From mboxrd@z Thu Jan 1 00:00:00 1970 From: Joe Hershberger Date: Wed, 7 Mar 2018 12:10:45 -0600 Subject: [U-Boot] [PATCH v7 2/4] Integrating TCP and wget into u-boot v7 In-Reply-To: <20180302030000.8129-2-DH@synoia.com> References: <20180302030000.8129-1-DH@synoia.com> <20180302030000.8129-2-DH@synoia.com> Message-ID: List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: u-boot@lists.denx.de Hi Duncan, The subject of all these patches ("Integrating TCP and wget into u-boot") is what should be the title of a cover letter. You need to give each patch a reasonable subject about that patch, not all of them together. On Thu, Mar 1, 2018 at 8:59 PM, wrote: > From: Duncan Hare > > All the code is new, and not copied from any source. This is what should be inside the "Commit-notes:" section... All the things you currently have there should be outside. > >>>>>> > <<<< > > Commit-notes You are missing a ':' so the tool is not detecting this. > > Routine tcp_print_buffer() is used to print portions of > non zero terminated buffers. If there is an existing routine > please let me know. I'm from the world of length fields > not zero terminated strings (zOS). > > TCP Prtocol stack, > > Why TCP: > > Currently u-boot file transfers are done using tftp or NFS, > both over udp. This requires a request to be sent from client > (u-boot) to the boot server for every block of data. > > For a 4 Mbyte kernel, with a 1k block size this requires > 4,000 request for a block. > > Using a large block size, one greater than the Ethernet > maximum frame size limitation, would require fragmentation, > which u-boot supports. However, missing fragment recovery > requires timeout detection and re-transmission requests > for missing fragments. > > UDP is ideally suited to fast single packet exchanges, > inquiry/response, for example dns, becuse of the lack of > connection overhead. > > UDP as a file transport mechanism is slow, even in low > latency networks, because file transfer with udp requires > poll/response mechanism to provide transfer integrity. > > In networks with large latency, for example: the internet, > UDP is even slower. What is a 30 second transfer on a local > boot server and LAN increase to over 3 minutes, because of > all the requests/response traffic. > > This was anticipated in the evolution of the IP protocols > and TCP was developed and then enhanced for high latency high > bandwidth networks. > > The current standard is TCP with selective acknowledgment. > > In our testing we have reduce kernel transmission time to > around 0.4 seconds for a 4Mbyte kernel, with a 100 Mbps > downlink. > > END > > Signed-off-by: Duncan Hare > --- > > Changes in v7: > Integrating TCP and wget into u-boot A much more appropriate thing to have put here would be something like "Fixed naming of x variable" and "Changed file location of z.h". The point of this is what you changed in this version of the patch. > > include/net.h | 8 +- > include/net/tcp.h | 218 ++++++++++++++++ > net/Kconfig | 35 +-- > net/Makefile | 2 +- > net/net.c | 51 +++- > net/tcp.c | 749 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ > 6 files changed, 1024 insertions(+), 39 deletions(-) > create mode 100644 include/net/tcp.h > create mode 100644 net/tcp.c > > diff --git a/include/net.h b/include/net.h > index 7e5f5a6a5b..e29d804a23 100644 > --- a/include/net.h > +++ b/include/net.h > @@ -548,7 +548,7 @@ extern int net_restart_wrap; /* Tried all network devices */ > > enum proto_t { > BOOTP, RARP, ARP, TFTPGET, DHCP, PING, DNS, NFS, CDP, NETCONS, SNTP, > - TFTPSRV, TFTPPUT, LINKLOCAL > + TFTPSRV, TFTPPUT, LINKLOCAL, WGET > }; > > extern char net_boot_file_name[1024];/* Boot File name */ > @@ -681,11 +681,15 @@ static inline void net_send_packet(uchar *pkt, int len) > * @param payload_len Length of data after the UDP header > */ > int net_send_ip_packet(uchar *ether, struct in_addr dest, int dport, int sport, > - int payload_len, int proto); > + int payload_len, int proto, u8 action, u32 tcp_seq_num, > + u32 tcp_ack_num); > > int net_send_udp_packet(uchar *ether, struct in_addr dest, int dport, > int sport, int payload_len); > > +int net_send_tcp_packet(int payload_len, int