Patchwork [5/7] Packet abstraction for VMWARE network devices

login
register
mail settings
Submitter Stefan Hajnoczi
Date March 25, 2013, 12:26 p.m.
Message ID <1364214373-2444-6-git-send-email-stefanha@redhat.com>
Download mbox | patch
Permalink /patch/230702/
State New
Headers show

Comments

Stefan Hajnoczi - March 25, 2013, 12:26 p.m.
From: Dmitry Fleytman <dmitry@daynix.com>

Signed-off-by: Dmitry Fleytman <dmitry@daynix.com>
Signed-off-by: Yan Vugenfirer <yan@daynix.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
---
 hw/Makefile.objs   |   1 +
 hw/vmxnet_rx_pkt.c | 187 ++++++++++++++++++
 hw/vmxnet_rx_pkt.h | 174 ++++++++++++++++
 hw/vmxnet_tx_pkt.c | 567 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 hw/vmxnet_tx_pkt.h | 148 ++++++++++++++
 5 files changed, 1077 insertions(+)
 create mode 100644 hw/vmxnet_rx_pkt.c
 create mode 100644 hw/vmxnet_rx_pkt.h
 create mode 100644 hw/vmxnet_tx_pkt.c
 create mode 100644 hw/vmxnet_tx_pkt.h

Patch

diff --git a/hw/Makefile.objs b/hw/Makefile.objs
index 11812c6..1a988ca 100644
--- a/hw/Makefile.objs
+++ b/hw/Makefile.objs
@@ -119,6 +119,7 @@  common-obj-$(CONFIG_PCNET_PCI) += pcnet-pci.o
 common-obj-$(CONFIG_PCNET_COMMON) += pcnet.o
 common-obj-$(CONFIG_E1000_PCI) += e1000.o
 common-obj-$(CONFIG_RTL8139_PCI) += rtl8139.o
+common-obj-$(CONFIG_VMXNET3_PCI) += vmxnet_tx_pkt.o vmxnet_rx_pkt.o
 
 common-obj-$(CONFIG_SMC91C111) += smc91c111.o
 common-obj-$(CONFIG_LAN9118) += lan9118.o
diff --git a/hw/vmxnet_rx_pkt.c b/hw/vmxnet_rx_pkt.c
new file mode 100644
index 0000000..a40e346
--- /dev/null
+++ b/hw/vmxnet_rx_pkt.c
@@ -0,0 +1,187 @@ 
+/*
+ * QEMU VMWARE VMXNET* paravirtual NICs - RX packets abstractions
+ *
+ * Copyright (c) 2012 Ravello Systems LTD (http://ravellosystems.com)
+ *
+ * Developed by Daynix Computing LTD (http://www.daynix.com)
+ *
+ * Authors:
+ * Dmitry Fleytman <dmitry@daynix.com>
+ * Tamir Shomer <tamirs@daynix.com>
+ * Yan Vugenfirer <yan@daynix.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "vmxnet_rx_pkt.h"
+#include "net/eth.h"
+#include "qemu-common.h"
+#include "qemu/iov.h"
+#include "net/checksum.h"
+#include "net/tap.h"
+
+/*
+ * RX packet may contain up to 2 fragments - rebuilt eth header
+ * in case of VLAN tag stripping
+ * and payload received from QEMU - in any case
+ */
+#define VMXNET_MAX_RX_PACKET_FRAGMENTS (2)
+
+struct VmxnetRxPkt {
+    struct virtio_net_hdr virt_hdr;
+    uint8_t ehdr_buf[ETH_MAX_L2_HDR_LEN];
+    struct iovec vec[VMXNET_MAX_RX_PACKET_FRAGMENTS];
+    uint16_t vec_len;
+    uint32_t tot_len;
+    uint16_t tci;
+    bool vlan_stripped;
+    bool has_virt_hdr;
+    eth_pkt_types_e packet_type;
+
+    /* Analysis results */
+    bool isip4;
+    bool isip6;
+    bool isudp;
+    bool istcp;
+};
+
+void vmxnet_rx_pkt_init(struct VmxnetRxPkt **pkt, bool has_virt_hdr)
+{
+    struct VmxnetRxPkt *p = g_malloc0(sizeof *p);
+    p->has_virt_hdr = has_virt_hdr;
+    *pkt = p;
+}
+
+void vmxnet_rx_pkt_uninit(struct VmxnetRxPkt *pkt)
+{
+    g_free(pkt);
+}
+
+struct virtio_net_hdr *vmxnet_rx_pkt_get_vhdr(struct VmxnetRxPkt *pkt)
+{
+    assert(pkt);
+    return &pkt->virt_hdr;
+}
+
+void vmxnet_rx_pkt_attach_data(struct VmxnetRxPkt *pkt, const void *data,
+                               size_t len, bool strip_vlan)
+{
+    uint16_t tci = 0;
+    uint16_t ploff;
+    assert(pkt);
+    pkt->vlan_stripped = false;
+
+    if (strip_vlan) {
+        pkt->vlan_stripped = eth_strip_vlan(data, pkt->ehdr_buf, &ploff, &tci);
+    }
+
+    if (pkt->vlan_stripped) {
+        pkt->vec[0].iov_base = pkt->ehdr_buf;
+        pkt->vec[0].iov_len = ploff - sizeof(struct vlan_header);
+        pkt->vec[1].iov_base = (uint8_t *) data + ploff;
+        pkt->vec[1].iov_len = len - ploff;
+        pkt->vec_len = 2;
+        pkt->tot_len = len - ploff + sizeof(struct eth_header);
+    } else {
+        pkt->vec[0].iov_base = (void *)data;
+        pkt->vec[0].iov_len = len;
+        pkt->vec_len = 1;
