2011-05-13 04:54:01 +00:00
|
|
|
/*-
|
|
|
|
* Copyright (c) 2011 NetApp, Inc.
|
|
|
|
* All rights reserved.
|
|
|
|
*
|
|
|
|
* Redistribution and use in source and binary forms, with or without
|
|
|
|
* modification, are permitted provided that the following conditions
|
|
|
|
* are met:
|
|
|
|
* 1. Redistributions of source code must retain the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer.
|
|
|
|
* 2. Redistributions in binary form must reproduce the above copyright
|
|
|
|
* notice, this list of conditions and the following disclaimer in the
|
|
|
|
* documentation and/or other materials provided with the distribution.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY NETAPP, INC ``AS IS'' AND
|
|
|
|
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
|
|
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
|
|
* ARE DISCLAIMED. IN NO EVENT SHALL NETAPP, INC OR CONTRIBUTORS BE LIABLE
|
|
|
|
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|
|
|
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|
|
|
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|
|
|
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
|
|
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|
|
|
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|
|
|
* SUCH DAMAGE.
|
|
|
|
*
|
|
|
|
* $FreeBSD$
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <sys/cdefs.h>
|
|
|
|
__FBSDID("$FreeBSD$");
|
|
|
|
|
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/linker_set.h>
|
|
|
|
#include <sys/select.h>
|
|
|
|
#include <sys/uio.h>
|
|
|
|
#include <sys/ioctl.h>
|
2013-07-04 05:35:56 +00:00
|
|
|
#include <net/ethernet.h>
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
#include <errno.h>
|
|
|
|
#include <fcntl.h>
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <stdint.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <strings.h>
|
|
|
|
#include <unistd.h>
|
|
|
|
#include <assert.h>
|
|
|
|
#include <md5.h>
|
|
|
|
#include <pthread.h>
|
2013-04-26 05:13:48 +00:00
|
|
|
#include <pthread_np.h>
|
2011-05-13 04:54:01 +00:00
|
|
|
|
2012-12-13 01:58:11 +00:00
|
|
|
#include "bhyverun.h"
|
2011-05-13 04:54:01 +00:00
|
|
|
#include "pci_emul.h"
|
|
|
|
#include "mevent.h"
|
|
|
|
#include "virtio.h"
|
|
|
|
|
2013-04-26 05:13:48 +00:00
|
|
|
#define VTNET_RINGSZ 1024
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
#define VTNET_MAXSEGS 32
|
|
|
|
|
|
|
|
/*
|
2013-07-17 23:37:33 +00:00
|
|
|
* Host capabilities. Note that we only offer a few of these.
|
2011-05-13 04:54:01 +00:00
|
|
|
*/
|
2013-07-17 23:37:33 +00:00
|
|
|
#define VIRTIO_NET_F_CSUM (1 << 0) /* host handles partial cksum */
|
|
|
|
#define VIRTIO_NET_F_GUEST_CSUM (1 << 1) /* guest handles partial cksum */
|
|
|
|
#define VIRTIO_NET_F_MAC (1 << 5) /* host supplies MAC */
|
|
|
|
#define VIRTIO_NET_F_GSO_DEPREC (1 << 6) /* deprecated: host handles GSO */
|
|
|
|
#define VIRTIO_NET_F_GUEST_TSO4 (1 << 7) /* guest can rcv TSOv4 */
|
|
|
|
#define VIRTIO_NET_F_GUEST_TSO6 (1 << 8) /* guest can rcv TSOv6 */
|
|
|
|
#define VIRTIO_NET_F_GUEST_ECN (1 << 9) /* guest can rcv TSO with ECN */
|
|
|
|
#define VIRTIO_NET_F_GUEST_UFO (1 << 10) /* guest can rcv UFO */
|
|
|
|
#define VIRTIO_NET_F_HOST_TSO4 (1 << 11) /* host can rcv TSOv4 */
|
|
|
|
#define VIRTIO_NET_F_HOST_TSO6 (1 << 12) /* host can rcv TSOv6 */
|
|
|
|
#define VIRTIO_NET_F_HOST_ECN (1 << 13) /* host can rcv TSO with ECN */
|
|
|
|
#define VIRTIO_NET_F_HOST_UFO (1 << 14) /* host can rcv UFO */
|
|
|
|
#define VIRTIO_NET_F_MRG_RXBUF (1 << 15) /* host can merge RX buffers */
|
|
|
|
#define VIRTIO_NET_F_STATUS (1 << 16) /* config status field available */
|
|
|
|
#define VIRTIO_NET_F_CTRL_VQ (1 << 17) /* control channel available */
|
|
|
|
#define VIRTIO_NET_F_CTRL_RX (1 << 18) /* control channel RX mode support */
|
|
|
|
#define VIRTIO_NET_F_CTRL_VLAN (1 << 19) /* control channel VLAN filtering */
|
|
|
|
#define VIRTIO_NET_F_GUEST_ANNOUNCE \
|
|
|
|
(1 << 21) /* guest can send gratuitous pkts */
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
#define VTNET_S_HOSTCAPS \
|
2013-07-17 23:37:33 +00:00
|
|
|
( VIRTIO_NET_F_MAC | VIRTIO_NET_F_MRG_RXBUF | VIRTIO_NET_F_STATUS | \
|
2013-05-03 01:16:18 +00:00
|
|
|
VIRTIO_F_NOTIFY_ON_EMPTY)
|
2011-05-13 04:54:01 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
/*
|
|
|
|
* PCI config-space "registers"
|
|
|
|
*/
|
|
|
|
struct virtio_net_config {
|
|
|
|
uint8_t mac[6];
|
|
|
|
uint16_t status;
|
|
|
|
} __packed;
|
|
|
|
|
2011-05-13 04:54:01 +00:00
|
|
|
/*
|
|
|
|
* Queue definitions.