dport, int sport, u8 action, > + u32 tcp_seq_num, u32 tcp_ack_num); > + > /* Processes a received packet */ > void net_process_received_packet(uchar *in_packet, int len); > > diff --git a/include/net/tcp.h b/include/net/tcp.h > new file mode 100644 > index 0000000000..81f263351e > --- /dev/null > +++ b/include/net/tcp.h > @@ -0,0 +1,218 @@ > +/* > + * TCP Support for file transfer. > + * > + * Copyright 2017 Duncan Hare, All rights reserved. > + * > + * SPDX-License-Identifier: GPL-2.0 > + */ > + > +#define TCP_ACTIVITY 127 /* Activity on downloading */ > + > +struct ip_tcp_hdr { > + u8 ip_hl_v; /* header length and version */ > + u8 ip_tos; /* type of service */ > + u16 ip_len; /* total length */ > + u16 ip_id; /* identification */ > + u16 ip_off; /* fragment offset field */ > + u8 ip_ttl; /* time to live */ > + u8 ip_p; /* protocol */ > + u16 ip_sum; /* checksum */ > + struct in_addr ip_src; /* Source IP address */ > + struct in_addr ip_dst; /* Destination IP address */ > + u16 tcp_src; /* TCP source port */ > + u16 tcp_dst; /* TCP destination port */ > + u32 tcp_seq; /* TCP sequence number */ > + u32 tcp_ack; /* TCP Acknowledgment number */ > + u8 tcp_hlen; /* 4 bits TCP header Length/4 */ > + /* 4 bits Reserved */ > + /* 2 more bits reserved */ > + u8 tcp_flags; /* see defines */ > + u16 tcp_win; /* TCP windows size */ > + u16 tcp_xsum; /* Checksum */ > + u16 tcp_ugr; /* Pointer to urgent data */ > +} __packed; > + > +#define IP_TCP_HDR_SIZE (sizeof(struct ip_tcp_hdr)) > +#define TCP_HDR_SIZE (IP_TCP_HDR_SIZE - IP_HDR_SIZE) > + > +#define TCP_DATA 0x00 /* Data Packet - internal use only */ > +#define TCP_FIN 0x01 /* Finish flag */ > +#define TCP_SYN 0x02 /* Synch (start) flag */ > +#define TCP_RST 0x04 /* reset flag */ > +#define TCP_PUSH 0x08 /* Push - Notify app */ > +#define TCP_ACK 0x10 /* Acknowledgment of data received */ > +#define TCP_URG 0x20 /* Urgent */ > +#define TCP_ECE 0x40 /* Congestion control */ > +#define TCP_CWR 0x80 /* Congestion Control */ > + > +/* > + * TCP header options, Seq, MSS, and SACK > + */ > + > +#define TCP_SACK 32 /* Number of packets analyzed */ > + /* on leading edge of stream */ > + > +#define TCP_O_END 0x00 /* End of option list */ > +#define TCP_1_NOP 0x01 /* Single padding NOP */ > +#define TCP_O_NOP 0x01010101 /* NOPs pad to 32 bit boundary */ > +#define TCP_O_MSS 0x02 /* MSS Size option */ > +#define TCP_O_SCL 0x03 /* Window Scale option */ > +#define TCP_P_SACK 0x04 /* SACK permitted */ > +#define TCP_V_SACK 0x05 /* SACK values */ > +#define TCP_O_TS 0x08 /* Timestanp option */ > +#define TCP_OPT_LEN_2 0x02 > +#define TCP_OPT_LEN_3 0x03 > +#define TCP_OPT_LEN_4 0x04 > +#define TCP_OPT_LEN_6 0x06 > +#define TCP_OPT_LEN_8 0x08 > +#define TCP_OPT_LEN_A 0x0a /* Timestamp Length */ > + > +/* > + * Please review the warning in net.c about these two parameters. > + * They are part of a promise of RX buffer size to the sending TCP > + */ > + > +#define TCP_MSS 1460 /* Max segment size - 1460 */ > +#define TCP_SCALE 0x01 /* Scale 1 */ > + > +struct tcp_mss { /* TCP Mex Segment size */ > + u8 kind; /* 0x02 */ > + u8 len; /* 0x04 */ > + u16 mss; /* 1460 - Max segment size */ > +} __packed; > + > +struct tcp_scale { /* TCP Windows Scale */ > + u8 kind; /* 0x03 */ > + u8 len; /* 0x03 */ > + u8 scale; /* win shift fat fast networks */ > +} __packed; > + > +struct tcp_sack_p { /* SACK permitted */ > + u8 kind; /* 0x04 */ > + u8 len; /* Length */ > +} __packed; > + > +struct sack_edges { > + u32 l; > + u32 r; > +} __packed; > + > +#define TCP_SACK_SIZE (sizeof(struct sack_edges)) > + > +#define TCP_SACK_HILLS 4 > + > +struct tcp_sack_v { > + u8 kind; /* 0x05 */ > + u8 len; /* Length */ > + struct sack_edges hill[TCP_SACK_HILLS]; /* L & R widow edges */ > +} __packed; > + > +struct tcp_t_opt { /* TCP time stamps option */ > + u8 kind; /* 0x08 */ > + u8 len; /* 0x0a */ > + u32 t_snd; /* Sender timestamp */ > + u32 t_rcv; /* Receiver timestamp */ > +} __packed; > + > +#define TCP_TSOPT_SIZE (sizeof(struct tcp_t_opt)) > + > +/* > + * ip tcp structure with options > + */ > + > +struct ip_tcp_hdr_o { > + struct ip_tcp_hdr hdr; > + struct tcp_mss mss; > + struct tcp_scale scale; > + struct tcp_sack_p