Here's the patch to get the KVM backend to do GSO. Please note
that this was a quick hack and I haven't even tested the case
where the tun device doesn't support GSO so it'll probably break
there.
It does the stupidest thing possible for guest => host by copying
the data so that it can use write(2).
diff --git a/qemu/hw/virtio-net.c b/qemu/hw/virtio-net.c
index 86f9e5a..b92e7c5 100644
--- a/qemu/hw/virtio-net.c
+++ b/qemu/hw/virtio-net.c
@@ -22,9 +22,9 @@
#define VIRTIO_ID_NET 1
/* The feature bitmap for virtio net */
-#define VIRTIO_NET_F_NO_CSUM 0
+#define VIRTIO_NET_F_CSUM 0
#define VIRTIO_NET_F_MAC 5
-#define VIRTIO_NET_F_GS0 6
+#define VIRTIO_NET_F_GSO 6
#define TX_TIMER_INTERVAL (1000 / 500)
@@ -87,7 +87,10 @@ static void virtio_net_update_config(VirtIODevice *vdev, uint8_t *config)
static uint32_t virtio_net_get_features(VirtIODevice *vdev)
{
- return (1 << VIRTIO_NET_F_MAC);
+ int gso = tap_has_gso(to_virtio_net(vdev)->vc->vlan->first_client);
+
+ return (1 << VIRTIO_NET_F_MAC) |
+ (gso ? (1 << VIRTIO_NET_F_CSUM) | (1 << VIRTIO_NET_F_GSO) : 0);
}
/* RX */
@@ -112,6 +115,7 @@ static void virtio_net_receive(void *opaque, const uint8_t *buf, int size)
VirtQueueElement elem;
struct virtio_net_hdr *hdr;
int offset, i;
+ int total;
/* FIXME: the drivers really need to set their status better */
if (n->rx_vq->vring.avail == NULL) {
@@ -129,18 +133,26 @@ static void virtio_net_receive(void *opaque, const uint8_t *buf, int size)
hdr->flags = 0;
hdr->gso_type = VIRTIO_NET_HDR_GSO_NONE;
- /* copy in packet. ugh */
offset = 0;
+ total = sizeof(*hdr);
+
+ if (tap_has_gso(n->vc->vlan->first_client)) {
+ memcpy(hdr, buf, sizeof(*hdr));
+ offset += total;
+ }
+
+ /* copy in packet. ugh */
i = 1;
while (offset < size && i < elem.in_num) {
int len = MIN(elem.in_sg[i].iov_len, size - offset);
memcpy(elem.in_sg[i].iov_base, buf + offset, len);
offset += len;
+ total += len;
i++;
}
/* signal other side */
- virtqueue_push(n->rx_vq, &elem, sizeof(*hdr) + offset);
+ virtqueue_push(n->rx_vq, &elem, total);
virtio_notify(&n->vdev, n->rx_vq);
}
@@ -201,7 +213,10 @@ void virtio_net_poll(void)
hdr->flags = 0;
hdr->gso_type = VIRTIO_NET_HDR_GSO_NONE;
again:
- len = readv(vnet->tap_fd, &elem.in_sg[1], elem.in_num - 1);
+ if (tap_has_gso(vnet->vc->vlan->first_client))
+ len = readv(vnet->tap_fd, &elem.in_sg[0], elem.in_num);
+ else
+ len = readv(vnet->tap_fd, &elem.in_sg[1], elem.in_num - 1);
if (len == -1) {
if (errno == EINTR || errno == EAGAIN)
goto again;
@@ -229,25 +244,32 @@ again:
static void virtio_net_flush_tx(VirtIONet *n, VirtQueue *vq)
{
VirtQueueElement elem;
+ uint8_t buf[sizeof(struct virtio_net_hdr) + 65536];
+ int size = sizeof(buf);
int count = 0;
+ int nogso = !tap_has_gso(n->vc->vlan->first_client);
if (!(n->vdev.status & VIRTIO_CONFIG_S_DRIVER_OK))
return;
while (virtqueue_pop(vq, &elem)) {
int i;
- size_t len = 0;
+ int offset = 0;
- /* ignore the header for now */
- for (i = 1; i < elem.out_num; i++) {
- qemu_send_packet(n->vc, elem.out_sg[i].iov_base,
- elem.out_sg[i].iov_len);
- len += elem.out_sg[i].iov_len;
+ /* Ignore the header if GSO is not supported. */
+ for (i = nogso; i < elem.out_num; i++) {
+ int len = MIN(elem.out_sg[i].iov_len, size - offset);
+ memcpy(buf + offset, elem.out_sg[i].iov_base, len);
+ offset += len;
}
+ qemu_send_packet(n->vc, buf, offset);
+
count++;
- virtqueue_push(vq, &elem, sizeof(struct virtio_net_hdr) + len);
+ if (nogso)
+ offset += elem.out_sg[0].iov_len;
+ virtqueue_push(vq, &elem, offset);
virtio_notify(&n->vdev, vq);
}
}