+        pkt->tot_len = len;
+    }
+
+    pkt->tci = tci;
+
+    eth_get_protocols(data, len, &pkt->isip4, &pkt->isip6,
+        &pkt->isudp, &pkt->istcp);
+}
+
+void vmxnet_rx_pkt_dump(struct VmxnetRxPkt *pkt)
+{
+#ifdef VMXNET_RX_PKT_DEBUG
+    VmxnetRxPkt *pkt = (VmxnetRxPkt *)pkt;
+    assert(pkt);
+
+    printf("RX PKT: tot_len: %d, vlan_stripped: %d, vlan_tag: %d\n",
+              pkt->tot_len, pkt->vlan_stripped, pkt->tci);
+#endif
+}
+
+void vmxnet_rx_pkt_set_packet_type(struct VmxnetRxPkt *pkt,
+    eth_pkt_types_e packet_type)
+{
+    assert(pkt);
+
+    pkt->packet_type = packet_type;
+
+}
+
+eth_pkt_types_e vmxnet_rx_pkt_get_packet_type(struct VmxnetRxPkt *pkt)
+{
+    assert(pkt);
+
+    return pkt->packet_type;
+}
+
+size_t vmxnet_rx_pkt_get_total_len(struct VmxnetRxPkt *pkt)
+{
+    assert(pkt);
+
+    return pkt->tot_len;
+}
+
+void vmxnet_rx_pkt_get_protocols(struct VmxnetRxPkt *pkt,
+                                 bool *isip4, bool *isip6,
+                                 bool *isudp, bool *istcp)
+{
+    assert(pkt);
+
+    *isip4 = pkt->isip4;
+    *isip6 = pkt->isip6;
+    *isudp = pkt->isudp;
+    *istcp = pkt->istcp;
+}
+
+struct iovec *vmxnet_rx_pkt_get_iovec(struct VmxnetRxPkt *pkt)
+{
+    assert(pkt);
+
+    return pkt->vec;
+}
+
+void vmxnet_rx_pkt_set_vhdr(struct VmxnetRxPkt *pkt,
+                            struct virtio_net_hdr *vhdr)
+{
+    assert(pkt);
+
+    memcpy(&pkt->virt_hdr, vhdr, sizeof pkt->virt_hdr);
+}
+
+bool vmxnet_rx_pkt_is_vlan_stripped(struct VmxnetRxPkt *pkt)
+{
+    assert(pkt);
+
+    return pkt->vlan_stripped;
+}
+
+bool vmxnet_rx_pkt_has_virt_hdr(struct VmxnetRxPkt *pkt)
+{
+    assert(pkt);
+
+    return pkt->has_virt_hdr;
+}
+
+uint16_t vmxnet_rx_pkt_get_num_frags(struct VmxnetRxPkt *pkt)
+{
+    assert(pkt);
+
+    return pkt->vec_len;
+}
+
+uint16_t vmxnet_rx_pkt_get_vlan_tag(struct VmxnetRxPkt *pkt)
+{
+    assert(pkt);
+
+    return pkt->tci;
+}
diff --git a/hw/vmxnet_rx_pkt.h b/hw/vmxnet_rx_pkt.h
new file mode 100644
index 0000000..6b2c60e
--- /dev/null
+++ b/hw/vmxnet_rx_pkt.h
@@ -0,0 +1,174 @@ 
+/*
+ * QEMU VMWARE VMXNET* paravirtual NICs - RX packets abstraction
+ *
+ * Copyright (c) 2012 Ravello Systems LTD (http://ravellosystems.com)
+ *
+ * Developed by Daynix Computing LTD (http://www.daynix.com)
+ *
+ * Authors:
+ * Dmitry Fleytman <dmitry@daynix.com>
+ * Tamir Shomer <tamirs@daynix.com>
+ * Yan Vugenfirer <yan@daynix.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef VMXNET_RX_PKT_H
+#define VMXNET_RX_PKT_H
+
+#include "stdint.h"
+#include "stdbool.h"
+#include "net/eth.h"
+
+/* defines to enable packet dump functions */
+/*#define VMXNET_RX_PKT_DEBUG*/
+
+struct VmxnetRxPkt;
+
+/**
+ * Clean all rx packet resources
+ *
+ * @pkt:            packet
+ *
+ */
+void vmxnet_rx_pkt_uninit(struct VmxnetRxPkt *pkt);
+
+/**
+ * Init function for rx packet functionality
+ *
+ * @pkt:            packet pointer
+ * @has_virt_hdr:   device uses virtio header
+ *
+ */
+void vmxnet_rx_pkt_init(struct VmxnetRxPkt **pkt, bool has_virt_hdr);
+
+/**
+ * returns total length of data attached to rx context
+ *
+ * @pkt:            packet
+ *
+ * Return:  nothing
+ *
+ */
+size_t vmxnet_rx_pkt_get_total_len(struct VmxnetRxPkt *pkt);
+
+/**
+ * fetches packet analysis results
+ *
+ * @pkt:            packet
+ * @isip4:          whether the packet given is IPv4
+ * @isip6:          whether the packet given is IPv6
+ * @isudp:          whether the packet given is UDP
+ * @istcp:          whether the packet given is TCP
+ *
+ */
+void vmxnet_rx_pkt_get_protocols(struct VmxnetRxPkt *pkt,
+                                 bool *isip4, bool *isip6,
+                                 bool *isudp, bool *istcp);
+
+/**
+ * returns virtio header stored in rx context
+ *
+ * @pkt:            packet
+ * @ret:            virtio header
+ *
+ */
+struct virtio_net_hdr *vmxnet_rx_pkt_get_vhdr(struct VmxnetRxPkt *pkt);
+
+/**
