mm: Restrict _shared_alloc_tag static definition to CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU

PATCH Superseded Hao Ge 3 participants 4 messages Jun 17, 2025

Subsystems (7)

Memory management Library code MM core s390 +3 more subsystems

lore

Share & tools

lore

Hao Ge

author reviewer

2025-06-17 07:11 UTC #0 lore raw

v5 cover mm: Restrict _shared_alloc_tag static definition to CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU

From: Hao Ge <[email protected]>

Recently discovered this entry while checking kallsyms on ARM64:
ffff800083e509c0 D _shared_alloc_tag

If ARCH_NEEDS_WEAK_PER_CPU is not defined(it is only defined for s390 and alpha architectures), there's no need to statically define the percpu variable _shared_alloc_tag.

Therefore, we need to implement isolation for this purpose.

However, currently ARCH_NEEDS_WEAK_PER_CPU is a #define and is enclosed within the #if defined(MODULE) conditional block.

When building the core kernel code for s390 or alpha architectures, ARCH_NEEDS_WEAK_PER_CPU remains undefined (as it is gated by #if defined(MODULE)). However, when building modules for these architectures, the macro is explicitly defined.

Therefore, we remove all instances of ARCH_NEEDS_WEAK_PER_CPU from the code and introduced CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU to replace the relevant logic. We can now conditionally define the perpcu variable _shared_alloc_tag based on CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU. This allows architectures (such as s390/alpha) that require weak definitions for percpu variables in modules to include the definition, while others can omit it via compile-time exclusion.

The following version can be regarded as the most original version:
https://lore.kernel.org/all/[email protected]/
But unfortunately, it caused build errors on s390.
Based on Suren's guidance and suggestions,
I've refined it into this patch series.
Many thanks to Suren for his patient instruction.

Verify:

     1. On Arm64:
        nm vmlinux | grep "_shared_alloc_tag",no output is returned.
     2. On S390:
        Compile tested.
        nm vmlinux | grep "_shared_alloc_tag"
        00000000015605b4 r __crc__shared_alloc_tag
        0000000001585fef r __kstrtab__shared_alloc_tag
        0000000001586897 r __kstrtabns__shared_alloc_tag
        00000000014f6548 r __ksymtab__shared_alloc_tag
        0000000001a8fa28 D _shared_alloc_tag
        nm net/ceph/libceph.ko | grep "_shared"
        U _shared_alloc_tag
     3. On alpha
        Compile tested.
        nm vmlinux | grep "_shared_alloc_tag"
        fffffc0000b080fa r __kstrtab__shared_alloc_tag
        fffffc0000b07ee7 r __kstrtabns__shared_alloc_tag
        fffffc0000adee98 r __ksymtab__shared_alloc_tag
        fffffc0000b83d38 D _shared_alloc_tag
        nm crypto/cryptomgr.ko | grep "_share"
        U _shared_alloc_tag

v5: Regarding the omission of defined(MODULE) in alloc_tag.h where
only #ifdef CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU was used, I apologize for this error.
Please find version 5 attached to address this issue.

v4:
Merge previous patches into a single patch.
Remove all instances of ARCH_MODULE_NEEDS_WEAK_PER_CPU from v3 and use CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU to
replace the relevant logic.
Replace CONFIG_ARCH_NEEDS_WEAK_PER_CPU with CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU in v3, as weak percpu support is only required for modules ,making the name more semantically accurate.
David, Mike, Matthew, Kent, Heiko and Suren have all provided valuable input. Thanks for this.

v3:
Suren pointed out that patches 1-2 can be merged into a single patch in version 2. And the commit message for patch 3 can be made more concise.Make corresponding modifications based on the pointed-out issues and update the corresponding commit message.

v2:
Heiko pointed out that when defining MODULE_NEEDS_WEAK_PER_CPU, the CONFIG_ARCH_NEEDS_WEAK_PER_CPU condition in the v1 version should be removed,as it is always true for s390 and alpha architectures.And He also pointed out that patches 2-4 need to be merged into one patch. Modify the code according to the suggestions and update the corresponding commit message

