The vma_lock and hugetlb_fault_mutex are dropped before handling
userfault and reacquire them again after handle_userfault(), but
reacquire the vma_lock could lead to UAF[1,2] due to the following
race,
hugetlb_fault
hugetlb_no_page
/*unlock vma_lock */
hugetlb_handle_userfault
handle_userfault
/* unlock mm->mmap_lock*/
vm_mmap_pgoff
do_mmap
mmap_region
munmap_vma_range
/* clean old vma */
/* lock vma_lock again <--- UAF */
/* unlock vma_lock */
Since the vma_lock will unlock immediately after hugetlb_handle_userfault(),
let's drop the unneeded lock and unlock in hugetlb_handle_userfault() to fix
the issue.
[1] https://lore.kernel.org/linux-mm/[email protected]/
[2] https://lore.kernel.org/linux-mm/[email protected]/
Reported-by: [email protected]
Reported-by: Liu Zixian <[email protected]>
Fixes: 1a1aad8a9b7b ("userfaultfd: hugetlbfs: add userfaultfd hugetlb hook")
CC: [email protected] # 4.14+
Signed-off-by: Liu Shixin <[email protected]>
Signed-off-by: Kefeng Wang <[email protected]>
Reviewed-by: Mike Kravetz <[email protected]>
---
v1->v2: add reported-by and link [1].
v2->v3: add comment to explain why unlock in hugetlb_no_page.
v3->v4: Rebase on mainline.
mm/hugetlb.c | 37 +++++++++++++++++--------------------
1 file changed, 17 insertions(+), 20 deletions(-)
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index e070b8593b37..477e3c2369d2 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -5476,7 +5476,6 @@ static inline vm_fault_t hugetlb_handle_userfault(struct vm_area_struct *vma,
unsigned long addr,
unsigned long reason)
{
- vm_fault_t ret;
u32 hash;
struct vm_fault vmf = {
.vma = vma,
@@ -5494,18 +5493,14 @@ static inline vm_fault_t hugetlb_handle_userfault(struct vm_area_struct *vma,
};
/*
- * hugetlb_fault_mutex and i_mmap_rwsem must be
- * dropped before handling userfault. Reacquire
- * after handling fault to make calling code simpler.
+ * vma_lock and hugetlb_fault_mutex must be dropped before handling
+ * userfault. Also mmap_lock will be dropped during handling
+ * userfault, any vma operation should be careful from here.
*/
hash = hugetlb_fault_mutex_hash(mapping, idx);
mutex_unlock(&hugetlb_fault_mutex_table[hash]);
i_mmap_unlock_read(mapping);
- ret = handle_userfault(&vmf, reason);
- i_mmap_lock_read(mapping);
- mutex_lock(&hugetlb_fault_mutex_table[hash]);
-
- return ret;
+ return handle_userfault(&vmf, reason);
}
static vm_fault_t hugetlb_no_page(struct mm_struct *mm,
@@ -5523,6 +5518,7 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm,
spinlock_t *ptl;
unsigned long haddr = address & huge_page_mask(h);
bool new_page, new_pagecache_page = false;
+ u32 hash = hugetlb_fault_mutex_hash(mapping, idx);
/*
* Currently, we are forced to kill the process in the event the
@@ -5533,7 +5529,7 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm,
if (is_vma_resv_set(vma, HPAGE_RESV_UNMAPPED)) {
pr_warn_ratelimited("PID %d killed due to inadequate hugepage pool\n",
current->pid);
- return ret;
+ goto out;
}
/*
@@ -5550,12 +5546,10 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm,
page = find_lock_page(mapping, idx);
if (!page) {
/* Check for page in userfault range */
- if (userfaultfd_missing(vma)) {
- ret = hugetlb_handle_userfault(vma, mapping, idx,
+ if (userfaultfd_missing(vma))
+ return hugetlb_handle_userfault(vma, mapping, idx,
