163,64 +174,109 @@ config CC_OPT_NO_ADDRSIG
 	def_string "-fno-addrsig" if CC_IS_CLANG
 
 config CC_OPT_DIAGNOSTICS_SHOW_CONTEXT
+	# gcc-16: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=6faa3cfe60ff9769d1bebfffdd2c7325217d7389
+	# clang: not supported
 	def_string "$(cc-option-str,-fdiagnostics-show-context=2)"
 
 config CC_OPT_DIAGNOSTICS_SHOW_INLINING_CHAIN
+	# clang-23: https://github.com/llvm/llvm-project/commit/f9ad232421e37e4168f6937abdd0a1665a590f12
+	# gcc: not needed, it does this by default
 	def_string "$(cc-option-str,-fdiagnostics-show-inlining-chain)"
 
 config CC_OPT_STRICT_FLEX_ARRAYS
+	# gcc-13: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=1879e48f3d8595bc9e7f583bbd12df3c6f5c42dc
+	# clang-16: https://github.com/llvm/llvm-project/commit/7f93ae808634e33e4dc9bce753c909aa5f9a6eb4
 	def_string "$(cc-option-str,-fstrict-flex-arrays=3)"
 
 config CC_OPT_NO_ALLOW_STORE_DATA_RACES
-	# gcc-10 renamed --param=allow-store-data-races=0 to -fno-allow-store-data-races
+	# gcc-10 renamed --param=allow-store-data-races=0 to -fno-allow-store-data-races:
+	#   https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=e622a32db78300821fc1327637ec6413febc2c66
+	# clang: not needed
 	def_string "$(cc-option-str,-fno-allow-store-data-races,--param=allow-store-data-races=0)" if CC_IS_GCC
 
 config CC_HAS_MNOP_MCOUNT
+	# s390 support
+	#   gcc-9: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=79c2f6d7de59d5580d224b491d5600d20e3101fe
+	#   clang-10: https://github.com/llvm/llvm-project/commit/93767143147b7d765c6ce8123a4226d449228649
+	# x86 support
+	#   gcc-5: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=ecc81e33123d7ac9c11742161e128858d844b99d
+	#   clang-24: https://github.com/llvm/llvm-project/commit/f514b16972f785fdd3fad1d19195aad5c9324fb5
 	def_bool $(cc-option,-mnop-mcount)
 
 config LD_OPT_FORCE_GROUP_ALLOCATION
+	# GNU ld 2.29: https://sourceware.org/git/gitweb.cgi?h=7bdf4127c38081c7764c0bf1db55d0b2e56391dc&p=binutils-gdb.git
+	# ld.lld 19: https://github.com/llvm/llvm-project/commit/4d9020ca0b125a936d5e26b48f07fcfa4bd78004
 	def_string "$(ld-option-str,--force-group-allocation)"
 
 config LD_OPT_NO_WARN_RWX_SEGMENTS
+	# GNU ld 2.39: https://sourceware.org/git/gitweb.cgi?h=ba951afb99912da01a6e8434126b8fac7aa75107&p=binutils-gdb.git
+	# ld.lld: not supported
 	def_string "$(ld-option-str,--no-warn-rwx-segments)"
 
 config CC_OPT_WNO_ADDRESS_OF_PACKED_MEMBER
+	# gcc-9: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=da77eace90fd99b38cd89ce281033d86ebfda434
+	# clang-4: https://github.com/llvm/llvm-project/commit/585ea9ddce50f3c3d3fd6dae0f9f148c074bcc2b
 	def_string "$(cc-option-str,-Wno-address-of-packed-member)"
 
 config CC_OPT_WNO_FORMAT_OVERFLOW_NON_KPRINTF
+	# clang-18: https://github.com/llvm/llvm-project/commit/56c3b8e997d065b568964f71715ecbbd28c2e4a4
+	# gcc: not needed, it has this behavior by default
 	def_string "$(cc-option-str,-Wno-format-overflow-non-kprintf)"
 
 config CC_OPT_WNO_FORMAT_TRUNCATION_NON_KPRINTF
+	# clang-18: https://github.com/llvm/llvm-project/commit/56c3b8e997d065b568964f71715ecbbd28c2e4a4
+	# gcc: not needed, it has this behavior by default
 	def_string "$(cc-option-str,-Wno-format-truncation-non-kprintf)"
 
 config CC_OPT_WNO_DEFAULT_CONST_INIT_UNSAFE
+	# clang-21: https://github.com/llvm/llvm-project/commit/00f9ef282c7482754a0fea497417604d1deca9fa
+	# gcc: not supported
 	def_string "$(cc-option-str,-Wno-default-const-init-unsafe)"
 
 config CC_OPT_WNO_DANGLING_POINTER
+	# gcc-12: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=9d6a0f388eb048f8d87f47af78f07b5ce513bfe6
+	# clang: not supported
 	def_string "$(cc-option-str,-Wno-dangling-pointer)"
 
