Skip to contentSkip to conversation

powerpc/64s: __find_linux_pte synchronization vs pmdp_invalidate

RFC PATCH In review Nicholas Piggin 2 participants 2 messages Jun 3, 2019 PowerPC

lore
Share & tools
lore

Nicholas Piggin

author reviewer
#0 lore raw

RFC PATCH 1/1 powerpc/64s: __find_linux_pte synchronization vs pmdp_invalidate

The pmd_none check does not catch hugepage collapse, nor does the pmd_present check in pmd_trans_huge, because hugepage collapse sets !_PAGE_PRESENT && _PAGE_INVALID (which results in !pmd_none and pmd_present).

Aneesh noticed we might need this check as well.

---
 arch/powerpc/mm/pgtable.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)
arch/powerpc/mm/pgtable.c +11 -4
index db4a6253df92..7a702d21400a 100644
@@ -372,13 +372,20 @@ pte_t *__find_linux_pte(pgd_t *pgdir, unsigned long ea,
 	pdshift = PMD_SHIFT;
 	pmdp = pmd_offset(&pud, ea);
 	pmd  = READ_ONCE(*pmdp);
-	/*
-	 * A hugepage collapse is captured by pmd_none, because
-	 * it mark the pmd none and do a hpte invalidate.
-	 */
+
 	if (pmd_none(pmd))
 		return NULL;
 
+#ifdef CONFIG_PPC_BOOK3S_64
+	if (pmd_val(pmd) & (_PAGE_PRESENT|_PAGE_INVALID) == _PAGE_INVALID) {
+		/*
+		 * A hugepage collapse is captured by this condition, see
+		 * pmdp_invalidate.
+		 */
+		return NULL;
+	}
+#endif
+
 	if (pmd_trans_huge(pmd) || pmd_devmap(pmd)) {
 		if (is_thp)
 			*is_thp = true;
Signature · 1 line
2.20.1
#1 lore raw

Le 03/06/2019 à 08:44, Nicholas Piggin a écrit :

Quoted 18 lines

The pmd_none check does not catch hugepage collapse, nor does the pmd_present check in pmd_trans_huge, because hugepage collapse sets !_PAGE_PRESENT && _PAGE_INVALID (which results in !pmd_none and pmd_present).

Aneesh noticed we might need this check as well.

---
  arch/powerpc/mm/pgtable.c | 15 +++++++++++----
  1 file changed, 11 insertions(+), 4 deletions(-)
arch/powerpc/mm/pgtable.c +0 -0
index db4a6253df92..7a702d21400a 100644
@@ -372,13 +372,20 @@ pte_t *__find_linux_pte(pgd_t *pgdir, unsigned long ea,
  	pdshift = PMD_SHIFT;
  	pmdp = pmd_offset(&pud, ea);

arch/powerpc/mm/pgtable.c pte_t *__find_linux_pte(pgd_t *pgdir, unsigned long ea, · line 379

Quoted 10 lines
arch/powerpc/mm/pgtable.c +2 -4
  	pmd  = READ_ONCE(*pmdp);
-	/*
-	 * A hugepage collapse is captured by pmd_none, because
-	 * it mark the pmd none and do a hpte invalidate.
-	 */
+
  	if (pmd_none(pmd))
  		return NULL;
 
+#ifdef CONFIG_PPC_BOOK3S_64

I can't see anything that would build fail on other subarches. Wouldn't be better to use

if (IS_ENABLED(CONFIG_PPC_BOOK3S_64) && (pmd_val(pmd) & (_PAGE_PRESENT|_PAGE_INVALID) == _PAGE_INVALID))

arch/powerpc/mm/pgtable.c pte_t *__find_linux_pte(pgd_t *pgdir, unsigned long ea, · line 380

Quoted 1 line
arch/powerpc/mm/pgtable.c +1 -0
+	if (pmd_val(pmd) & (_PAGE_PRESENT|_PAGE_INVALID) == _PAGE_INVALID) {

Maybe using pmd_raw() instead as in all the book3s64 helpers ?

Christophe

Quoted 12 lines
arch/powerpc/mm/pgtable.c +8 -0
+		/*
+		 * A hugepage collapse is captured by this condition, see
+		 * pmdp_invalidate.
+		 */
+		return NULL;
+	}
+#endif
+
  	if (pmd_trans_huge(pmd) || pmd_devmap(pmd)) {
  		if (is_thp)
  			*is_thp = true;