On Tue, Aug 9, 2016 at 12:34 AM, Jason Wang <jasow...@redhat.com> wrote:
> From: chaojianhu <chaojia...@hotmail.com>
>
> The .receive callback of xlnx.xps-ethernetlite doesn't check the length
> of data before calling memcpy. As a result, the NetClientState object in
> heap will be overflowed. All versions of qemu with xlnx.xps-ethernetlite
> will be affected.
>
> Reported-by: chaojianhu <chaojia...@hotmail.com>
> Signed-off-by: chaojianhu <chaojia...@hotmail.com>
> Signed-off-by: Jason Wang <jasow...@redhat.com>
> ---
> hw/net/xilinx_ethlite.c | 4 ++++
> 1 file changed, 4 insertions(+)
>
> diff --git a/hw/net/xilinx_ethlite.c b/hw/net/xilinx_ethlite.c
> index 54db2b8..35de353 100644
> --- a/hw/net/xilinx_ethlite.c
> +++ b/hw/net/xilinx_ethlite.c
> @@ -197,6 +197,10 @@ static ssize_t eth_rx(NetClientState *nc, const uint8_t
> *buf, size_t size)
> }
>
> D(qemu_log("%s %zd rxbase=%x\n", __func__, size, rxbase));
This might be too late. A new line would be great here, but no big deal.
Reviewed-by: Alistair Francis <alistair.fran...@xilinx.com>
Thanks,
Alistair
> + if (size > (R_MAX - R_RX_BUF0 - rxbase) * 4) {
> + D(qemu_log("ethlite packet is too big, size=%x\n", size));
> + return -1;
> + }
> memcpy(&s->regs[rxbase + R_RX_BUF0], buf, size);
>
> s->regs[rxbase + R_RX_CTRL0] |= CTRL_S;
> --
> 2.7.4
>
>