|
|
|
|
*/
|
|
|
|
#define VTNET_RXQ 0
|
|
|
|
#define VTNET_TXQ 1
|
2013-07-17 23:37:33 +00:00
|
|
|
#define VTNET_CTLQ 2 /* NB: not yet supported */
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
#define VTNET_MAXQ 3
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Fixed network header size
|
|
|
|
*/
|
|
|
|
struct virtio_net_rxhdr {
|
|
|
|
uint8_t vrh_flags;
|
|
|
|
uint8_t vrh_gso_type;
|
|
|
|
uint16_t vrh_hdr_len;
|
|
|
|
uint16_t vrh_gso_size;
|
|
|
|
uint16_t vrh_csum_start;
|
|
|
|
uint16_t vrh_csum_offset;
|
|
|
|
uint16_t vrh_bufs;
|
|
|
|
} __packed;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Debug printf
|
|
|
|
*/
|
|
|
|
static int pci_vtnet_debug;
|
|
|
|
#define DPRINTF(params) if (pci_vtnet_debug) printf params
|
|
|
|
#define WPRINTF(params) printf params
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Per-device softc
|
|
|
|
*/
|
|
|
|
struct pci_vtnet_softc {
|
2013-07-17 23:37:33 +00:00
|
|
|
struct virtio_softc vsc_vs;
|
|
|
|
struct vqueue_info vsc_queues[VTNET_MAXQ - 1];
|
2011-05-13 04:54:01 +00:00
|
|
|
pthread_mutex_t vsc_mtx;
|
|
|
|
struct mevent *vsc_mevp;
|
|
|
|
|
|
|
|
int vsc_tapfd;
|
|
|
|
int vsc_rx_ready;
|
2013-07-17 23:37:33 +00:00
|
|
|
volatile int resetting; /* set and checked outside lock */
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
uint32_t vsc_features;
|
2013-07-17 23:37:33 +00:00
|
|
|
struct virtio_net_config vsc_config;
|
2013-04-30 00:36:16 +00:00
|
|
|
|
|
|
|
pthread_mutex_t rx_mtx;
|
|
|
|
int rx_in_progress;
|
|
|
|
|
2013-04-26 05:13:48 +00:00
|
|
|
pthread_t tx_tid;
|
|
|
|
pthread_mutex_t tx_mtx;
|
|
|
|
pthread_cond_t tx_cond;
|
2013-04-30 00:36:16 +00:00
|
|
|
int tx_in_progress;
|
2011-05-13 04:54:01 +00:00
|
|
|
};
|
2012-12-12 19:45:36 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
static void pci_vtnet_reset(void *);
|
|
|
|
/* static void pci_vtnet_notify(void *, struct vqueue_info *); */
|
|
|
|
static int pci_vtnet_cfgread(void *, int, int, uint32_t *);
|
|
|
|
static int pci_vtnet_cfgwrite(void *, int, int, uint32_t);
|
|
|
|
|
|
|
|
static struct virtio_consts vtnet_vi_consts = {
|
|
|
|
"vtnet", /* our name */
|
|
|
|
VTNET_MAXQ - 1, /* we currently support 2 virtqueues */
|
|
|
|
sizeof(struct virtio_net_config), /* config reg size */
|
|
|
|
pci_vtnet_reset, /* reset */
|
|
|
|
NULL, /* device-wide qnotify -- not used */
|
|
|
|
pci_vtnet_cfgread, /* read PCI config */
|
|
|
|
pci_vtnet_cfgwrite, /* write PCI config */
|
|
|
|
VTNET_S_HOSTCAPS, /* our capabilities */
|
|
|
|
};
|
2012-12-12 19:45:36 +00:00
|
|
|
|
2013-04-30 00:36:16 +00:00
|
|
|
/*
|
|
|
|
* If the transmit thread is active then stall until it is done.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
pci_vtnet_txwait(struct pci_vtnet_softc *sc)
|
|
|
|
{
|
|
|
|
|
|
|
|
pthread_mutex_lock(&sc->tx_mtx);
|
|
|
|
while (sc->tx_in_progress) {
|
|
|
|
pthread_mutex_unlock(&sc->tx_mtx);
|
|
|
|
usleep(10000);
|
|
|
|
pthread_mutex_lock(&sc->tx_mtx);
|
|
|
|
}
|
|
|
|
pthread_mutex_unlock(&sc->tx_mtx);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the receive thread is active then stall until it is done.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
pci_vtnet_rxwait(struct pci_vtnet_softc *sc)
|
|
|
|
{
|
|
