[PATCH 1/3] mm: Allow pagewalk without locks
Dev Jain
dev.jain at arm.com
Fri May 30 02:04:05 PDT 2025
It is noted at [1] that KFENCE can manipulate kernel pgtable entries during
softirqs. It does this by calling set_memory_valid() -> __change_memory_common().
This being a non-sleepable context, we cannot take the init_mm mmap lock.
Therefore, add PGWALK_NOLOCK to enable walk_page_range_novma() usage without
locks.
[1] https://lore.kernel.org/linux-arm-kernel/89d0ad18-4772-4d8f-ae8a-7c48d26a927e@arm.com/
Signed-off-by: Dev Jain <dev.jain at arm.com>
---
include/linux/pagewalk.h | 2 ++
mm/pagewalk.c | 12 ++++++++----
2 files changed, 10 insertions(+), 4 deletions(-)
diff --git a/include/linux/pagewalk.h b/include/linux/pagewalk.h
index 9700a29f8afb..9bc8853ed3de 100644
--- a/include/linux/pagewalk.h
+++ b/include/linux/pagewalk.h
@@ -14,6 +14,8 @@ enum page_walk_lock {
PGWALK_WRLOCK = 1,
/* vma is expected to be already write-locked during the walk */
PGWALK_WRLOCK_VERIFY = 2,
+ /* no lock is needed */
+ PGWALK_NOLOCK = 3,
};
/**
diff --git a/mm/pagewalk.c b/mm/pagewalk.c
index e478777c86e1..9657cf4664b2 100644
--- a/mm/pagewalk.c
+++ b/mm/pagewalk.c
@@ -440,6 +440,8 @@ static inline void process_vma_walk_lock(struct vm_area_struct *vma,
case PGWALK_RDLOCK:
/* PGWALK_RDLOCK is handled by process_mm_walk_lock */
break;
+ default:
+ break;
}
#endif
}
@@ -640,10 +642,12 @@ int walk_page_range_novma(struct mm_struct *mm, unsigned long start,
* specified address range from being freed. The caller should take
* other actions to prevent this race.
*/
- if (mm == &init_mm)
- mmap_assert_locked(walk.mm);
- else
- mmap_assert_write_locked(walk.mm);
+ if (ops->walk_lock != PGWALK_NOLOCK) {
+ if (mm == &init_mm)
+ mmap_assert_locked(walk.mm);
+ else
+ mmap_assert_write_locked(walk.mm);
+ }
return walk_pgd_range(start, end, &walk);
}
--
2.30.2
More information about the linux-arm-kernel
mailing list