Re: [PATCH v2 1/5] KVM: do not allow mapping valid but non-refcounted pages

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 25/06/21 09:58, Christian Borntraeger wrote:


On 25.06.21 09:36, David Stevens wrote:
From: Nicholas Piggin <npiggin@xxxxxxxxx>

It's possible to create a region which maps valid but non-refcounted
pages (e.g., tail pages of non-compound higher order allocations). These
host pages can then be returned by gfn_to_page, gfn_to_pfn, etc., family
of APIs, which take a reference to the page, which takes it from 0 to 1.
When the reference is dropped, this will free the page incorrectly.

Fix this by only taking a reference on the page if it was non-zero,
which indicates it is participating in normal refcounting (and can be
released with put_page).

Signed-off-by: Nicholas Piggin <npiggin@xxxxxxxxx>

I guess this would be the small fix for stable? Do we want to add that cc?

Reviewed-by: Christian Borntraeger <borntraeger@xxxxxxxxxx>

Yes, this one is going to Linus today.  The rest is for 5.15.

Paolo

---
  virt/kvm/kvm_main.c | 19 +++++++++++++++++--
  1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 3dcc2abbfc60..f7445c3bcd90 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -2175,6 +2175,13 @@ static bool vma_is_valid(struct vm_area_struct *vma, bool write_fault)
      return true;
  }

+static int kvm_try_get_pfn(kvm_pfn_t pfn)
+{
+    if (kvm_is_reserved_pfn(pfn))
+        return 1;
+    return get_page_unless_zero(pfn_to_page(pfn));
+}
+
  static int hva_to_pfn_remapped(struct vm_area_struct *vma,
                     unsigned long addr, bool *async,
                     bool write_fault, bool *writable,
@@ -2224,13 +2231,21 @@ static int hva_to_pfn_remapped(struct vm_area_struct *vma,
       * Whoever called remap_pfn_range is also going to call e.g.
       * unmap_mapping_range before the underlying pages are freed,
       * causing a call to our MMU notifier.
+     *
+     * Certain IO or PFNMAP mappings can be backed with valid
+     * struct pages, but be allocated without refcounting e.g.,
+     * tail pages of non-compound higher order allocations, which
+     * would then underflow the refcount when the caller does the
+     * required put_page. Don't allow those pages here.
       */
-    kvm_get_pfn(pfn);
+    if (!kvm_try_get_pfn(pfn))
+        r = -EFAULT;

  out:
      pte_unmap_unlock(ptep, ptl);
      *p_pfn = pfn;
-    return 0;
+
+    return r;
  }

  /*






[Index of Archives]     [LKML Archive]     [Linux ARM Kernel]     [Linux ARM]     [Git]     [Yosemite News]     [Linux SCSI]     [Linux Hams]

  Powered by Linux