/*************************************************************************** * Copyright (c) 2024 Microsoft Corporation * Copyright (c) 2025-present Eclipse ThreadX Contributors * * This program and the accompanying materials are made available under the * terms of the MIT License which is available at * https://opensource.org/licenses/MIT. * * SPDX-License-Identifier: MIT **************************************************************************/ /**************************************************************************/ /**************************************************************************/ /** */ /** NetX Component */ /** */ /** MINRES ETH Network */ /** */ /**************************************************************************/ /**************************************************************************/ /* Include necessary system files. */ #include "gen/ethmac.h" #include "nx_api.h" #include "platform.h" #include "tx_port.h" #include #if defined(NX_DEBUG) || defined(NX_DEBUG_PACKET) #include #endif extern int register_irq_handler(unsigned irq_num, void (*handler)()); /* Define the Link MTU. Note this is not the same as the IP MTU. The Link MTU includes the addition of the Physical Network header (usually Ethernet). This should be larger than the IP instance MTU by the size of the physical header. */ #define NX_LINK_MTU 1514 #define NX_MAX_PACKET_SIZE 1536 /* Define Ethernet address format. This is prepended to the incoming IP and ARP/RARP messages. The frame beginning is 14 bytes, but for speed purposes, we are going to assume there are 16 bytes free in front of the prepend pointer and that the prepend pointer is 32-bit aligned. Byte Offset Size Meaning 0 6 Destination Ethernet Address 6 6 Source Ethernet Address 12 2 Ethernet Frame Type, where: 0x0800 -> IP Datagram 0x0806 -> ARP Request/Reply 0x0835 -> RARP request reply 42 18 Padding on ARP and RARP messages only. */ #define NX_ETHERNET_IP 0x0800 #define NX_ETHERNET_ARP 0x0806 #define NX_ETHERNET_RARP 0x8035 #define NX_ETHERNET_IPV6 0x86DD #define NX_ETHERNET_SIZE 14 /* For the ethernet driver, physical addresses are allocated starting at the preset value and then incremented before the next allocation. */ // Locally Administered Addresses that can be assigned by network, range is 02-00-00 to 02-00-5E const ULONG mnrs_mac_address_msw = 0x0200; const ULONG mnrs_mac_address_lsw = 0x00334450; /* Define driver prototypes. */ VOID _nx_mnrs_network_driver(NX_IP_DRIVER *driver_req_ptr); void _nx_mnrs_network_driver_output(NX_PACKET *packet_ptr, UINT interface_instance_id); void _nx_mnrs_network_driver_receive(NX_IP *ip_ptr, NX_PACKET *packet_ptr, UINT interface_instance_id); UINT _nx_mnrs_eth_send_packet(NX_PACKET *packet_ptr, volatile ethmac_t* ethmac); VOID _nx_mnrs_eth_recv_packet(UINT id, volatile ethmac_t* ethmac); VOID _nx_mnrs_eth_recv_packet_eth0(VOID); VOID _nx_mnrs_eth_recv_packet_eth1(VOID); #define NX_MNRS_DRIVER_MAX_MCAST_ADDRESSES 3 typedef struct MAC_ADDRESS_STRUCT { ULONG nx_mac_address_msw; ULONG nx_mac_address_lsw; } MAC_ADDRESS; /* Define an application-specific data structure that holds internal data (such as the state information) of a device driver. The example below applies to the MNRS ETH driver. User shall replace its content with information related to the actual driver being used. */ typedef struct _nx_mnrs_network_driver_instance_type { UINT nx_mnrs_network_driver_in_use; UINT nx_mnrs_network_driver_id; NX_INTERFACE *nx_mnrs_driver_interface_ptr; NX_IP *nx_mnrs_driver_ip_ptr; MAC_ADDRESS nx_mnrs_driver_mac_address; MAC_ADDRESS nx_mnrs_driver_mcast_address[NX_MNRS_DRIVER_MAX_MCAST_ADDRESSES]; volatile ethmac_t* ethmac; } _nx_mnrs_network_driver_instance_type; /* In this example, there are four instances of the MNRS ETH driver. Therefore an array of four driver instances are created to keep track of the interface information of each driver. */ static _nx_mnrs_network_driver_instance_type nx_mnrs_driver[NX_MAX_PHYSICAL_INTERFACES]; /**************************************************************************/ /* */ /* FUNCTION RELEASE */ /* */ /* _nx_mnrs_network_driver PORTABLE C */ /* 6.4.3 */ /* AUTHOR */ /* */ /* Yuxin Zhou, Microsoft Corporation */ /* */ /* DESCRIPTION */ /* */ /* This function acts as a virtual network for testing the NetX source */ /* and driver concepts. User application may use this routine as */ /* a template for the actual network driver. Note that this driver */ /* simulates Ethernet operation. Some of the parameters don't apply */ /* for non-Ethernet interfaces. */ /* */ /* INPUT */ /* */ /* ip_ptr Pointer to IP protocol block */ /* */ /* OUTPUT */ /* */ /* None */ /* */ /* CALLS */ /* */ /* _nx_mnrs_network_driver_output Send physical packet out */ /* */ /* CALLED BY */ /* */ /* NetX IP processing */ /* */ /* RELEASE HISTORY */ /* */ /* DATE NAME DESCRIPTION */ /* */ /* 05-19-2020 Yuxin Zhou Initial Version 6.0 */ /* 09-30-2020 Yuxin Zhou Modified comment(s), */ /* resulting in version 6.1 */ /* 10-15-2021 Yuxin Zhou Modified comment(s), and */ /* added sample of returning */ /* link's interface type, */ /* resulting in version 6.1.9 */ /* 12-31-2023 Yajun Xia Modified comment(s), */ /* supported VLAN and generic */ /* link layer, */ /* resulting in version 6.4.0 */ /* */ /**************************************************************************/ VOID _nx_mnrs_network_driver(NX_IP_DRIVER* driver_req_ptr) { UINT i = 0; NX_IP* ip_ptr; NX_PACKET* packet_ptr; NX_INTERFACE* interface_ptr; UINT interface_index; USHORT ether_type; #ifndef NX_ENABLE_VLAN ULONG *ethernet_frame_ptr; #endif /* NX_ENABLE_VLAN */ /* Setup the IP pointer from the driver request. */ ip_ptr = driver_req_ptr -> nx_ip_driver_ptr; /* Default to successful return. */ driver_req_ptr -> nx_ip_driver_status = NX_SUCCESS; #ifdef NX_ENABLE_VLAN /* Let link layer to preprocess the driver request and return actual interface. */ if (nx_link_driver_request_preprocess(driver_req_ptr, &interface_ptr) != NX_SUCCESS) { return; } #else /* Setup interface pointer. */ interface_ptr = driver_req_ptr -> nx_ip_driver_interface; #endif /* NX_ENABLE_VLAN */ /* Obtain the index number of the network interface. */ interface_index = interface_ptr -> nx_interface_index; /* Find out the driver interface if the driver command is not ATTACH. */ if (driver_req_ptr->nx_ip_driver_command != NX_LINK_INTERFACE_ATTACH) { for (i = 0; i < NX_MAX_PHYSICAL_INTERFACES; i++) { if (nx_mnrs_driver[i].nx_mnrs_network_driver_in_use == 0) { continue; } if (nx_mnrs_driver[i].nx_mnrs_driver_ip_ptr != ip_ptr) { continue; } if (nx_mnrs_driver[i].nx_mnrs_driver_interface_ptr == interface_ptr) { break; } } if (i == NX_MAX_PHYSICAL_INTERFACES) { driver_req_ptr->nx_ip_driver_status = NX_INVALID_INTERFACE; return; } } /* Process according to the driver request type in the IP control block. */ switch (driver_req_ptr->nx_ip_driver_command) { case NX_LINK_INTERFACE_ATTACH: { /* Find an available driver instance to attach the interface. */ for (i = 0; i < NX_MAX_PHYSICAL_INTERFACES; i++) { if (nx_mnrs_driver[i].nx_mnrs_network_driver_in_use == 0) { break; } } /* An available entry is found. */ if (i < NX_MAX_PHYSICAL_INTERFACES) { /* Set the IN USE flag.*/ nx_mnrs_driver[i].nx_mnrs_network_driver_in_use = 1; nx_mnrs_driver[i].nx_mnrs_network_driver_id = i; /* Record the interface attached to the IP instance. */ nx_mnrs_driver[i].nx_mnrs_driver_interface_ptr = interface_ptr; /* Record the IP instance. */ nx_mnrs_driver[i].nx_mnrs_driver_ip_ptr = ip_ptr; /* store the mac address */ nx_mnrs_driver[i].nx_mnrs_driver_mac_address.nx_mac_address_msw = mnrs_mac_address_msw; nx_mnrs_driver[i].nx_mnrs_driver_mac_address.nx_mac_address_lsw = mnrs_mac_address_lsw + i; /* Record the peripheral address */ switch (i) { case 0: nx_mnrs_driver[i].ethmac = ethmac0; break; #if defined(ethmac1) case 1: nx_mnrs_driver[i].ethmac = ethmac1; break; #endif #if defined(ethmac2) case 2: nx_mnrs_driver[i].ethmac = ethmac2; #endif #if defined(ethmac3) case 2: nx_mnrs_driver[i].ethmac = ethmac3; #endif default: } } else { driver_req_ptr->nx_ip_driver_status = NX_INVALID_INTERFACE; } break; } case NX_LINK_INTERFACE_DETACH: { /* Zero out the driver instance. */ memset(&(nx_mnrs_driver[i]), 0, sizeof(_nx_mnrs_network_driver_instance_type)); break; } case NX_LINK_INITIALIZE: { /* Device driver shall initialize the Ethernet Controller here. */ #ifdef NX_DEBUG printf("NetX MNRS ETH Driver Initialization - %s\n", ip_ptr->nx_ip_name); printf(" IP Address =%08X\n", ip_ptr->nx_ip_address); #endif // enable TX & RX path set_ethmac_mac_ctrl(nx_mnrs_driver[i].ethmac, get_ethmac_mac_ctrl(nx_mnrs_driver[i].ethmac) & ~(ETHMAC_MAC_CTRL_TX_FLUSH(1) | ETHMAC_MAC_CTRL_RX_FLUSH(1))); register_irq_handler(ETH0_IRQ, &_nx_mnrs_eth_recv_packet_eth0); register_irq_handler(ETH1_IRQ, &_nx_mnrs_eth_recv_packet_eth1); csr_read_set_bits_mie(1 << ETH0_IRQ | 1 << ETH1_IRQ); set_ethmac_mac_intr_rx_data_avail_intr_enable(nx_mnrs_driver[i].ethmac, 1); /* Once the Ethernet controller is initialized, the driver needs to configure the NetX Interface Control block, as outlined below. */ /* The nx_interface_ip_mtu_size should be the MTU for the IP payload. For regular Ethernet, the IP MTU is 1500. */ nx_ip_interface_mtu_set(ip_ptr, interface_index, (NX_LINK_MTU - NX_ETHERNET_SIZE)); /* Set the physical address (MAC address) of this IP instance. */ /* For this MNRS ETH driver, the MAC address is constructed by incrementing a base lsw value, to simulate multiple nodes on the ethernet. */ nx_ip_interface_physical_address_set(ip_ptr, interface_index, nx_mnrs_driver[i].nx_mnrs_driver_mac_address.nx_mac_address_msw, nx_mnrs_driver[i].nx_mnrs_driver_mac_address.nx_mac_address_lsw, NX_FALSE); /* Indicate to the IP software that IP to physical mapping is required. */ nx_ip_interface_address_mapping_configure(ip_ptr, interface_index, NX_TRUE); break; } case NX_LINK_UNINITIALIZE: { // disable TX & RX path set_ethmac_mac_intr_rx_data_avail_intr_enable(nx_mnrs_driver[i].ethmac, 0); csr_read_clr_bits_mie(1 << ETH0_IRQ | 1 << ETH1_IRQ); register_irq_handler(ETH0_IRQ, NULL); register_irq_handler(ETH1_IRQ, NULL); // set_ethmac_mac_ctrl_tx_flush(nx_mnrs_driver[i].ethmac, 1); // set_ethmac_mac_ctrl_rx_flush(nx_mnrs_driver[i].ethmac, 1); set_ethmac_mac_ctrl(nx_mnrs_driver[i].ethmac, get_ethmac_mac_ctrl(nx_mnrs_driver[i].ethmac) | ETHMAC_MAC_CTRL_TX_FLUSH(1) | ETHMAC_MAC_CTRL_RX_FLUSH(1)); /* Zero out the driver instance. */ memset(&(nx_mnrs_driver[i]), 0, sizeof(_nx_mnrs_network_driver_instance_type)); break; } case NX_LINK_ENABLE: { /* Process driver link enable. An Ethernet driver shall enable the transmit and reception logic. Once the IP stack issues the LINK_ENABLE command, the stack may start transmitting IP packets. */ // set_ethmac_int_ctrl_reg_tx_free_intr_enable(nx_mnrs_driver[i].ethmac, 1); set_ethmac_mac_intr_rx_data_avail_intr_enable(nx_mnrs_driver[i].ethmac, 1); /* In the MNRS ETH driver, just set the enabled flag. */ interface_ptr->nx_interface_link_up = NX_TRUE; #ifdef NX_DEBUG printf("NetX MNRS ETH Driver Link Enabled - %s\n", ip_ptr->nx_ip_name); #endif break; } case NX_LINK_DISABLE: { /* Process driver link disable. This command indicates the IP layer is not going to transmit any IP datagrams, nor does it expect any IP datagrams from the interface. Therefore after processing this command, the device driver shall not send any incoming packets to the IP layer. Optionally the device driver may turn off the interface. */ // set_ethmac_int_ctrl_reg_tx_free_intr_enable(nx_mnrs_driver[i].ethmac, 0); set_ethmac_mac_intr_rx_data_avail_intr_enable(nx_mnrs_driver[i].ethmac, 0); /* In the MNRS ETH driver, just clear the enabled flag. */ interface_ptr->nx_interface_link_up = NX_FALSE; #ifdef NX_DEBUG printf("NetX MNRS ETH Driver Link Disabled - %s\n", ip_ptr->nx_ip_name); #endif break; } case NX_LINK_PACKET_SEND: case NX_LINK_PACKET_BROADCAST: case NX_LINK_ARP_SEND: case NX_LINK_ARP_RESPONSE_SEND: case NX_LINK_RARP_SEND: { /* The IP stack sends down a data packet for transmission. The device driver needs to prepend a MAC header, and fill in