sack_p; > + struct tcp_t_opt t_opt; > + u8 end; > +} __packed; > + > +#define IP_TCP_O_SIZE (sizeof(struct ip_tcp_hdr_o)) > + > +struct ip_tcp_hdr_s { > + struct ip_tcp_hdr hdr; > + struct tcp_t_opt t_opt; > + struct tcp_sack_v sack_v; > + u8 end; > +} __packed; > + > +#define IP_TCP_SACK_SIZE (sizeof(struct ip_tcp_hdr_s)) > + > +/* > + * TCP pseudo header definitions > + */ > +#define PSEUDO_PAD_SIZE 8 > + > +struct pseudo_hdr { > + u8 padding[PSEUDO_PAD_SIZE]; /* pseudo hdr size = ip_tcp hdr size */ > + struct in_addr p_src; > + struct in_addr p_dst; > + u8 rsvd; > + u8 p; > + u16 len; > +} __packed; > + > +#define PSEUDO_HDR_SIZE (sizeof(struct pseudo_hdr)) - PSEUDO_PAD_SIZE > + > +/* > + * union for building IP/TCP packet. > + * build Pseudo header in packed buffer first, calculate TCP checksum > + * then build IP header in packed buffer. > + */ > + > +union tcp_build_pkt { > + struct pseudo_hdr ph; > + struct ip_tcp_hdr_o ip; > + struct ip_tcp_hdr_s sack; > + uchar raw[1600]; > +} __packed; > + > +/* > + * TCP STATE MACHINE STATES FOR SOCKET > + */ > + > +enum TCP_STATE { > + TCP_CLOSED, /* Need to send SYN to connect */ > + TCP_SYN_SENT, /* Trying to connect, waiting for SYN ACK */ > + TCP_ESTABLISHED, /* both server & client have a connection */ > + TCP_CLOSE_WAIT, /* Rec FIN, passed to app for FIN, ACK rsp*/ > + TCP_CLOSING, /* Rec FIN, sent FIN, ACK waiting for ACK */ > + TCP_FIN_WAIT_1, /* Sent FIN waiting for response */ > + TCP_FIN_WAIT_2 /* Rec ACK from FIN sent, waiting for FIN */ > +}; > + > +int tcp_find_in_buffer(uchar raw[], int payload_len, uchar field[], > + int field_len); > +void tcp_print_buffer(uchar raw[], int pkt_len, int payload_len, > + int hdr_len, bool hide); > +enum TCP_STATE tcp_get_tcp_state(void); > +void tcp_set_tcp_state(enum TCP_STATE new_state); > +int tcp_set_tcp_header(uchar *pkt, int dport, int sport, int payload_len, > + u8 action, u32 tcp_seq_num, u32 tcp_ack_num); > + > +/* > + * An incoming packet handler. > + * @param pkt pointer to the application packet > + * @param dport destination UDP port > + * @param sip source IP address > + * @param sport source UDP port > + * @param len packet length > + */ > +typedef void rxhand_tcp(uchar *pkt, unsigned int dport, > + struct in_addr sip, unsigned int sport, > + unsigned int len); > +void tcp_set_tcp_handler(rxhand_tcp *f); > + > +void rxhand_tcp_f(union tcp_build_pkt *b, unsigned int len); > + > +/* > + * An incoming TCP packet handler for the TCP protocol. > + * There is also a dynamic function pointer for TCP based commands to > + * receive incoming traffic after the TCP protocol code has done its work. > + */ > + > +void rxhand_action(u8 tcp_action, int payload_len, u32 tcp_seq_num, > + u32 tcp_ack_num, unsigned int pkt_len, > + union tcp_build_pkt *b); > diff --git a/net/Kconfig b/net/Kconfig > index 143c4416cd..895b93dea9 100644 > --- a/net/Kconfig > +++ b/net/Kconfig > @@ -4,12 +4,12 @@ > > menuconfig NET > bool "Networking support" > - default y > > if NET > > config NET_RANDOM_ETHADDR > bool "Random ethaddr if unset" > + select LIB_RAND > help > Selecting this will allow the Ethernet interface to function > even when the ethaddr variable for that interface is unset. > @@ -24,7 +24,6 @@ config NETCONSOLE > > config NET_TFTP_VARS > bool "Control TFTP timeout and count through environment" > - depends on CMD_NET > default y > help > If set, allows controlling the TFTP timeout through the > @@ -33,44 +32,22 @@ config NET_TFTP_VARS > If unset, timeout and maximum are hard-defined as 1 second > and 10 timouts per TFTP transfer. > > -config BOOTP_BOOTPATH > - bool "Enable BOOTP BOOTPATH" > - depends on CMD_NET > - > -config BOOTP_DNS > - bool "Enable bootp DNS" > - depends on CMD_NET > - > -config BOOTP_GATEWAY > - bool "Enable BOOTP gateway" > - depends on CMD_NET > - > -config BOOTP_HOSTNAME > - bool "Enable BOOTP hostname" > - depends on CMD_NET > - > -config BOOTP_PXE > - bool "Enable BOOTP PXE" > - depends on CMD_NET > - > -config BOOTP_SUBNETMASK > - bool "Enable BOOTP subnetmask" > - depends on CMD_NET > - depends on CMD_NET > - > config BOOTP_PXE_CLIENTARCH > hex > - depends on CMD_NET > default 0x16 if ARM64 > default 0x15 