Skip to contentSkip to conversation

enable PT_RECLAIM on more 64-bit architectures

PATCH In review Qi Zheng 6 participants 19 messages Jan 27 – Feb 7

Subsystems (12)
Memory management x86 PowerPC MIPS +8 more subsystems
lore
Share & tools
lore

Qi Zheng

author
#0 lore raw

v4 cover enable PT_RECLAIM on more 64-bit architectures

From: Qi Zheng <[email protected]>

Changes in v4:
- convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config
- fix a WARN_ON_ONCE() on sparc64 (and on ppc)
(reported by Andreas Larsson)
- collect Acked-by (Hi David, I've kept your Acked-by, feel free to drop it)
- rebase onto the v6.19-rc7

Changes in v3:
- modify the commit message in [PATCH v3 1/7] (suggested by David Hildenbrand)
- make PT_RECLAIM depends on MMU_GATHER_RCU_TABLE_FREE instead of 64BIT
- collect Acked-by
- rebase onto the next-20251217

Changelog in v2:
- fix compilation errors (reported by Magnus Lindholm and kernel test robot)
- adjust some code style (suggested by Huacai Chen)
- make PT_RECLAIM user-unselectable (suggested by David Hildenbrand)
- rebase onto the next-20251119

Hi all,

This series aims to enable PT_RECLAIM on more 64-bit architectures.

On a 64-bit system, madvise(MADV_DONTNEED) may cause a large number of empty PTE page table pages (such as 100GB+). To resolve this problem, we need to enable PT_RECLAIM, which depends on MMU_GATHER_RCU_TABLE_FREE.

For these architectures that define its own __tlb_remove_table(), since their page tables are not of type struct ptdesc, they cannot be supported PT_RECLAIM.

Therefore, this series first enables MMU_GATHER_RCU_TABLE_FREE on all 64-bit architectures, then converts __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config, and finally makes PT_RECLAIM depend on MMU_GATHER_RCU_TABLE_FREE && !HAVE_ARCH_TLB_REMOVE_TABLE. This way, PT_RECLAIM can be enabled by default on most 64-bit architectures.

Of course, this will also be enabled on some 32-bit architectures that already support MMU_GATHER_RCU_TABLE_FREE. That's fine, PT_RECLAIM works well on all 32-bit architectures as well. Although the benefit isn't significant, there's still memory that can be reclaimed. Perhaps PT_RECLAIM can be enabled on all 32-bit architectures in the future.

Comments and suggestions are welcome!

Thanks,
Qi

Qi Zheng (8):
mm: change mm/pt_reclaim.c to use asm/tlb.h instead of
asm-generic/tlb.h
alpha: mm: enable MMU_GATHER_RCU_TABLE_FREE
LoongArch: mm: enable MMU_GATHER_RCU_TABLE_FREE
mips: mm: enable MMU_GATHER_RCU_TABLE_FREE
parisc: mm: enable MMU_GATHER_RCU_TABLE_FREE
um: mm: enable MMU_GATHER_RCU_TABLE_FREE
mm: convert __HAVE_ARCH_TLB_REMOVE_TABLE to
CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config
mm: make PT_RECLAIM depends on MMU_GATHER_RCU_TABLE_FREE

 arch/alpha/Kconfig                   | 1 +
 arch/alpha/include/asm/tlb.h         | 6 +++---
 arch/loongarch/Kconfig               | 1 +
 arch/loongarch/include/asm/pgalloc.h | 7 +++----
 arch/mips/Kconfig                    | 1 +
 arch/mips/include/asm/pgalloc.h      | 7 +++----
 arch/parisc/Kconfig                  | 1 +
 arch/parisc/include/asm/tlb.h        | 4 ++--
 arch/powerpc/Kconfig                 | 1 +
 arch/powerpc/include/asm/tlb.h       | 1 -
 arch/sparc/Kconfig                   | 1 +
 arch/sparc/include/asm/tlb_64.h      | 1 -
 arch/um/Kconfig                      | 1 +
 arch/x86/Kconfig                     | 1 -
 include/asm-generic/tlb.h            | 2 +-
 mm/Kconfig                           | 8 +++-----
 mm/pt_reclaim.c                      | 2 +-
 17 files changed, 23 insertions(+), 23 deletions(-)
Signature · 1 line
2.20.1
author
#1 lore raw

PATCH v4 1/8 mm: change mm/pt_reclaim.c to use asm/tlb.h instead of asm-generic/tlb.h

From: Qi Zheng <[email protected]>

Generally, the asm/tlb.h will include asm-generic/tlb.h, so change mm/pt_reclaim.c to use asm/tlb.h instead of asm-generic/tlb.h. This is a preparation for enabling CONFIG_PT_RECLAIM on other architectures, such as alpha.

---
 mm/pt_reclaim.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
mm/pt_reclaim.c +1 -1
index 0d9cfbf4fe5d8..46771cfff8239 100644
@@ -2,7 +2,7 @@
 #include <linux/hugetlb.h>
 #include <linux/pgalloc.h>
 
-#include <asm-generic/tlb.h>
+#include <asm/tlb.h>
 
 #include "internal.h"
 
Signature · 1 line
2.20.1
author
#2 lore raw

PATCH v4 2/8 alpha: mm: enable MMU_GATHER_RCU_TABLE_FREE

In reply to Qi Zheng

From: Qi Zheng <[email protected]>

On a 64-bit system, madvise(MADV_DONTNEED) may cause a large number of empty PTE page table pages (such as 100GB+). To resolve this problem, first enable MMU_GATHER_RCU_TABLE_FREE to prepare for enabling the PT_RECLAIM feature, which resolves this problem.

