diff options
| author | Michael S. Tsirkin <mst@redhat.com> | 2026-06-03 11:57:33 -0400 |
|---|---|---|
| committer | Sean Christopherson <seanjc@google.com> | 2026-06-08 09:15:43 -0700 |
| commit | 48dbe473219832788c3940c84faa7f1df4375d5d (patch) | |
| tree | 19c773cda3d8e5c4d86da0551f255471ad84f3af /virt | |
| parent | b408b52e7111c560dcd7d69612c2e07c59c36ae3 (diff) | |
| download | linux-48dbe473219832788c3940c84faa7f1df4375d5d.tar.gz linux-48dbe473219832788c3940c84faa7f1df4375d5d.zip | |
KVM: guest_memfd: fix NUMA interleave index double-counting
kvm_gmem_get_policy() sets the interleave index (the output param that's
typically named "ilx") to the full page offset (vm_pgoff + vma offset).
But get_vma_policy() adds the page offset on top of the interleave index,
and so the offset is counted twice. This causes NUMA interleaving to skip
nodes: for order-0 pages the effective index jumps by 2 for each
consecutive page.
The vm_op.get_policy() implementation should return only a per-file bias in
the interleave index (like shmem_get_policy does with inode->i_ino),
letting get_vma_policy() add the page-offset component.
Fix by setting the output interleave index to the inode number (a la shmem)
instead of the full page offset, as the index is intended to be a constant,
semi-random value for a given file, e.g. so that interleaving doesn't start
at the same node for every file, and so that allocations are round-robined
across nodes based on the page offset (the selected node would bounce/skip
around if the index isn't constant).
Found by Sashiko (sashiko.dev) AI code review.
Fixes: ed1ffa810bd6 ("KVM: guest_memfd: Enforce NUMA mempolicy using shared policy")
Cc: Sean Christopherson <seanjc@google.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Assisted-by: Claude:claude-opus-4-6
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: David Hildenbrand (Arm) <david@kernel.org>
Reviewed-by: Shivank Garg <shivankg@amd.com>
Tested-by: Shivank Garg <shivankg@amd.com>
Fixes: 7f3779a3ac3e ("mm/filemap: Add NUMA mempolicy support to filemap_alloc_folio()")
Link: https://patch.msgid.link/0eff0a90667b900bee837d06b5db5025e1f304b5.1780501924.git.mst@redhat.com
[sean: use reverse fir-tree, massage changelog]
Signed-off-by: Sean Christopherson <seanjc@google.com>
Diffstat (limited to 'virt')
| -rw-r--r-- | virt/kvm/guest_memfd.c | 7 |
1 files changed, 4 insertions, 3 deletions
diff --git a/virt/kvm/guest_memfd.c b/virt/kvm/guest_memfd.c index f54502640b08..557aac91fc0a 100644 --- a/virt/kvm/guest_memfd.c +++ b/virt/kvm/guest_memfd.c @@ -438,11 +438,12 @@ static int kvm_gmem_set_policy(struct vm_area_struct *vma, struct mempolicy *mpo } static struct mempolicy *kvm_gmem_get_policy(struct vm_area_struct *vma, - unsigned long addr, pgoff_t *pgoff) + unsigned long addr, pgoff_t *ilx) { + pgoff_t pgoff = vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT); struct inode *inode = file_inode(vma->vm_file); - *pgoff = vma->vm_pgoff + ((addr - vma->vm_start) >> PAGE_SHIFT); + *ilx = inode->i_ino; /* * Return the memory policy for this index, or NULL if none is set. @@ -453,7 +454,7 @@ static struct mempolicy *kvm_gmem_get_policy(struct vm_area_struct *vma, * can then replace NULL with the default memory policy instead of the * current task's memory policy. */ - return mpol_shared_policy_lookup(&GMEM_I(inode)->policy, *pgoff); + return mpol_shared_policy_lookup(&GMEM_I(inode)->policy, pgoff); } #endif /* CONFIG_NUMA */ |