|
|
|
|
|
|
pthread_mutex_lock(&sc->rx_mtx);
|
|
|
|
while (sc->rx_in_progress) {
|
|
|
|
pthread_mutex_unlock(&sc->rx_mtx);
|
|
|
|
usleep(10000);
|
|
|
|
pthread_mutex_lock(&sc->rx_mtx);
|
|
|
|
}
|
|
|
|
pthread_mutex_unlock(&sc->rx_mtx);
|
|
|
|
}
|
|
|
|
|
2011-05-13 04:54:01 +00:00
|
|
|
static void
|
2013-07-17 23:37:33 +00:00
|
|
|
pci_vtnet_reset(void *vsc)
|
2011-05-13 04:54:01 +00:00
|
|
|
{
|
2013-07-17 23:37:33 +00:00
|
|
|
struct pci_vtnet_softc *sc = vsc;
|
2013-04-26 05:13:48 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
DPRINTF(("vtnet: device reset requested !\n"));
|
2013-04-30 00:36:16 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
sc->resetting = 1;
|
2013-04-30 01:14:54 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
/*
|
|
|
|
* Wait for the transmit and receive threads to finish their
|
|
|
|
* processing.
|
|
|
|
*/
|
|
|
|
pci_vtnet_txwait(sc);
|
|
|
|
pci_vtnet_rxwait(sc);
|
2011-05-13 04:54:01 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
sc->vsc_rx_ready = 0;
|
2011-05-13 04:54:01 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
/* now reset rings, MSI-X vectors, and negotiated capabilities */
|
|
|
|
vi_reset_dev(&sc->vsc_vs);
|
2013-05-03 01:16:18 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
sc->resetting = 0;
|
2013-05-03 01:16:18 +00:00
|
|
|
}
|
|
|
|
|
2011-05-13 04:54:01 +00:00
|
|
|
/*
|
|
|
|
* Called to send a buffer chain out to the tap device
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
pci_vtnet_tap_tx(struct pci_vtnet_softc *sc, struct iovec *iov, int iovcnt,
|
|
|
|
int len)
|
|
|
|
{
|
2013-07-17 23:37:33 +00:00
|
|
|
static char pad[60]; /* all zero bytes */
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
if (sc->vsc_tapfd == -1)
|
|
|
|
return;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the length is < 60, pad out to that and add the
|
|
|
|
* extra zero'd segment to the iov. It is guaranteed that
|
|
|
|
* there is always an extra iov available by the caller.
|
|
|
|
*/
|
|
|
|
if (len < 60) {
|
|
|
|
iov[iovcnt].iov_base = pad;
|
|
|
|
iov[iovcnt].iov_len = 60 - len;
|
|
|
|
iovcnt++;
|
|
|
|
}
|
|
|
|
(void) writev(sc->vsc_tapfd, iov, iovcnt);
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Called when there is read activity on the tap file descriptor.
|
|
|
|
* Each buffer posted by the guest is assumed to be able to contain
|
|
|
|
* an entire ethernet frame + rx header.
|
|
|
|
* MP note: the dummybuf is only used for discarding frames, so there
|
|
|
|
* is no need for it to be per-vtnet or locked.
|
|
|
|
*/
|
|
|
|
static uint8_t dummybuf[2048];
|
|
|
|
|
|
|
|
static void
|
|
|
|
pci_vtnet_tap_rx(struct pci_vtnet_softc *sc)
|
|
|
|
{
|
2013-07-17 23:37:33 +00:00
|
|
|
struct vqueue_info *vq;
|
2011-05-13 04:54:01 +00:00
|
|
|
struct virtio_net_rxhdr *vrx;
|
|
|
|
uint8_t *buf;
|
|
|
|
int len;
|
2013-07-17 23:37:33 +00:00
|
|
|
struct iovec iov;
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Should never be called without a valid tap fd
|
|
|
|
*/
|
|
|
|
assert(sc->vsc_tapfd != -1);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* But, will be called when the rx ring hasn't yet
|
2013-04-30 00:36:16 +00:00
|
|
|
* been set up or the guest is resetting the device.