Cc Richard Henderson <[email protected]>
Cc Matt Turner <[email protected]>
---
 arch/alpha/Kconfig           | 1 +
 arch/alpha/include/asm/tlb.h | 6 +++---
 2 files changed, 4 insertions(+), 3 deletions(-)
arch/alpha/Kconfig +1 -0
index 80367f2cf821c..6c7dbf0adad62 100644
@@ -38,6 +38,7 @@ config ALPHA
 	select OLD_SIGSUSPEND
 	select CPU_NO_EFFICIENT_FFS if !ALPHA_EV67
 	select MMU_GATHER_NO_RANGE
+	select MMU_GATHER_RCU_TABLE_FREE
 	select SPARSEMEM_EXTREME if SPARSEMEM
 	select ZONE_DMA
 	help
arch/alpha/include/asm/tlb.h +3 -3
index 4f79e331af5ea..ad586b898fd6b 100644
@@ -4,7 +4,7 @@
 
 #include <asm-generic/tlb.h>
 
-#define __pte_free_tlb(tlb, pte, address)		pte_free((tlb)->mm, pte)
-#define __pmd_free_tlb(tlb, pmd, address)		pmd_free((tlb)->mm, pmd)
- 
+#define __pte_free_tlb(tlb, pte, address)	tlb_remove_ptdesc((tlb), page_ptdesc(pte))
+#define __pmd_free_tlb(tlb, pmd, address)	tlb_remove_ptdesc((tlb), virt_to_ptdesc(pmd))
+
 #endif
Signature · 1 line
2.20.1
maintainer Acked-by on v4 2/8
#3 lore raw

Re: [PATCH v4 2/8] alpha: mm: enable MMU_GATHER_RCU_TABLE_FREE

Quoted 44 lines · Qi Zheng

From: Qi Zheng <[email protected]>

On a 64-bit system, madvise(MADV_DONTNEED) may cause a large number of empty PTE page table pages (such as 100GB+). To resolve this problem, first enable MMU_GATHER_RCU_TABLE_FREE to prepare for enabling the PT_RECLAIM feature, which resolves this problem.

Cc Richard Henderson <[email protected]>
Cc Matt Turner <[email protected]>
---
 arch/alpha/Kconfig           | 1 +
 arch/alpha/include/asm/tlb.h | 6 +++---
 2 files changed, 4 insertions(+), 3 deletions(-)
arch/alpha/Kconfig +1 -0
index 80367f2cf821c..6c7dbf0adad62 100644
@@ -38,6 +38,7 @@ config ALPHA
        select OLD_SIGSUSPEND
        select CPU_NO_EFFICIENT_FFS if !ALPHA_EV67
        select MMU_GATHER_NO_RANGE
+       select MMU_GATHER_RCU_TABLE_FREE
        select SPARSEMEM_EXTREME if SPARSEMEM
        select ZONE_DMA
        help
arch/alpha/include/asm/tlb.h +3 -4
index 4f79e331af5ea..ad586b898fd6b 100644
@@ -4,7 +4,7 @@
 
 #include <asm-generic/tlb.h>
 
-#define __pte_free_tlb(tlb, pte, address)              pte_free((tlb)->mm, pte)
-#define __pmd_free_tlb(tlb, pmd, address)              pmd_free((tlb)->mm, pmd)
-
+#define __pte_free_tlb(tlb, pte, address)      tlb_remove_ptdesc((tlb), page_ptdesc(pte))
+#define __pmd_free_tlb(tlb, pmd, address)      tlb_remove_ptdesc((tlb), virt_to_ptdesc(pmd))
+
 #endif
--
Quoted 2 lines

2.20.1

Looks good from an Alpha perspective.

author
#4 lore raw

PATCH v4 3/8 LoongArch: mm: enable MMU_GATHER_RCU_TABLE_FREE

In reply to Qi Zheng

From: Qi Zheng <[email protected]>

On a 64-bit system, madvise(MADV_DONTNEED) may cause a large number of empty PTE page table pages (such as 100GB+). To resolve this problem, first enable MMU_GATHER_RCU_TABLE_FREE to prepare for enabling the PT_RECLAIM feature, which resolves this problem.

Cc Huacai Chen <[email protected]>
Cc WANG Xuerui <[email protected]>
---
 arch/loongarch/Kconfig               | 1 +
 arch/loongarch/include/asm/pgalloc.h | 7 +++----
 2 files changed, 4 insertions(+), 4 deletions(-)
arch/loongarch/Kconfig +1 -0
index 730f342145197..43d5b863e1fb2 100644
@@ -187,6 +187,7 @@ config LOONGARCH
 	select IRQ_LOONGARCH_CPU
 	select LOCK_MM_AND_FIND_VMA
 	select MMU_GATHER_MERGE_VMAS if MMU
+	select MMU_GATHER_RCU_TABLE_FREE
 	select MODULES_USE_ELF_RELA if MODULES
 	select NEED_PER_CPU_EMBED_FIRST_CHUNK
 	select NEED_PER_CPU_PAGE_FIRST_CHUNK
arch/loongarch/include/asm/pgalloc.h +3 -4
index 08dcc698ec184..248f62d0b590e 100644
@@ -55,8 +55,7 @@ static inline pte_t *pte_alloc_one_kernel(struct mm_struct *mm)
 	return pte;
 }
 
-#define __pte_free_tlb(tlb, pte, address)	\
-	tlb_remove_ptdesc((tlb), page_ptdesc(pte))
+#define __pte_free_tlb(tlb, pte, address)	tlb_remove_ptdesc((tlb), page_ptdesc(pte))
 
 #ifndef __PAGETABLE_PMD_FOLDED
 
@@ -79,7 +78,7 @@ static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long address)
 	return pmd;
 }
 
-#define __pmd_free_tlb(tlb, x, addr)	pmd_free((tlb)->mm, x)
+#define __pmd_free_tlb(tlb, x, addr)	tlb_remove_ptdesc((tlb), virt_to_ptdesc(x))
 
 #endif
 
@@ -99,7 +98,7 @@ static inline pud_t *pud_alloc_one(struct mm_struct *mm, unsigned long address)
 	return pud;
 }
 
-#define __pud_free_tlb(tlb, x, addr)	pud_free((tlb)->mm, x)
+#define __pud_free_tlb(tlb, x, addr)	tlb_remove_ptdesc((tlb), virt_to_ptdesc(x))
 
 #endif /* __PAGETABLE_PUD_FOLDED */
 
Signature · 1 line
2.20.1
author
#5 lore raw

PATCH v4 4/8 mips: mm: enable MMU_GATHER_RCU_TABLE_FREE

In reply to Qi Zheng

From: Qi Zheng <[email protected]>

On a 64-bit system, madvise(MADV_DONTNEED) may cause a large number of empty PTE page table pages (such as 100GB+). To resolve this problem, first enable MMU_GATHER_RCU_TABLE_FREE to prepare for enabling the PT_RECLAIM feature, which resolves this problem.