+ * returns packet type
+ *
+ * @pkt:            packet
+ * @ret:            packet type
+ *
+ */
+eth_pkt_types_e vmxnet_rx_pkt_get_packet_type(struct VmxnetRxPkt *pkt);
+
+/**
+ * returns vlan tag
+ *
+ * @pkt:            packet
+ * @ret:            VLAN tag
+ *
+ */
+uint16_t vmxnet_rx_pkt_get_vlan_tag(struct VmxnetRxPkt *pkt);
+
+/**
+ * tells whether vlan was stripped from the packet
+ *
+ * @pkt:            packet
+ * @ret:            VLAN stripped sign
+ *
+ */
+bool vmxnet_rx_pkt_is_vlan_stripped(struct VmxnetRxPkt *pkt);
+
+/**
+ * notifies caller if the packet has virtio header
+ *
+ * @pkt:            packet
+ * @ret:            true if packet has virtio header, false otherwize
+ *
+ */
+bool vmxnet_rx_pkt_has_virt_hdr(struct VmxnetRxPkt *pkt);
+
+/**
+ * returns number of frags attached to the packet
+ *
+ * @pkt:            packet
+ * @ret:            number of frags
+ *
+ */
+uint16_t vmxnet_rx_pkt_get_num_frags(struct VmxnetRxPkt *pkt);
+
+/**
+ * attach data to rx packet
+ *
+ * @pkt:            packet
+ * @data:           pointer to the data buffer
+ * @len:            data length
+ * @strip_vlan:     should the module strip vlan from data
+ *
+ */
+void vmxnet_rx_pkt_attach_data(struct VmxnetRxPkt *pkt, const void *data,
+    size_t len, bool strip_vlan);
+
+/**
+ * returns io vector that holds the attached data
+ *
+ * @pkt:            packet
+ * @ret:            pointer to IOVec
+ *
+ */
+struct iovec *vmxnet_rx_pkt_get_iovec(struct VmxnetRxPkt *pkt);
+
+/**
+ * prints rx packet data if debug is enabled
+ *
+ * @pkt:            packet
+ *
+ */
+void vmxnet_rx_pkt_dump(struct VmxnetRxPkt *pkt);
+
+/**
+ * copy passed vhdr data to packet context
+ *
+ * @pkt:            packet
+ * @vhdr:           VHDR buffer
+ *
+ */
+void vmxnet_rx_pkt_set_vhdr(struct VmxnetRxPkt *pkt,
+    struct virtio_net_hdr *vhdr);
+
+/**
+ * save packet type in packet context
+ *
+ * @pkt:            packet
+ * @packet_type:    the packet type
+ *
+ */
+void vmxnet_rx_pkt_set_packet_type(struct VmxnetRxPkt *pkt,
+    eth_pkt_types_e packet_type);
+
+#endif
diff --git a/hw/vmxnet_tx_pkt.c b/hw/vmxnet_tx_pkt.c
new file mode 100644
index 0000000..b1e795b
--- /dev/null
+++ b/hw/vmxnet_tx_pkt.c
@@ -0,0 +1,567 @@ 
+/*
+ * QEMU VMWARE VMXNET* paravirtual NICs - TX packets abstractions
+ *
+ * Copyright (c) 2012 Ravello Systems LTD (http://ravellosystems.com)
+ *
+ * Developed by Daynix Computing LTD (http://www.daynix.com)
+ *
+ * Authors:
+ * Dmitry Fleytman <dmitry@daynix.com>
+ * Tamir Shomer <tamirs@daynix.com>
+ * Yan Vugenfirer <yan@daynix.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#include "vmxnet_tx_pkt.h"
+#include "net/eth.h"
+#include "qemu-common.h"
+#include "qemu/iov.h"
+#include "net/checksum.h"
+#include "net/tap.h"
+#include "net/net.h"
+#include "exec/cpu-common.h"
+
+enum {
+    VMXNET_TX_PKT_VHDR_FRAG = 0,
+    VMXNET_TX_PKT_L2HDR_FRAG,
+    VMXNET_TX_PKT_L3HDR_FRAG,
+    VMXNET_TX_PKT_PL_START_FRAG
+};
+
+/* TX packet private context */
+struct VmxnetTxPkt {
+    struct virtio_net_hdr virt_hdr;
+    bool has_virt_hdr;
+
+    struct iovec *raw;
+    uint32_t raw_frags;
+    uint32_t max_raw_frags;
+
+    struct iovec *vec;
+
+    uint8_t l2_hdr[ETH_MAX_L2_HDR_LEN];
+
+    uint32_t payload_len;
+
+    uint32_t payload_frags;
+    uint32_t max_payload_frags;
+
+    uint16_t hdr_len;
+    eth_pkt_types_e packet_type;
+    uint8_t l4proto;
+};
+
+void vmxnet_tx_pkt_init(struct VmxnetTxPkt **pkt, uint32_t max_frags,
+    bool has_virt_hdr)
+{
+    struct VmxnetTxPkt *p = g_malloc0(sizeof *p);
+
+    p->vec = g_malloc((sizeof *p->vec) *
+        (max_frags + VMXNET_TX_PKT_PL_START_FRAG));
+
+    p->raw = g_malloc((sizeof *p->raw) * max_frags);
+
+    p->max_payload_frags = max_frags;
+    p->max_raw_frags = max_frags;
+    p->has_virt_hdr = has_virt_hdr;
+    p->vec[VMXNET_TX_PKT_VHDR_FRAG].iov_base = &p->virt_hdr;
+    p->vec[VMXNET_TX_PKT_VHDR_FRAG].iov_len =