Hao Ge (1):
mm/percpu: Conditionally define _shared_alloc_tag via
CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU

 arch/alpha/Kconfig              | 1 +
 arch/alpha/include/asm/percpu.h | 5 ++---
 arch/s390/Kconfig               | 1 +
 arch/s390/include/asm/percpu.h  | 5 ++---
 include/linux/alloc_tag.h       | 6 +++---
 include/linux/percpu-defs.h     | 7 ++++---
 lib/alloc_tag.c                 | 2 ++
 mm/Kconfig                      | 7 +++++++
 8 files changed, 22 insertions(+), 12 deletions(-)

Signature · 1 line

2.25.1

Hao Ge

author reviewer

2025-06-17 07:12 UTC #1 lore raw

PATCH v5 1/1 mm/percpu: Conditionally define _shared_alloc_tag via CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU

From: Hao Ge <[email protected]>

Recently discovered this entry while checking kallsyms on ARM64:
ffff800083e509c0 D _shared_alloc_tag

If ARCH_NEEDS_WEAK_PER_CPU is not defined(it is only defined for s390 and alpha architectures), there's no need to statically define the percpu variable _shared_alloc_tag.

Therefore, we need to implement isolation for this purpose.

Signed-off-by Hao Ge <[email protected]>

---
 arch/alpha/Kconfig              | 1 +
 arch/alpha/include/asm/percpu.h | 5 ++---
 arch/s390/Kconfig               | 1 +
 arch/s390/include/asm/percpu.h  | 5 ++---
 include/linux/alloc_tag.h       | 6 +++---
 include/linux/percpu-defs.h     | 7 ++++---
 lib/alloc_tag.c                 | 2 ++
 mm/Kconfig                      | 7 +++++++
 8 files changed, 22 insertions(+), 12 deletions(-)

arch/alpha/Kconfig +1 -0

index 109a4cddcd13..80367f2cf821 100644
@@ -7,6 +7,7 @@ config ALPHA
 	select ARCH_HAS_DMA_OPS if PCI
 	select ARCH_MIGHT_HAVE_PC_PARPORT
 	select ARCH_MIGHT_HAVE_PC_SERIO
+	select ARCH_MODULE_NEEDS_WEAK_PER_CPU if SMP
 	select ARCH_NO_PREEMPT
 	select ARCH_NO_SG_CHAIN
 	select ARCH_USE_CMPXCHG_LOCKREF

arch/alpha/include/asm/percpu.h +2 -3

index 6923249f2d49..4383d66341dc 100644
@@ -9,10 +9,9 @@
  * way above 4G.
  *
  * Always use weak definitions for percpu variables in modules.
+ * Therefore, we have enabled CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU
+ * in the Kconfig.
  */
-#if defined(MODULE) && defined(CONFIG_SMP)
-#define ARCH_NEEDS_WEAK_PER_CPU
-#endif
 
 #include <asm-generic/percpu.h>

arch/s390/Kconfig +1 -0

index 0c16dc443e2f..b652cb952f31 100644
@@ -132,6 +132,7 @@ config S390
 	select ARCH_INLINE_WRITE_UNLOCK_IRQ
 	select ARCH_INLINE_WRITE_UNLOCK_IRQRESTORE
 	select ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
+	select ARCH_MODULE_NEEDS_WEAK_PER_CPU
 	select ARCH_STACKWALK
 	select ARCH_SUPPORTS_ATOMIC_RMW
 	select ARCH_SUPPORTS_DEBUG_PAGEALLOC

arch/s390/include/asm/percpu.h +2 -3

index 84f6b8357b45..96af7d964014 100644
@@ -16,10 +16,9 @@
  * For 64 bit module code, the module may be more than 4G above the
  * per cpu area, use weak definitions to force the compiler to
  * generate external references.
+ * Therefore, we have enabled CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU
+ * in the Kconfig.
  */
-#if defined(MODULE)
-#define ARCH_NEEDS_WEAK_PER_CPU
-#endif
 
 /*
  * We use a compare-and-swap loop since that uses less cpu cycles than

include/linux/alloc_tag.h +3 -3

index 8f7931eb7d16..9ef2633e2c08 100644
@@ -88,7 +88,7 @@ static inline struct alloc_tag *ct_to_alloc_tag(struct codetag *ct)
 	return container_of(ct, struct alloc_tag, ct);
 }
 
-#ifdef ARCH_NEEDS_WEAK_PER_CPU
+#if defined(CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU) && defined(MODULE)
 /*
  * When percpu variables are required to be defined as weak, static percpu
  * variables can't be used inside a function (see comments for DECLARE_PER_CPU_SECTION).

