Backports of 84ffc79bfbf7 for 6.12 and earlier

Hi Greg and Sasha,

Please find attached backports for commit 84ffc79bfbf7 ("kbuild: Add '-fno-builtin-wcslen'") to 6.12 and earlier. It has already been queued for 6.13 and 6.14 since it applied cleanly. If there are any issues, please let me know.

Cheers,
Nathan

From 5dc6f440b2309877ab27879fc327ef0dfc58b682 Mon Sep 17 00:00:00 2001
From: Nathan Chancellor <[email protected]>
Date: Mon, 7 Apr 2025 16:22:12 -0700
Subject: [PATCH 5.4] kbuild: Add '-fno-builtin-wcslen'

commit 84ffc79bfbf70c779e60218563f2f3ad45288671 upstream.

A recent optimization change in LLVM [1] aims to transform certain loop idioms into calls to strlen() or wcslen(). This change transforms the first while loop in UniStrcat() into a call to wcslen(), breaking the build when UniStrcat() gets inlined into alloc_path_with_tree_prefix():

ld.lld: error: undefined symbol: wcslen

Disable this optimization with '-fno-builtin-wcslen', which prevents the compiler from assuming that wcslen() is available in the kernel's C library.

[ More to the point - it's not that we couldn't implement wcslen(), it's
that this isn't an optimization at all in the context of the kernel.

Replacing a simple inlined loop with a function call to the same loop is just stupid and pointless if you don't have long strings and fancy libraries with vectorization support etc.

For the regular 'strlen()' cases, we want the compiler to do this in order to handle the trivial case of constant strings. And we do have optimized versions of 'strlen()' on some architectures. But for
wcslen? Just no. - Linus ]

---
 Makefile | 3 +++
 1 file changed, 3 insertions(+)

index 6cb965fe2cb3..05eed26d3e4b 100644
@@ -930,6 +930,9 @@ KBUILD_CFLAGS   += $(call cc-option,-Werror=incompatible-pointer-types)
 # Require designated initializers for all marked structures
 KBUILD_CFLAGS   += $(call cc-option,-Werror=designated-init)
 
+# Ensure compilers do not transform certain loops into calls to wcslen()
+KBUILD_CFLAGS += -fno-builtin-wcslen
+
 # change __FILE__ to the relative path from the srctree
 KBUILD_CFLAGS	+= $(call cc-option,-fmacro-prefix-map=$(srctree)/=)
 

base-commit: 1b01d9c341770d4dab5a1a04a706a8eb6c389bb1

2.49.0

From b0f27298358b4b6f1b44a586fa0de88ce6ca74d0 Mon Sep 17 00:00:00 2001
From: Nathan Chancellor <[email protected]>
Date: Mon, 7 Apr 2025 16:22:12 -0700
Subject: [PATCH 5.10] kbuild: Add '-fno-builtin-wcslen'

commit 84ffc79bfbf70c779e60218563f2f3ad45288671 upstream.

A recent optimization change in LLVM [1] aims to transform certain loop idioms into calls to strlen() or wcslen(). This change transforms the first while loop in UniStrcat() into a call to wcslen(), breaking the build when UniStrcat() gets inlined into alloc_path_with_tree_prefix():

ld.lld: error: undefined symbol: wcslen

Disable this optimization with '-fno-builtin-wcslen', which prevents the compiler from assuming that wcslen() is available in the kernel's C library.

[ More to the point - it's not that we couldn't implement wcslen(), it's
that this isn't an optimization at all in the context of the kernel.

Replacing a simple inlined loop with a function call to the same loop is just stupid and pointless if you don't have long strings and fancy libraries with vectorization support etc.

For the regular 'strlen()' cases, we want the compiler to do this in order to handle the trivial case of constant strings. And we do have optimized versions of 'strlen()' on some architectures. But for
wcslen? Just no. - Linus ]

---
 Makefile | 3 +++
 1 file changed, 3 insertions(+)

index e13f8f8a9ea9..6f7a5e72bbb9 100644
@@ -976,6 +976,9 @@ KBUILD_CFLAGS   += $(call cc-option,-Werror=incompatible-pointer-types)
 # Require designated initializers for all marked structures
 KBUILD_CFLAGS   += $(call cc-option,-Werror=designated-init)
 