if ARM > default 0 if X86 > > config BOOTP_VCI_STRING > string > - depends on CMD_NET > default "U-Boot.armv7" if CPU_V7 || CPU_V7M > default "U-Boot.armv8" if ARM64 > default "U-Boot.arm" if ARM > default "U-Boot" > > +config TCP > + bool "Include TCP for receiving files" > + help > + TCP support for receiving files. > + > endif # if NET > diff --git a/net/Makefile b/net/Makefile > index ae54eee5af..25729ebeeb 100644 > --- a/net/Makefile > +++ b/net/Makefile > @@ -25,7 +25,7 @@ obj-$(CONFIG_CMD_PING) += ping.o > obj-$(CONFIG_CMD_RARP) += rarp.o > obj-$(CONFIG_CMD_SNTP) += sntp.o > obj-$(CONFIG_CMD_NET) += tftp.o > - > +obj-$(CONFIG_TCP) += tcp.o > # Disable this warning as it is triggered by: > # sprintf(buf, index ? "foo%d" : "foo", index) > # and this is intentional usage. > diff --git a/net/net.c b/net/net.c > index df4e7317e7..7adc4b472c 100644 > --- a/net/net.c > +++ b/net/net.c > @@ -107,6 +107,9 @@ > #if defined(CONFIG_CMD_SNTP) > #include "sntp.h" > #endif > +#if defined(CONFIG_TCP) > +#include > +#endif > > DECLARE_GLOBAL_DATA_PTR; > > @@ -382,6 +385,9 @@ void net_init(void) > > /* Only need to setup buffer pointers once. */ > first_call = 0; > +#if defined(CONFIG_TCP) > + tcp_set_tcp_state(TCP_CLOSED); > +#endif > } > > net_init_loop(); > @@ -781,11 +787,22 @@ int net_send_udp_packet(uchar *ether, struct in_addr dest, int dport, int sport, > int payload_len) > { > return net_send_ip_packet(ether, dest, dport, sport, payload_len, > - IPPROTO_UDP); > + IPPROTO_UDP, 0, 0, 0); > +} > + > +#if defined(CONFIG_TCP) > +int net_send_tcp_packet(int payload_len, int dport, int sport, u8 action, > + u32 tcp_seq_num, u32 tcp_ack_num) > +{ > + return net_send_ip_packet(net_server_ethaddr, net_server_ip, dport, > + sport, payload_len, IPPROTO_TCP, action, > + tcp_seq_num, tcp_ack_num); > } > +#endif > > int net_send_ip_packet(uchar *ether, struct in_addr dest, int dport, int sport, > - int payload_len, int proto) > + int payload_len, int proto, u8 action, u32 tcp_seq_num, > + u32 tcp_ack_num) > { > uchar *pkt; > int eth_hdr_size; > @@ -794,6 +811,13 @@ int net_send_ip_packet(uchar *ether, struct in_addr dest, int dport, int sport, > debug_cond(DEBUG_DEV_PKT, > "UDP Send (to=%pI4, from=%pI4, len=%d)\n", > &dest, &net_ip, payload_len); > +#if defined(CONFIG_TCP) > + > + } else { > + debug_cond(DEBUG_DEV_PKT, > + "TCP Send (%pI4, %pI4, len=%d, A=%x)\n", > + &dest, &net_ip, payload_len, action); > +#endif > } > > /* make sure the net_tx_packet is initialized (net_init() was called) */ > @@ -812,15 +836,19 @@ int net_send_ip_packet(uchar *ether, struct in_addr dest, int dport, int sport, > pkt = (uchar *)net_tx_packet; > > eth_hdr_size = net_set_ether(pkt, ether, PROT_IP); > - switch (proto) { > - case IPPROTO_UDP: > + if (proto == IPPROTO_UDP) { > net_set_udp_header(pkt + eth_hdr_size, dest, > dport, sport, payload_len); > pkt_hdr_size = eth_hdr_size + IP_UDP_HDR_SIZE; > - break; > - default: > - return -1; > + > +#if defined(CONFIG_TCP) > + } else { > + pkt_hdr_size = eth_hdr_size + > + tcp_set_tcp_header(pkt + eth_hdr_size, dport, sport, > + payload_len, action, > + tcp_seq_num, tcp_ack_num); > } > +#endif > > /* if MAC address was not discovered yet, do an ARP request */ > if (memcmp(ether, net_null_ethaddr, 6) == 0) { > @@ -1225,6 +1253,15 @@ void net_process_received_packet(uchar *in_packet, int len) > if (ip->ip_p == IPPROTO_ICMP) { > receive_icmp(ip, len, src_ip, et); > return; > +#if defined(CONFIG_TCP) > + } else if (ip->ip_p == IPPROTO_TCP) { > + debug_cond(DEBUG_DEV_PKT, > + "TCP PH (to=%pI4, from=%pI4, len=%d)\n", > + &dst_ip, &src_ip, len); > + > + rxhand_tcp_f((union tcp_build_pkt *)ip, len); > + return; > +#endif > } else if (ip->ip_p != IPPROTO_UDP) { /* Only UDP packets */ > return; > } > diff --git a/net/tcp.c b/net/tcp.c > new file mode 100644 > index 0000000000..612a6a4b7d > --- /dev/null > +++ b/net/tcp.c > @@ -0,0 +1,749 @@ > +/* > + * Copyright 2017 Duncan Hare, all rights reserved. > + * SPDX-License-Identifier: GPL-2.0 > + */ > + > +/* > + * General Desription: > + * > + * TCP support for the wget command, for fast file downloading. > + * > + * HTTP/TCP Receiver: > + * > + * Prequeisites: - own ethernet