Cc Thomas Bogendoerfer <[email protected]>
---
 arch/mips/Kconfig               | 1 +
 arch/mips/include/asm/pgalloc.h | 7 +++----
 2 files changed, 4 insertions(+), 4 deletions(-)
arch/mips/Kconfig +1 -0
index b88b97139fa8e..c0c94e26ce396 100644
@@ -99,6 +99,7 @@ config MIPS
 	select IRQ_FORCED_THREADING
 	select ISA if EISA
 	select LOCK_MM_AND_FIND_VMA
+	select MMU_GATHER_RCU_TABLE_FREE
 	select MODULES_USE_ELF_REL if MODULES
 	select MODULES_USE_ELF_RELA if MODULES && 64BIT
 	select PERF_USE_VMALLOC
arch/mips/include/asm/pgalloc.h +3 -4
index 7a04381efa0b5..895bf79e76762 100644
@@ -48,8 +48,7 @@ static inline void pud_populate(struct mm_struct *mm, pud_t *pud, pmd_t *pmd)
 extern void pgd_init(void *addr);
 extern pgd_t *pgd_alloc(struct mm_struct *mm);
 
-#define __pte_free_tlb(tlb, pte, address)	\
-	tlb_remove_ptdesc((tlb), page_ptdesc(pte))
+#define __pte_free_tlb(tlb, pte, address)	tlb_remove_ptdesc((tlb), page_ptdesc(pte))
 
 #ifndef __PAGETABLE_PMD_FOLDED
 
@@ -72,7 +71,7 @@ static inline pmd_t *pmd_alloc_one(struct mm_struct *mm, unsigned long address)
 	return pmd;
 }
 
-#define __pmd_free_tlb(tlb, x, addr)	pmd_free((tlb)->mm, x)
+#define __pmd_free_tlb(tlb, x, addr)	tlb_remove_ptdesc((tlb), virt_to_ptdesc(x))
 
 #endif
 
@@ -97,7 +96,7 @@ static inline void p4d_populate(struct mm_struct *mm, p4d_t *p4d, pud_t *pud)
 	set_p4d(p4d, __p4d((unsigned long)pud));
 }
 
-#define __pud_free_tlb(tlb, x, addr)	pud_free((tlb)->mm, x)
+#define __pud_free_tlb(tlb, x, addr)	tlb_remove_ptdesc((tlb), virt_to_ptdesc(x))
 
 #endif /* __PAGETABLE_PUD_FOLDED */
 
Signature · 1 line
2.20.1
author
#6 lore raw

PATCH v4 5/8 parisc: mm: enable MMU_GATHER_RCU_TABLE_FREE

In reply to Qi Zheng

From: Qi Zheng <[email protected]>

On a 64-bit system, madvise(MADV_DONTNEED) may cause a large number of empty PTE page table pages (such as 100GB+). To resolve this problem, first enable MMU_GATHER_RCU_TABLE_FREE to prepare for enabling the PT_RECLAIM feature, which resolves this problem.

Cc "James E.J. Bottomley" <[email protected]>
Cc Helge Deller <[email protected]>
---
 arch/parisc/Kconfig           | 1 +
 arch/parisc/include/asm/tlb.h | 4 ++--
 2 files changed, 3 insertions(+), 2 deletions(-)
arch/parisc/Kconfig +1 -0
index 47fd9662d8005..62d5a89d5c7bc 100644
@@ -79,6 +79,7 @@ config PARISC
 	select GENERIC_CLOCKEVENTS
 	select CPU_NO_EFFICIENT_FFS
 	select THREAD_INFO_IN_TASK
+	select MMU_GATHER_RCU_TABLE_FREE
 	select NEED_DMA_MAP_STATE
 	select NEED_SG_DMA_LENGTH
 	select HAVE_ARCH_KGDB
arch/parisc/include/asm/tlb.h +2 -2
index 44235f367674d..4501fee0a8fa4 100644
@@ -5,8 +5,8 @@
 #include <asm-generic/tlb.h>
 
 #if CONFIG_PGTABLE_LEVELS == 3
-#define __pmd_free_tlb(tlb, pmd, addr)	pmd_free((tlb)->mm, pmd)
+#define __pmd_free_tlb(tlb, pmd, addr)	tlb_remove_ptdesc((tlb), virt_to_ptdesc(pmd))
 #endif
-#define __pte_free_tlb(tlb, pte, addr)	pte_free((tlb)->mm, pte)
+#define __pte_free_tlb(tlb, pte, addr)	tlb_remove_ptdesc((tlb), page_ptdesc(pte))
 
 #endif
Signature · 1 line
2.20.1
maintainer Acked-by Tested-by on v4 5/8
#7 lore raw

Re: [PATCH v4 5/8] parisc: mm: enable MMU_GATHER_RCU_TABLE_FREE

Quoted 7 lines · Qi Zheng

From: Qi Zheng <[email protected]>

On a 64-bit system, madvise(MADV_DONTNEED) may cause a large number of empty PTE page table pages (such as 100GB+). To resolve this problem, first enable MMU_GATHER_RCU_TABLE_FREE to prepare for enabling the PT_RECLAIM feature, which resolves this problem.

