xfs: handle NULL b_addr in xfs_buf_free

[ Upstream commit d852729c5f ]

When xfs_buf_alloc_backing_mem() fails, xfs_buf_free() is called with
bp->b_addr still NULL.  The code falls through to the folio_put path
which calls virt_to_folio(NULL), dereferencing an invalid address and
causing a kernel crash.

 Call Trace:
  xfs_buf_free+0x25f/0x510
  xfs_buf_alloc+0xc98/0x19b0
  xfs_buf_find_insert+0x55/0x14d0
  xfs_buf_get_map+0x122b/0x17c0
  xfbtree_init_leaf_block+0x11c/0x4a0
  xfbtree_init+0x1bb/0x460
  xrep_rmap_setup_scan+0x100/0x1f0
  xrep_rmapbt+0x41/0xc0

Fix this by skipping folio_put() when bp->b_addr is NULL.

Fixes: 5076a6040c ("xfs: support in-memory buffer cache targets")
Reported-by: syzbot+94c22d92f72f5a235b7d@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=94c22d92f72f5a235b7d
Signed-off-by: Yun Zhou <yun.zhou@windriver.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Carlos Maiolino <cem@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

fs/xfs/xfs_buf.c

@@ -117,7 +117,7 @@ xfs_buf_free(
 		vfree(bp->b_addr);
 	else if (bp->b_flags & _XBF_KMEM)
 		kfree(bp->b_addr);
-	else
+	else if (bp->b_addr)
 		folio_put(virt_to_folio(bp->b_addr));
 
 	call_rcu(&bp->b_rcu, xfs_buf_free_callback);