flags, haddr, address,
VM_UFFD_MISSING);
- goto out;
- }
page = alloc_huge_page(vma, haddr, 0);
if (IS_ERR(page)) {
@@ -5615,10 +5609,9 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm,
if (userfaultfd_minor(vma)) {
unlock_page(page);
put_page(page);
- ret = hugetlb_handle_userfault(vma, mapping, idx,
+ return hugetlb_handle_userfault(vma, mapping, idx,
flags, haddr, address,
VM_UFFD_MINOR);
- goto out;
}
}
@@ -5676,6 +5669,8 @@ static vm_fault_t hugetlb_no_page(struct mm_struct *mm,
unlock_page(page);
out:
+ mutex_unlock(&hugetlb_fault_mutex_table[hash]);
+ i_mmap_unlock_read(mapping);
return ret;
backout:
@@ -5774,11 +5769,13 @@ vm_fault_t hugetlb_fault(struct mm_struct *mm, struct vm_area_struct *vma,
entry = huge_ptep_get(ptep);
/* PTE markers should be handled the same way as none pte */
- if (huge_pte_none_mostly(entry)) {
- ret = hugetlb_no_page(mm, vma, mapping, idx, address, ptep,
+ if (huge_pte_none_mostly(entry))
+ /*
+ * hugetlb_no_page will drop vma lock and hugetlb fault
+ * mutex internally, which make us return immediately.
+ */
+ return hugetlb_no_page(mm, vma, mapping, idx, address, ptep,
entry, flags);
- goto out_mutex;
- }
ret = 0;
--
2.25.1
On Sat, 24 Sep 2022 11:49:05 +0800 Liu Shixin <[email protected]> wrote:
> The vma_lock and hugetlb_fault_mutex are dropped before handling
> userfault and reacquire them again after handle_userfault(), but
> reacquire the vma_lock could lead to UAF[1,2] due to the following
> race,
>
> hugetlb_fault
> hugetlb_no_page
> /*unlock vma_lock */
> hugetlb_handle_userfault
> handle_userfault
> /* unlock mm->mmap_lock*/
> vm_mmap_pgoff
> do_mmap
> mmap_region
> munmap_vma_range
> /* clean old vma */
> /* lock vma_lock again <--- UAF */
> /* unlock vma_lock */
>
> Since the vma_lock will unlock immediately after hugetlb_handle_userfault(),
> let's drop the unneeded lock and unlock in hugetlb_handle_userfault() to fix
> the issue.
>
Thanks. Turns out that porting all the pending material on top of this
change was not a confidence-inspiring activity. So I ended up merging
your v3. Please work with Greg on the backporting when he gets on to
it? Hopefully that will merely involve sending him this v4.
On 2022/9/26 2:28, Andrew Morton wrote:
> On Sat, 24 Sep 2022 11:49:05 +0800 Liu Shixin <[email protected]> wrote:
>
>> The vma_lock and hugetlb_fault_mutex are dropped before handling
>> userfault and reacquire them again after handle_userfault(), but
>> reacquire the vma_lock could lead to UAF[1,2] due to the following
>> race,
>>
>> hugetlb_fault
>> hugetlb_no_page
>> /*unlock vma_lock */
>> hugetlb_handle_userfault
>> handle_userfault
>> /* unlock mm->mmap_lock*/
>> vm_mmap_pgoff
>> do_mmap
>> mmap_region
>> munmap_vma_range
>> /* clean old vma */
>> /* lock vma_lock again <--- UAF */
>> /* unlock vma_lock */
>>
>> Since the vma_lock will unlock immediately after hugetlb_handle_userfault(),
>> let's drop the unneeded lock and unlock in hugetlb_handle_userfault() to fix
>> the issue.
>>
> Thanks. Turns out that porting all the pending material on top of this
> change was not a confidence-inspiring activity. So I ended up merging
> your v3. Please work with Greg on the backporting when he gets on to
> it? Hopefully that will merely involve sending him this v4.
OK, I'll do it.
> .
>