Quoted 7 lines
Cc "James E.J. Bottomley" <[email protected]>
Cc Helge Deller <[email protected]>
---
  arch/parisc/Kconfig           | 1 +
  arch/parisc/include/asm/tlb.h | 4 ++--
  2 files changed, 3 insertions(+), 2 deletions(-)

Thanks!
Helge

author
#8 lore raw

PATCH v4 6/8 um: mm: enable MMU_GATHER_RCU_TABLE_FREE

In reply to Qi Zheng

From: Qi Zheng <[email protected]>

On a 64-bit system, madvise(MADV_DONTNEED) may cause a large number of empty PTE page table pages (such as 100GB+). To resolve this problem, first enable MMU_GATHER_RCU_TABLE_FREE to prepare for enabling the PT_RECLAIM feature, which resolves this problem.

Cc Richard Weinberger <[email protected]>
Cc Anton Ivanov <[email protected]>
Cc Johannes Berg <[email protected]>
---
 arch/um/Kconfig | 1 +
 1 file changed, 1 insertion(+)
arch/um/Kconfig +1 -0
index 8415d39b0d430..098cda44db225 100644
@@ -42,6 +42,7 @@ config UML
 	select HAVE_SYSCALL_TRACEPOINTS
 	select THREAD_INFO_IN_TASK
 	select SPARSE_IRQ
+	select MMU_GATHER_RCU_TABLE_FREE
 
 config MMU
 	bool
Signature · 1 line
2.20.1
author
#9 lore raw

PATCH v4 7/8 mm: convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config

In reply to Qi Zheng

From: Qi Zheng <[email protected]>

For architectures that define __HAVE_ARCH_TLB_REMOVE_TABLE, the page tables at the pmd/pud level are generally not of struct ptdesc type, and do not have pt_rcu_head member, thus these architectures cannot support PT_RECLAIM.

In preparation for enabling PT_RECLAIM on more architectures, convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config, so that we can make conditional judgments in Kconfig.

---
 arch/powerpc/Kconfig            | 1 +
 arch/powerpc/include/asm/tlb.h  | 1 -
 arch/sparc/Kconfig              | 1 +
 arch/sparc/include/asm/tlb_64.h | 1 -
 include/asm-generic/tlb.h       | 2 +-
 mm/Kconfig                      | 3 +++
 6 files changed, 6 insertions(+), 3 deletions(-)
arch/powerpc/Kconfig +1 -0
index 9537a61ebae02..b47aa8fd62742 100644
@@ -304,6 +304,7 @@ config PPC
 	select LOCK_MM_AND_FIND_VMA
 	select MMU_GATHER_PAGE_SIZE
 	select MMU_GATHER_RCU_TABLE_FREE
+	select HAVE_ARCH_TLB_REMOVE_TABLE
 	select MMU_GATHER_MERGE_VMAS
 	select MMU_LAZY_TLB_SHOOTDOWN		if PPC_BOOK3S_64
 	select MODULES_USE_ELF_RELA
arch/powerpc/include/asm/tlb.h +0 -1
index 2058e8d3e0138..1ca7d4c4b90db 100644
@@ -37,7 +37,6 @@ extern void tlb_flush(struct mmu_gather *tlb);
  */
 #define tlb_needs_table_invalidate()	radix_enabled()
 
-#define __HAVE_ARCH_TLB_REMOVE_TABLE
 /* Get the generic bits... */
 #include <asm-generic/tlb.h>
 
arch/sparc/Kconfig +1 -0
index a630d373e6453..25fa2908d6152 100644
@@ -74,6 +74,7 @@ config SPARC64
 	select HAVE_KRETPROBES
 	select HAVE_KPROBES
 	select MMU_GATHER_RCU_TABLE_FREE if SMP
+	select HAVE_ARCH_TLB_REMOVE_TABLE if SMP
 	select MMU_GATHER_MERGE_VMAS
 	select MMU_GATHER_NO_FLUSH_CACHE
 	select HAVE_ARCH_TRANSPARENT_HUGEPAGE
arch/sparc/include/asm/tlb_64.h +0 -1
index 1a6e694418e39..3037187482db7 100644
@@ -33,7 +33,6 @@ void flush_tlb_pending(void);
 #define tlb_needs_table_invalidate()	(false)
 #endif
 
-#define __HAVE_ARCH_TLB_REMOVE_TABLE
 #include <asm-generic/tlb.h>
 
 #endif /* _SPARC64_TLB_H */
include/asm-generic/tlb.h +1 -1
index 4d679d2a206b4..3427b8036ba33 100644
@@ -213,7 +213,7 @@ struct mmu_table_batch {
 #define MAX_TABLE_BATCH		\
 	((PAGE_SIZE - sizeof(struct mmu_table_batch)) / sizeof(void *))
 
-#ifndef __HAVE_ARCH_TLB_REMOVE_TABLE
+#ifndef CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE
 static inline void __tlb_remove_table(void *table)
 {
 	struct ptdesc *ptdesc = (struct ptdesc *)table;
mm/Kconfig +3 -0
index a992f2203eb91..456e9eaedca39 100644
@@ -1451,6 +1451,9 @@ config ARCH_HAS_USER_SHADOW_STACK
 	  The architecture has hardware support for userspace shadow call
           stacks (eg, x86 CET, arm64 GCS or RISC-V Zicfiss).
 
+config HAVE_ARCH_TLB_REMOVE_TABLE
+	def_bool n
+
 config ARCH_SUPPORTS_PT_RECLAIM
 	def_bool n
 
Signature · 1 line
2.20.1
#10 lore raw

Re: [PATCH v4 7/8] mm: convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config

Quoted 50 lines · Qi Zheng

From: Qi Zheng <[email protected]>

For architectures that define __HAVE_ARCH_TLB_REMOVE_TABLE, the page tables at the pmd/pud level are generally not of struct ptdesc type, and do not have pt_rcu_head member, thus these architectures cannot support PT_RECLAIM.

In preparation for enabling PT_RECLAIM on more architectures, convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config, so that we can make conditional judgments in Kconfig.