@@ -102,7 +102,7 @@ DECLARE_PER_CPU(struct alloc_tag_counters, _shared_alloc_tag);
 		.ct = CODE_TAG_INIT,						\
 		.counters = &_shared_alloc_tag };
 
-#else /* ARCH_NEEDS_WEAK_PER_CPU */
+#else /* CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU && MODULE */
 
 #ifdef MODULE

@@ -123,7 +123,7 @@ DECLARE_PER_CPU(struct alloc_tag_counters, _shared_alloc_tag);
 
 #endif /* MODULE */
 
-#endif /* ARCH_NEEDS_WEAK_PER_CPU */
+#endif /* CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU && MODULE */
 
 DECLARE_STATIC_KEY_MAYBE(CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT,
 			mem_alloc_profiling_key);

include/linux/percpu-defs.h +4 -3

index 0aeb0e276a3e..5931fcad9a91 100644
@@ -63,14 +63,15 @@
  * 1. The symbol must be globally unique, even the static ones.
  * 2. Static percpu variables cannot be defined inside a function.
  *
- * Archs which need weak percpu definitions should define
- * ARCH_NEEDS_WEAK_PER_CPU in asm/percpu.h when necessary.
+ * Archs which need weak percpu definitions should set
+ * CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU when necessary.
  *
  * To ensure that the generic code observes the above two
  * restrictions, if CONFIG_DEBUG_FORCE_WEAK_PER_CPU is set weak
  * definition is used for all cases.
  */
-#if defined(ARCH_NEEDS_WEAK_PER_CPU) || defined(CONFIG_DEBUG_FORCE_WEAK_PER_CPU)
+#if defined(CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU) && defined(MODULE) || \
+	defined(CONFIG_DEBUG_FORCE_WEAK_PER_CPU)
 /*
  * __pcpu_scope_* dummy variable is used to enforce scope.  It
  * receives the static modifier when it's used in front of

lib/alloc_tag.c +2 -0

index c7f602fa7b23..ab0936ebf38e 100644
@@ -24,8 +24,10 @@ static bool mem_profiling_support;
 
 static struct codetag_type *alloc_tag_cttype;
 
+#ifdef CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU
 DEFINE_PER_CPU(struct alloc_tag_counters, _shared_alloc_tag);
 EXPORT_SYMBOL(_shared_alloc_tag);
+#endif
 
 DEFINE_STATIC_KEY_MAYBE(CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT,
 			mem_alloc_profiling_key);

mm/Kconfig +7 -0

index e113f713b493..00514df3eae4 100644
@@ -929,6 +929,13 @@ config ARCH_SUPPORTS_PUD_PFNMAP
 	def_bool y
 	depends on ARCH_SUPPORTS_HUGE_PFNMAP && HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
 
+#
+# Architectures that always use weak definitions for percpu
+# variables in modules should set this.
+#
+config ARCH_MODULE_NEEDS_WEAK_PER_CPU
+       bool
+
 #
 # UP and nommu archs use km based percpu allocator
 #

Signature · 1 line

2.25.1

Alexander Gordeev

maintainer Acked-by on v5 1/1

2025-06-17 13:08 UTC #2 lore raw

Re: [PATCH v5 1/1] mm/percpu: Conditionally define _shared_alloc_tag via CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU

On Tue, Jun 17, 2025 at 03:10:52PM +0800, Hao Ge wrote:

Hi Hao,

Quoted 36 lines

From: Hao Ge <[email protected]>
Recently discovered this entry while checking kallsyms on ARM64:
ffff800083e509c0 D _shared_alloc_tag
If ARCH_NEEDS_WEAK_PER_CPU is not defined(it is only defined for s390 and alpha architectures), there's no need to statically define the percpu variable _shared_alloc_tag.
Therefore, we need to implement isolation for this purpose.
When building the core kernel code for s390 or alpha architectures, ARCH_NEEDS_WEAK_PER_CPU remains undefined (as it is gated by #if defined(MODULE)). However, when building modules for these architectures, the macro is explicitly defined.
Therefore, we remove all instances of ARCH_NEEDS_WEAK_PER_CPU from the code and introduced CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU to replace the relevant logic. We can now conditionally define the perpcu variable _shared_alloc_tag based on CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU. This allows architectures (such as s390/alpha) that require weak definitions for percpu variables in modules to include the definition, while others can omit it via compile-time exclusion.
Suggested-by Suren Baghdasaryan <[email protected]>
Signed-off-by Hao Ge <[email protected]>
---
 arch/alpha/Kconfig              | 1 +
 arch/alpha/include/asm/percpu.h | 5 ++---
 arch/s390/Kconfig               | 1 +
 arch/s390/include/asm/percpu.h  | 5 ++---
 include/linux/alloc_tag.h       | 6 +++---
 include/linux/percpu-defs.h     | 7 ++++---
 lib/alloc_tag.c                 | 2 ++
 mm/Kconfig                      | 7 +++++++
 8 files changed, 22 insertions(+), 12 deletions(-)

...

Acked-by Alexander Gordeev <[email protected]> # s390

Thanks!

Suren Baghdasaryan

maintainer

2025-06-17 14:12 UTC #3 lore raw

Re: [PATCH v5 1/1] mm/percpu: Conditionally define _shared_alloc_tag via CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU

In reply to Hao Ge

Quoted 137 lines · Hao Ge

From: Hao Ge <[email protected]>

Recently discovered this entry while checking kallsyms on ARM64:
ffff800083e509c0 D _shared_alloc_tag

If ARCH_NEEDS_WEAK_PER_CPU is not defined(it is only defined for s390 and alpha architectures), there's no need to statically define the percpu variable _shared_alloc_tag.

Therefore, we need to implement isolation for this purpose.

Signed-off-by Hao Ge <[email protected]>

---
 arch/alpha/Kconfig              | 1 +
 arch/alpha/include/asm/percpu.h | 5 ++---
 arch/s390/Kconfig               | 1 +
 arch/s390/include/asm/percpu.h  | 5 ++---
 include/linux/alloc_tag.h       | 6 +++---
 include/linux/percpu-defs.h     | 7 ++++---
 lib/alloc_tag.c                 | 2 ++
 mm/Kconfig                      | 7 +++++++
 8 files changed, 22 insertions(+), 12 deletions(-)

arch/alpha/Kconfig +1 -0

index 109a4cddcd13..80367f2cf821 100644
@@ -7,6 +7,7 @@ config ALPHA
        select ARCH_HAS_DMA_OPS if PCI
        select ARCH_MIGHT_HAVE_PC_PARPORT
        select ARCH_MIGHT_HAVE_PC_SERIO
+       select ARCH_MODULE_NEEDS_WEAK_PER_CPU if SMP
        select ARCH_NO_PREEMPT
        select ARCH_NO_SG_CHAIN
        select ARCH_USE_CMPXCHG_LOCKREF

arch/alpha/include/asm/percpu.h +2 -3

index 6923249f2d49..4383d66341dc 100644
@@ -9,10 +9,9 @@
  * way above 4G.
  *
  * Always use weak definitions for percpu variables in modules.
+ * Therefore, we have enabled CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU
+ * in the Kconfig.
  */
-#if defined(MODULE) && defined(CONFIG_SMP)
-#define ARCH_NEEDS_WEAK_PER_CPU
-#endif
 
 #include <asm-generic/percpu.h>

arch/s390/Kconfig +1 -0

index 0c16dc443e2f..b652cb952f31 100644
@@ -132,6 +132,7 @@ config S390
        select ARCH_INLINE_WRITE_UNLOCK_IRQ
        select ARCH_INLINE_WRITE_UNLOCK_IRQRESTORE
        select ARCH_MHP_MEMMAP_ON_MEMORY_ENABLE
+       select ARCH_MODULE_NEEDS_WEAK_PER_CPU
        select ARCH_STACKWALK
        select ARCH_SUPPORTS_ATOMIC_RMW
        select ARCH_SUPPORTS_DEBUG_PAGEALLOC

arch/s390/include/asm/percpu.h +2 -3

index 84f6b8357b45..96af7d964014 100644
@@ -16,10 +16,9 @@
  * For 64 bit module code, the module may be more than 4G above the
  * per cpu area, use weak definitions to force the compiler to
  * generate external references.
+ * Therefore, we have enabled CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU
+ * in the Kconfig.
  */
-#if defined(MODULE)
-#define ARCH_NEEDS_WEAK_PER_CPU
-#endif
 