+# Ensure compilers do not transform certain loops into calls to wcslen()
+KBUILD_CFLAGS += -fno-builtin-wcslen
+
 # change __FILE__ to the relative path from the srctree
 KBUILD_CPPFLAGS += $(call cc-option,-fmacro-prefix-map=$(srctree)/=)
 

base-commit: 7222a9d1c9246944d284902fb9aed25613dd0ece

2.49.0

From bc78bacca3f3f9f7367afc6dbf56a4733f47e3f6 Mon Sep 17 00:00:00 2001
From: Nathan Chancellor <[email protected]>
Date: Mon, 7 Apr 2025 16:22:12 -0700
Subject: [PATCH 5.15] kbuild: Add '-fno-builtin-wcslen'

commit 84ffc79bfbf70c779e60218563f2f3ad45288671 upstream.

A recent optimization change in LLVM [1] aims to transform certain loop idioms into calls to strlen() or wcslen(). This change transforms the first while loop in UniStrcat() into a call to wcslen(), breaking the build when UniStrcat() gets inlined into alloc_path_with_tree_prefix():

ld.lld: error: undefined symbol: wcslen

Disable this optimization with '-fno-builtin-wcslen', which prevents the compiler from assuming that wcslen() is available in the kernel's C library.

[ More to the point - it's not that we couldn't implement wcslen(), it's
that this isn't an optimization at all in the context of the kernel.

Replacing a simple inlined loop with a function call to the same loop is just stupid and pointless if you don't have long strings and fancy libraries with vectorization support etc.

For the regular 'strlen()' cases, we want the compiler to do this in order to handle the trivial case of constant strings. And we do have optimized versions of 'strlen()' on some architectures. But for
wcslen? Just no. - Linus ]

---
 Makefile | 3 +++
 1 file changed, 3 insertions(+)

index 619f4ec0f613..035578b9fafc 100644
@@ -1066,6 +1066,9 @@ KBUILD_CFLAGS   += $(call cc-option,-Werror=incompatible-pointer-types)
 # Require designated initializers for all marked structures
 KBUILD_CFLAGS   += $(call cc-option,-Werror=designated-init)
 
+# Ensure compilers do not transform certain loops into calls to wcslen()
+KBUILD_CFLAGS += -fno-builtin-wcslen
+
 # change __FILE__ to the relative path from the srctree
 KBUILD_CPPFLAGS += $(call cc-option,-fmacro-prefix-map=$(srctree)/=)
 

base-commit: f7347f4005727f3155551c0550f4deb9c40b56c2

2.49.0

From a87930300e5abe2d100b17d7941eb2a4357b88b6 Mon Sep 17 00:00:00 2001
From: Nathan Chancellor <[email protected]>
Date: Mon, 7 Apr 2025 16:22:12 -0700
Subject: [PATCH 6.1] kbuild: Add '-fno-builtin-wcslen'

commit 84ffc79bfbf70c779e60218563f2f3ad45288671 upstream.

A recent optimization change in LLVM [1] aims to transform certain loop idioms into calls to strlen() or wcslen(). This change transforms the first while loop in UniStrcat() into a call to wcslen(), breaking the build when UniStrcat() gets inlined into alloc_path_with_tree_prefix():

ld.lld: error: undefined symbol: wcslen

Disable this optimization with '-fno-builtin-wcslen', which prevents the compiler from assuming that wcslen() is available in the kernel's C library.

[ More to the point - it's not that we couldn't implement wcslen(), it's
that this isn't an optimization at all in the context of the kernel.

Replacing a simple inlined loop with a function call to the same loop is just stupid and pointless if you don't have long strings and fancy libraries with vectorization support etc.

For the regular 'strlen()' cases, we want the compiler to do this in order to handle the trivial case of constant strings. And we do have optimized versions of 'strlen()' on some architectures. But for
wcslen? Just no. - Linus ]

---
 Makefile | 3 +++
 1 file changed, 3 insertions(+)

index 50f1a283b67d..eae097d3db6c 100644
@@ -1075,6 +1075,9 @@ KBUILD_CFLAGS   += $(call cc-option,-Werror=incompatible-pointer-types)
 # Require designated initializers for all marked structures
 KBUILD_CFLAGS   += $(call cc-option,-Werror=designated-init)
 
+# Ensure compilers do not transform certain loops into calls to wcslen()
+KBUILD_CFLAGS += -fno-builtin-wcslen
+
 # change __FILE__ to the relative path from the srctree
 KBUILD_CPPFLAGS += $(call cc-option,-fmacro-prefix-map=$(srctree)/=)
 

base-commit: 420102835862f49ec15c545594278dc5d2712f42

2.49.0

From 1b9fbad11767ed56fdc7da1b252913e2f2a155e9 Mon Sep 17 00:00:00 2001
From: Nathan Chancellor <[email protected]>
Date: Mon, 7 Apr 2025 16:22:12 -0700
Subject: [PATCH 6.6] kbuild: Add '-fno-builtin-wcslen'

commit 84ffc79bfbf70c779e60218563f2f3ad45288671 upstream.