---
  arch/powerpc/Kconfig            | 1 +
  arch/powerpc/include/asm/tlb.h  | 1 -
  arch/sparc/Kconfig              | 1 +
  arch/sparc/include/asm/tlb_64.h | 1 -
  include/asm-generic/tlb.h       | 2 +-
  mm/Kconfig                      | 3 +++
  6 files changed, 6 insertions(+), 3 deletions(-)
arch/powerpc/Kconfig +1 -0
index 9537a61ebae02..b47aa8fd62742 100644
@@ -304,6 +304,7 @@ config PPC
  	select LOCK_MM_AND_FIND_VMA
  	select MMU_GATHER_PAGE_SIZE
  	select MMU_GATHER_RCU_TABLE_FREE
+	select HAVE_ARCH_TLB_REMOVE_TABLE
  	select MMU_GATHER_MERGE_VMAS
  	select MMU_LAZY_TLB_SHOOTDOWN		if PPC_BOOK3S_64
  	select MODULES_USE_ELF_RELA
arch/powerpc/include/asm/tlb.h +0 -1
index 2058e8d3e0138..1ca7d4c4b90db 100644
@@ -37,7 +37,6 @@ extern void tlb_flush(struct mmu_gather *tlb);
   */
  #define tlb_needs_table_invalidate()	radix_enabled()
 
-#define __HAVE_ARCH_TLB_REMOVE_TABLE
  /* Get the generic bits... */
  #include <asm-generic/tlb.h>
 
arch/sparc/Kconfig +0 -0
index a630d373e6453..25fa2908d6152 100644

arch/sparc/Kconfig config SPARC64 · line 77

Quoted 4 lines
arch/sparc/Kconfig +1 -0
@@ -74,6 +74,7 @@ config SPARC64
  	select HAVE_KRETPROBES
  	select HAVE_KPROBES
  	select MMU_GATHER_RCU_TABLE_FREE if SMP
+	select HAVE_ARCH_TLB_REMOVE_TABLE if SMP

Why the SMP? At least in tlb_64.h, it was not protected by SMP IIUC.

Apart from that LGTM.

Signature · 3 lines
Cheers,

David
author
#11 lore raw

Re: [PATCH v4 7/8] mm: convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config

Quoted 47 lines · David Hildenbrand (Arm)
Quoted 46 lines · Qi Zheng

From: Qi Zheng <[email protected]>

For architectures that define __HAVE_ARCH_TLB_REMOVE_TABLE, the page tables at the pmd/pud level are generally not of struct ptdesc type, and do not have pt_rcu_head member, thus these architectures cannot support PT_RECLAIM.

In preparation for enabling PT_RECLAIM on more architectures, convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config, so that we can make conditional judgments in Kconfig.

---
  arch/powerpc/Kconfig            | 1 +
  arch/powerpc/include/asm/tlb.h  | 1 -
  arch/sparc/Kconfig              | 1 +
  arch/sparc/include/asm/tlb_64.h | 1 -
  include/asm-generic/tlb.h       | 2 +-
  mm/Kconfig                      | 3 +++

  6 files changed, 6 insertions(+), 3 deletions(-)
arch/powerpc/Kconfig +0 -0
index 9537a61ebae02..b47aa8fd62742 100644
@@ -304,6 +304,7 @@ config PPC

      select LOCK_MM_AND_FIND_VMA
      select MMU_GATHER_PAGE_SIZE
      select MMU_GATHER_RCU_TABLE_FREE
+    select HAVE_ARCH_TLB_REMOVE_TABLE
      select MMU_GATHER_MERGE_VMAS
      select MMU_LAZY_TLB_SHOOTDOWN        if PPC_BOOK3S_64 select MODULES_USE_ELF_RELA

arch/powerpc/include/asm/tlb.h +0 -1
index 2058e8d3e0138..1ca7d4c4b90db 100644
@@ -37,7 +37,6 @@ extern void tlb_flush(struct mmu_gather *tlb);
    */
   #define tlb_needs_table_invalidate()    radix_enabled()
-#define __HAVE_ARCH_TLB_REMOVE_TABLE
   /* Get the generic bits... */
   #include <asm-generic/tlb.h>
arch/sparc/Kconfig +0 -1
index a630d373e6453..25fa2908d6152 100644
--- a/arch/sparc/Kconfig
Quoted 8 lines
Quoted 6 lines

+++ b/arch/sparc/Kconfig

arch/sparc/Kconfig +0 -0
@@ -74,6 +74,7 @@ config SPARC64

      select HAVE_KRETPROBES
      select HAVE_KPROBES
      select MMU_GATHER_RCU_TABLE_FREE if SMP
+    select HAVE_ARCH_TLB_REMOVE_TABLE if SMP

Why the SMP? At least in tlb_64.h, it was not protected by SMP IIUC.

Ah, It's just simply aligned with the MMU_GATHER_RCU_TABLE_FREE above.

Quoted 2 lines

Apart from that LGTM.

Thanks!

Quoted 1 line
#12 lore raw

Re: [PATCH v4 7/8] mm: convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config

Quoted 59 lines · Qi Zheng
Quoted 47 lines · David Hildenbrand (Arm)
Quoted 46 lines · Qi Zheng

From: Qi Zheng <[email protected]>

For architectures that define __HAVE_ARCH_TLB_REMOVE_TABLE, the page tables at the pmd/pud level are generally not of struct ptdesc type, and do not have pt_rcu_head member, thus these architectures cannot support PT_RECLAIM.

In preparation for enabling PT_RECLAIM on more architectures, convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config, so that we can make conditional judgments in Kconfig.

