| Message ID | ef8f6538b83b7fc3372602f90375348f9b4f3596.1744128123.git.agordeev@linux.ibm.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | mm: Fix apply_to_pte_range() vs lazy MMU mode | expand |
On Tue, Apr 08, 2025 at 06:07:32PM +0200, Alexander Gordeev wrote: Hi Andrew, > The lazy MMU mode can only be entered and left under the protection > of the page table locks for all page tables which may be modified. Heiko Carstens noticed that the above claim is not valid, since v6.15-rc1 commit 691ee97e1a9d ("mm: fix lazy mmu docs and usage"), which restates it to: "In the general case, no lock is guaranteed to be held between entry and exit of the lazy mode. So the implementation must assume preemption may be enabled" That effectively invalidates this patch, so it needs to be dropped. Patch 2 still could be fine, except -stable and Fixes tags and it does not need to aim 6.15-rcX. Do you want me to repost it? Thanks!
On Thu, 10 Apr 2025 16:50:33 +0200 Alexander Gordeev <agordeev@linux.ibm.com> wrote: > On Tue, Apr 08, 2025 at 06:07:32PM +0200, Alexander Gordeev wrote: > > Hi Andrew, > > > The lazy MMU mode can only be entered and left under the protection > > of the page table locks for all page tables which may be modified. > > Heiko Carstens noticed that the above claim is not valid, since > v6.15-rc1 commit 691ee97e1a9d ("mm: fix lazy mmu docs and usage"), > which restates it to: > > "In the general case, no lock is guaranteed to be held between entry and exit > of the lazy mode. So the implementation must assume preemption may be enabled" > > That effectively invalidates this patch, so it needs to be dropped. > > Patch 2 still could be fine, except -stable and Fixes tags and it does > not need to aim 6.15-rcX. Do you want me to repost it? I dropped the whole series - let's start again.
diff --git a/mm/kasan/shadow.c b/mm/kasan/shadow.c index edfa77959474..6531a7aa8562 100644 --- a/mm/kasan/shadow.c +++ b/mm/kasan/shadow.c @@ -308,14 +308,14 @@ static int kasan_populate_vmalloc_pte(pte_t *ptep, unsigned long addr, __memset((void *)page, KASAN_VMALLOC_INVALID, PAGE_SIZE); pte = pfn_pte(PFN_DOWN(__pa(page)), PAGE_KERNEL); - spin_lock(&init_mm.page_table_lock); if (likely(pte_none(ptep_get(ptep)))) { set_pte_at(&init_mm, addr, ptep, pte); page = 0; } - spin_unlock(&init_mm.page_table_lock); + if (page) free_page(page); + return 0; } @@ -401,13 +401,10 @@ static int kasan_depopulate_vmalloc_pte(pte_t *ptep, unsigned long addr, page = (unsigned long)__va(pte_pfn(ptep_get(ptep)) << PAGE_SHIFT); - spin_lock(&init_mm.page_table_lock); - if (likely(!pte_none(ptep_get(ptep)))) { pte_clear(&init_mm, addr, ptep); free_page(page); } - spin_unlock(&init_mm.page_table_lock); return 0; } diff --git a/mm/memory.c b/mm/memory.c index f0201c8ec1ce..1f3727104e99 100644 --- a/mm/memory.c +++ b/mm/memory.c @@ -2926,6 +2926,7 @@ static int apply_to_pte_range(struct mm_struct *mm, pmd_t *pmd, pte = pte_offset_kernel(pmd, addr); if (!pte) return err; + spin_lock(&init_mm.page_table_lock); } else { if (create) pte = pte_alloc_map_lock(mm, pmd, addr, &ptl); @@ -2951,7 +2952,9 @@ static int apply_to_pte_range(struct mm_struct *mm, pmd_t *pmd, arch_leave_lazy_mmu_mode(); - if (mm != &init_mm) + if (mm == &init_mm) + spin_unlock(&init_mm.page_table_lock); + else pte_unmap_unlock(mapped_pte, ptl); *mask |= PGTBL_PTE_MODIFIED;
The lazy MMU mode can only be entered and left under the protection of the page table locks for all page tables which may be modified. Yet, when it comes to kernel mappings apply_to_pte_range() does not take any locks. That does not conform arch_enter|leave_lazy_mmu_mode() semantics and could potentially lead to re-schedulling a process while in lazy MMU mode or racing on a kernel page table updates. Cc: stable@vger.kernel.org Fixes: 38e0edb15bd0 ("mm/apply_to_range: call pte function with lazy updates") Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com> --- mm/kasan/shadow.c | 7 ++----- mm/memory.c | 5 ++++- 2 files changed, 6 insertions(+), 6 deletions(-)