[PATCH] hugetlb: supports unmapping private mappings

[Edward Adam Davis <eadavis@sina.com>]

If a file has a private mapping, then truncating the file's original logic
will only unmap it according to the shared mapping method, potentially
hitting:

refcount_t: addition on 0; use-after-free.
kernel BUG at lib/refcount.c:34!
RIP: 0010:refcount_warn_saturate+0xf0/0x110 lib/refcount.c:34
Call Trace:
 kref_get include/linux/kref.h:45 [inline]
 hugetlb_unmap_file_folio fs/hugetlbfs/inode.c:413 [inline]
 remove_inode_single_folio fs/hugetlbfs/inode.c:512 [inline]
 remove_inode_hugepages+0xbdb/0x10f0 fs/hugetlbfs/inode.c:577
 hugetlb_vmtruncate fs/hugetlbfs/inode.c:621 [inline]
 hugetlbfs_setattr+0x4f0/0x6d0 fs/hugetlbfs/inode.c:856
 notify_change+0xbb8/0xec0 fs/attr.c:556
 do_truncate+0x1c2/0x250 fs/open.c:68

Add logic to handle the unmapping of private mappings.

Fixes: 40549ba8f8e0 ("hugetlb: use new vma_lock for pmd sharing synchronization")
Reported-by: syzbot+2fcec0eca7f4cade53b2@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=2fcec0eca7f4cade53b2
Signed-off-by: Edward Adam Davis <eadavis@sina.com>
---
 fs/hugetlbfs/inode.c    | 58 +++++++++++++++++++++++++----------------
 include/linux/hugetlb.h |  6 +++++
 mm/hugetlb.c            |  3 +--
 3 files changed, 43 insertions(+), 24 deletions(-)

diff --git a/fs/hugetlbfs/inode.c b/fs/hugetlbfs/inode.c
index 6656807ce437..166183b7ee36 100644
--- a/fs/hugetlbfs/inode.c
+++ b/fs/hugetlbfs/inode.c
@@ -382,6 +382,7 @@ static void hugetlb_unmap_file_folio(struct hstate *h,
 					struct folio *folio, pgoff_t index)
 {
 	struct hugetlb_vma_lock *vma_lock;
+	struct resv_map *resv_map;
 	unsigned long pfn = folio_pfn(folio);
 	struct vm_area_struct *vma;
 	unsigned long v_start;
@@ -394,6 +395,7 @@ static void hugetlb_unmap_file_folio(struct hstate *h,
 	i_mmap_lock_write(mapping);
 retry:
 	vma_lock = NULL;
+	resv_map = NULL;
 	mapping_rmap_tree_foreach(vma, mapping, start, end - 1) {
 		v_start = vma_offset_start(vma, start);
 		v_end = vma_offset_end(vma, end);
@@ -402,15 +404,21 @@ static void hugetlb_unmap_file_folio(struct hstate *h,
 			continue;
 
 		if (!hugetlb_vma_trylock_write(vma)) {
-			vma_lock = vma->vm_private_data;
-			/*
-			 * If we can not get vma lock, we need to drop
-			 * immap_sema and take locks in order.  First,
-			 * take a ref on the vma_lock structure so that
-			 * we can be guaranteed it will not go away when
-			 * dropping immap_sema.
-			 */
-			kref_get(&vma_lock->refs);
+			if (vma->vm_flags & VM_MAYSHARE) {
+				vma_lock = vma->vm_private_data;
+				/*
+				 * If we can not get vma lock, we need to drop
+				 * immap_sema and take locks in order.  First,
+				 * take a ref on the vma_lock structure so that
+				 * we can be guaranteed it will not go away when
+				 * dropping immap_sema.
+				 */
+				kref_get(&vma_lock->refs);
+			} else {
+				resv_map = vma_resv_map(vma);
+				kref_get(&resv_map->refs);
+			}
+
 			break;
 		}
 
@@ -421,25 +429,28 @@ static void hugetlb_unmap_file_folio(struct hstate *h,
 
 	i_mmap_unlock_write(mapping);
 
