[PATCH v2] io_uring: fix dangling iovec after provided-buffer bundle grow failure

[Hao-Yu Yang <naup96721@gmail.com>]

When growing a provided-buffer bundle, the old cached iovec is freed
before the new buffers have all been validated. If validation fails, the
request still points at the freed iovec, which can be freed again during
completion cleanup.

BUG: KASAN: double-free in io_vec_free+0x2c/0x90
Freed by task 73:
 kfree+0x104/0x3b0
 io_vec_free+0x2c/0x90
 __io_submit_flush_completions+0xc03/0x1e40
 io_submit_sqes+0xdb5/0x2310

Allocated by task 73:
 io_ring_buffers_peek+0x559/0xc60
 io_buffers_select+0x1c1/0x460
 io_send+0x770/0x1050

Fix this by deferring the free of the old cached iovec until validation
has succeeded. On failure, free the newly allocated iovec and leave the
request pointing at the original one.

change log:
 v2: slimming v1 patch

Fixes: 46800585ae04 ("io_uring/kbuf: validate ring provided buffer addresses with access_ok()")
Signed-off-by: Hao-Yu Yang <naup96721@gmail.com>
---
 io_uring/kbuf.c | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/io_uring/kbuf.c b/io_uring/kbuf.c
index 3cd29477fff2..b6b969b55e12 100644
--- a/io_uring/kbuf.c
+++ b/io_uring/kbuf.c
@@ -287,8 +287,6 @@ static int io_ring_buffers_peek(struct io_kiocb *req, struct buf_sel_arg *arg,
 		iov = kmalloc_objs(struct iovec, nr_avail);
 		if (unlikely(!iov))
 			return -ENOMEM;
-		if (arg->mode & KBUF_MODE_FREE)
-			kfree(arg->iovs);
 		arg->iovs = iov;
 		nr_iovs = nr_avail;
 	} else if (nr_avail < nr_iovs) {
@@ -330,6 +328,9 @@ static int io_ring_buffers_peek(struct io_kiocb *req, struct buf_sel_arg *arg,
 		buf = io_ring_head_to_buf(br, ++head, bl->mask);
 	} while (--nr_iovs);
 
+	if (arg->mode & KBUF_MODE_FREE)
+		kfree(arg->iovs);
+
 	if (head == tail)
 		req->flags |= REQ_F_BL_EMPTY;
 
-- 
2.34.1