the Ethernet frame type (assuming Ethernet protocol for network transmission) based on the type of packet being transmitted. The following sequence illustrates this process. */ /* Place the ethernet frame at the front of the packet. */ packet_ptr = driver_req_ptr->nx_ip_driver_packet; /* Get Ethernet type. */ if (driver_req_ptr->nx_ip_driver_command == NX_LINK_ARP_SEND) { ether_type = NX_ETHERNET_ARP; } else if (driver_req_ptr->nx_ip_driver_command == NX_LINK_ARP_RESPONSE_SEND) { ether_type = NX_ETHERNET_ARP; } else if (driver_req_ptr->nx_ip_driver_command == NX_LINK_RARP_SEND) { ether_type = NX_ETHERNET_RARP; } else if (packet_ptr->nx_packet_ip_version == 4) { ether_type = NX_ETHERNET_IP; } else { ether_type = NX_ETHERNET_IPV6; } #ifdef NX_ENABLE_VLAN /* Add Ethernet header. */ if (nx_link_ethernet_header_add(ip_ptr, driver_req_ptr->nx_ip_driver_interface->nx_interface_index, packet_ptr, driver_req_ptr->nx_ip_driver_physical_address_msw, driver_req_ptr->nx_ip_driver_physical_address_lsw, (UINT)ether_type)) { /* Release the packet. */ nx_packet_transmit_release(packet_ptr); break; } #else /* Adjust the prepend pointer. */ packet_ptr->nx_packet_prepend_ptr = packet_ptr->nx_packet_prepend_ptr - NX_ETHERNET_SIZE; /* Adjust the packet length. */ packet_ptr->nx_packet_length = packet_ptr->nx_packet_length + NX_ETHERNET_SIZE; /* Setup the ethernet frame pointer to build the ethernet frame. Backup another 2 bytes to get 32-bit word alignment. */ /*lint -e{927} -e{826} suppress cast of pointer to pointer, since it is necessary */ ethernet_frame_ptr = (ULONG*)(packet_ptr->nx_packet_prepend_ptr - 2); /* Build the ethernet frame. */ *ethernet_frame_ptr = driver_req_ptr->nx_ip_driver_physical_address_msw; *(ethernet_frame_ptr + 1) = driver_req_ptr->nx_ip_driver_physical_address_lsw; *(ethernet_frame_ptr + 2) = (interface_ptr->nx_interface_physical_address_msw << 16) | (interface_ptr->nx_interface_physical_address_lsw >> 16); *(ethernet_frame_ptr + 3) = (interface_ptr->nx_interface_physical_address_lsw << 16) | ether_type; /* Endian swapping if NX_LITTLE_ENDIAN is defined. */ NX_CHANGE_ULONG_ENDIAN(*(ethernet_frame_ptr)); NX_CHANGE_ULONG_ENDIAN(*(ethernet_frame_ptr + 1)); NX_CHANGE_ULONG_ENDIAN(*(ethernet_frame_ptr + 2)); NX_CHANGE_ULONG_ENDIAN(*(ethernet_frame_ptr + 3)); #endif /* NX_ENABLE_VLAN */ #ifdef NX_DEBUG_PACKET printf("NetX MNRS ETH Driver Packet Send - %s\n", ip_ptr->nx_ip_name); #endif /* At this point, the packet is a complete Ethernet frame, ready to be transmitted. The driver shall call the actual Ethernet transmit routine and put the packet on the wire. In this example, the MNRS ETH network transmit routine is called. */ _nx_mnrs_network_driver_output(packet_ptr, i); break; } #ifdef NX_ENABLE_VLAN case NX_LINK_RAW_PACKET_SEND: { /* Send raw packet out directly. */ _nx_mnrs_network_driver_output(driver_req_ptr->nx_ip_driver_packet, i); break; } #endif /* NX_ENABLE_VLAN */ case NX_LINK_MULTICAST_JOIN: { UINT mcast_index; /* The IP layer issues this command to join a multicast group. Note that multicast operation is required for IPv6. On a typically Ethernet controller, the driver computes a hash value based on MAC address, and programs the hash table. It is likely the driver also needs to maintain an internal MAC address table. Later if a multicast address is removed, the driver needs to reprogram the hash table based on the remaining multicast MAC addresses. */ /* The following procedure only applies to our MNRS ETH network driver, which manages multicast MAC addresses by a simple look up table. */ for (mcast_index = 0; mcast_index < NX_MNRS_DRIVER_MAX_MCAST_ADDRESSES; mcast_index++) { if (nx_mnrs_driver[i].nx_mnrs_driver_mcast_address[mcast_index].nx_mac_address_msw == 0 && nx_mnrs_driver[i].nx_mnrs_driver_mcast_address[mcast_index].nx_mac_address_lsw == 0) { nx_mnrs_driver[i].nx_mnrs_driver_mcast_address[mcast_index].nx_mac_address_msw = driver_req_ptr->nx_ip_driver_physical_address_msw; nx_mnrs_driver[i].nx_mnrs_driver_mcast_address[mcast_index].nx_mac_address_lsw = driver_req_ptr->nx_ip_driver_physical_address_lsw; break; } } if (mcast_index == NX_MNRS_DRIVER_MAX_MCAST_ADDRESSES) { driver_req_ptr->nx_ip_driver_status = NX_NO_MORE_ENTRIES; } break; } case NX_LINK_MULTICAST_LEAVE: { UINT mcast_index; /* The IP layer issues this command to remove a multicast MAC address from the receiving list. A device driver shall properly remove the multicast address from the hash table, so the hardware does not receive such traffic. Note that in order to reprogram the hash table, the device driver may have to keep track of current active multicast MAC addresses. */ /* The following procedure only applies to our MNRS ETH network driver, which manages multicast MAC addresses by a simple look up table. */ for (mcast_index = 0; mcast_index < NX_MNRS_DRIVER_MAX_MCAST_ADDRESSES; mcast_index++) { if (nx_mnrs_driver[i].nx_mnrs_driver_mcast_address[mcast_index].nx_mac_address_msw == driver_req_ptr->nx_ip_driver_physical_address_msw && nx_mnrs_driver[i].nx_mnrs_driver_mcast_address[mcast_index].nx_mac_address_lsw == driver_req_ptr->nx_ip_driver_physical_address_lsw) { nx_mnrs_driver[i].nx_mnrs_driver_mcast_address[mcast_index].nx_mac_address_msw = 0; nx_mnrs_driver[i].nx_mnrs_driver_mcast_address[mcast_index].nx_mac_address_lsw = 0; break; } } if (mcast_index == NX_MNRS_DRIVER_MAX_MCAST_ADDRESSES) { driver_req_ptr->nx_ip_driver_status = NX_ENTRY_NOT_FOUND; } break; } case NX_LINK_GET_STATUS: { /* Return the link status in the supplied return pointer. */ *(driver_req_ptr->nx_ip_driver_return_ptr) = ip_ptr->nx_ip_interface[0].nx_interface_link_up; break; } case NX_LINK_GET_SPEED: { /* Return the link's line speed in the supplied return pointer. Unsupported feature. */ *(driver_req_ptr->nx_ip_driver_return_ptr) = 0; break; } case NX_LINK_GET_DUPLEX_TYPE: { /* Return the link's line speed in the supplied return pointer. Unsupported feature. */ *(driver_req_ptr->nx_ip_driver_return_ptr) = 0; break; } case NX_LINK_GET_ERROR_COUNT: { /* Return the link's line speed in the supplied return pointer. Unsupported feature. */ *(driver_req_ptr->nx_ip_driver_return_ptr) = 0; break; } case NX_LINK_GET_RX_COUNT: { /* Return the link's line speed in the supplied return pointer. Unsupported feature. */ *(driver_req_ptr->nx_ip_driver_return_ptr) = 0; break; } case NX_LINK_GET_TX_COUNT: { /* Return the link's line speed in the supplied return pointer. Unsupported feature. */ *(driver_req_ptr->nx_ip_driver_return_ptr) = 0; break; } case NX_LINK_GET_ALLOC_ERRORS: { /* Return the link's line speed in the supplied return pointer. Unsupported feature. */ *(driver_req_ptr->nx_ip_driver_return_ptr) = 0; break; } case NX_LINK_GET_INTERFACE_TYPE: { /* Return the link's interface type in the supplied return pointer. Unsupported feature. */ *(driver_req_ptr->nx_ip_driver_return_ptr) = NX_INTERFACE_TYPE_UNKNOWN; break; } case NX_LINK_DEFERRED_PROCESSING: { /* Driver defined deferred processing. This is typically used to defer interrupt processing to the thread level. A typical use case of this command is: On receiving an Ethernet frame, the RX ISR does not process the received frame, but instead records such an event in its internal data structure, and issues a notification to the IP stack (the driver sends the notification to the IP helping thread by calling "_nx_ip_driver_deferred_processing()". When the IP stack gets a notification of a pending driver deferred process, it calls the driver with the NX_LINK_DEFERRED_PROCESSING command. The driver shall complete the pending receive process. */ /* The MNRS ETH driver doesn't require a deferred process so it breaks out of the switch case. */ break; } case NX_LINK_SET_PHYSICAL_ADDRESS: { /* Find an driver instance to attach the interface. */ for (i = 0; i < NX_MAX_PHYSICAL_INTERFACES; i++) { if (nx_mnrs_driver[i].nx_mnrs_driver_interface_ptr == interface_ptr) { break; } } /* An available entry is found. */ if (i < NX_MAX_PHYSICAL_INTERFACES) { /* Set the physical address. */ nx_mnrs_driver[i].nx_mnrs_driver_mac_address.nx_mac_address_msw = driver_req_ptr->nx_ip_driver_physical_address_msw; nx_mnrs_driver[i].nx_mnrs_driver_mac_address.nx_mac_address_lsw = driver_req_ptr->nx_ip_driver_physical_address_lsw; } else { driver_req_ptr->nx_ip_driver_status = NX_INVALID_INTERFACE; } break; } #ifdef NX_ENABLE_INTERFACE_CAPABILITY case NX_INTERFACE_CAPABILITY_GET: { /* Return the capability of the Ethernet controller speed in the supplied return pointer. Unsupported * feature. */ *(driver_req_ptr->nx_ip_driver_return_ptr) = 0; break; } case NX_INTERFACE_CAPABILITY_SET: { /* Set the capability of the Ethernet controller. Unsupported feature. */ break; } #endif /* NX_ENABLE_INTERFACE_CAPABILITY */ default: /* Invalid driver request. */ /* Return the unhandled command status. */ driver_req_ptr->nx_ip_driver_status = NX_UNHANDLED_COMMAND; #ifdef NX_DEBUG printf("NetX MNRS ETH Driver Received invalid request - %s\n", ip_ptr->nx_ip_name); #endif break; } } /**************************************************************************/ /* */ /* FUNCTION RELEASE */ /* */ /* _nx_mnrs_network_driver_output PORTABLE C */ /* 6.4.3 */ /* AUTHOR */ /* */ /* Yuxin Zhou, Microsoft Corporation */ /* */ /* DESCRIPTION */ /* */ /* This function simply sends the packet to the IP instance on the */ /* created IP list that matches the physical destination specified in */ /* the Ethernet packet. In a real hardware setting, this routine */ /* would simply put the packet out on the wire. */ /* */ /* INPUT */ /* */ /* packet_ptr Packet pointer */ /* interface_instance_id ID of driver instance */ /* */ /* OUTPUT */ /* */ /* None */ /* */ /* CALLS */ /* */ /* nx_packet_copy Copy a packet */ /* nx_packet_transmit_release Release a packet */ /* _nx_mnrs_network_driver_receive ETH driver receive processing */ /* */ /* CALLED BY */ /* */ /* NetX IP processing */ /* */ /* RELEASE HISTORY */ /* */ /* DATE NAME DESCRIPTION */ /* */ /* 05-19-2020 Yuxin Zhou Initial Version 6.0 */ /* 09-30-2020 Yuxin Zhou Modified comment(s), */ /* resulting in version 6.1 */ /* 12-31-2023 Yajun Xia Modified comment(s), */ /* supported VLAN and generic */ /* link layer, */ /* resulting in version 6.4.0 */ /* */ /**************************************************************************/ void _nx_mnrs_network_driver_output(NX_PACKET* packet_ptr, UINT interface_instance_id) { UINT old_threshold = 0; #ifdef NX_DEBUG_PACKET UCHAR *ptr; UINT i, j; ptr = packet_ptr -> nx_packet_prepend_ptr; printf("Ethernet Packet: "); for (j = 0; j < 6; j++) { printf("%02X", *ptr++); } printf(" "); for (j = 0; j < 6; j++) { printf("%02X", *ptr++); } printf(" %02X", *ptr++); printf("%02X ", *ptr++); i = 0; for (j = 0; j < (packet_ptr->nx_packet_length - NX_ETHERNET_SIZE); j++) { printf("%02X", *ptr++); i++; if (i > 3) { i = 0; printf(" "); } } printf("\n"); #endif /* Disable preemption. */ tx_thread_preemption_change(tx_thread_identify(), 0, &old_threshold); _nx_mnrs_eth_send_packet(packet_ptr, nx_mnrs_driver[interface_instance_id].ethmac); #ifdef NX_ENABLE_VLAN /* Release the packet. */ nx_link_packet_transmitted(nx_mnrs_driver[interface_instance_id].nx_mnrs_driver_ip_ptr, nx_mnrs_driver[interface_instance_id].nx_mnrs_driver_interface_ptr->nx_interface_index, packet_ptr, NX_NULL); #else /* Remove the Ethernet header. In real hardware environments, this is typically done after a transmit complete interrupt. */ packet_ptr -> nx_packet_prepend_ptr = packet_ptr -> nx_packet_prepend_ptr + NX_ETHERNET_SIZE; /* Adjust the packet length. */ packet_ptr -> nx_packet_length = packet_ptr -> nx_packet_length - NX_ETHERNET_SIZE; /* Now that the Ethernet frame has been removed, release the packet. */ nx_packet_transmit_release(packet_ptr); #endif /* NX_ENABLE_VLAN */ /* Restore preemption. */ tx_thread_preemption_change(tx_thread_identify(), old_threshold, &old_threshold); } /**************************************************************************/ /* */ /* FUNCTION RELEASE */ /* */ /* _nx_mnrs_network_driver_receive PORTABLE C */ /* 6.4.3 */ /* AUTHOR */ /* */ /* Yuxin Zhou, Microsoft