@@ -291,7 +313,7 @@ void *virtio_net_init(PCIBus *bus, NICInfo *nd, int devfn)
n->vdev.update_config = virtio_net_update_config;
n->vdev.get_features = virtio_net_get_features;
n->rx_vq = virtio_add_queue(&n->vdev, 512, virtio_net_handle_rx);
- n->tx_vq = virtio_add_queue(&n->vdev, 128, virtio_net_handle_tx);
+ n->tx_vq = virtio_add_queue(&n->vdev, 512, virtio_net_handle_tx);
n->can_receive = 0;
memcpy(n->mac, nd->macaddr, 6);
n->vc = qemu_new_vlan_client(nd->vlan, virtio_net_receive,
diff --git a/qemu/net.h b/qemu/net.h
index c8ff6d6..c3a75d6 100644
--- a/qemu/net.h
+++ b/qemu/net.h
@@ -36,6 +36,7 @@ void do_info_network(void);
/* virtio hack for zero copy receive */
int hack_around_tap(void *opaque);
+int tap_has_gso(void *opaque);
/* NIC info */
diff --git a/qemu/vl.c b/qemu/vl.c
index 21c9b53..3f10d0a 100644
--- a/qemu/vl.c
+++ b/qemu/vl.c
@@ -3962,6 +3962,7 @@ typedef struct TAPState {
int fd;
char down_script[1024];
int no_poll;
+ int gso;
} TAPState;
static int tap_read_poll(void *opaque)
@@ -4019,6 +4020,14 @@ int hack_around_tap(void *opaque)
return -1;
}
+int tap_has_gso(void *opaque)
+{
+ VLANClientState *vc = opaque;
+ TAPState *ts = vc->opaque;
+
+ return ts ? ts->gso : 0;
+}
+
/* fd support */
static TAPState *net_tap_fd_init(VLANState *vlan, int fd)
@@ -4038,7 +4047,7 @@ static TAPState *net_tap_fd_init(VLANState *vlan, int fd)
}
#if defined (_BSD) || defined (__FreeBSD_kernel__)
-static int tap_open(char *ifname, int ifname_size)
+static int tap_open(char *ifname, int ifname_size, int *gso)
{
int fd;
char *dev;
@@ -4180,7 +4189,7 @@ int tap_alloc(char *dev)
return tap_fd;
}
-static int tap_open(char *ifname, int ifname_size)
+static int tap_open(char *ifname, int ifname_size, int *gso)
{
char dev[10]="";
int fd;
@@ -4193,18 +4202,30 @@ static int tap_open(char *ifname, int ifname_size)
return fd;
}
#else
-static int tap_open(char *ifname, int ifname_size)
+static int tap_open(char *ifname, int ifname_size, int *gso)
{
struct ifreq ifr;
int fd, ret;
+ unsigned int features;
TFR(fd = open("/dev/net/tun", O_RDWR));
if (fd < 0) {
fprintf(stderr, "warning: could not open /dev/net/tun: no virtual network emulation\n");
return -1;
}
+
+ if (ioctl(fd, TUNGETFEATURES, &features))
+ features = IFF_TUN | IFF_TAP | IFF_NO_PI | IFF_ONE_QUEUE;
+
memset(&ifr, 0, sizeof(ifr));
ifr.ifr_flags = IFF_TAP | IFF_NO_PI;
+
+ if (features & IFF_VIRTIO_HDR && features & IFF_RECV_CSUM &&
+ features & IFF_RECV_GSO) {
+ *gso = 1;
+ ifr.ifr_flags |= IFF_VIRTIO_HDR | IFF_RECV_CSUM | IFF_RECV_GSO;
+ }
+
if (ifname[0] != '\0')
pstrcpy(ifr.ifr_name, IFNAMSIZ, ifname);
else
@@ -4262,13 +4283,15 @@ static int net_tap_init(VLANState *vlan, const char *ifname1,
{
TAPState *s;
int fd;
+ int gso;
char ifname[128];
if (ifname1 != NULL)
pstrcpy(ifname, sizeof(ifname), ifname1);
else
ifname[0] = '\0';
- TFR(fd = tap_open(ifname, sizeof(ifname)));
+ gso = 0;
+ TFR(fd = tap_open(ifname, sizeof(ifname), &gso));
if (fd < 0)
return -1;
@@ -4281,6 +4304,8 @@ static int net_tap_init(VLANState *vlan, const char *ifname1,
s = net_tap_fd_init(vlan, fd);
if (!s)
return -1;
+
+ s->gso = gso;
snprintf(s->vc->info_str, sizeof(s->vc->info_str),
"tap: ifname=%s setup_script=%s", ifname, setup_script);
if (down_script && strcmp(down_script, "no"))
--
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@xxxxxxxxxxxxxxxxxxx>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt
_______________________________________________
Virtualization mailing list
Virtualization@xxxxxxxxxxxxxxxxxxxxxxxxxx
https://lists.linux-foundation.org/mailman/listinfo/virtualization
