Skip to contentSkip to conversation

mm/huge_memory: simplify pgtable deposit detection

PATCH Applied Lance Yang 6 participants 8 messages Sep 17 – Sep 18

Subsystems (2)
lore
Share & tools
lore

Lance Yang

author reviewer
#0 lore raw

PATCH v2 1/1 mm/huge_memory: simplify pgtable deposit detection

Whether a PMD has a deposited PTE page table only depends on whether the architecture requires deposits or the VMA is anonymous.

Implement this rule directly in vma_has_deposited_pgtable(), avoiding mistaking a non-anonymous raw PFN PMD for one with a deposit and attempting to withdraw a page table that was never deposited.

---
v1 -> v2:
- Simplify the pgtable deposit check and rename the helper (per David) - thanks!
- https://lore.kernel.org/linux-mm/[email protected]

 mm/huge_memory.c | 22 +++++-----------------
 1 file changed, 5 insertions(+), 17 deletions(-)
mm/huge_memory.c +5 -17
index 30b7c63b0e35..bfd972b997f9 100644
@@ -2521,25 +2521,13 @@ static struct folio *normal_or_softleaf_folio_pmd(struct vm_area_struct *vma,
 	return pmd_to_softleaf_folio(pmdval);
 }
 
-static bool has_deposited_pgtable(struct vm_area_struct *vma, pmd_t pmdval,
-		struct folio *folio)
+static bool vma_has_deposited_pgtable(struct vm_area_struct *vma)
 {
-	/* Some architectures require unconditional depositing. */
-	if (arch_needs_pgtable_deposit())
-		return true;
-
-	/*
-	 * Huge zero always deposited except for DAX which handles itself, see
-	 * set_huge_zero_folio().
-	 */
-	if (is_huge_zero_pmd(pmdval))
-		return !vma_is_dax(vma);
-
 	/*
-	 * Otherwise, only anonymous folios are deposited, see
-	 * __do_huge_pmd_anonymous_page().
+	 * PMDs in anonymous VMAs always have a deposited page table. PMDs in
+	 * other VMAs only have one when required by the architecture.
 	 */
-	return folio && folio_test_anon(folio);
+	return arch_needs_pgtable_deposit() || vma_is_anonymous(vma);
 }
 
 /**
@@ -2579,7 +2567,7 @@ bool zap_huge_pmd(struct mmu_gather *tlb, struct vm_area_struct *vma,
 
 	is_present = pmd_present(orig_pmd);
 	folio = normal_or_softleaf_folio_pmd(vma, addr, orig_pmd, is_present);
-	has_deposit = has_deposited_pgtable(vma, orig_pmd, folio);
+	has_deposit = vma_has_deposited_pgtable(vma);
 	if (folio)
 		zap_huge_pmd_folio(mm, vma, orig_pmd, folio, is_present);
 	if (has_deposit)
Signature · 1 line
2.39.3 (Apple Git-146)
reviewer Reviewed-by on v2 1/1
#1 lore raw
Quoted 9 lines · Lance Yang

Whether a PMD has a deposited PTE page table only depends on whether the architecture requires deposits or the VMA is anonymous.

Implement this rule directly in vma_has_deposited_pgtable(), avoiding mistaking a non-anonymous raw PFN PMD for one with a deposit and attempting to withdraw a page table that was never deposited.

Signature · 1 line
  Kiryl Shutsemau / Kirill A. Shutemov
maintainer Acked-by on v2 1/1
#2 lore raw
In reply to Lance Yang
Quoted 10 lines · Lance Yang

Whether a PMD has a deposited PTE page table only depends on whether the architecture requires deposits or the VMA is anonymous.

Implement this rule directly in vma_has_deposited_pgtable(), avoiding mistaking a non-anonymous raw PFN PMD for one with a deposit and attempting to withdraw a page table that was never deposited.

---

Signature · 3 lines
Cheers,

David
reviewer Acked-by on v2 1/1
#3 lore raw
In reply to Lance Yang
Quoted 9 lines · Lance Yang

Whether a PMD has a deposited PTE page table only depends on whether the architecture requires deposits or the VMA is anonymous.