|
2011-05-13 04:54:01 +00:00
|
|
|
*/
|
2013-04-30 00:36:16 +00:00
|
|
|
if (!sc->vsc_rx_ready || sc->resetting) {
|
2011-05-13 04:54:01 +00:00
|
|
|
/*
|
|
|
|
* Drop the packet and try later.
|
|
|
|
*/
|
|
|
|
(void) read(sc->vsc_tapfd, dummybuf, sizeof(dummybuf));
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2013-07-17 23:37:33 +00:00
|
|
|
* Check for available rx buffers
|
2011-05-13 04:54:01 +00:00
|
|
|
*/
|
2013-07-17 23:37:33 +00:00
|
|
|
vq = &sc->vsc_queues[VTNET_RXQ];
|
|
|
|
vq_startchains(vq);
|
|
|
|
if (!vq_has_descs(vq)) {
|
2011-05-13 04:54:01 +00:00
|
|
|
/*
|
2013-07-17 23:37:33 +00:00
|
|
|
* Drop the packet and try later. Interrupt on
|
|
|
|
* empty, if that's negotiated.
|
2011-05-13 04:54:01 +00:00
|
|
|
*/
|
|
|
|
(void) read(sc->vsc_tapfd, dummybuf, sizeof(dummybuf));
|
2013-07-17 23:37:33 +00:00
|
|
|
vq_endchains(vq, 1);
|
2011-05-13 04:54:01 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
do {
|
2011-05-13 04:54:01 +00:00
|
|
|
/*
|
2013-07-17 23:37:33 +00:00
|
|
|
* Get descriptor chain, which should have just
|
|
|
|
* one descriptor in it.
|
|
|
|
* ??? allow guests to use multiple descs?
|
2011-05-13 04:54:01 +00:00
|
|
|
*/
|
2013-07-17 23:37:33 +00:00
|
|
|
assert(vq_getchain(vq, &iov, 1, NULL) == 1);
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Get a pointer to the rx header, and use the
|
|
|
|
* data immediately following it for the packet buffer.
|
|
|
|
*/
|
2013-07-17 23:37:33 +00:00
|
|
|
vrx = iov.iov_base;
|
2011-05-13 04:54:01 +00:00
|
|
|
buf = (uint8_t *)(vrx + 1);
|
|
|
|
|
|
|
|
len = read(sc->vsc_tapfd, buf,
|
2013-07-17 23:37:33 +00:00
|
|
|
iov.iov_len - sizeof(struct virtio_net_rxhdr));
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
if (len < 0 && errno == EWOULDBLOCK) {
|
2013-07-17 23:37:33 +00:00
|
|
|
/*
|
|
|
|
* No more packets, but still some avail ring
|
|
|
|
* entries. Interrupt if needed/appropriate.
|
|
|
|
*/
|
|
|
|
vq_endchains(vq, 0);
|
|
|
|
return;
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The only valid field in the rx packet header is the
|
|
|
|
* number of buffers, which is always 1 without TSO
|
|
|
|
* support.
|
|
|
|
*/
|
|
|
|
memset(vrx, 0, sizeof(struct virtio_net_rxhdr));
|
|
|
|
vrx->vrh_bufs = 1;
|
|
|
|
|
|
|
|
/*
|
2013-07-17 23:37:33 +00:00
|
|
|
* Release this chain and handle more chains.
|
2011-05-13 04:54:01 +00:00
|
|
|
*/
|
2013-07-17 23:37:33 +00:00
|
|
|
vq_relchain(vq, len + sizeof(struct virtio_net_rxhdr));
|
|
|
|
} while (vq_has_descs(vq));
|
2011-05-13 04:54:01 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
/* Interrupt if needed, including for NOTIFY_ON_EMPTY. */
|
|
|
|
vq_endchains(vq, 1);
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
pci_vtnet_tap_callback(int fd, enum ev_type type, void *param)
|
|
|
|
{
|
|
|
|
struct pci_vtnet_softc *sc = param;
|
|
|
|
|
2013-04-30 00:36:16 +00:00
|
|
|
pthread_mutex_lock(&sc->rx_mtx);
|
|
|
|
sc->rx_in_progress = 1;
|
2011-05-13 04:54:01 +00:00
|
|
|
pci_vtnet_tap_rx(sc);
|
2013-04-30 00:36:16 +00:00
|
|
|
sc->rx_in_progress = 0;
|
|
|
|
pthread_mutex_unlock(&sc->rx_mtx);
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
2013-07-17 23:37:33 +00:00
|
|
|
pci_vtnet_ping_rxq(void *vsc, struct vqueue_info *vq)
|
2011-05-13 04:54:01 +00:00
|
|
|
{
|
2013-07-17 23:37:33 +00:00
|
|
|
struct pci_vtnet_softc *sc = vsc;
|
|
|
|
|
2011-05-13 04:54:01 +00:00
|
|
|
/*
|
|
|
|
* A qnotify means that the rx process can now begin
|
|
|
|
*/
|
|
|
|
if (sc->vsc_rx_ready == 0) {
|
|
|
|
sc->vsc_rx_ready = 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
2013-07-17 23:37:33 +00:00
|
|
|
pci_vtnet_proctx(struct pci_vtnet_softc *sc, struct vqueue_info *vq)
|
2011-05-13 04:54:01 +00:00
|
|
|
{
|
|
|
|
struct iovec iov[VTNET_MAXSEGS + 1];
|
2013-07-17 23:37:33 +00:00
|
|
|
int i, n;
|
|
|
|
int plen, tlen;
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
/*
|
2013-07-17 23:37:33 +00:00
|
|
|
* Obtain chain of descriptors. The first one is
|
|
|
|
* really the header descriptor, so we need to sum
|
|
|
|
* up two lengths: packet length and transfer length.
