Skip to contentSkip to conversation

FAILED: patch "[PATCH] KVM: s390: Fix guest migration for huge guests resulting in" failed to apply to 4.4-stable tree

PATCH New [email protected] 2 participants 3 messages Mar 15, 2017 – Mar 16, 2017

Subsystems (3)
lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

The patch below does not apply to the 4.4-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

Quoted 1 line

From 2e4d88009f57057df7672fa69a32b5224af54d37 Mon Sep 17 00:00:00 2001

From: Janosch Frank <[email protected]>
Date: Thu, 2 Mar 2017 15:23:42 +0100
Subject: [PATCH] KVM: s390: Fix guest migration for huge guests resulting in
panic

While we can technically not run huge page guests right now, we can setup a guest with huge pages. Trying to migrate it will trigger a VM_BUG_ON and, if the kernel is not configured to panic on a BUG, it will happily try to work on non-existing page table entries.

With this patch, we always return "dirty" if we encounter a large page when migrating. This at least fixes the immediate problem until we have proper handling for both kind of pages.

Fixes 15f36eb ("KVM: s390: Add proper dirty bitmap support to S390 kvm.")
Cc <[email protected]> # 3.16+
arch/s390/mm/pgtable.c +18 -1
index b48dc5f1900b..463e5ef02304 100644
@@ -608,12 +608,29 @@ void ptep_zap_key(struct mm_struct *mm, unsigned long addr, pte_t *ptep)
 bool test_and_clear_guest_dirty(struct mm_struct *mm, unsigned long addr)
 {
 	spinlock_t *ptl;
+	pgd_t *pgd;
+	pud_t *pud;
+	pmd_t *pmd;
 	pgste_t pgste;
 	pte_t *ptep;
 	pte_t pte;
 	bool dirty;
 
-	ptep = get_locked_pte(mm, addr, &ptl);
+	pgd = pgd_offset(mm, addr);
+	pud = pud_alloc(mm, pgd, addr);
+	if (!pud)
+		return false;
+	pmd = pmd_alloc(mm, pud, addr);
+	if (!pmd)
+		return false;
+	/* We can't run guests backed by huge pages, but userspace can
+	 * still set them up and then try to migrate them without any
+	 * migration support.
+	 */
+	if (pmd_large(*pmd))
+		return true;
+
+	ptep = pte_alloc_map_lock(mm, pmd, addr, &ptl);
 	if (unlikely(!ptep))
 		return false;
 
#1 lore raw

PATCH 1/1 Backport KVM: s390: Fix guest migration for huge guests resulting in panic

Dear Greg,

Here's the backport for the KVM migration fix that failed to apply on 4.4. It is applicable to 3.16 - 4.4.

Cheers,
Janosch

Janosch Frank (1):
KVM: s390: Fix guest migration for huge guests resulting in panic

 arch/s390/mm/pgtable.c | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)
Signature · 1 line
2.7.4
#2 lore raw

PATCH 1/1 KVM: s390: Fix guest migration for huge guests resulting in panic

While we can technically not run huge page guests right now, we can setup a guest with huge pages. Trying to migrate it will trigger a VM_BUG_ON and, if the kernel is not configured to panic on a BUG, it will happily try to work on non-existing page table entries.

With this patch, we always return "dirty" if we encounter a large page when migrating. This at least fixes the immediate problem until we have proper handling for both kind of pages.

Fixes 15f36eb ("KVM: s390: Add proper dirty bitmap support to S390 kvm.")

Upstream-ID: 2e4d88009f57057df7672fa69a32b5224af54d37

---
 arch/s390/mm/pgtable.c | 19 ++++++++++++++++++-
 1 file changed, 18 insertions(+), 1 deletion(-)
arch/s390/mm/pgtable.c +18 -1
index 80fcdfb..6802530 100644
@@ -1411,11 +1411,28 @@ EXPORT_SYMBOL_GPL(s390_enable_skey);
  */
 bool gmap_test_and_clear_dirty(unsigned long address, struct gmap *gmap)
 {
+	pgd_t *pgd;
+	pud_t *pud;
+	pmd_t *pmd;
 	pte_t *pte;
 	spinlock_t *ptl;
 	bool dirty = false;
 
-	pte = get_locked_pte(gmap->mm, address, &ptl);
+	pgd = pgd_offset(gmap->mm, address);
+	pud = pud_alloc(gmap->mm, pgd, address);
+	if (!pud)
+		return false;
+	pmd = pmd_alloc(gmap->mm, pud, address);
+	if (!pmd)
+		return false;
+	/* We can't run guests backed by huge pages, but userspace can
+	 * still set them up and then try to migrate them without any
+	 * migration support.
+	 */
+	if (pmd_large(*pmd))
+		return true;
+
+	pte = pte_alloc_map_lock(gmap->mm, pmd, address, &ptl);
 	if (unlikely(!pte))
 		return false;
 
Signature · 1 line
2.7.4