Implement this rule directly in vma_has_deposited_pgtable(), avoiding mistaking a non-anonymous raw PFN PMD for one with a deposit and attempting to withdraw a page table that was never deposited.

Quoted 8 lines

---
v1 -> v2:
- Simplify the pgtable deposit check and rename the helper (per David) - thanks!
- https://lore.kernel.org/linux-mm/[email protected]

 mm/huge_memory.c | 22 +++++-----------------
 1 file changed, 5 insertions(+), 17 deletions(-)

LGTM.

Best Regards,
Yan, Zi

maintainer
#4 lore raw
In reply to Lance Yang
Quoted 6 lines · On Thu, 17 Sep 2026 13:40:15 +0800 Lance Yang <[email protected]> wrote:

Whether a PMD has a deposited PTE page table only depends on whether the architecture requires deposits or the VMA is anonymous.

Implement this rule directly in vma_has_deposited_pgtable(), avoiding mistaking a non-anonymous raw PFN PMD for one with a deposit and attempting to withdraw a page table that was never deposited.

A large part of my life is asking $LLM "what are the userspace-visible runtime effects of this bug". Because NOBODY EVER TELLS ME!!!

This time it says "Processes using PMD-sized raw PFN mappings can trigger a kernel NULL-pointer dereference when the mapping is torn down", which sounds kinda important if true.

Help me out here?

Also, thanks ;)

author reviewer
#5 lore raw
Quoted 11 lines · Andrew Morton
Quoted 6 lines · On Thu, 17 Sep 2026 13:40:15 +0800 Lance Yang <[email protected]> wrote:

Whether a PMD has a deposited PTE page table only depends on whether the architecture requires deposits or the VMA is anonymous.

Implement this rule directly in vma_has_deposited_pgtable(), avoiding mistaking a non-anonymous raw PFN PMD for one with a deposit and attempting to withdraw a page table that was never deposited.

A large part of my life is asking $LLM "what are the userspace-visible runtime effects of this bug". Because NOBODY EVER TELLS ME!!!

That's on me ... I trimmed too much from v1.

Quoted 3 lines

This time it says "Processes using PMD-sized raw PFN mappings can trigger a kernel NULL-pointer dereference when the mapping is torn down", which sounds kinda important if true.

There is no known in-tree workload that triggers it, though. Mostly a defensive/simplifying change.

Quoted 3 lines

Help me out here?

Also, thanks ;)

Thanks!

reviewer Reviewed-by on v2 1/1
#7 lore raw
In reply to Lance Yang
Quoted 10 lines · Lance Yang

Whether a PMD has a deposited PTE page table only depends on whether the architecture requires deposits or the VMA is anonymous.

Implement this rule directly in vma_has_deposited_pgtable(), avoiding mistaking a non-anonymous raw PFN PMD for one with a deposit and attempting to withdraw a page table that was never deposited.

---

LGTM.

#6 lore raw
In reply to Lance Yang
Quoted 14 lines · Lance Yang
Quoted 11 lines · Andrew Morton
Quoted 6 lines · On Thu, 17 Sep 2026 13:40:15 +0800 Lance Yang <[email protected]> wrote:

Whether a PMD has a deposited PTE page table only depends on whether the architecture requires deposits or the VMA is anonymous.

Implement this rule directly in vma_has_deposited_pgtable(), avoiding mistaking a non-anonymous raw PFN PMD for one with a deposit and attempting to withdraw a page table that was never deposited.

A large part of my life is asking $LLM "what are the userspace-visible runtime effects of this bug". Because NOBODY EVER TELLS ME!!!

Quoted 8 lines

That's on me ... I trimmed too much from v1.

Quoted 3 lines

This time it says "Processes using PMD-sized raw PFN mappings can trigger a kernel NULL-pointer dereference when the mapping is torn down", which sounds kinda important if true.

There is no known in-tree workload that triggers it, though. Mostly a defensive/simplifying change.

Yeah, it's not a fix.

Signature · 3 lines
Cheers,

David