Initial API changes to ipv4 and freq
Change-Id: I5a0e3e97ab4df4a6455169e23d814d4db330889d
Signed-off-by: Thomas Wu <wthomas@codeaurora.org>
diff --git a/nss_ipv4.c b/nss_ipv4.c
new file mode 100755
index 0000000..b561904
--- /dev/null
+++ b/nss_ipv4.c
@@ -0,0 +1,271 @@
+/*
+ **************************************************************************
+ * Copyright (c) 2013, The Linux Foundation. All rights reserved.
+ * Permission to use, copy, modify, and/or distribute this software for
+ * any purpose with or without fee is hereby granted, provided that the
+ * above copyright notice and this permission notice appear in all copies.
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT
+ * OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ **************************************************************************
+ */
+
+/*
+ * nss_tx_rx_ipv4.c
+ * NSS IPv4 APIs
+ */
+#include <linux/ppp_channel.h>
+#include "nss_tx_rx_common.h"
+#include "nss_ipv4.h"
+
+extern void nss_rx_metadata_ipv4_rule_establish(struct nss_ctx_instance *nss_ctx, struct nss_ipv4_rule_establish *nire);
+extern void nss_rx_ipv4_sync(struct nss_ctx_instance *nss_ctx, struct nss_ipv4_conn_sync *nirs);
+
+/*
+ * nss_ipv4_driver_update()
+ * Update driver specific information from the messsage.
+ */
+#if 0
+static void nss_ipv4_driver_update(void)
+{
+ /*
+ * Update statistics maintained by NSS driver
+ */
+ spin_lock_bh(&nss_top->stats_lock);
+
+ nss_top->stats_ipv4[NSS_STATS_IPV4_ACCELERATED_RX_PKTS] += nirs->flow_rx_packet_count + nirs->return_rx_packet_count;
+ nss_top->stats_ipv4[NSS_STATS_IPV4_ACCELERATED_RX_BYTES] += nirs->flow_rx_byte_count + nirs->return_rx_byte_count;
+ nss_top->stats_ipv4[NSS_STATS_IPV4_ACCELERATED_TX_PKTS] += nirs->flow_tx_packet_count + nirs->return_tx_packet_count;
+ nss_top->stats_ipv4[NSS_STATS_IPV4_ACCELERATED_TX_BYTES] += nirs->flow_tx_byte_count + nirs->return_tx_byte_count;
+
+ /*
+ * Update the PPPoE interface stats, if there is any PPPoE session on the interfaces.
+ */
+ if (nirs->flow_pppoe_session_id) {
+ pppoe_dev = ppp_session_to_netdev(nirs->flow_pppoe_session_id, (uint8_t *)nirs->flow_pppoe_remote_mac);
+ if (pppoe_dev) {
+ ppp_update_stats(pppoe_dev, nirs->flow_rx_packet_count, nirs->flow_rx_byte_count,
+ nirs->flow_tx_packet_count, nirs->flow_tx_byte_count);
+ dev_put(pppoe_dev);
+ }
+ }
+
+ if (nirs->return_pppoe_session_id) {
+ pppoe_dev = ppp_session_to_netdev(nirs->return_pppoe_session_id, (uint8_t *)nirs->return_pppoe_remote_mac);
+ if (pppoe_dev) {
+ ppp_update_stats(pppoe_dev, nirs->return_rx_packet_count, nirs->return_rx_byte_count,
+ nirs->return_tx_packet_count, nirs->return_tx_byte_count);
+ dev_put(pppoe_dev);
+ }
+ }
+
+ /*
+ * TODO: Update per dev accelerated statistics
+ */
+ spin_unlock_bh(&nss_top->stats_lock);
+}
+#endif
+
+/*
+ * nss_ipv4_handler()
+ * Handle NSS -> HLOS messages for IPv4 bridge/route
+ */
+static void nss_ipv4_handler(struct nss_ctx_instance *nss_ctx, struct nss_cmn_msg *ncm, __attribute__((unused))void *app_data)
+{
+ struct nss_ipv4_msg *nim = (struct nss_ipv4_msg *)ncm;
+
+/*
+ * TODO: Turn back on for new APIs
+ */
+// nss_ipv4_msg_callback_t cb;
+
+ BUG_ON(ncm->interface != NSS_IPV4_RX_INTERFACE);
+
+ /*
+ * Sanity check the message type
+ */
+ if (ncm->type > NSS_IPV4_MAX_MSG_TYPES) {
+ nss_warning("%p: message type out of range: %d", nss_ctx, ncm->type);
+ return;
+ }
+
+ if (ncm->len > sizeof(struct nss_ipv4_msg)) {
+ nss_warning("%p: tx request for another interface: %d", nss_ctx, ncm->interface);
+ return;
+ }
+
+
+ /*
+ * Update the callback and app_data for NOTIFY messages, IPv4 sends all notify messages
+ * to the same callback/app_data.