|
2011-05-13 04:54:01 +00:00
|
|
|
*/
|
2013-07-17 23:37:33 +00:00
|
|
|
n = vq_getchain(vq, iov, VTNET_MAXSEGS, NULL);
|
|
|
|
assert(n >= 1 && n <= VTNET_MAXSEGS);
|
|
|
|
plen = 0;
|
|
|
|
tlen = iov[0].iov_len;
|
|
|
|
for (i = 1; i < n; i++) {
|
|
|
|
plen += iov[i].iov_len;
|
|
|
|
tlen += iov[i].iov_len;
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
DPRINTF(("virtio: packet send, %d bytes, %d segs\n\r", plen, n));
|
|
|
|
pci_vtnet_tap_tx(sc, &iov[1], n - 1, plen);
|
2011-05-13 04:54:01 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
/* chain is processed, release it and set tlen */
|
|
|
|
vq_relchain(vq, tlen);
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
2013-07-17 23:37:33 +00:00
|
|
|
pci_vtnet_ping_txq(void *vsc, struct vqueue_info *vq)
|
2011-05-13 04:54:01 +00:00
|
|
|
{
|
2013-07-17 23:37:33 +00:00
|
|
|
struct pci_vtnet_softc *sc = vsc;
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
/*
|
2013-07-17 23:37:33 +00:00
|
|
|
* Any ring entries to process?
|
2011-05-13 04:54:01 +00:00
|
|
|
*/
|
2013-07-17 23:37:33 +00:00
|
|
|
if (!vq_has_descs(vq))
|
2011-05-13 04:54:01 +00:00
|
|
|
return;
|
|
|
|
|
2013-04-26 05:13:48 +00:00
|
|
|
/* Signal the tx thread for processing */
|
|
|
|
pthread_mutex_lock(&sc->tx_mtx);
|
|
|
|
if (sc->tx_in_progress == 0)
|
|
|
|
pthread_cond_signal(&sc->tx_cond);
|
|
|
|
pthread_mutex_unlock(&sc->tx_mtx);
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
|
|
|
|
2013-04-26 05:13:48 +00:00
|
|
|
/*
|
|
|
|
* Thread which will handle processing of TX desc
|
|
|
|
*/
|
|
|
|
static void *
|
|
|
|
pci_vtnet_tx_thread(void *param)
|
|
|
|
{
|
2013-07-17 23:37:33 +00:00
|
|
|
struct pci_vtnet_softc *sc = param;
|
|
|
|
struct vqueue_info *vq;
|
|
|
|
int have_work, error;
|
|
|
|
|
|
|
|
vq = &sc->vsc_queues[VTNET_TXQ];
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Let us wait till the tx queue pointers get initialised &
|
|
|
|
* first tx signaled
|
2013-04-26 05:13:48 +00:00
|
|
|
*/
|
|
|
|
pthread_mutex_lock(&sc->tx_mtx);
|
|
|
|
error = pthread_cond_wait(&sc->tx_cond, &sc->tx_mtx);
|
|
|
|
assert(error == 0);
|
2013-07-17 23:37:33 +00:00
|
|
|
|
2013-04-26 05:13:48 +00:00
|
|
|
for (;;) {
|
2013-07-17 23:37:33 +00:00
|
|
|
/* note - tx mutex is locked here */
|
|
|
|
do {
|
2013-05-03 01:16:18 +00:00
|
|
|
if (sc->resetting)
|
2013-07-17 23:37:33 +00:00
|
|
|
have_work = 0;
|
2013-05-03 01:16:18 +00:00
|
|
|
else
|
2013-07-17 23:37:33 +00:00
|
|
|
have_work = vq_has_descs(vq);
|
|
|
|
|
|
|
|
if (!have_work) {
|
|
|
|
sc->tx_in_progress = 0;
|
|
|
|
error = pthread_cond_wait(&sc->tx_cond,
|
|
|
|
&sc->tx_mtx);
|
|
|
|
assert(error == 0);
|
|
|
|
}
|
|
|
|
} while (!have_work);
|
2013-04-26 05:13:48 +00:00
|
|
|
sc->tx_in_progress = 1;
|
|
|
|
pthread_mutex_unlock(&sc->tx_mtx);
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
vq_startchains(vq);
|
|
|
|
do {
|
2013-04-26 05:13:48 +00:00
|
|
|
/*
|
2013-07-17 23:37:33 +00:00
|
|
|
* Run through entries, placing them into
|
|
|
|
* iovecs and sending when an end-of-packet
|
|
|
|
* is found
|
2013-04-26 05:13:48 +00:00
|
|
|
*/
|
2013-07-17 23:37:33 +00:00
|
|
|
pci_vtnet_proctx(sc, vq);
|
|
|
|
} while (vq_has_descs(vq));
|
2013-05-03 01:16:18 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Generate an interrupt if needed.