address > + * - own IP address > + * - Server IP address > + * - Server with TCP > + * - TCP application (eg wget) > + * Next Step HTTPS? > + */ > +#include > +#include > +#include > +#include > +#include > +#include > +#include > +#if defined(CONFIG_CMD_WGET) > +#include "net/wget.h" > +#endif > + > +/* > + * TCP sliding window control > + * used by us to request re-TX > + */ > + > +static struct tcp_sack_v tcp_lost; > + > +/* TCP option timestamp */ > +static u32 loc_timestamp; > +static u32 rmt_timestamp; > + > +u32 tcp_seq_init; > +u32 tcp_ack_edge; > +u32 tcp_seq_max; > + > +int tcp_activity_count; > + > +/* > + * Search for TCP_SACK and review the > + * comments before the code section > + * TCP_SACK is the number of packets > + * at the front of the stream > + */ > + > +enum pkt_state {PKT, NOPKT}; > +struct sack_r { > + struct sack_edges se; > + enum pkt_state st; > +}; > + > +struct sack_r edge_a[TCP_SACK]; > +unsigned int sack_idx; > +unsigned int prev_len; > + > +/* TCP connection state */ > +static enum TCP_STATE tcp_state; > + > +/* > + * An incoming TCP packet handler for the TCP protocol. > + * There is also a dymanic function pointer for TCP based commads to > + * receive incoming traffic after the TCP protocol code has done its work. > + */ > + > +/* Current TCP RX packet handler */ > +static rxhand_tcp *tcp_packet_handler; > + > +enum TCP_STATE tcp_get_tcp_state(void) > +{ > + return tcp_state; > +} > + > +void tcp_set_tcp_state(enum TCP_STATE new_state) > +{ > + tcp_state = new_state; > +} > + > +void tcp_print_buffer(uchar raw[], int pkt_len, int payload_len, > + int hdr_len, bool hide) > +{ > + int i; > + > + for (i = pkt_len - payload_len; i < pkt_len; i++) { > + if (i <= hdr_len) > + printf("%02X", raw[i]); > + else if ((raw[i] > 0x19) && (raw[i] < 0x7f)) > + putc(raw[i]); > + else if (hide == 0) > + putc(raw[i]); > + else > + printf("%02X", raw[i]); > + } > + printf("%s", "\n"); > +} > + > +int tcp_find_in_buffer(uchar raw[], int payload_len, uchar field[], > + int field_len) > +{ > + int i, j; > + > + for (i = 0; i < payload_len; i++) { > + if (raw[i] == field[0]) { > + for (j = 1; j < field_len; j++) { > + if (raw[i + j] != field[j]) > + break; > + if (j == field_len) > + return i; > + } > + } > + } > + return 0; > +} > + > +static void dummy_handler(uchar *pkt, unsigned int dport, > + struct in_addr sip, unsigned int sport, > + unsigned int len) > +{ > +} > + > +void tcp_set_tcp_handler(rxhand_tcp *f) > +{ > + debug_cond(DEBUG_INT_STATE, "--- net_loop TCP handler set (%p)\n", f); > + if (!f) > + tcp_packet_handler = dummy_handler; > + else > + tcp_packet_handler = f; > +} > + > +u16 tcp_set_pseudo_header(uchar *pkt, struct in_addr src, struct in_addr dest, > + int tcp_len, int pkt_len) > +{ > + union tcp_build_pkt *b = (union tcp_build_pkt *)pkt; > + int checksum_len; > + > + /* > + * Pseudo header > + * > + * Zero the byte after the last byte so that the header checksum > + * will always work. > + */ > + > + pkt[pkt_len] = 0x00; > + > + net_copy_ip((void *)&b->ph.p_src, &src); > + net_copy_ip((void *)&b->ph.p_dst, &dest); > + b->ph.rsvd = 0x00; > + b->ph.p = IPPROTO_TCP; > + b->ph.len = htons(tcp_len); > + checksum_len = tcp_len + PSEUDO_HDR_SIZE; > + > + debug_cond(DEBUG_DEV_PKT, > + "TCP Pesudo Header (to=%pI4, from=%pI4, Len=%d)\n", > + &b->ph.p_dst, &b->ph.p_src, checksum_len); > + > + return compute_ip_checksum(pkt + PSEUDO_PAD_SIZE, checksum_len); > +} > + > +int net_set_ack_options(union tcp_build_pkt *b) > +{ > + b->sack.hdr.tcp_hlen = (TCP_HDR_SIZE >> 2) << 4; > + > + b->sack.t_opt.kind = TCP_O_TS; > + b->sack.t_opt.len = TCP_OPT_LEN_A; > + b->sack.t_opt.t_snd = htons(loc_timestamp); > + b->sack.t_opt.t_rcv = rmt_timestamp; > + b->sack.sack_v.kind = 0x01; > + b->sack.sack_v.len = 0x00; > + > + if (tcp_lost.len > TCP_OPT_LEN_2) { > + debug_cond(DEBUG_DEV_PKT, "TCP ack opt lost.len %x\n", > + tcp_lost.len); > + b->sack.sack_v.len = tcp_lost.len; > + b->sack.sack_v.kind = TCP_V_SACK; > + b->sack.sack_v.hill[0].l = htonl(tcp_lost.hill[0].l); > + b->sack.sack_v.hill[0].r = htonl(tcp_lost.hill[0].r); > + > + /* > + * These SACK structures are initialized with NOPs to > + * provide TCP header