+        p->has_virt_hdr ? sizeof p->virt_hdr : 0;
+    p->vec[VMXNET_TX_PKT_L2HDR_FRAG].iov_base = &p->l2_hdr;
+    p->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_base = NULL;
+    p->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_len = 0;
+
+    *pkt = p;
+}
+
+void vmxnet_tx_pkt_uninit(struct VmxnetTxPkt *pkt)
+{
+    if (pkt) {
+        g_free(pkt->vec);
+        g_free(pkt->raw);
+        g_free(pkt);
+    }
+}
+
+void vmxnet_tx_pkt_update_ip_checksums(struct VmxnetTxPkt *pkt)
+{
+    uint16_t csum;
+    uint32_t ph_raw_csum;
+    assert(pkt);
+    uint8_t gso_type = pkt->virt_hdr.gso_type & ~VIRTIO_NET_HDR_GSO_ECN;
+    struct ip_header *ip_hdr;
+
+    if (VIRTIO_NET_HDR_GSO_TCPV4 != gso_type &&
+        VIRTIO_NET_HDR_GSO_UDP != gso_type) {
+        return;
+    }
+
+    ip_hdr = pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_base;
+
+    if (pkt->payload_len + pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_len >
+        ETH_MAX_IP_DGRAM_LEN) {
+        return;
+    }
+
+    ip_hdr->ip_len = cpu_to_be16(pkt->payload_len +
+        pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_len);
+
+    /* Calculate IP header checksum                    */
+    ip_hdr->ip_sum = 0;
+    csum = net_raw_checksum((uint8_t *)ip_hdr,
+        pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_len);
+    ip_hdr->ip_sum = cpu_to_be16(csum);
+
+    /* Calculate IP pseudo header checksum             */
+    ph_raw_csum = eth_calc_pseudo_hdr_csum(ip_hdr, pkt->payload_len);
+    csum = cpu_to_be16(~net_checksum_finish(ph_raw_csum));
+    iov_from_buf(&pkt->vec[VMXNET_TX_PKT_PL_START_FRAG], pkt->payload_frags,
+                 pkt->virt_hdr.csum_offset, &csum, sizeof(csum));
+}
+
+static void vmxnet_tx_pkt_calculate_hdr_len(struct VmxnetTxPkt *pkt)
+{
+    pkt->hdr_len = pkt->vec[VMXNET_TX_PKT_L2HDR_FRAG].iov_len +
+        pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_len;
+}
+
+static bool vmxnet_tx_pkt_parse_headers(struct VmxnetTxPkt *pkt)
+{
+    struct iovec *l2_hdr, *l3_hdr;
+    size_t bytes_read;
+    size_t full_ip6hdr_len;
+    uint16_t l3_proto;
+
+    assert(pkt);
+
+    l2_hdr = &pkt->vec[VMXNET_TX_PKT_L2HDR_FRAG];
+    l3_hdr = &pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG];
+
+    bytes_read = iov_to_buf(pkt->raw, pkt->raw_frags, 0, l2_hdr->iov_base,
+                            ETH_MAX_L2_HDR_LEN);
+    if (bytes_read < ETH_MAX_L2_HDR_LEN) {
+        l2_hdr->iov_len = 0;
+        return false;
+    } else {
+        l2_hdr->iov_len = eth_get_l2_hdr_length(l2_hdr->iov_base);
+    }
+
+    l3_proto = eth_get_l3_proto(l2_hdr->iov_base, l2_hdr->iov_len);
+
+    switch (l3_proto) {
+    case ETH_P_IP:
+        l3_hdr->iov_base = g_malloc(ETH_MAX_IP4_HDR_LEN);
+
+        bytes_read = iov_to_buf(pkt->raw, pkt->raw_frags, l2_hdr->iov_len,
+                                l3_hdr->iov_base, sizeof(struct ip_header));
+
+        if (bytes_read < sizeof(struct ip_header)) {
+            l3_hdr->iov_len = 0;
+            return false;
+        }
+
+        l3_hdr->iov_len = IP_HDR_GET_LEN(l3_hdr->iov_base);
+        pkt->l4proto = ((struct ip_header *) l3_hdr->iov_base)->ip_p;
+
+        /* copy optional IPv4 header data */
+        bytes_read = iov_to_buf(pkt->raw, pkt->raw_frags,
+                                l2_hdr->iov_len + sizeof(struct ip_header),
+                                l3_hdr->iov_base + sizeof(struct ip_header),
+                                l3_hdr->iov_len - sizeof(struct ip_header));
+        if (bytes_read < l3_hdr->iov_len - sizeof(struct ip_header)) {
+            l3_hdr->iov_len = 0;
+            return false;
+        }
+        break;
+
+    case ETH_P_IPV6:
+        if (!eth_parse_ipv6_hdr(pkt->raw, pkt->raw_frags, l2_hdr->iov_len,
+                               &pkt->l4proto, &full_ip6hdr_len)) {
+            l3_hdr->iov_len = 0;
+            return false;
+        }
+
+        l3_hdr->iov_base = g_malloc(full_ip6hdr_len);