 config CC_OPT_WVLA_LARGER_THAN
+	# gcc-7 for -Wvla-larger-than=: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=adc577c558a4b2e3b604511927cce17039d57c11
+	# gcc-9 for -Wvla-larger-than=0: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=00abf86c47606f8220d6508654ef52ef11d76e3d
+	# clang: not supported
 	def_string "$(cc-option-str,-Wvla-larger-than=1)"
 
 config CC_HAS_WSTRINGOP_OVERFLOW
+	# gcc-7: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=ee92e7bafb52d267f52e4c48edada5f279bcf591
+	# clang: not supported
 	def_bool $(cc-option,-Wstringop-overflow)
 
 config CC_OPT_WNO_UNTERMINATED_STRING_INITIALIZATION
+	# gcc-15: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=44c9403ed1833ae71a59e84f9e37af3182be0df5
+	# clang-21: https://github.com/llvm/llvm-project/commit/e8ae77947154e10dbc05cbb95ec9e10d3b0be13e
 	def_string "$(cc-option-str,-Wno-unterminated-string-initialization)"
 
 config CC_OPT_WERROR_DESIGNATED_INIT
+	# gcc-5: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=976d5a222fc23279e239ad926472c7810869e494
+	# clang: not supported
 	def_string "$(cc-option-str,-Werror=designated-init)"
 
 config CC_OPT_WENUM_CONVERSION
+	# gcc-10: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=68e2c1996ec6bde27363ce0db15233ac8cae1c4d
+	# clang-3.2: https://github.com/llvm/llvm-project/commit/68e2c7bee3ab337cc3134ff99fdd1ce2ddb33c2f
 	def_string "$(cc-option-str,-Wenum-conversion)"
 
 config CC_OPT_WNO_PACKED_NOT_ALIGNED
+	# gcc-8: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=00aa1fa221744dc57f6c6cea8baa669b732d1101
+	# clang: not supported
 	def_string "$(cc-option-str,-Wno-packed-not-aligned)"
 
 config CC_OPT_WNO_FORMAT_OVERFLOW
+	# gcc-7: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=bce6f76013fd5f2e88a235231231c7879597220c
+	# clang-18: https://github.com/llvm/llvm-project/commit/56c3b8e997d065b568964f71715ecbbd28c2e4a4
 	def_string "$(cc-option-str,-Wno-format-overflow)"
 
 config CC_OPT_WNO_STRINGOP_TRUNCATION
+	# gcc-8: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=025d57f037ad13eb479818b677ef4be4d97b639c
+	# clang: not supported
 	def_string "$(cc-option-str,-Wno-stringop-truncation)"
 
 config CC_HAS_COUNTED_BY
@@ -247,14 +303,20 @@ config CC_HAS_BROKEN_COUNTED_BY_REF
 	default y if CC_IS_CLANG && CLANG_VERSION < 220100
 
 config CC_HAS_ALLOC_TOKEN
+	# clang-22: https://github.com/llvm/llvm-project/commit/774ffe5cce7392e6b4e29c83e725337727c2c994
+	# gcc: not supported
 	def_bool $(cc-option,-falloc-token-max=123)
 
 config CC_HAS_MULTIDIMENSIONAL_NONSTRING
+	# gcc-15: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=ab714e60870d1bb98039652b760918b5e680ac10
+	# clang-21: https://github.com/llvm/llvm-project/commit/3d4f979e271d2a1fe0906f4d1b16db108838f98f
 	def_bool $(success,echo 'char tag[][4] __attribute__((__nonstring__)) = { };' | $(CC) $(CLANG_FLAGS) -x c - -c -o /dev/null -Werror)
 
 config CC_OPT_INLINE_MEMSET
 	string
+	# gcc-14: https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=1ff6d9f7428b0668cd8ab0b3e3ab94f1d733124d
 	default "-finline-stringops=memset" if $(cc-option,-finline-stringops=memset)
+	# clang-22: https://github.com/llvm/llvm-project/commit/b28eeb28bea39148738dc375e8a97072a1907e64
 	default "-mllvm -max-store-memset=4294967294" if $(cc-option,-mllvm -max-store-memset=4294967294)
 
 config LD_CAN_USE_KEEP_IN_OVERLAY

-- 
2.56.0

Re: [PATCH v2 6/6] scripts/Kconfig.toolchain: Document minimum supported versions for dynamic checks

[Nicolas Schier]

On Wed, Nov 11, 2026 at 05:52:08PM -0700, Nathan Chancellor wrote:
> To make it easier to clean up unnecessary dynamic checks when the
> minimum supported version of a toolchain component is updated, add
> comments documenting when features were first supported. Request that
> future additions document this as well.
> 
> Signed-off-by: Nathan Chancellor <nathan@kernel.org>
> ---
>  scripts/Kconfig.toolchain | 64 ++++++++++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 63 insertions(+), 1 deletion(-)
> 

Great, I really appreciate that.  Thanks a million!

Acked-by: Nicolas Schier <nsc@kernel.org>

-- 
Nicolas