Corporation */ /* */ /* DESCRIPTION */ /* */ /* This function processing incoming packets. This routine needs to be */ /* be called from the receive packet ISR. */ /* */ /* INPUT */ /* */ /* ip_ptr Pointer to IP protocol block */ /* packet_ptr Packet pointer */ /* interface_instance_id The interface ID the packet is*/ /* destined for */ /* */ /* OUTPUT */ /* */ /* None */ /* */ /* CALLS */ /* */ /* _nx_ip_packet_receive IP receive packet processing */ /* _nx_ip_packet_deferred_receive IP deferred receive packet */ /* processing */ /* _nx_arp_packet_deferred_receive ARP receive processing */ /* _nx_rarp_packet_deferred_receive RARP receive processing */ /* nx_packet_release Packet release */ /* */ /* CALLED BY */ /* */ /* NetX IP processing */ /* */ /* RELEASE HISTORY */ /* */ /* DATE NAME DESCRIPTION */ /* */ /* 05-19-2020 Yuxin Zhou Initial Version 6.0 */ /* 09-30-2020 Yuxin Zhou Modified comment(s), */ /* resulting in version 6.1 */ /* 12-31-2023 Yajun Xia Modified comment(s), */ /* supported VLAN and generic */ /* link layer, */ /* resulting in version 6.4.0 */ /* */ /**************************************************************************/ void _nx_mnrs_network_driver_receive(NX_IP* ip_ptr, NX_PACKET* packet_ptr, UINT interface_instance_id) { UINT packet_type; #ifdef NX_ENABLE_VLAN nx_link_ethernet_packet_received(ip_ptr, nx_mnrs_driver[interface_instance_id].nx_mnrs_driver_interface_ptr->nx_interface_index, packet_ptr, NX_NULL); #else /* Pickup the packet header to determine where the packet needs to be sent. */ packet_type = (((UINT)(*(packet_ptr->nx_packet_prepend_ptr + 12))) << 8) | ((UINT)(*(packet_ptr->nx_packet_prepend_ptr + 13))); /* Setup interface pointer. */ packet_ptr->nx_packet_address.nx_packet_interface_ptr = nx_mnrs_driver[interface_instance_id].nx_mnrs_driver_interface_ptr; /* Route the incoming packet according to its ethernet type. */ /* The MNRS ETH driver accepts both IPv4 and IPv6 frames. */ if ((packet_type == NX_ETHERNET_IP) || (packet_type == NX_ETHERNET_IPV6)) { /* Note: The length reported by some Ethernet hardware includes bytes after the packet as well as the Ethernet header. In some cases, the actual packet length after the Ethernet header should be derived from the length in the IP header (lower 16 bits of the first 32-bit word). */ /* Clean off the Ethernet header. */ packet_ptr -> nx_packet_prepend_ptr = packet_ptr -> nx_packet_prepend_ptr + NX_ETHERNET_SIZE; /* Adjust the packet length. */ packet_ptr -> nx_packet_length = packet_ptr -> nx_packet_length - NX_ETHERNET_SIZE; /* Route to the ip receive function. */ #ifdef NX_DEBUG_PACKET printf("NetX MNRS ETH Driver IP Packet Receive - %s\n", ip_ptr -> nx_ip_name); #endif #ifdef NX_DIRECT_ISR_CALL _nx_ip_packet_receive(ip_ptr, packet_ptr); #else _nx_ip_packet_deferred_receive(ip_ptr, packet_ptr); #endif } #ifndef NX_DISABLE_IPV4 else if (packet_type == NX_ETHERNET_ARP) { /* Clean off the Ethernet header. */ packet_ptr -> nx_packet_prepend_ptr = packet_ptr -> nx_packet_prepend_ptr + NX_ETHERNET_SIZE; /* Adjust the packet length. */ packet_ptr -> nx_packet_length = packet_ptr -> nx_packet_length - NX_ETHERNET_SIZE; /* Route to the ARP receive function. */ #ifdef NX_DEBUG printf("NetX MNRS ETH Driver ARP Receive - %s\n", ip_ptr -> nx_ip_name); #endif _nx_arp_packet_deferred_receive(ip_ptr, packet_ptr); } else if (packet_type == NX_ETHERNET_RARP) { /* Clean off the Ethernet header. */ packet_ptr -> nx_packet_prepend_ptr = packet_ptr -> nx_packet_prepend_ptr + NX_ETHERNET_SIZE; /* Adjust the packet length. */ packet_ptr -> nx_packet_length = packet_ptr -> nx_packet_length - NX_ETHERNET_SIZE; /* Route to the RARP receive function. */ #ifdef NX_DEBUG printf("NetX MNRS ETH Driver RARP Receive - %s\n", ip_ptr -> nx_ip_name); #endif _nx_rarp_packet_deferred_receive(ip_ptr, packet_ptr); } #endif /* !NX_DISABLE_IPV4 */ else { /* Invalid ethernet header... release the packet. */ nx_packet_release(packet_ptr); } #endif /* NX_ENABLE_VLAN */ } UINT _nx_mnrs_eth_send_packet(NX_PACKET* packet_ptr, volatile