alignment padding. There are 4 > + * SACK structures used for both header padding and > + * internally. > + */ > + > + b->sack.sack_v.hill[1].l = htonl(tcp_lost.hill[1].l); > + b->sack.sack_v.hill[1].r = htonl(tcp_lost.hill[1].r); > + b->sack.sack_v.hill[2].l = htonl(tcp_lost.hill[2].l); > + b->sack.sack_v.hill[2].r = htonl(tcp_lost.hill[2].r); > + b->sack.sack_v.hill[3].l = TCP_O_NOP; > + b->sack.sack_v.hill[3].r = TCP_O_NOP; > + } > + > + /* > + * TCP lengths are stored as a rounded up number of 32 bit words > + * Add 3 to length round up, rounded, then divided into the length > + * in 32 bit words. > + */ > + > + b->sack.hdr.tcp_hlen = (((TCP_HDR_SIZE + TCP_TSOPT_SIZE > + + tcp_lost.len + 3) >> 2) << 4); > + > + /* > + * This returns the actual rounded up length of the > + * TCP header to add to the total packet length > + */ > + > + return b->sack.hdr.tcp_hlen >> 2; > +} > + > +void net_set_syn_options(union tcp_build_pkt *b) > +{ > + tcp_lost.len = 0; > + b->ip.hdr.tcp_hlen = 0xa0; > + > + b->ip.mss.kind = TCP_O_MSS; > + b->ip.mss.len = TCP_OPT_LEN_4; > + b->ip.mss.mss = htons(TCP_MSS); > + b->ip.scale.kind = TCP_O_SCL; > + b->ip.scale.scale = TCP_SCALE; > + b->ip.scale.len = TCP_OPT_LEN_3; > + b->ip.sack_p.kind = TCP_P_SACK; > + b->ip.sack_p.len = TCP_OPT_LEN_2; > + b->ip.t_opt.kind = TCP_O_TS; > + b->ip.t_opt.len = TCP_OPT_LEN_A; > + loc_timestamp = get_ticks() % 3072; > + rmt_timestamp = 0x00000000; > + b->ip.t_opt.t_snd = 0; > + b->ip.t_opt.t_rcv = 0x00000000; > + b->ip.end = TCP_O_END; > +} > + > +int tcp_set_tcp_header(uchar *pkt, int dport, int sport, int payload_len, > + u8 action, u32 tcp_seq_num, u32 tcp_ack_num) > +{ > + union tcp_build_pkt *b = (union tcp_build_pkt *)pkt; > + int pkt_hdr_len; > + int pkt_len; > + int tcp_len; > + > + b->ip.hdr.tcp_flags = action; > + pkt_hdr_len = IP_TCP_HDR_SIZE; > + b->ip.hdr.tcp_hlen = 0x50; /* Header is 5 32 bit words */ > + /* 4 bits TCP header Length/4*/ > + /* 4 bits Reserved */ > + /* For options */ > + switch (action) { > + case TCP_SYN: > + debug_cond(DEBUG_DEV_PKT, > + "TCP Hdr:SYN (%pI4, %pI4, sq=%d, ak=%d)\n", > + &net_server_ip, &net_ip, > + tcp_seq_num, tcp_ack_num); > + tcp_activity_count = 0; > + net_set_syn_options(b); > + tcp_seq_num = 0; > + tcp_ack_num = 0; > + pkt_hdr_len = IP_TCP_O_SIZE; > + if (tcp_state == TCP_SYN_SENT) { /* Too many sins */ > + action = TCP_FIN; > + tcp_state = TCP_FIN_WAIT_1; > + } else { > + tcp_state = TCP_SYN_SENT; > + } > + break; > + case TCP_ACK: > + pkt_hdr_len = IP_HDR_SIZE + > + net_set_ack_options(b); > + b->ip.hdr.tcp_flags = action; > + debug_cond(DEBUG_DEV_PKT, > + "TCP Hdr:ACK (%pI4, %pI4, s=%d, a=%d, A=%x)\n", > + &net_server_ip, &net_ip, tcp_seq_num, tcp_ack_num, > + action); > + break; > + case TCP_FIN: > + debug_cond(DEBUG_DEV_PKT, > + "TCP Hdr:FIN (%pI4, %pI4, s=%d, a=%d)\n", > + &net_server_ip, &net_ip, tcp_seq_num, tcp_ack_num); > + payload_len = 0; > + pkt_hdr_len = IP_TCP_HDR_SIZE; > + tcp_state = TCP_FIN_WAIT_1; > + > + break; > + > + /* Notify connection closing */ > + > + case (TCP_FIN | TCP_ACK): > + case ((TCP_FIN | TCP_ACK) | TCP_PUSH): > + if (tcp_state == TCP_CLOSE_WAIT) > + tcp_state = TCP_CLOSING; > + tcp_ack_edge++; > + debug_cond(DEBUG_DEV_PKT, > + "TCP Hdr:FIN ACK PSH(%pI4, %pI4, s=%d, a=%d, A=%x)\n", > + &net_server_ip, &net_ip, > + tcp_seq_num, tcp_ack_edge, action); > + /* FALLTHRU */ > + default: > + pkt_hdr_len = IP_HDR_SIZE + > + net_set_ack_options(b); > + b->ip.hdr.tcp_flags = action | TCP_PUSH | TCP_ACK; > + debug_cond(DEBUG_DEV_PKT, > + "TCP Hdr:dft (%pI4, %pI4, s=%d, a=%d, A=%x)\n", > + &net_server_ip, &net_ip, > + tcp_seq_num, tcp_ack_num, action); > + } > + > + pkt_len = pkt_hdr_len + payload_len; > + tcp_len = pkt_len - IP_HDR_SIZE; > + > + /* TCP Header */ > + b->ip.hdr.tcp_ack = htonl(tcp_ack_edge); > + b->ip.hdr.tcp_src = htons(sport); > + b->ip.hdr.tcp_dst = htons(dport); > + b->ip.hdr.tcp_seq = htonl(tcp_seq_num); > + tcp_seq_num = tcp_seq_num + payload_len; > + > + /* > + * TCP window size - TCP header variable tcp_win. > + * Change tcp_win only if you have an understanding of network > + * overrun, congestion, TCP segment sizes, TCP windows, TCP scale, > + * queuing theory