A recent optimization change in LLVM [1] aims to transform certain loop idioms into calls to strlen() or wcslen(). This change transforms the first while loop in UniStrcat() into a call to wcslen(), breaking the build when UniStrcat() gets inlined into alloc_path_with_tree_prefix():

ld.lld: error: undefined symbol: wcslen

Disable this optimization with '-fno-builtin-wcslen', which prevents the compiler from assuming that wcslen() is available in the kernel's C library.

[ More to the point - it's not that we couldn't implement wcslen(), it's
that this isn't an optimization at all in the context of the kernel.

Replacing a simple inlined loop with a function call to the same loop is just stupid and pointless if you don't have long strings and fancy libraries with vectorization support etc.

For the regular 'strlen()' cases, we want the compiler to do this in order to handle the trivial case of constant strings. And we do have optimized versions of 'strlen()' on some architectures. But for
wcslen? Just no. - Linus ]

---
 Makefile | 3 +++
 1 file changed, 3 insertions(+)

index 45f6b7d3d51e..18ca8d458fa4 100644
@@ -1004,6 +1004,9 @@ ifdef CONFIG_CC_IS_GCC
 KBUILD_CFLAGS   += -fconserve-stack
 endif
 
+# Ensure compilers do not transform certain loops into calls to wcslen()
+KBUILD_CFLAGS += -fno-builtin-wcslen
+
 # change __FILE__ to the relative path from the srctree
 KBUILD_CPPFLAGS += $(call cc-option,-fmacro-prefix-map=$(srctree)/=)
 

base-commit: 814637ca257f4faf57a73fd4e38888cce88b5911

2.49.0

From 99d01048fc148dc10471b67da642df190a9727bb Mon Sep 17 00:00:00 2001
From: Nathan Chancellor <[email protected]>
Date: Mon, 7 Apr 2025 16:22:12 -0700
Subject: [PATCH 6.12] kbuild: Add '-fno-builtin-wcslen'

commit 84ffc79bfbf70c779e60218563f2f3ad45288671 upstream.

A recent optimization change in LLVM [1] aims to transform certain loop idioms into calls to strlen() or wcslen(). This change transforms the first while loop in UniStrcat() into a call to wcslen(), breaking the build when UniStrcat() gets inlined into alloc_path_with_tree_prefix():

ld.lld: error: undefined symbol: wcslen

Disable this optimization with '-fno-builtin-wcslen', which prevents the compiler from assuming that wcslen() is available in the kernel's C library.

[ More to the point - it's not that we couldn't implement wcslen(), it's
that this isn't an optimization at all in the context of the kernel.

Replacing a simple inlined loop with a function call to the same loop is just stupid and pointless if you don't have long strings and fancy libraries with vectorization support etc.

For the regular 'strlen()' cases, we want the compiler to do this in order to handle the trivial case of constant strings. And we do have optimized versions of 'strlen()' on some architectures. But for
wcslen? Just no. - Linus ]

---
 Makefile | 3 +++
 1 file changed, 3 insertions(+)

index 6a2a60eb67a3..1e0ec6317c17 100644
@@ -1013,6 +1013,9 @@ ifdef CONFIG_CC_IS_GCC
 KBUILD_CFLAGS   += -fconserve-stack
 endif
 
+# Ensure compilers do not transform certain loops into calls to wcslen()
+KBUILD_CFLAGS += -fno-builtin-wcslen
+
 # change __FILE__ to the relative path from the srctree
 KBUILD_CPPFLAGS += $(call cc-option,-fmacro-prefix-map=$(srctree)/=)
 

base-commit: 83b4161a63b87ce40d9f24f09b5b006f63d95b7c

2.49.0

Now applied, thanks.

greg k-h