+ */
+ if (nim->cm.response == NSS_CMM_RESPONSE_NOTIFY) {
+ ncm->cb = (uint32_t)nss_ctx->nss_top->ipv4_callback;
+ //ncm->app_data = nss_ctx->nss_top->ipv4_app_data;
+ }
+
+ /*
+ * Log failures
+ */
+ nss_core_log_msg_failures(nss_ctx, ncm);
+
+ /*
+ * Handle deprecated messages. Eventually these messages should be removed.
+ */
+ switch (nim->cm.type) {
+ case NSS_IPV4_RX_ESTABLISH_RULE_MSG:
+ return nss_rx_metadata_ipv4_rule_establish(nss_ctx, &nim->msg.rule_establish);
+ break;
+
+ case NSS_IPV4_RX_CONN_STATS_SYNC_MSG:
+ return nss_rx_ipv4_sync(nss_ctx, &nim->msg.conn_stats);
+ break;
+ }
+
+ /*
+ * Local driver updates for ipv4.
+ */
+#if 0
+ /*
+ * Do we have a callback?
+ */
+ if (!ncm->cb) {
+ return;
+ }
+
+ /*
+ * Callback
+ */
+ cb = (nss_ipv4_msg_callback_t)ncm->cb;
+ cb((void *)ncm->app_data, nim);
+#endif
+}
+
+/*
+ * nss_ipv4_tx()
+ * Transmit an ipv4 message to the FW.
+ */
+nss_tx_status_t nss_ipv4_tx(struct nss_ctx_instance *nss_ctx, struct nss_ipv4_msg *nim)
+{
+ struct nss_ipv4_msg *nim2;
+ struct nss_cmn_msg *ncm = &nim->cm;
+ struct sk_buff *nbuf;
+ int32_t status;
+
+ NSS_VERIFY_CTX_MAGIC(nss_ctx);
+ if (unlikely(nss_ctx->state != NSS_CORE_STATE_INITIALIZED)) {
+ nss_warning("%p: ipv4 msg dropped as core not ready", nss_ctx);
+ return NSS_TX_FAILURE_NOT_READY;
+ }
+
+ /*
+ * Sanity check the message
+ */
+ if (ncm->interface != NSS_IPV4_RX_INTERFACE) {
+ nss_warning("%p: tx request for another interface: %d", nss_ctx, ncm->interface);
+ return NSS_TX_FAILURE;
+ }
+
+ if (ncm->type > NSS_IPV4_MAX_MSG_TYPES) {
+ nss_warning("%p: message type out of range: %d", nss_ctx, ncm->type);
+ return NSS_TX_FAILURE;
+ }
+
+ if (ncm->len > sizeof(struct nss_ipv4_msg)) {
+ nss_warning("%p: tx request for another interface: %d", nss_ctx, ncm->interface);
+ return NSS_TX_FAILURE;
+ }
+
+ nbuf = dev_alloc_skb(NSS_NBUF_PAYLOAD_SIZE);
+ if (unlikely(!nbuf)) {
+ spin_lock_bh(&nss_ctx->nss_top->stats_lock);
+ nss_ctx->nss_top->stats_drv[NSS_STATS_DRV_NBUF_ALLOC_FAILS]++;
+ spin_unlock_bh(&nss_ctx->nss_top->stats_lock);
+ nss_warning("%p: msg dropped as command allocation failed", nss_ctx);
+ return NSS_TX_FAILURE;
+ }
+
+ /*
+ * Copy the message to our skb.