+
+        bytes_read = iov_to_buf(pkt->raw, pkt->raw_frags, l2_hdr->iov_len,
+                                l3_hdr->iov_base, full_ip6hdr_len);
+
+        if (bytes_read < full_ip6hdr_len) {
+            l3_hdr->iov_len = 0;
+            return false;
+        } else {
+            l3_hdr->iov_len = full_ip6hdr_len;
+        }
+        break;
+
+    default:
+        l3_hdr->iov_len = 0;
+        break;
+    }
+
+    vmxnet_tx_pkt_calculate_hdr_len(pkt);
+    pkt->packet_type = get_eth_packet_type(l2_hdr->iov_base);
+    return true;
+}
+
+static bool vmxnet_tx_pkt_rebuild_payload(struct VmxnetTxPkt *pkt)
+{
+    size_t payload_len = iov_size(pkt->raw, pkt->raw_frags) - pkt->hdr_len;
+
+    pkt->payload_frags = iov_copy(&pkt->vec[VMXNET_TX_PKT_PL_START_FRAG],
+                                pkt->max_payload_frags,
+                                pkt->raw, pkt->raw_frags,
+                                pkt->hdr_len, payload_len);
+
+    if (pkt->payload_frags != (uint32_t) -1) {
+        pkt->payload_len = payload_len;
+        return true;
+    } else {
+        return false;
+    }
+}
+
+bool vmxnet_tx_pkt_parse(struct VmxnetTxPkt *pkt)
+{
+    return vmxnet_tx_pkt_parse_headers(pkt) &&
+           vmxnet_tx_pkt_rebuild_payload(pkt);
+}
+
+struct virtio_net_hdr *vmxnet_tx_pkt_get_vhdr(struct VmxnetTxPkt *pkt)
+{
+    assert(pkt);
+    return &pkt->virt_hdr;
+}
+
+static uint8_t vmxnet_tx_pkt_get_gso_type(struct VmxnetTxPkt *pkt,
+                                          bool tso_enable)
+{
+    uint8_t rc = VIRTIO_NET_HDR_GSO_NONE;
+    uint16_t l3_proto;
+
+    l3_proto = eth_get_l3_proto(pkt->vec[VMXNET_TX_PKT_L2HDR_FRAG].iov_base,
+        pkt->vec[VMXNET_TX_PKT_L2HDR_FRAG].iov_len);
+
+    if (!tso_enable) {
+        goto func_exit;
+    }
+
+    rc = eth_get_gso_type(l3_proto, pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_base,
+                          pkt->l4proto);
+
+func_exit:
+    return rc;
+}
+
+void vmxnet_tx_pkt_build_vheader(struct VmxnetTxPkt *pkt, bool tso_enable,
+    bool csum_enable, uint32_t gso_size)
+{
+    struct tcp_hdr l4hdr;
+    assert(pkt);
+
+    /* csum has to be enabled if tso is. */
+    assert(csum_enable || !tso_enable);
+
+    pkt->virt_hdr.gso_type = vmxnet_tx_pkt_get_gso_type(pkt, tso_enable);
+
+    switch (pkt->virt_hdr.gso_type & ~VIRTIO_NET_HDR_GSO_ECN) {
+    case VIRTIO_NET_HDR_GSO_NONE:
+        pkt->virt_hdr.hdr_len = 0;
+        pkt->virt_hdr.gso_size = 0;
+        break;
+
+    case VIRTIO_NET_HDR_GSO_UDP:
+        pkt->virt_hdr.gso_size = IP_FRAG_ALIGN_SIZE(gso_size);
+        pkt->virt_hdr.hdr_len = pkt->hdr_len + sizeof(struct udp_header);
+        break;
+
+    case VIRTIO_NET_HDR_GSO_TCPV4:
+    case VIRTIO_NET_HDR_GSO_TCPV6:
+        iov_to_buf(&pkt->vec[VMXNET_TX_PKT_PL_START_FRAG], pkt->payload_frags,
+                   0, &l4hdr, sizeof(l4hdr));
+        pkt->virt_hdr.hdr_len = pkt->hdr_len + l4hdr.th_off * sizeof(uint32_t);
+        pkt->virt_hdr.gso_size = IP_FRAG_ALIGN_SIZE(gso_size);
+        break;
+
+    default:
+        assert(false);
+    }
+
+    if (csum_enable) {
+        switch (pkt->l4proto) {
+        case IP_PROTO_TCP:
+            pkt->virt_hdr.flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
+            pkt->virt_hdr.csum_start = pkt->hdr_len;
+            pkt->virt_hdr.csum_offset = offsetof(struct tcp_hdr, th_sum);
+            break;
+        case IP_PROTO_UDP:
+            pkt->virt_hdr.flags = VIRTIO_NET_HDR_F_NEEDS_CSUM;
+            pkt->virt_hdr.csum_start = pkt->hdr_len;
+            pkt->virt_hdr.csum_offset = offsetof(struct udp_hdr, uh_sum);
+            break;
+        default:
+            break;
+        }
+    }
+}
+
+void vmxnet_tx_pkt_setup_vlan_header(struct VmxnetTxPkt *pkt, uint16_t vlan)
+{
+    bool is_new;
+    assert(pkt);
+
+    eth_setup_vlan_headers(pkt->vec[VMXNET_TX_PKT_L2HDR_FRAG].iov_base,
+        vlan, &is_new);
+
+    /* update l2hdrlen */