|
|
|
|
*/
|
2013-07-17 23:37:33 +00:00
|
|
|
vq_endchains(vq, 1);
|
2013-04-26 05:13:48 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
pthread_mutex_lock(&sc->tx_mtx);
|
|
|
|
}
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
#ifdef notyet
|
2011-05-13 04:54:01 +00:00
|
|
|
static void
|
2013-07-17 23:37:33 +00:00
|
|
|
pci_vtnet_ping_ctlq(void *vsc, struct vqueue_info *vq)
|
2011-05-13 04:54:01 +00:00
|
|
|
{
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
DPRINTF(("vtnet: control qnotify!\n\r"));
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
2013-07-17 23:37:33 +00:00
|
|
|
#endif
|
2011-05-13 04:54:01 +00:00
|
|
|
|
2013-07-04 05:35:56 +00:00
|
|
|
static int
|
|
|
|
pci_vtnet_parsemac(char *mac_str, uint8_t *mac_addr)
|
|
|
|
{
|
|
|
|
struct ether_addr *ea;
|
|
|
|
char *tmpstr;
|
|
|
|
char zero_addr[ETHER_ADDR_LEN] = { 0, 0, 0, 0, 0, 0 };
|
|
|
|
|
|
|
|
tmpstr = strsep(&mac_str,"=");
|
|
|
|
|
|
|
|
if ((mac_str != NULL) && (!strcmp(tmpstr,"mac"))) {
|
|
|
|
ea = ether_aton(mac_str);
|
|
|
|
|
|
|
|
if (ea == NULL || ETHER_IS_MULTICAST(ea->octet) ||
|
|
|
|
memcmp(ea->octet, zero_addr, ETHER_ADDR_LEN) == 0) {
|
|
|
|
fprintf(stderr, "Invalid MAC %s\n", mac_str);
|
|
|
|
return (EINVAL);
|
|
|
|
} else
|
|
|
|
memcpy(mac_addr, ea->octet, ETHER_ADDR_LEN);
|
|
|
|
}
|
|
|
|
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2011-05-13 04:54:01 +00:00
|
|
|
static int
|
|
|
|
pci_vtnet_init(struct vmctx *ctx, struct pci_devinst *pi, char *opts)
|
|
|
|
{
|
|
|
|
MD5_CTX mdctx;
|
|
|
|
unsigned char digest[16];
|
|
|
|
char nstr[80];
|
2013-04-26 05:13:48 +00:00
|
|
|
char tname[MAXCOMLEN + 1];
|
2011-05-13 04:54:01 +00:00
|
|
|
struct pci_vtnet_softc *sc;
|
2013-07-04 05:35:56 +00:00
|
|
|
char *devname;
|
|
|
|
char *vtopts;
|
|
|
|
int mac_provided;
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
sc = malloc(sizeof(struct pci_vtnet_softc));
|
|
|
|
memset(sc, 0, sizeof(struct pci_vtnet_softc));
|
|
|
|
|
|
|
|
pthread_mutex_init(&sc->vsc_mtx, NULL);
|
2013-07-17 23:37:33 +00:00
|
|
|
|
|
|
|
vi_softc_linkup(&sc->vsc_vs, &vtnet_vi_consts, sc, pi, sc->vsc_queues);
|
2014-01-29 14:56:48 +00:00
|
|
|
sc->vsc_vs.vs_mtx = &sc->vsc_mtx;
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
sc->vsc_queues[VTNET_RXQ].vq_qsize = VTNET_RINGSZ;
|
|
|
|
sc->vsc_queues[VTNET_RXQ].vq_notify = pci_vtnet_ping_rxq;
|
|
|
|
sc->vsc_queues[VTNET_TXQ].vq_qsize = VTNET_RINGSZ;
|
|
|
|
sc->vsc_queues[VTNET_TXQ].vq_notify = pci_vtnet_ping_txq;
|
|
|
|
#ifdef notyet
|
|
|
|
sc->vsc_queues[VTNET_CTLQ].vq_qsize = VTNET_RINGSZ;
|
|
|
|
sc->vsc_queues[VTNET_CTLQ].vq_notify = pci_vtnet_ping_ctlq;
|
|
|
|
#endif
|
2013-01-30 04:30:36 +00:00
|
|
|
|
2011-05-13 04:54:01 +00:00
|
|
|
/*
|
2013-07-04 05:35:56 +00:00
|
|
|
* Attempt to open the tap device and read the MAC address
|
|
|
|
* if specified
|
2011-05-13 04:54:01 +00:00
|
|
|
*/
|
2013-07-04 05:35:56 +00:00