---
  arch/powerpc/Kconfig            | 1 +
  arch/powerpc/include/asm/tlb.h  | 1 -
  arch/sparc/Kconfig              | 1 +
  arch/sparc/include/asm/tlb_64.h | 1 -
  include/asm-generic/tlb.h       | 2 +-
  mm/Kconfig                      | 3 +++

  6 files changed, 6 insertions(+), 3 deletions(-)
arch/powerpc/Kconfig +0 -0
index 9537a61ebae02..b47aa8fd62742 100644
@@ -304,6 +304,7 @@ config PPC

      select LOCK_MM_AND_FIND_VMA
      select MMU_GATHER_PAGE_SIZE
      select MMU_GATHER_RCU_TABLE_FREE
+    select HAVE_ARCH_TLB_REMOVE_TABLE
      select MMU_GATHER_MERGE_VMAS
      select MMU_LAZY_TLB_SHOOTDOWN        if PPC_BOOK3S_64 select MODULES_USE_ELF_RELA

arch/powerpc/include/asm/tlb.h +0 -1
index 2058e8d3e0138..1ca7d4c4b90db 100644
@@ -37,7 +37,6 @@ extern void tlb_flush(struct mmu_gather *tlb);
    */
   #define tlb_needs_table_invalidate()    radix_enabled()
-#define __HAVE_ARCH_TLB_REMOVE_TABLE
   /* Get the generic bits... */
   #include <asm-generic/tlb.h>
arch/sparc/Kconfig +0 -1
index a630d373e6453..25fa2908d6152 100644
--- a/arch/sparc/Kconfig
Quoted 8 lines
Quoted 6 lines

+++ b/arch/sparc/Kconfig

arch/sparc/Kconfig +0 -0
@@ -74,6 +74,7 @@ config SPARC64

      select HAVE_KRETPROBES
      select HAVE_KPROBES
      select MMU_GATHER_RCU_TABLE_FREE if SMP
+    select HAVE_ARCH_TLB_REMOVE_TABLE if SMP

Why the SMP? At least in tlb_64.h, it was not protected by SMP IIUC.

Quoted 1 line

Ah, It's just simply aligned with the MMU_GATHER_RCU_TABLE_FREE above.

But does that work as expected for !SMP?

Signature · 3 lines
Cheers,

David
author
#13 lore raw

Re: [PATCH v4 7/8] mm: convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config

Quoted 60 lines · David Hildenbrand (Arm)
Quoted 59 lines · Qi Zheng
Quoted 55 lines · David Hildenbrand (Arm)
Quoted 48 lines · Qi Zheng

From: Qi Zheng <[email protected]>

For architectures that define __HAVE_ARCH_TLB_REMOVE_TABLE, the page tables at the pmd/pud level are generally not of struct ptdesc type, and do not have pt_rcu_head member, thus these architectures cannot support PT_RECLAIM.

In preparation for enabling PT_RECLAIM on more architectures, convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config, so that we can make conditional judgments in Kconfig.

---
  arch/powerpc/Kconfig            | 1 +
  arch/powerpc/include/asm/tlb.h  | 1 -
  arch/sparc/Kconfig              | 1 +
  arch/sparc/include/asm/tlb_64.h | 1 -
  include/asm-generic/tlb.h       | 2 +-
  mm/Kconfig                      | 3 +++

  6 files changed, 6 insertions(+), 3 deletions(-)
arch/powerpc/Kconfig +0 -0
index 9537a61ebae02..b47aa8fd62742 100644
@@ -304,6 +304,7 @@ config PPC

      select LOCK_MM_AND_FIND_VMA
      select MMU_GATHER_PAGE_SIZE
      select MMU_GATHER_RCU_TABLE_FREE
+    select HAVE_ARCH_TLB_REMOVE_TABLE
      select MMU_GATHER_MERGE_VMAS
      select MMU_LAZY_TLB_SHOOTDOWN        if PPC_BOOK3S_64 select MODULES_USE_ELF_RELA

arch/powerpc/include/asm/tlb.h +0 -1
index 2058e8d3e0138..1ca7d4c4b90db 100644
@@ -37,7 +37,6 @@ extern void tlb_flush(struct mmu_gather *tlb);
    */
   #define tlb_needs_table_invalidate()    radix_enabled()
-#define __HAVE_ARCH_TLB_REMOVE_TABLE
   /* Get the generic bits... */
   #include <asm-generic/tlb.h>
arch/sparc/Kconfig +0 -0
index a630d373e6453..25fa2908d6152 100644
@@ -74,6 +74,7 @@ config SPARC64
Quoted 4 lines

      select HAVE_KRETPROBES
      select HAVE_KPROBES
      select MMU_GATHER_RCU_TABLE_FREE if SMP
+    select HAVE_ARCH_TLB_REMOVE_TABLE if SMP

Why the SMP? At least in tlb_64.h, it was not protected by SMP IIUC.

Quoted 3 lines
Quoted 1 line

Ah, It's just simply aligned with the MMU_GATHER_RCU_TABLE_FREE above.

But does that work as expected for !SMP?

In the case of !SMP, tlb_remove_table() will not be called:

static inline void pgtable_free_tlb(struct mmu_gather *tlb, void *table, bool is_page)

{
	pgtable_free(table, is_page);
}

so I think it's okay to only select it in case of SMP.

Quoted 1 line
maintainer Acked-by
#14 lore raw

Re: [PATCH v4 7/8] mm: convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config

Quoted 9 lines · Qi Zheng
Quoted 6 lines · David Hildenbrand (Arm)
Quoted 4 lines · Qi Zheng

Ah, It's just simply aligned with the MMU_GATHER_RCU_TABLE_FREE above.

Quoted 8 lines
Quoted 1 line

But does that work as expected for !SMP?

In the case of !SMP, tlb_remove_table() will not be called:

static inline void pgtable_free_tlb(struct mmu_gather *tlb, void *table, bool is_page) {
    pgtable_free(table, is_page);
}

Ah, okay. Confusing stuff. Would have been nice to document/mention that in the patch description.

So if it compiles, all good

Signature · 3 lines
Cheers,

David
maintainer Tested-by Acked-by
#15 lore raw

Re: [PATCH v4 7/8] mm: convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config

Quoted 19 lines · David Hildenbrand (Arm)
Quoted 9 lines · Qi Zheng
Quoted 6 lines · David Hildenbrand (Arm)
Quoted 4 lines · Qi Zheng

Ah, It's just simply aligned with the MMU_GATHER_RCU_TABLE_FREE above.