+ */
+ nim2 = (struct nss_ipv4_msg *)skb_put(nbuf, sizeof(struct nss_ipv4_msg));
+ memcpy(nim2, nim, sizeof(struct nss_ipv4_msg));
+
+ status = nss_core_send_buffer(nss_ctx, 0, nbuf, NSS_IF_CMD_QUEUE, H2N_BUFFER_CTRL, 0);
+ if (status != NSS_CORE_STATUS_SUCCESS) {
+ dev_kfree_skb_any(nbuf);
+ nss_warning("%p: Unable to enqueue 'Destroy IPv4' rule\n", nss_ctx);
+ return NSS_TX_FAILURE;
+ }
+
+ nss_hal_send_interrupt(nss_ctx->nmap, nss_ctx->h2n_desc_rings[NSS_IF_CMD_QUEUE].desc_ring.int_bit,
+ NSS_REGS_H2N_INTR_STATUS_DATA_COMMAND_QUEUE);
+
+ NSS_PKT_STATS_INCREMENT(nss_ctx, &nss_ctx->nss_top->stats_drv[NSS_STATS_DRV_TX_CMD_REQ]);
+ return NSS_TX_SUCCESS;
+}
+
+/*
+ **********************************
+ Register/Unregister/Miscellaneous APIs
+ **********************************
+ */
+
+/*
+ * nss_ipv4_notify_register()
+ * Register to received IPv4 events.
+ *
+ * NOTE: Do we want to pass an nss_ctx here so that we can register for ipv4 on any core?
+ */
+struct nss_ctx_instance *nss_ipv4_notify_register(nss_ipv4_msg_callback_t cb, void *app_data)
+{
+ /*
+ * TODO: We need to have a new array in support of the new API
+ * TODO: If we use a per-context array, we would move the array into nss_ctx based.
+ */
+ nss_top_main.ipv4_callback = cb;
+ // nss_top_main.ipv4_app_data = app_data;
+ return &nss_top_main.nss[nss_top_main.ipv4_handler_id];
+}
+
+/*
+ * nss_ipv4_notify_unregister()
+ * Unregister to received IPv4 events.
+ *
+ * NOTE: Do we want to pass an nss_ctx here so that we can register for ipv4 on any core?
+ */
+void nss_ipv4_notify_unregister(void)
+{
+ nss_top_main.ipv4_callback = NULL;
+}
+
+/*
+ * nss_ipv4_get_mgr()
+ *
+ * TODO: This only suppports a single ipv4, do we ever want to support more?
+ */
+struct nss_ctx_instance *nss_ipv4_get_mgr(void)
+{
+ return (void *)&nss_top_main.nss[nss_top_main.ipv4_handler_id];
+}
+
+/*
+ * nss_ipv4_register_handler()
+ * Register our handler to receive messages for this interface
+ */
+void nss_ipv4_register_handler()
+{
+ if (nss_core_register_handler(NSS_IPV4_RX_INTERFACE, nss_ipv4_handler, NULL) != NSS_CORE_STATUS_SUCCESS) {
+ nss_warning("IPv4 handler failed to register");
+ }
+}
+
+EXPORT_SYMBOL(nss_ipv4_tx);
+EXPORT_SYMBOL(nss_ipv4_notify_register);
+EXPORT_SYMBOL(nss_ipv4_notify_unregister);
+EXPORT_SYMBOL(nss_ipv4_get_mgr);
+EXPORT_SYMBOL(nss_ipv4_register_handler);