|
|
|
mac_provided = 0;
|
2011-05-13 04:54:01 +00:00
|
|
|
sc->vsc_tapfd = -1;
|
|
|
|
if (opts != NULL) {
|
|
|
|
char tbuf[80];
|
2013-07-04 05:35:56 +00:00
|
|
|
int err;
|
|
|
|
|
|
|
|
devname = vtopts = strdup(opts);
|
|
|
|
(void) strsep(&vtopts, ",");
|
|
|
|
|
|
|
|
if (vtopts != NULL) {
|
2013-07-17 23:37:33 +00:00
|
|
|
err = pci_vtnet_parsemac(vtopts, sc->vsc_config.mac);
|
2013-07-04 05:35:56 +00:00
|
|
|
if (err != 0) {
|
|
|
|
free(devname);
|
|
|
|
return (err);
|
|
|
|
}
|
|
|
|
mac_provided = 1;
|
|
|
|
}
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
strcpy(tbuf, "/dev/");
|
2013-07-04 05:35:56 +00:00
|
|
|
strlcat(tbuf, devname, sizeof(tbuf));
|
|
|
|
|
|
|
|
free(devname);
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
sc->vsc_tapfd = open(tbuf, O_RDWR);
|
|
|
|
if (sc->vsc_tapfd == -1) {
|
|
|
|
WPRINTF(("open of tap device %s failed\n", tbuf));
|
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* Set non-blocking and register for read
|
|
|
|
* notifications with the event loop
|
|
|
|
*/
|
|
|
|
int opt = 1;
|
|
|
|
if (ioctl(sc->vsc_tapfd, FIONBIO, &opt) < 0) {
|
|
|
|
WPRINTF(("tap device O_NONBLOCK failed\n"));
|
|
|
|
close(sc->vsc_tapfd);
|
|
|
|
sc->vsc_tapfd = -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
sc->vsc_mevp = mevent_add(sc->vsc_tapfd,
|
|
|
|
EVF_READ,
|
|
|
|
pci_vtnet_tap_callback,
|
|
|
|
sc);
|
|
|
|
if (sc->vsc_mevp == NULL) {
|
|
|
|
WPRINTF(("Could not register event\n"));
|
|
|
|
close(sc->vsc_tapfd);
|
|
|
|
sc->vsc_tapfd = -1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2013-07-04 05:35:56 +00:00
|
|
|
* The default MAC address is the standard NetApp OUI of 00-a0-98,
|
|
|
|
* followed by an MD5 of the PCI slot/func number and dev name
|
2011-05-13 04:54:01 +00:00
|
|
|
*/
|
2013-07-04 05:35:56 +00:00
|
|
|
if (!mac_provided) {
|
2012-12-12 19:25:48 +00:00
|
|
|
snprintf(nstr, sizeof(nstr), "%d-%d-%s", pi->pi_slot,
|
2013-11-06 00:25:17 +00:00
|
|
|
pi->pi_func, vmname);
|
2013-07-04 05:35:56 +00:00
|
|
|
|
|
|
|
MD5Init(&mdctx);
|
|
|
|
MD5Update(&mdctx, nstr, strlen(nstr));
|
|
|
|
MD5Final(digest, &mdctx);
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
sc->vsc_config.mac[0] = 0x00;
|
|
|
|
sc->vsc_config.mac[1] = 0xa0;
|
|
|
|
sc->vsc_config.mac[2] = 0x98;
|
|
|
|
sc->vsc_config.mac[3] = digest[0];
|
|
|
|
sc->vsc_config.mac[4] = digest[1];
|
|
|
|
sc->vsc_config.mac[5] = digest[2];
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* initialize config space */
|
|
|
|
pci_set_cfgdata16(pi, PCIR_DEVICE, VIRTIO_DEV_NET);
|
|
|
|
pci_set_cfgdata16(pi, PCIR_VENDOR, VIRTIO_VENDOR);
|
|
|
|
pci_set_cfgdata8(pi, PCIR_CLASS, PCIC_NETWORK);
|
|
|
|
pci_set_cfgdata16(pi, PCIR_SUBDEV_0, VIRTIO_TYPE_NET);
|
2013-01-30 04:30:36 +00:00
|
|
|
|
2014-01-29 14:56:48 +00:00
|
|
|
pci_lintr_request(pi);
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
/* link always up */
|
|
|
|
