The only safe way for non core/arch code to use follow_pfn() is
together with an mmu_notifier subscription. follow_pfn() is already
marked as _GPL and the kerneldoc explains this restriction.
This patch here enforces all this by adding a mmu_notifier argument
and verifying that it is registered for the correct mm_struct.
Motivated by discussions with Christoph Hellwig and Jason Gunthorpe.
Since requiring an mmu_notifier makes it very clear that follow_pfn()
cannot be used on !CONFIG_MMU hardware, remove it from there. The sole
user kvm not existing on such hardware also supports that.
Signed-off-by: Daniel Vetter <daniel.vetter@xxxxxxxxx>
Cc: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Cc: Jason Gunthorpe <jgg@xxxxxxxx>
Cc: Kees Cook <keescook@xxxxxxxxxxxx>
Cc: Dan Williams <dan.j.williams@xxxxxxxxx>
Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
Cc: John Hubbard <jhubbard@xxxxxxxxxx>
Cc: Jérôme Glisse <jglisse@xxxxxxxxxx>
Cc: Jan Kara <jack@xxxxxxx>
Cc: Dan Williams <dan.j.williams@xxxxxxxxx>
Cc: linux-mm@xxxxxxxxx
Cc: linux-arm-kernel@xxxxxxxxxxxxxxxxxxx
Cc: linux-samsung-soc@xxxxxxxxxxxxxxx
Cc: linux-media@xxxxxxxxxxxxxxx
Cc: kvm@xxxxxxxxxxxxxxx
Signed-off-by: Daniel Vetter <daniel.vetter@xxxxxxxx>
--
v7: Comments from Jason:
- ditch follow_pfn from nommu.c
- simplify mmu_notifer->mm check
---
include/linux/mm.h | 3 ++-
mm/memory.c | 38 ++++++++++++++++++++++++--------------
mm/nommu.c | 27 +++++----------------------
virt/kvm/kvm_main.c | 4 ++--
4 files changed, 33 insertions(+), 39 deletions(-)
diff --git a/include/linux/mm.h b/include/linux/mm.h
index bb3e926afd91..2a564bfd818c 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -1651,6 +1651,7 @@ void unmap_vmas(struct mmu_gather *tlb, struct vm_area_struct *start_vma,
unsigned long start, unsigned long end);
struct mmu_notifier_range;
+struct mmu_notifier;
void free_pgd_range(struct mmu_gather *tlb, unsigned long addr,
unsigned long end, unsigned long floor, unsigned long ceiling);
@@ -1660,7 +1661,7 @@ int follow_pte_pmd(struct mm_struct *mm, unsigned long address,
struct mmu_notifier_range *range,
pte_t **ptepp, pmd_t **pmdpp, spinlock_t **ptlp);
int follow_pfn(struct vm_area_struct *vma, unsigned long address,
- unsigned long *pfn);
+ unsigned long *pfn, struct mmu_notifier *subscription);
int unsafe_follow_pfn(struct vm_area_struct *vma, unsigned long address,
unsigned long *pfn);
int follow_phys(struct vm_area_struct *vma, unsigned long address,
diff --git a/mm/memory.c b/mm/memory.c
index 0db0c5e233fd..a27b9b9c22c2 100644
--- a/mm/memory.c
+++ b/mm/memory.c
@@ -4789,11 +4789,30 @@ int follow_pte_pmd(struct mm_struct *mm, unsigned long address,
}
EXPORT_SYMBOL(follow_pte_pmd);
+static int __follow_pfn(struct vm_area_struct *vma, unsigned long address,
+ unsigned long *pfn)
+{
+ int ret = -EINVAL;
+ spinlock_t *ptl;
+ pte_t *ptep;
+
+ if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
+ return ret;
+
+ ret = follow_pte(vma->vm_mm, address, &ptep, &ptl);
+ if (ret)
+ return ret;
+ *pfn = pte_pfn(*ptep);
+ pte_unmap_unlock(ptep, ptl);
+ return 0;
+}
+
/**
* follow_pfn - look up PFN at a user virtual address
* @vma: memory mapping
* @address: user virtual address
* @pfn: location to store found PFN
+ * @subscription: mmu_notifier subscription for the mm @vma is part of
*
* Only IO mappings and raw PFN mappings are allowed. Note that callers must
* ensure coherency with pte updates by using a &mmu_notifier to follow updates.
