qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Qemu-devel] Re: [RFC][RESEND][PATCH v1 09/15] virtproxy: add handler fo


From: Adam Litke
Subject: [Qemu-devel] Re: [RFC][RESEND][PATCH v1 09/15] virtproxy: add handler for data packets
Date: Wed, 03 Nov 2010 19:46:18 -0500

On Wed, 2010-11-03 at 10:28 -0500, Michael Roth wrote:
> Process VPPackets coming in from channel and send them to the
> appropriate server/client connections.
> 
> Signed-off-by: Michael Roth <address@hidden>
> ---
>  virtproxy.c |   42 ++++++++++++++++++++++++++++++++++++++++++
>  1 files changed, 42 insertions(+), 0 deletions(-)
> 
> diff --git a/virtproxy.c b/virtproxy.c
> index 6c3611b..57ab2b0 100644
> --- a/virtproxy.c
> +++ b/virtproxy.c
> @@ -235,6 +235,48 @@ static void vp_channel_accept(void *opaque)
>      vp_set_fd_handler(drv->listen_fd, NULL, NULL, NULL);
>  }
> 
> +/* handle data packets
> + *
> + * process VPPackets containing data and send them to the corresponding
> + * FDs
> + */
> +static int vp_handle_data_packet(void *drv, const VPPacket *pkt)
> +{
> +    int fd, ret;
> +
> +    TRACE("called with drv: %p", drv);
> +
> +    if (pkt->type == VP_PKT_CLIENT) {
> +        TRACE("recieved client packet, client fd: %d, server fd: %d",
> +              pkt->payload.proxied.client_fd, 
> pkt->payload.proxied.server_fd);
> +        fd = pkt->payload.proxied.server_fd;
> +    } else if (pkt->type == VP_PKT_SERVER) {
> +        TRACE("recieved server packet, client fd: %d, server fd: %d",
> +              pkt->payload.proxied.client_fd, 
> pkt->payload.proxied.server_fd);
> +        fd = pkt->payload.proxied.client_fd;
> +    } else {
> +        TRACE("unknown packet type");
> +        return -1;
> +    }
> +
> +    /* TODO: proxied in non-blocking mode can causes us to spin here
> +     * for slow servers/clients. need to use write()'s and maintain
> +     * a per-conn write queue that we clear out before sending any
> +     * more data to the fd
> +     */

Hmm, so a guest can cause a denial of service in the host qemu process?
Are you working on adding the write queues?

> +    ret = vp_send_all(fd, (void *)pkt->payload.proxied.data,
> +            pkt->payload.proxied.bytes);
> +    if (ret == -1) {
> +        LOG("error sending data over channel");
> +        return -1;
> +    } else if (ret != pkt->payload.proxied.bytes) {
> +        TRACE("buffer full?");

This can happen?  Does this bring the whole connection down?  The whole
virtproxy instance?

-- 
Thanks,
Adam




reply via email to

[Prev in Thread] Current Thread [Next in Thread]