On Sat, Mar 07, 2020 at 09:49:25AM +0800, yangerkun wrote:
On 2020/3/6 21:39, Sasha Levin wrote:
On Thu, Mar 05, 2020 at 04:57:55PM +0800, yangerkun wrote:
Nowdays, we trigger a oops:
...
kasan: GPF could be caused by NULL-ptr deref or user memory
accessgeneral protection fault: 0000 [#1] SMP KASAN
...
Call Trace:
[<ffffffff81a26fb1>] skcipher_recvmsg_async+0x3f1/0x1400
x86/../crypto/algif_skcipher.c:543
[<ffffffff81a28053>] skcipher_recvmsg+0x93/0x7f0
x86/../crypto/algif_skcipher.c:723
[<ffffffff823e43a4>] sock_recvmsg_nosec x86/../net/socket.c:702 [inline]
[<ffffffff823e43a4>] sock_recvmsg x86/../net/socket.c:710 [inline]
[<ffffffff823e43a4>] sock_recvmsg+0x94/0xc0 x86/../net/socket.c:705
[<ffffffff823e464b>] sock_read_iter+0x27b/0x3a0 x86/../net/socket.c:787
[<ffffffff817f479b>] aio_run_iocb+0x21b/0x7a0 x86/../fs/aio.c:1520
[<ffffffff817f57c9>] io_submit_one x86/../fs/aio.c:1630 [inline]
[<ffffffff817f57c9>] do_io_submit+0x6b9/0x10b0 x86/../fs/aio.c:1688
[<ffffffff817f902d>] SYSC_io_submit x86/../fs/aio.c:1713 [inline]
[<ffffffff817f902d>] SyS_io_submit+0x2d/0x40 x86/../fs/aio.c:1710
[<ffffffff828b33c3>] tracesys_phase2+0x90/0x95
In skcipher_recvmsg_async, we use '!sreq->tsg' to determine does we
calloc fail. However, kcalloc may return ZERO_SIZE_PTR, and with this,
the latter sg_init_table will trigger the bug. Fix it be use
ZERO_OF_NULL_PTR.
This function was introduced with ' commit a596999b7ddf ("crypto:
algif - change algif_skcipher to be asynchronous")', and has been removed
with 'commit e870456d8e7c ("crypto: algif_skcipher - overhaul memory
management")'.
Reported-by: Hulk Robot <hulkci@xxxxxxxxxx>
Signed-off-by: yangerkun <yangerkun@xxxxxxxxxx>
---
crypto/algif_skcipher.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
v1->v2:
update the commit message
diff --git a/crypto/algif_skcipher.c b/crypto/algif_skcipher.c
index d12782dc9683..9bd4691cc5c5 100644
--- a/crypto/algif_skcipher.c
+++ b/crypto/algif_skcipher.c
@@ -538,7 +538,7 @@ static int skcipher_recvmsg_async(struct
socket *sock, struct msghdr *msg,
lock_sock(sk);
tx_nents = skcipher_all_sg_nents(ctx);
sreq->tsg = kcalloc(tx_nents, sizeof(*sg), GFP_KERNEL);
- if (unlikely(!sreq->tsg))
+ if (unlikely(ZERO_OR_NULL_PTR(sreq->tsg)))
I'm a bit confused: kcalloc() will return ZERO_SIZE_PTR for allocations
that ask for 0 bytes, but here we ask for "sizeof(*sg)" bytes, which is
guaranteed to be more than 0, no?
Actually, the size need to calloc is (tx_nents * sizeof(*sg)), and
tx_nents is 0.
Makes sense. This is also needed on 4.9, right?
--
Thanks,
Sasha
