Re: [syzbot] kernel BUG in remove_inode_hugepages (2)

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

From: Edward Aadm Davis <eadavis@sina.com>

#syz test: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git 9ba46df34488

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,