 /*
  * We use a compare-and-swap loop since that uses less cpu cycles than

include/linux/alloc_tag.h +3 -3

index 8f7931eb7d16..9ef2633e2c08 100644
@@ -88,7 +88,7 @@ static inline struct alloc_tag *ct_to_alloc_tag(struct codetag *ct)
        return container_of(ct, struct alloc_tag, ct);
 }
 
-#ifdef ARCH_NEEDS_WEAK_PER_CPU
+#if defined(CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU) && defined(MODULE)
 /*
  * When percpu variables are required to be defined as weak, static percpu
  * variables can't be used inside a function (see comments for DECLARE_PER_CPU_SECTION).

@@ -102,7 +102,7 @@ DECLARE_PER_CPU(struct alloc_tag_counters, _shared_alloc_tag);
                .ct = CODE_TAG_INIT,                                            \
                .counters = &_shared_alloc_tag };
 
-#else /* ARCH_NEEDS_WEAK_PER_CPU */
+#else /* CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU && MODULE */
 
 #ifdef MODULE

@@ -123,7 +123,7 @@ DECLARE_PER_CPU(struct alloc_tag_counters, _shared_alloc_tag);
 
 #endif /* MODULE */
 
-#endif /* ARCH_NEEDS_WEAK_PER_CPU */
+#endif /* CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU && MODULE */
 
 DECLARE_STATIC_KEY_MAYBE(CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT,
                        mem_alloc_profiling_key);

include/linux/percpu-defs.h +0 -2

index 0aeb0e276a3e..5931fcad9a91 100644
@@ -63,14 +63,15 @@
  * 1. The symbol must be globally unique, even the static ones.
  * 2. Static percpu variables cannot be defined inside a function.
  *
- * Archs which need weak percpu definitions should define
- * ARCH_NEEDS_WEAK_PER_CPU in asm/percpu.h when necessary.

include/linux/percpu-defs.h line 74

Quoted 10 lines

include/linux/percpu-defs.h +4 -1

+ * Archs which need weak percpu definitions should set
+ * CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU when necessary.
  *
  * To ensure that the generic code observes the above two
  * restrictions, if CONFIG_DEBUG_FORCE_WEAK_PER_CPU is set weak
  * definition is used for all cases.
  */
-#if defined(ARCH_NEEDS_WEAK_PER_CPU) || defined(CONFIG_DEBUG_FORCE_WEAK_PER_CPU)
+#if defined(CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU) && defined(MODULE) || \
+       defined(CONFIG_DEBUG_FORCE_WEAK_PER_CPU)

Please enclose defined(CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU) && defined(MODULE) part of the condition in parentheses.

Quoted 39 lines

include/linux/percpu-defs.h +0 -0

 /*
  * __pcpu_scope_* dummy variable is used to enforce scope.  It
  * receives the static modifier when it's used in front of

lib/alloc_tag.c +2 -0

index c7f602fa7b23..ab0936ebf38e 100644
@@ -24,8 +24,10 @@ static bool mem_profiling_support;
 
 static struct codetag_type *alloc_tag_cttype;
 
+#ifdef CONFIG_ARCH_MODULE_NEEDS_WEAK_PER_CPU
 DEFINE_PER_CPU(struct alloc_tag_counters, _shared_alloc_tag);
 EXPORT_SYMBOL(_shared_alloc_tag);
+#endif
 
 DEFINE_STATIC_KEY_MAYBE(CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT,
                        mem_alloc_profiling_key);

mm/Kconfig +7 -1

index e113f713b493..00514df3eae4 100644
@@ -929,6 +929,13 @@ config ARCH_SUPPORTS_PUD_PFNMAP
        def_bool y
        depends on ARCH_SUPPORTS_HUGE_PFNMAP && HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD
 
+#
+# Architectures that always use weak definitions for percpu
+# variables in modules should set this.
+#
+config ARCH_MODULE_NEEDS_WEAK_PER_CPU
+       bool
+
 #
 # UP and nommu archs use km based percpu allocator
 #
--

2.25.1