+    if (is_new) {
+        pkt->hdr_len += sizeof(struct vlan_header);
+        pkt->vec[VMXNET_TX_PKT_L2HDR_FRAG].iov_len +=
+            sizeof(struct vlan_header);
+    }
+}
+
+bool vmxnet_tx_pkt_add_raw_fragment(struct VmxnetTxPkt *pkt, hwaddr pa,
+    size_t len)
+{
+    hwaddr mapped_len = 0;
+    struct iovec *ventry;
+    assert(pkt);
+    assert(pkt->max_raw_frags > pkt->raw_frags);
+
+    if (!len) {
+        return true;
+     }
+
+    ventry = &pkt->raw[pkt->raw_frags];
+    mapped_len = len;
+
+    ventry->iov_base = cpu_physical_memory_map(pa, &mapped_len, false);
+    ventry->iov_len = mapped_len;
+    pkt->raw_frags += !!ventry->iov_base;
+
+    if ((ventry->iov_base == NULL) || (len != mapped_len)) {
+        return false;
+    }
+
+    return true;
+}
+
+eth_pkt_types_e vmxnet_tx_pkt_get_packet_type(struct VmxnetTxPkt *pkt)
+{
+    assert(pkt);
+
+    return pkt->packet_type;
+}
+
+size_t vmxnet_tx_pkt_get_total_len(struct VmxnetTxPkt *pkt)
+{
+    assert(pkt);
+
+    return pkt->hdr_len + pkt->payload_len;
+}
+
+void vmxnet_tx_pkt_dump(struct VmxnetTxPkt *pkt)
+{
+#ifdef VMXNET_TX_PKT_DEBUG
+    assert(pkt);
+
+    printf("TX PKT: hdr_len: %d, pkt_type: 0x%X, l2hdr_len: %lu, "
+        "l3hdr_len: %lu, payload_len: %u\n", pkt->hdr_len, pkt->packet_type,
+        pkt->vec[VMXNET_TX_PKT_L2HDR_FRAG].iov_len,
+        pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_len, pkt->payload_len);
+#endif
+}
+
+void vmxnet_tx_pkt_reset(struct VmxnetTxPkt *pkt)
+{
+    int i;
+
+    /* no assert, as reset can be called before tx_pkt_init */
+    if (!pkt) {
+        return;
+    }
+
+    memset(&pkt->virt_hdr, 0, sizeof(pkt->virt_hdr));
+
+    g_free(pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_base);
+    pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_base = NULL;
+
+    assert(pkt->vec);
+    for (i = VMXNET_TX_PKT_L2HDR_FRAG;
+         i < pkt->payload_frags + VMXNET_TX_PKT_PL_START_FRAG; i++) {
+        pkt->vec[i].iov_len = 0;
+    }
+    pkt->payload_len = 0;
+    pkt->payload_frags = 0;
+
+    assert(pkt->raw);
+    for (i = 0; i < pkt->raw_frags; i++) {
+        assert(pkt->raw[i].iov_base);
+        cpu_physical_memory_unmap(pkt->raw[i].iov_base, pkt->raw[i].iov_len,
+                                  false, pkt->raw[i].iov_len);
+        pkt->raw[i].iov_len = 0;
+    }
+    pkt->raw_frags = 0;
+
+    pkt->hdr_len = 0;
+    pkt->packet_type = 0;
+    pkt->l4proto = 0;
+}
+
+static void vmxnet_tx_pkt_do_sw_csum(struct VmxnetTxPkt *pkt)
+{
+    struct iovec *iov = &pkt->vec[VMXNET_TX_PKT_L2HDR_FRAG];
+    uint32_t csum_cntr;
+    uint16_t csum = 0;
+    /* num of iovec without vhdr */
+    uint32_t iov_len = pkt->payload_frags + VMXNET_TX_PKT_PL_START_FRAG - 1;
+    uint16_t csl;
+    struct ip_header *iphdr;
+    size_t csum_offset = pkt->virt_hdr.csum_start + pkt->virt_hdr.csum_offset;
+
+    /* Put zero to checksum field */
+    iov_from_buf(iov, iov_len, csum_offset, &csum, sizeof csum);
+
+    /* Calculate L4 TCP/UDP checksum */
+    csl = pkt->payload_len;
+
+    /* data checksum */
+    csum_cntr =
+        net_checksum_add_iov(iov, iov_len, pkt->virt_hdr.csum_start, csl);
+    /* add pseudo header to csum */
+    iphdr = pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_base;
+    csum_cntr += eth_calc_pseudo_hdr_csum(iphdr, csl);
+
+    /* Put the checksum obtained into the packet */
+    csum = cpu_to_be16(net_checksum_finish(csum_cntr));
+    iov_from_buf(iov, iov_len, csum_offset, &csum, sizeof csum);
+}
+
+enum {
+    VMXNET_TX_PKT_FRAGMENT_L2_HDR_POS = 0,
+    VMXNET_TX_PKT_FRAGMENT_L3_HDR_POS,
+    VMXNET_TX_PKT_FRAGMENT_HEADER_NUM
+};
+
+#define VMXNET_MAX_FRAG_SG_LIST (64)
+
+static size_t vmxnet_tx_pkt_fetch_fragment(struct VmxnetTxPkt *pkt,
+    int *src_idx, size_t *src_offset, struct iovec *dst, int *dst_idx)
+{
+    size_t fetched = 0;
+    struct iovec *src = pkt->vec;
+
+    *dst_idx = VMXNET_TX_PKT_FRAGMENT_HEADER_NUM;