-	if (vma_lock) {
+	if (vma_lock || resv_map) {
 		/*
 		 * Wait on vma_lock.  We know it is still valid as we have
 		 * a reference.  We must 'open code' vma locking as we do
 		 * not know if vma_lock is still attached to vma.
 		 */
-		down_write(&vma_lock->rw_sema);
+		hugetlb_vma_lock_write(vma);
 		i_mmap_lock_write(mapping);
 
-		vma = vma_lock->vma;
-		if (!vma) {
-			/*
-			 * If lock is no longer attached to vma, then just
-			 * unlock, drop our reference and retry looking for
-			 * other vmas.
-			 */
-			up_write(&vma_lock->rw_sema);
-			kref_put(&vma_lock->refs, hugetlb_vma_lock_release);
-			goto retry;
+		if (vma_lock) {
+			vma = vma_lock->vma;
+			if (!vma) {
+				/*
+				 * If lock is no longer attached to vma, then
+				 * just unlock, drop our reference and retry
+				 * looking for other vmas.
+				 */
+				up_write(&vma_lock->rw_sema);
+				kref_put(&vma_lock->refs,
+					 hugetlb_vma_lock_release);
+				goto retry;
+			}
 		}
 
 		/*
@@ -452,7 +463,10 @@ static void hugetlb_unmap_file_folio(struct hstate *h,
 			unmap_hugepage_range(vma, v_start, v_end, NULL,
 					     ZAP_FLAG_DROP_MARKER);
 
-		kref_put(&vma_lock->refs, hugetlb_vma_lock_release);
+		if (vma_lock)
+			kref_put(&vma_lock->refs, hugetlb_vma_lock_release);
+		else
+			kref_put(&resv_map->refs, resv_map_release);
 		hugetlb_vma_unlock_write(vma);
 
 		goto retry;
diff --git a/include/linux/hugetlb.h b/include/linux/hugetlb.h
index 80a5a03e9cee..f980191c651b 100644
--- a/include/linux/hugetlb.h
+++ b/include/linux/hugetlb.h
@@ -268,6 +268,7 @@ void hugetlb_vma_unlock_read(struct vm_area_struct *vma);
 void hugetlb_vma_lock_write(struct vm_area_struct *vma);
 void hugetlb_vma_unlock_write(struct vm_area_struct *vma);
 int hugetlb_vma_trylock_write(struct vm_area_struct *vma);
+struct resv_map *vma_resv_map(struct vm_area_struct *vma);
 void hugetlb_vma_assert_locked(struct vm_area_struct *vma);
 void hugetlb_vma_lock_release(struct kref *kref);
 long hugetlb_change_protection(struct vm_area_struct *vma,
@@ -392,6 +393,11 @@ static inline int is_hugepage_only_range(struct mm_struct *mm,
 	return 0;
 }
 
+static inline struct resv_map *vma_resv_map(struct vm_area_struct *vma)
+{
+	return NULL;
+}
+
 #ifdef CONFIG_USERFAULTFD
 static inline int hugetlb_mfill_atomic_pte(pte_t *dst_pte,
 					   struct vm_area_struct *dst_vma,
diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index b017ae1dfd6f..def6aebc44fd 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -134,7 +134,6 @@ static int __huge_pmd_unshare(struct mmu_gather *tlb,
 		bool check_locks);
 static void hugetlb_unshare_pmds(struct vm_area_struct *vma,
 		unsigned long start, unsigned long end, bool take_locks);
-static struct resv_map *vma_resv_map(struct vm_area_struct *vma);
 
 static inline bool subpool_is_free(struct hugepage_subpool *spool)
 {
@@ -1145,7 +1144,7 @@ static inline struct resv_map *inode_resv_map(struct inode *inode)
 	return HUGETLBFS_I(inode)->resv_map;
 }
 
-static struct resv_map *vma_resv_map(struct vm_area_struct *vma)
+struct resv_map *vma_resv_map(struct vm_area_struct *vma)
 {
 	VM_BUG_ON_VMA(!is_vm_hugetlb_page(vma), vma);
 	if (vma->vm_flags & VM_MAYSHARE) {
-- 
2.43.0