Quoted 8 lines
Quoted 1 line

But does that work as expected for !SMP?

In the case of !SMP, tlb_remove_table() will not be called:

static inline void pgtable_free_tlb(struct mmu_gather *tlb, void *table, bool is_page) {
     pgtable_free(table, is_page);
}

Quoted 6 lines

Ah, okay. Confusing stuff. Would have been nice to document/mention that in the patch description.

So if it compiles, all good

It looks a bit weird, but it is actually more logically correct than before the patch. Before this in the !SMP sparc64 case __HAVE_ARCH_TLB_REMOVE_TABLE was defined even with no actual __tlb_remove_table() present. But it did not matter in practice. But I agree that a mention would not hurt.

Tested both with SMP and !SMP

Cheers,
Andreas

author
#16 lore raw

PATCH v4 8/8 mm: make PT_RECLAIM depends on MMU_GATHER_RCU_TABLE_FREE

In reply to Qi Zheng

From: Qi Zheng <[email protected]>

The PT_RECLAIM can work on all architectures that support MMU_GATHER_RCU_TABLE_FREE, except for those that have selected HAVE_ARCH_TLB_REMOVE_TABLE,so make PT_RECLAIM depends on MMU_GATHER_RCU_TABLE_FREE && !HAVE_ARCH_TLB_REMOVE_TABLE.

BTW, change PT_RECLAIM to be enabled by default, since nobody should want to turn it off.

---
 arch/x86/Kconfig | 1 -
 mm/Kconfig       | 9 ++-------
 2 files changed, 2 insertions(+), 8 deletions(-)
arch/x86/Kconfig +0 -1
index 80527299f859a..0d22da56a71b0 100644
@@ -331,7 +331,6 @@ config X86
 	select FUNCTION_ALIGNMENT_4B
 	imply IMA_SECURE_AND_OR_TRUSTED_BOOT    if EFI
 	select HAVE_DYNAMIC_FTRACE_NO_PATCHABLE
-	select ARCH_SUPPORTS_PT_RECLAIM		if X86_64
 	select ARCH_SUPPORTS_SCHED_SMT		if SMP
 	select SCHED_SMT			if SMP
 	select ARCH_SUPPORTS_SCHED_CLUSTER	if SMP
mm/Kconfig +2 -7
index 456e9eaedca39..1f3109b164d86 100644
@@ -1454,14 +1454,9 @@ config ARCH_HAS_USER_SHADOW_STACK
 config HAVE_ARCH_TLB_REMOVE_TABLE
 	def_bool n
 
-config ARCH_SUPPORTS_PT_RECLAIM
-	def_bool n
-
 config PT_RECLAIM
-	bool "reclaim empty user page table pages"
-	default y
-	depends on ARCH_SUPPORTS_PT_RECLAIM && MMU && SMP
-	select MMU_GATHER_RCU_TABLE_FREE
+	def_bool y
+	depends on MMU_GATHER_RCU_TABLE_FREE && !HAVE_ARCH_TLB_REMOVE_TABLE
 	help
 	  Try to reclaim empty user page table pages in paths other than munmap
 	  and exit_mmap path.
Signature · 1 line
2.20.1
maintainer
#17 lore raw
In reply to Qi Zheng
Quoted 1 line · On Tue, 27 Jan 2026 20:12:53 +0800 Qi Zheng <[email protected]> wrote:

This series aims to enable PT_RECLAIM on more 64-bit architectures.

Thanks, I updated mm.git's mm-unstable branch to v4.

Quoted 7 lines

Changes in v4:
- convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config
- fix a WARN_ON_ONCE() on sparc64 (and on ppc)
(reported by Andreas Larsson)
- collect Acked-by (Hi David, I've kept your Acked-by, feel free to drop it)
- rebase onto the v6.19-rc7

Below is how v4 altered mm.git.

I'm not seeing the WARN_ON_ONCE() fix. I assume that was due to the Kconfig alterations?

arch/powerpc/include/asm/tlb.h renamed from arch/powerpc/include/asm/tlb.h~b +0 -1
@@ -37,7 +37,6 @@ extern void tlb_flush(struct mmu_gather
  */
 #define tlb_needs_table_invalidate()	radix_enabled()
 
-#define __HAVE_ARCH_TLB_REMOVE_TABLE
 /* Get the generic bits... */
 #include <asm-generic/tlb.h>
 
arch/powerpc/Kconfig renamed from arch/powerpc/Kconfig~b +1 -0
@@ -305,6 +305,7 @@ config PPC
 	select LOCK_MM_AND_FIND_VMA
 	select MMU_GATHER_PAGE_SIZE
 	select MMU_GATHER_RCU_TABLE_FREE
+	select HAVE_ARCH_TLB_REMOVE_TABLE
 	select MMU_GATHER_MERGE_VMAS
 	select MMU_LAZY_TLB_SHOOTDOWN		if PPC_BOOK3S_64
 	select MODULES_USE_ELF_RELA
arch/sparc/include/asm/tlb_64.h renamed from arch/sparc/include/asm/tlb_64.h~b +0 -1
@@ -33,7 +33,6 @@ void flush_tlb_pending(void);
 #define tlb_needs_table_invalidate()	(false)
 #endif
 
-#define __HAVE_ARCH_TLB_REMOVE_TABLE
 #include <asm-generic/tlb.h>
 
 #endif /* _SPARC64_TLB_H */
arch/sparc/Kconfig renamed from arch/sparc/Kconfig~b +1 -0
@@ -74,6 +74,7 @@ config SPARC64
 	select HAVE_KRETPROBES
 	select HAVE_KPROBES
 	select MMU_GATHER_RCU_TABLE_FREE if SMP
+	select HAVE_ARCH_TLB_REMOVE_TABLE if SMP
 	select MMU_GATHER_MERGE_VMAS
 	select MMU_GATHER_NO_FLUSH_CACHE
 	select HAVE_ARCH_TRANSPARENT_HUGEPAGE
include/asm-generic/tlb.h renamed from include/asm-generic/tlb.h~b +1 -1
@@ -213,7 +213,7 @@ struct mmu_table_batch {
 #define MAX_TABLE_BATCH		\
 	((PAGE_SIZE - sizeof(struct mmu_table_batch)) / sizeof(void *))
 
-#ifndef __HAVE_ARCH_TLB_REMOVE_TABLE
+#ifndef CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE
 static inline void __tlb_remove_table(void *table)
 {
 	struct ptdesc *ptdesc = (struct ptdesc *)table;
mm/Kconfig renamed from mm/Kconfig~b +4 -1
@@ -1448,9 +1448,12 @@ config ARCH_HAS_USER_SHADOW_STACK
 	  The architecture has hardware support for userspace shadow call
           stacks (eg, x86 CET, arm64 GCS or RISC-V Zicfiss).
 
