On 2/25/20 12:39 AM, Marc Kleine-Budde wrote:
> Syzbot found the use of uninitialzied memory when injecting non conformant
> CANFD frames via a tun device into the kernel:
>
> | BUG: KMSAN: uninit-value in number+0x9f8/0x2000 lib/vsprintf.c:459
> | CPU: 1 PID: 11897 Comm: syz-executor136 Not tainted 5.6.0-rc2-syzkaller #0
> | Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> | Call Trace:
> | __dump_stack lib/dump_stack.c:77 [inline]
> | dump_stack+0x1c9/0x220 lib/dump_stack.c:118
> | kmsan_report+0xf7/0x1e0 mm/kmsan/kmsan_report.c:118
> | __msan_warning+0x58/0xa0 mm/kmsan/kmsan_instr.c:215
> | number+0x9f8/0x2000 lib/vsprintf.c:459
> | vsnprintf+0x1d85/0x31b0 lib/vsprintf.c:2640
> | vscnprintf+0xc2/0x180 lib/vsprintf.c:2677
> | vprintk_store+0xef/0x11d0 kernel/printk/printk.c:1917
> | vprintk_emit+0x2c0/0x860 kernel/printk/printk.c:1984
> | vprintk_default+0x90/0xa0 kernel/printk/printk.c:2029
> | vprintk_func+0x636/0x820 kernel/printk/printk_safe.c:386
> | printk+0x18b/0x1d3 kernel/printk/printk.c:2062
> | canfd_rcv+0x370/0x3a0 net/can/af_can.c:697
> | __netif_receive_skb_one_core net/core/dev.c:5198 [inline]
> | __netif_receive_skb net/core/dev.c:5312 [inline]
> | netif_receive_skb_internal net/core/dev.c:5402 [inline]
> | netif_receive_skb+0xe77/0xf20 net/core/dev.c:5461
> | tun_rx_batched include/linux/skbuff.h:4321 [inline]
> | tun_get_user+0x6aef/0x6f60 drivers/net/tun.c:1997
> | tun_chr_write_iter+0x1f2/0x360 drivers/net/tun.c:2026
> | call_write_iter include/linux/fs.h:1901 [inline]
> | new_sync_write fs/read_write.c:483 [inline]
> | __vfs_write+0xa5a/0xca0 fs/read_write.c:496
> | vfs_write+0x44a/0x8f0 fs/read_write.c:558
> | ksys_write+0x267/0x450 fs/read_write.c:611
> | __do_sys_write fs/read_write.c:623 [inline]
> | __se_sys_write+0x92/0xb0 fs/read_write.c:620
> | __x64_sys_write+0x4a/0x70 fs/read_write.c:620
> | do_syscall_64+0xb8/0x160 arch/x86/entry/common.c:296
> | entry_SYSCALL_64_after_hwframe+0x44/0xa9
>
> In canfd_rcv() the non conformant CANFD frame (i.e. skb too short) is detected,
> but the pr_warn_once() accesses uninitialized memory or the skb->data out of
> bounds to print the warning message.
>
> This problem exists in both can_rcv() and canfd_rcv(). This patch removes the
> access to the skb->data from the pr_warn_once() in both functions.
>
> Reported-by: syzbot+9bcb0c9409066696d3aa@xxxxxxxxxxxxxxxxxxxxxxxxx
> Signed-off-by: Marc Kleine-Budde <mkl@xxxxxxxxxxxxxx>
> ---
> Hello,
>
> changes since RFC:
> - print cfd->len if backed by skb, -1 otherwise
> (Requested by Oliver)
>
> Marc
>
> net/can/af_can.c | 8 ++++++--
> 1 file changed, 6 insertions(+), 2 deletions(-)
>
> diff --git a/net/can/af_can.c b/net/can/af_can.c
> index 7511bfae2e0d..38b7aaa6b286 100644
> --- a/net/can/af_can.c
> +++ b/net/can/af_can.c
> @@ -678,7 +678,9 @@ static int can_rcv(struct sk_buff *skb, struct net_device *dev,
> if (unlikely(dev->type != ARPHRD_CAN || skb->len != CAN_MTU ||
> cfd->len > CAN_MAX_DLEN)) {
> pr_warn_once("PF_CAN: dropped non conform CAN skbuf: dev type %d, len %d, datalen %d\n",
> - dev->type, skb->len, cfd->len);
> + dev->type, skb->len,
> + skb->len >= offsetof(struct canfd_frame, len) +
> + sizeof(cfd->len) ? cfd->len : -1);
skb->len being big enough is not sufficient.
You also need to make sure the header is present in skb->head
Check for calls to pskb_may_pull() all over the stack :)
Alternative is to force skb_linearize(), since it seems net/can assumes
skbs are linear ?
> kfree_skb(skb);
> return NET_RX_DROP;
> }
> @@ -695,7 +697,9 @@ static int canfd_rcv(struct sk_buff *skb, struct net_device *dev,
> if (unlikely(dev->type != ARPHRD_CAN || skb->len != CANFD_MTU ||
> cfd->len > CANFD_MAX_DLEN)) {
> pr_warn_once("PF_CAN: dropped non conform CAN FD skbuf: dev type %d, len %d, datalen %d\n",
> - dev->type, skb->len, cfd->len);
> + dev->type, skb->len,
> + skb->len >= offsetof(struct canfd_frame, len) +
> + sizeof(cfd->len) ? cfd->len : -1);
> kfree_skb(skb);
> return NET_RX_DROP;
> }
>