sc->vsc_config.status = 1;
|
2013-01-30 04:30:36 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
/* use BAR 1 to map MSI-X table and PBA, if we're using MSI-X */
|
2013-10-17 22:01:17 +00:00
|
|
|
if (vi_intr_init(&sc->vsc_vs, 1, fbsdrun_virtio_msix()))
|
2013-07-17 23:37:33 +00:00
|
|
|
return (1);
|
|
|
|
|
|
|
|
/* use BAR 0 to map config regs in IO space */
|
|
|
|
vi_set_io_bar(&sc->vsc_vs, 0);
|
2013-04-30 00:36:16 +00:00
|
|
|
|
|
|
|
sc->resetting = 0;
|
|
|
|
|
|
|
|
sc->rx_in_progress = 0;
|
|
|
|
pthread_mutex_init(&sc->rx_mtx, NULL);
|
|
|
|
|
2013-04-26 05:13:48 +00:00
|
|
|
/*
|
2013-07-17 23:37:33 +00:00
|
|
|
* Initialize tx semaphore & spawn TX processing thread.
|
2013-04-26 05:13:48 +00:00
|
|
|
* As of now, only one thread for TX desc processing is
|
|
|
|
* spawned.
|
|
|
|
*/
|
|
|
|
sc->tx_in_progress = 0;
|
|
|
|
pthread_mutex_init(&sc->tx_mtx, NULL);
|
|
|
|
pthread_cond_init(&sc->tx_cond, NULL);
|
|
|
|
pthread_create(&sc->tx_tid, NULL, pci_vtnet_tx_thread, (void *)sc);
|
2013-11-06 00:25:17 +00:00
|
|
|
snprintf(tname, sizeof(tname), "vtnet-%d:%d tx", pi->pi_slot,
|
|
|
|
pi->pi_func);
|
2013-04-26 05:13:48 +00:00
|
|
|
pthread_set_name_np(sc->tx_tid, tname);
|
2011-05-13 04:54:01 +00:00
|
|
|
|
|
|
|
return (0);
|
|
|
|
}
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
static int
|
|
|
|
pci_vtnet_cfgwrite(void *vsc, int offset, int size, uint32_t value)
|
2011-05-13 04:54:01 +00:00
|
|
|
{
|
2013-07-17 23:37:33 +00:00
|
|
|
struct pci_vtnet_softc *sc = vsc;
|
2011-06-07 18:35:45 +00:00
|
|
|
void *ptr;
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
if (offset < 6) {
|
|
|
|
assert(offset + size <= 6);
|
2011-05-13 04:54:01 +00:00
|
|
|
/*
|
|
|
|
* The driver is allowed to change the MAC address
|
|
|
|
*/
|
2013-07-17 23:37:33 +00:00
|
|
|
ptr = &sc->vsc_config.mac[offset];
|
|
|
|
memcpy(ptr, &value, size);
|
|
|
|
} else {
|
|
|
|
DPRINTF(("vtnet: write to readonly reg %d\n\r", offset));
|
|
|
|
return (1);
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
2013-07-17 23:37:33 +00:00
|
|
|
return (0);
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
static int
|
|
|
|
pci_vtnet_cfgread(void *vsc, int offset, int size, uint32_t *retval)
|
2011-05-13 04:54:01 +00:00
|
|
|
{
|
2013-07-17 23:37:33 +00:00
|
|
|
struct pci_vtnet_softc *sc = vsc;
|
2011-06-07 18:35:45 +00:00
|
|
|
void *ptr;
|
2013-01-30 04:30:36 +00:00
|
|
|
|
2013-07-17 23:37:33 +00:00
|
|
|
ptr = (uint8_t *)&sc->vsc_config + offset;
|
|
|
|
memcpy(retval, ptr, size);
|
|
|
|
return (0);
|
2011-05-13 04:54:01 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
struct pci_devemu pci_de_vnet = {
|
2012-10-19 18:11:17 +00:00
|
|
|
.pe_emu = "virtio-net",
|
|
|
|
.pe_init = pci_vtnet_init,
|
2013-07-17 23:37:33 +00:00
|
|
|
.pe_barwrite = vi_pci_write,
|
|
|
|
.pe_barread = vi_pci_read
|
2011-05-13 04:54:01 +00:00
|
|
|
};
|
|
|
|
PCI_EMUL_SET(pci_de_vnet);
|