+config HAVE_ARCH_TLB_REMOVE_TABLE
+	def_bool n
+
 config PT_RECLAIM
 	def_bool y
-	depends on MMU_GATHER_RCU_TABLE_FREE
+	depends on MMU_GATHER_RCU_TABLE_FREE && !HAVE_ARCH_TLB_REMOVE_TABLE
 	help
 	  Try to reclaim empty user page table pages in paths other than munmap
 	  and exit_mmap path.

_

author
#18 lore raw
Quoted 14 lines · Andrew Morton
Quoted 1 line · On Tue, 27 Jan 2026 20:12:53 +0800 Qi Zheng <[email protected]> wrote:

This series aims to enable PT_RECLAIM on more 64-bit architectures.

Thanks, I updated mm.git's mm-unstable branch to v4.

Quoted 7 lines

Changes in v4:
- convert __HAVE_ARCH_TLB_REMOVE_TABLE to CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE config
- fix a WARN_ON_ONCE() on sparc64 (and on ppc)
(reported by Andreas Larsson)
- collect Acked-by (Hi David, I've kept your Acked-by, feel free to drop it)
- rebase onto the v6.19-rc7

Quoted 4 lines

Below is how v4 altered mm.git.

I'm not seeing the WARN_ON_ONCE() fix. I assume that was due to the Kconfig alterations?

Yes, sparc64 and ppc will select HAVE_ARCH_TLB_REMOVE_TABLE, so PT_RECLAIM will not be enabled. This means it won't affect sparc64 and ppc, and will not trigger WARN_ON_ONCE().

I don't have a testing environment for sparc64 and ppc, but I expect it should be able to fix the WARN_ON_ONCE().

Thanks,
Qi

Quoted 72 lines
arch/powerpc/include/asm/tlb.h renamed from arch/powerpc/include/asm/tlb.h~b +0 -1
@@ -37,7 +37,6 @@ extern void tlb_flush(struct mmu_gather
   */
  #define tlb_needs_table_invalidate()	radix_enabled()
 
-#define __HAVE_ARCH_TLB_REMOVE_TABLE
  /* Get the generic bits... */
  #include <asm-generic/tlb.h>
 
arch/powerpc/Kconfig renamed from arch/powerpc/Kconfig~b +1 -0
@@ -305,6 +305,7 @@ config PPC
  	select LOCK_MM_AND_FIND_VMA
  	select MMU_GATHER_PAGE_SIZE
  	select MMU_GATHER_RCU_TABLE_FREE
+	select HAVE_ARCH_TLB_REMOVE_TABLE
  	select MMU_GATHER_MERGE_VMAS
  	select MMU_LAZY_TLB_SHOOTDOWN		if PPC_BOOK3S_64
  	select MODULES_USE_ELF_RELA
arch/sparc/include/asm/tlb_64.h renamed from arch/sparc/include/asm/tlb_64.h~b +0 -1
@@ -33,7 +33,6 @@ void flush_tlb_pending(void);
  #define tlb_needs_table_invalidate()	(false)
  #endif
 
-#define __HAVE_ARCH_TLB_REMOVE_TABLE
  #include <asm-generic/tlb.h>
 
  #endif /* _SPARC64_TLB_H */
arch/sparc/Kconfig renamed from arch/sparc/Kconfig~b +1 -0
@@ -74,6 +74,7 @@ config SPARC64
  	select HAVE_KRETPROBES
  	select HAVE_KPROBES
  	select MMU_GATHER_RCU_TABLE_FREE if SMP
+	select HAVE_ARCH_TLB_REMOVE_TABLE if SMP
  	select MMU_GATHER_MERGE_VMAS
  	select MMU_GATHER_NO_FLUSH_CACHE
  	select HAVE_ARCH_TRANSPARENT_HUGEPAGE
include/asm-generic/tlb.h renamed from include/asm-generic/tlb.h~b +1 -1
@@ -213,7 +213,7 @@ struct mmu_table_batch {
  #define MAX_TABLE_BATCH		\
  	((PAGE_SIZE - sizeof(struct mmu_table_batch)) / sizeof(void *))
 
-#ifndef __HAVE_ARCH_TLB_REMOVE_TABLE
+#ifndef CONFIG_HAVE_ARCH_TLB_REMOVE_TABLE
  static inline void __tlb_remove_table(void *table)
  {
  	struct ptdesc *ptdesc = (struct ptdesc *)table;
mm/Kconfig renamed from mm/Kconfig~b +4 -1
@@ -1448,9 +1448,12 @@ config ARCH_HAS_USER_SHADOW_STACK
  	  The architecture has hardware support for userspace shadow call
            stacks (eg, x86 CET, arm64 GCS or RISC-V Zicfiss).
 
+config HAVE_ARCH_TLB_REMOVE_TABLE
+	def_bool n
+
  config PT_RECLAIM
  	def_bool y
-	depends on MMU_GATHER_RCU_TABLE_FREE
+	depends on MMU_GATHER_RCU_TABLE_FREE && !HAVE_ARCH_TLB_REMOVE_TABLE
  	help
  	  Try to reclaim empty user page table pages in paths other than munmap
  	  and exit_mmap path.

_