+
+    while (fetched < pkt->virt_hdr.gso_size) {
+
+        /* no more place in fragment iov */
+        if (*dst_idx == VMXNET_MAX_FRAG_SG_LIST) {
+            break;
+        }
+
+        /* no more data in iovec */
+        if (*src_idx == (pkt->payload_frags + VMXNET_TX_PKT_PL_START_FRAG)) {
+            break;
+        }
+
+
+        dst[*dst_idx].iov_base = src[*src_idx].iov_base + *src_offset;
+        dst[*dst_idx].iov_len = MIN(src[*src_idx].iov_len - *src_offset,
+            pkt->virt_hdr.gso_size - fetched);
+
+        *src_offset += dst[*dst_idx].iov_len;
+        fetched += dst[*dst_idx].iov_len;
+
+        if (*src_offset == src[*src_idx].iov_len) {
+            *src_offset = 0;
+            (*src_idx)++;
+        }
+
+        (*dst_idx)++;
+    }
+
+    return fetched;
+}
+
+static bool vmxnet_tx_pkt_do_sw_fragmentation(struct VmxnetTxPkt *pkt,
+    NetClientState *nc)
+{
+    struct iovec fragment[VMXNET_MAX_FRAG_SG_LIST];
+    size_t fragment_len = 0;
+    bool more_frags = false;
+
+    /* some pointers for shorter code */
+    void *l2_iov_base, *l3_iov_base;
+    size_t l2_iov_len, l3_iov_len;
+    int src_idx =  VMXNET_TX_PKT_PL_START_FRAG, dst_idx;
+    size_t src_offset = 0;
+    size_t fragment_offset = 0;
+
+    l2_iov_base = pkt->vec[VMXNET_TX_PKT_L2HDR_FRAG].iov_base;
+    l2_iov_len = pkt->vec[VMXNET_TX_PKT_L2HDR_FRAG].iov_len;
+    l3_iov_base = pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_base;
+    l3_iov_len = pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_len;
+
+    /* Copy headers */
+    fragment[VMXNET_TX_PKT_FRAGMENT_L2_HDR_POS].iov_base = l2_iov_base;
+    fragment[VMXNET_TX_PKT_FRAGMENT_L2_HDR_POS].iov_len = l2_iov_len;
+    fragment[VMXNET_TX_PKT_FRAGMENT_L3_HDR_POS].iov_base = l3_iov_base;
+    fragment[VMXNET_TX_PKT_FRAGMENT_L3_HDR_POS].iov_len = l3_iov_len;
+
+
+    /* Put as much data as possible and send */
+    do {
+        fragment_len = vmxnet_tx_pkt_fetch_fragment(pkt, &src_idx, &src_offset,
+            fragment, &dst_idx);
+
+        more_frags = (fragment_offset + fragment_len < pkt->payload_len);
+
+        eth_setup_ip4_fragmentation(l2_iov_base, l2_iov_len, l3_iov_base,
+            l3_iov_len, fragment_len, fragment_offset, more_frags);
+
+        eth_fix_ip4_checksum(l3_iov_base, l3_iov_len);
+
+        qemu_sendv_packet(nc, fragment, dst_idx);
+
+        fragment_offset += fragment_len;
+
+    } while (more_frags);
+
+    return true;
+}
+
+bool vmxnet_tx_pkt_send(struct VmxnetTxPkt *pkt, NetClientState *nc)
+{
+    assert(pkt);
+
+    if (!pkt->has_virt_hdr &&
+        pkt->virt_hdr.flags & VIRTIO_NET_HDR_F_NEEDS_CSUM) {
+        vmxnet_tx_pkt_do_sw_csum(pkt);
+    }
+
+    /*
+     * Since underlying infrastructure does not support IP datagrams longer
+     * than 64K we should drop such packets and don't even try to send
+     */
+    if (VIRTIO_NET_HDR_GSO_NONE != pkt->virt_hdr.gso_type) {
+        if (pkt->payload_len >
+            ETH_MAX_IP_DGRAM_LEN -
+            pkt->vec[VMXNET_TX_PKT_L3HDR_FRAG].iov_len) {
+            return false;
+        }
+    }
+
+    if (pkt->has_virt_hdr ||
+        pkt->virt_hdr.gso_type == VIRTIO_NET_HDR_GSO_NONE) {
+        qemu_sendv_packet(nc, pkt->vec,
+            pkt->payload_frags + VMXNET_TX_PKT_PL_START_FRAG);
+        return true;
+    }
+
+    return vmxnet_tx_pkt_do_sw_fragmentation(pkt, nc);
+}
diff --git a/hw/vmxnet_tx_pkt.h b/hw/vmxnet_tx_pkt.h
new file mode 100644
index 0000000..57121a6
--- /dev/null
+++ b/hw/vmxnet_tx_pkt.h
@@ -0,0 +1,148 @@ 
+/*
+ * QEMU VMWARE VMXNET* paravirtual NICs - TX packets abstraction
+ *
+ * Copyright (c) 2012 Ravello Systems LTD (http://ravellosystems.com)
+ *
+ * Developed by Daynix Computing LTD (http://www.daynix.com)
+ *
+ * Authors:
+ * Dmitry Fleytman <dmitry@daynix.com>
+ * Tamir Shomer <tamirs@daynix.com>
+ * Yan Vugenfirer <yan@daynix.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2 or later.