@@ -4805,21 +4824,12 @@ EXPORT_SYMBOL(follow_pte_pmd);
* Return: zero and the pfn at @pfn on success, -ve otherwise.
*/
int follow_pfn(struct vm_area_struct *vma, unsigned long address,
- unsigned long *pfn)
+ unsigned long *pfn, struct mmu_notifier *subscription)
{
- int ret = -EINVAL;
- spinlock_t *ptl;
- pte_t *ptep;
-
- if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
- return ret;
+ if (WARN_ON(subscription->mm != vma->vm_mm))
+ return -EINVAL;
- ret = follow_pte(vma->vm_mm, address, &ptep, &ptl);
- if (ret)
- return ret;
- *pfn = pte_pfn(*ptep);
- pte_unmap_unlock(ptep, ptl);
- return 0;
+ return __follow_pfn(vma, address, pfn);
}
EXPORT_SYMBOL_GPL(follow_pfn);
@@ -4844,7 +4854,7 @@ int unsafe_follow_pfn(struct vm_area_struct *vma, unsigned long address,
WARN_ONCE(1, "unsafe follow_pfn usage\n");
add_taint(TAINT_USER, LOCKDEP_STILL_OK);
- return follow_pfn(vma, address, pfn);
+ return __follow_pfn(vma, address, pfn);
}
EXPORT_SYMBOL(unsafe_follow_pfn);
diff --git a/mm/nommu.c b/mm/nommu.c
index 79fc98a6c94a..a1e178401146 100644
--- a/mm/nommu.c
+++ b/mm/nommu.c
@@ -111,27 +111,6 @@ unsigned int kobjsize(const void *objp)
return page_size(page);
}
-/**
- * follow_pfn - look up PFN at a user virtual address
- * @vma: memory mapping
- * @address: user virtual address
- * @pfn: location to store found PFN
- *
- * Only IO mappings and raw PFN mappings are allowed.
- *
- * Returns zero and the pfn at @pfn on success, -ve otherwise.
- */
-int follow_pfn(struct vm_area_struct *vma, unsigned long address,
- unsigned long *pfn)
-{
- if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
- return -EINVAL;
-
- *pfn = address >> PAGE_SHIFT;
- return 0;
-}
-EXPORT_SYMBOL_GPL(follow_pfn);
-
/**
* unsafe_follow_pfn - look up PFN at a user virtual address
* @vma: memory mapping
@@ -153,7 +132,11 @@ int unsafe_follow_pfn(struct vm_area_struct *vma, unsigned long address,
WARN_ONCE(1, "unsafe follow_pfn usage\n");
add_taint(TAINT_USER, LOCKDEP_STILL_OK);
- return follow_pfn(vma, address, pfn);
+ if (!(vma->vm_flags & (VM_IO | VM_PFNMAP)))
+ return -EINVAL;
+
+ *pfn = address >> PAGE_SHIFT;
+ return 0;
}
EXPORT_SYMBOL(unsafe_follow_pfn);
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index 417f3d470c3e..6f6786524eff 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -1891,7 +1891,7 @@ static int hva_to_pfn_remapped(struct kvm *kvm, struct vm_area_struct *vma,
unsigned long pfn;
int r;
- r = follow_pfn(vma, addr, &pfn);
+ r = follow_pfn(vma, addr, &pfn, &kvm->mmu_notifier);
if (r) {
/*
* get_user_pages fails for VM_IO and VM_PFNMAP vmas and does
@@ -1906,7 +1906,7 @@ static int hva_to_pfn_remapped(struct kvm *kvm, struct vm_area_struct *vma,
if (r)
return r;
- r = follow_pfn(vma, addr, &pfn);
+ r = follow_pfn(vma, addr, &pfn, &kvm->mmu_notifier);
if (r)
return r;
--
2.29.2