ethmac_t* ethmac) { ULONG size = 0; UCHAR* data; UINT i; ULONG buffer; ULONG words; /* Make sure the data length is less than MTU. */ if (packet_ptr->nx_packet_length > NX_MAX_PACKET_SIZE) { return NX_NOT_SUCCESSFUL; } /* get data pointer to be transmitted. */ data = packet_ptr->nx_packet_prepend_ptr; size = packet_ptr->nx_packet_length; words = (size + 3) / 4; while (get_ethmac_mac_ctrl_tx_ready(ethmac) == 0) ; set_ethmac_mac_tx(ethmac, size * 8); for (i = 0; i < size; i += 4) { if (i < size - 3) { memcpy(&buffer, data + i, 4); } else { memcpy(&buffer, data + i, size - i); } while (!get_ethmac_mac_ctrl_tx_ready(ethmac)) ; set_ethmac_mac_tx(ethmac, buffer); } return NX_SUCCESS; } UINT mac_addr_hit(UINT i, MAC_ADDRESS* mac_addr) { UINT mcast_index; if (nx_mnrs_driver[i].nx_mnrs_driver_mac_address.nx_mac_address_msw == mac_addr->nx_mac_address_msw && nx_mnrs_driver[i].nx_mnrs_driver_mac_address.nx_mac_address_lsw == mac_addr->nx_mac_address_lsw) { return 1; } return 0; } UINT mcast_mac_addr_hit_idx(UINT i, MAC_ADDRESS* mac_addr) { UINT mcast_index; for (mcast_index = 0; mcast_index < NX_MNRS_DRIVER_MAX_MCAST_ADDRESSES; mcast_index++) { if (nx_mnrs_driver[i].nx_mnrs_driver_mcast_address[mcast_index].nx_mac_address_msw == mac_addr->nx_mac_address_msw && nx_mnrs_driver[i].nx_mnrs_driver_mcast_address[mcast_index].nx_mac_address_lsw == mac_addr->nx_mac_address_lsw) { return i; } } return -1; } static UCHAR nx_mnrs_receive_buffer[NX_MAX_PACKET_SIZE]; VOID _nx_mnrs_eth_recv_packet(UINT id, volatile ethmac_t* ethmac) { UINT bits_received; UINT words_to_read; UINT i, j; UINT word; UINT status; NX_PACKET* packet_ptr; NX_IP* ip_ptr; UCHAR* data; UINT bytes_received; UINT address_len; UINT packet_type; set_ethmac_mac_intr_rx_data_avail_intr_enable(ethmac, 0); bits_received = get_ethmac_mac_rx(ethmac); words_to_read = (bits_received + 31) / 32; if (words_to_read < 4) { for (j = 0; j < words_to_read; j++) { while (!get_ethmac_mac_ctrl_rx_pending(ethmac)) ; // be carefull of unaligned accesses since data is 2byte aligned word = get_ethmac_mac_rx(ethmac); memcpy(data + j * sizeof(UINT), &word, sizeof(UINT)); } set_ethmac_mac_intr_rx_data_avail_intr_enable(ethmac, 1); return; } // find the diver instance belonging to our ethmac for (i = 0; i < NX_MAX_PHYSICAL_INTERFACES; ++i) { if (nx_mnrs_driver[i].ethmac == ethmac) { ip_ptr = nx_mnrs_driver[i].nx_mnrs_driver_ip_ptr; status = nx_packet_allocate(ip_ptr->nx_ip_default_packet_pool, &packet_ptr, NX_RECEIVE_PACKET, NX_NO_WAIT); if (status) { packet_ptr = NX_NULL; data = nx_mnrs_receive_buffer; } else if (ip_ptr->nx_ip_default_packet_pool->nx_packet_pool_payload_size >= (NX_LINK_MTU + 2)) { data = packet_ptr->nx_packet_prepend_ptr + 2; } else { data = nx_mnrs_receive_buffer; } /* now get the remaining bytes from the ethmac peripheral */ for (j = 0; j < words_to_read; j++) { while (!get_ethmac_mac_ctrl_rx_pending(ethmac)) ; // be carefull of unaligned accesses since data is 2byte aligned word = get_ethmac_mac_rx(ethmac); memcpy(data + j * sizeof(UINT), &word, sizeof(UINT)); } if (packet_ptr == NX_NULL) { /* No packet available. Drop it and continue. */ return; } /* Make sure IP header is 4-byte aligned. */ packet_ptr->nx_packet_prepend_ptr += 2; packet_ptr->nx_packet_append_ptr += 2; if (data == nx_mnrs_receive_buffer) { /* Copy data into packet. */ status = nx_packet_data_append( packet_ptr, (VOID*)data, bytes_received, ip_ptr->nx_ip_default_packet_pool, NX_NO_WAIT); if (status) { nx_packet_release(packet_ptr); return; } } else { packet_ptr->nx_packet_length = (ULONG)bytes_received; packet_ptr->nx_packet_append_ptr += (ULONG)bytes_received; } _nx_mnrs_network_driver_receive(ip_ptr, packet_ptr, i); break; } } set_ethmac_mac_intr_rx_data_avail_intr_enable(ethmac, 1); } VOID _nx_mnrs_eth_recv_packet_eth0(VOID) { if (get_ethmac_mac_ctrl_rx_pending(ethmac0)) { _nx_mnrs_eth_recv_packet(0, ethmac0); } } VOID _nx_mnrs_eth_recv_packet_eth1(VOID) { if (get_ethmac_mac_ctrl_rx_pending(ethmac1)) { _nx_mnrs_eth_recv_packet(1, ethmac1); } }