+ * See the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef VMXNET_TX_PKT_H
+#define VMXNET_TX_PKT_H
+
+#include "stdint.h"
+#include "stdbool.h"
+#include "net/eth.h"
+#include "exec/hwaddr.h"
+
+/* define to enable packet dump functions */
+/*#define VMXNET_TX_PKT_DEBUG*/
+
+struct VmxnetTxPkt;
+
+/**
+ * Init function for tx packet functionality
+ *
+ * @pkt:            packet pointer
+ * @max_frags:      max tx ip fragments
+ * @has_virt_hdr:   device uses virtio header.
+ */
+void vmxnet_tx_pkt_init(struct VmxnetTxPkt **pkt, uint32_t max_frags,
+    bool has_virt_hdr);
+
+/**
+ * Clean all tx packet resources.
+ *
+ * @pkt:            packet.
+ */
+void vmxnet_tx_pkt_uninit(struct VmxnetTxPkt *pkt);
+
+/**
+ * get virtio header
+ *
+ * @pkt:            packet
+ * @ret:            virtio header
+ */
+struct virtio_net_hdr *vmxnet_tx_pkt_get_vhdr(struct VmxnetTxPkt *pkt);
+
+/**
+ * build virtio header (will be stored in module context)
+ *
+ * @pkt:            packet
+ * @tso_enable:     TSO enabled
+ * @csum_enable:    CSO enabled
+ * @gso_size:       MSS size for TSO
+ *
+ */
+void vmxnet_tx_pkt_build_vheader(struct VmxnetTxPkt *pkt, bool tso_enable,
+    bool csum_enable, uint32_t gso_size);
+
+/**
+ * updates vlan tag, and adds vlan header in case it is missing
+ *
+ * @pkt:            packet
+ * @vlan:           VLAN tag
+ *
+ */
+void vmxnet_tx_pkt_setup_vlan_header(struct VmxnetTxPkt *pkt, uint16_t vlan);
+
+/**
+ * populate data fragment into pkt context.
+ *
+ * @pkt:            packet
+ * @pa:             physical address of fragment
+ * @len:            length of fragment
+ *
+ */
+bool vmxnet_tx_pkt_add_raw_fragment(struct VmxnetTxPkt *pkt, hwaddr pa,
+    size_t len);
+
+/**
+ * fix ip header fields and calculate checksums needed.
+ *
+ * @pkt:            packet
+ *
+ */
+void vmxnet_tx_pkt_update_ip_checksums(struct VmxnetTxPkt *pkt);
+
+/**
+ * get length of all populated data.
+ *
+ * @pkt:            packet
+ * @ret:            total data length
+ *
+ */
+size_t vmxnet_tx_pkt_get_total_len(struct VmxnetTxPkt *pkt);
+
+/**
+ * get packet type
+ *
+ * @pkt:            packet
+ * @ret:            packet type
+ *
+ */
+eth_pkt_types_e vmxnet_tx_pkt_get_packet_type(struct VmxnetTxPkt *pkt);
+
+/**
+ * prints packet data if debug is enabled
+ *
+ * @pkt:            packet
+ *
+ */
+void vmxnet_tx_pkt_dump(struct VmxnetTxPkt *pkt);
+
+/**
+ * reset tx packet private context (needed to be called between packets)
+ *
+ * @pkt:            packet
+ *
+ */
+void vmxnet_tx_pkt_reset(struct VmxnetTxPkt *pkt);
+
+/**
+ * Send packet to qemu. handles sw offloads if vhdr is not supported.
+ *
+ * @pkt:            packet
+ * @nc:             NetClientState
+ * @ret:            operation result
+ *
+ */
+bool vmxnet_tx_pkt_send(struct VmxnetTxPkt *pkt, NetClientState *nc);
+
+/**
+ * parse raw packet data and analyze offload requirements.
+ *
+ * @pkt:            packet
+ *
+ */
+bool vmxnet_tx_pkt_parse(struct VmxnetTxPkt *pkt);
+
+#endif