and packet buffering. If there are too few buffers, > + * there will be data loss, recovery may work or the sending TCP, > + * the server, could abort the stream transmission. > + * MSS is governed by maximum Ethernet frame length. > + * The number of buffers is governed by the desire to have a queue of > + * full buffers to be processed at the destination to maximize > + * throughput. Temporary memory use for the boot phase on modern > + * SOCs is may not be considered a constraint to buffer space, if > + * it is, then the u-boot tftp or nfs kernel netboot should be > + * considered. > + */ > + > + b->ip.hdr.tcp_win = htons(PKTBUFSRX * TCP_MSS >> TCP_SCALE); > + > + b->ip.hdr.tcp_xsum = 0x0000; > + b->ip.hdr.tcp_ugr = 0x0000; > + > + b->ip.hdr.tcp_xsum = tcp_set_pseudo_header(pkt, net_ip, net_server_ip, > + tcp_len, pkt_len); > + > + /* > + * IP Header > + */ > + > + net_set_ip_header((uchar *)&b->ip, net_server_ip, net_ip, > + pkt_len, IPPROTO_TCP); > + > + return pkt_hdr_len; > +} > + > +/* > + * Selective Acknowledgment (Essential for fast stream transfer) > + * > + * Before modifying this section of code, which the author found difficult > + * to write, please be familiar with the SACK (Selective Acknowledgment) RFCs. > + */ > + > +void tcp_hole(u32 tcp_seq_num, u32 len, u32 tcp_seq_max) > +{ > + unsigned int idx_sack; > + unsigned int sack_end = TCP_SACK - 1; > + unsigned int sack_in; > + unsigned int hill = 0; > + enum pkt_state expect = PKT; > + > + u32 seq = tcp_seq_num - tcp_seq_init; > + u32 hol_l = tcp_ack_edge - tcp_seq_init; > + u32 hol_r = 0; > + > + /* Place seq number in correct place */ > + > + if (prev_len == 0) > + prev_len = len; > + idx_sack = sack_idx + ((tcp_seq_num - tcp_ack_edge) / prev_len); > + if (idx_sack < TCP_SACK) { > + edge_a[idx_sack].se.l = tcp_seq_num; > + edge_a[idx_sack].se.r = tcp_seq_num + len; > + edge_a[idx_sack].st = PKT; > + > + /* The fin (last) packet is not the same length as data packets, > + * and if it's length is recorded and used for array index > + * the array index calculations break. > + */ > + if (prev_len < len) > + prev_len = len; > + } > + > + debug_cond(DEBUG_DEV_PKT, > + "TCP 1 seq %d, edg %d, len %d, sack_idx %d, sack_end %d\n", > + seq, hol_l, len, sack_idx, sack_end); > + > + /* Right edge of contiguous stream, is the left edge of first hill */ > + > + hol_l = tcp_seq_num - tcp_seq_init; > + hol_r = hol_l + len; > + > + tcp_lost.len = TCP_OPT_LEN_2; > + > + debug_cond(DEBUG_DEV_PKT, > + "TCP 1 in %d, seq %d, pkt_l %d, pkt_r %d, sack_idx %d, sack_end %d\n", > + idx_sack, seq, hol_l, hol_r, sack_idx, sack_end); > + > + for (sack_in = sack_idx; sack_in < sack_end && hill < TCP_SACK_HILLS; > + sack_in++) { > + switch (expect) { > + case NOPKT: > + switch (edge_a[sack_in].st) { > + case NOPKT: > + debug_cond(DEBUG_INT_STATE, "N"); > + break; > + case PKT: > + debug_cond(DEBUG_INT_STATE, "n"); > + tcp_lost.hill[hill].l = > + edge_a[sack_in].se.l; > + tcp_lost.hill[hill].r = > + edge_a[sack_in].se.r; > + expect = PKT; > + break; > + } > + break; > + case PKT: > + switch (edge_a[sack_in].st) { > + case NOPKT: > + debug_cond(DEBUG_INT_STATE, "p"); > + if ((sack_in > sack_idx) && > + (hill < TCP_SACK_HILLS)) { > + hill++; > + tcp_lost.len += TCP_OPT_LEN_8; > + } > + expect = NOPKT; > + break; > + case PKT: > + debug_cond(DEBUG_INT_STATE, "P"); > + > + if (tcp_ack_edge == edge_a[sack_in].se.l) { > + tcp_ack_edge = edge_a[sack_in].se.r; > + edge_a[sack_in].st = NOPKT; > + sack_idx++; > + } else { > + if (hill < TCP_SACK_HILLS) > + tcp_lost.hill[hill].r = > + edge_a[sack_in].se.r; > + if (sack_in == sack_end - 1) > + tcp_lost.hill[hill].r = > + edge_a[sack_in].se.r; > + } > + break; > + } > + break; > + } > + } > + debug_cond(DEBUG_INT_STATE, "\n"); > + if (tcp_lost.len <= TCP_OPT_LEN_2) > + sack_idx = 0; > +} > + > +void tcp_parse_options(uchar *o, int o_len) > +{ > + struct tcp_t_opt *tsopt; > + uchar *p = o; > + > + for (p = o; p < (o + o_len); p = p + p[1]) { > + if (p[1] != 0) { > + switch (p[0]) { > + case TCP_O_END: > + return; > + case TCP_O_MSS: > + break; > + case TCP_O_SCL: > + break; > + case TCP_P_SACK: > + break; > + case TCP_V_SACK: > + break; > + case TCP_O_TS: > + tsopt = (struct tcp_t_opt *)p; > + rmt_timestamp = tsopt->t_snd; > + return; > + break; > + } > + if (p[0] == TCP_O_NOP) > + p++; > + } else { > + return; > + } > + } > +} > + > +u8 tcp_state_machine(u8 tcp_flags, u32 *tcp_seq_num, int payload_len) > +{ > + u8 tcp_fin = tcp_flags & TCP_FIN; > + u8 tcp_syn = tcp_flags & TCP_SYN; > + u8 tcp_rst = tcp_flags & TCP_RST; > + u8 tcp_push = tcp_flags & TCP_PUSH; > + u8 tcp_ack = tcp_flags & TCP_ACK; > + u8 action = TCP_DATA; > + int i; > + > + /* > + * tcp_flags are examined to determine TX action in a given state > + * tcp_push is interpreted to mean "inform the app" > + * urg, ece, cer and nonce flags are not supported. > + * > + * exe and crw are use to signal and confirm knowledge of congestion. > + * This TCP only sends a file request and acks. If it generates > + * congestion, the network is broken. > + */ > + > + debug_cond(DEBUG_INT_STATE, "TCP STATE ENTRY %x\n", action); > + if (tcp_rst) { > + action = TCP_DATA; > + tcp_state = TCP_CLOSED; > + net_set_state(NETLOOP_FAIL); > + debug_cond(DEBUG_INT_STATE, "TCP Reset %x\n", tcp_flags); > + return TCP_RST; > + } > + > + switch (tcp_state) { > + case TCP_CLOSED: > + debug_cond(DEBUG_INT_STATE, "TCP CLOSED %x\n", tcp_flags); > + if (tcp_fin) > + action = TCP_DATA; > + if (tcp_syn) > + action = TCP_RST; > + if (tcp_ack) > + action = TCP_DATA; > + break; > + case TCP_SYN_SENT: > + debug_cond(DEBUG_INT_STATE, "TCP_SYN_SENT %x, %d\n", > + tcp_flags, *tcp_seq_num); > + if (tcp_fin) { > + action = action | TCP_PUSH; > + tcp_state = TCP_CLOSE_WAIT; > + } > + if (tcp_syn) { > + action = action | TCP_ACK | TCP_PUSH; > + if (tcp_ack) { > + tcp_seq_init = *tcp_seq_num; > + *tcp_seq_num = *tcp_seq_num + 1; > + tcp_seq_max = *tcp_seq_num; > + tcp_ack_edge = *tcp_seq_num; > + sack_idx = 0; > + edge_a[sack_idx].se.l = *tcp_seq_num; > + edge_a[sack_idx].se.r = *tcp_seq_num; > + prev_len = 0; > + tcp_state = TCP_ESTABLISHED; > + for (i = 0; i < TCP_SACK; i++) > + edge_a[i].st = NOPKT; > + } > + } else { > + if (tcp_ack) > + action = TCP_DATA; > + } > + break; > + case TCP_ESTABLISHED: > + debug_cond(DEBUG_INT_STATE, > + "TCP_ESTABLISHED %x\n", tcp_flags); > + if (*tcp_seq_num > tcp_seq_max) > + tcp_seq_max = *tcp_seq_num; > + if (payload_len > 0) { > + tcp_hole(*tcp_seq_num, payload_len, tcp_seq_max); > + tcp_fin = TCP_DATA; /* cause standalone FIN */ > + } > + > + if ((tcp_fin) && (tcp_lost.len <= TCP_OPT_LEN_2)) { > + action = action | TCP_FIN | TCP_PUSH | TCP_ACK; > + tcp_state = TCP_CLOSE_WAIT; > + } else { > + if (tcp_ack) > + action = TCP_DATA; > + } > + if (tcp_push) > + action = action | TCP_PUSH; > + if (tcp_syn) > + action = TCP_ACK + TCP_RST; > + break; > + case TCP_CLOSE_WAIT: > + debug_cond(DEBUG_INT_STATE, "TCP_CLOSE_WAIT (%x)\n", tcp_flags); > + action = TCP_DATA; /* Wait for app */ > + break; > + case TCP_FIN_WAIT_2: > + debug_cond(DEBUG_INT_STATE, "TCP_FIN_WAIT_2 (%x)\n", tcp_flags); > + if (tcp_fin) > + action = TCP_DATA; > + if (tcp_syn) > + action = TCP_DATA; > + if (tcp_ack) { > + action = TCP_PUSH | TCP_ACK; > + tcp_state = TCP_CLOSED; > + puts("\n"); > + } > + break; > + case TCP_FIN_WAIT_1: > + debug_cond(DEBUG_INT_STATE, "TCP_FIN_WAIT_1 (%x)\n", tcp_flags); > + if (tcp_fin) { > + action = TCP_ACK | TCP_FIN; > + tcp_state = TCP_FIN_WAIT_2; > + } > + if (tcp_syn) > + action = TCP_RST; > + if (tcp_ack) { > + tcp_state = TCP_CLOSED; > + tcp_seq_num = tcp_seq_num + 1; > + } > + break; > + case TCP_CLOSING: > + debug_cond(DEBUG_INT_STATE, "TCP_CLOSING (%x)\n", tcp_flags); > + if (tcp_fin) > + action = TCP_DATA; > + if (tcp_syn) > + action = TCP_RST; > + if (tcp_ack) { > + action = TCP_PUSH; > + tcp_state = TCP_CLOSED; > + puts("\n"); > + } > + break; > + } > + return action; > +} > + > +void rxhand_tcp_f(union tcp_build_pkt *b, unsigned int pkt_len) > +{ > + int tcp_len = pkt_len - IP_HDR_SIZE; > + u16 tcp_rx_xsum = b->ip.hdr.ip_sum; > + u8 tcp_action = TCP_DATA; > + u32 tcp_seq_num; > + u32 tcp_ack_num; > + struct in_addr action_and_state; > + > + int tcp_hdr_len; > + int payload_len; > + > + /* > + * Verify IP header > + */ > + debug_cond(DEBUG_DEV_PKT, > + "TCP RX in RX Sum (to=%pI4, from=%pI4, len=%d)\n", > + &b->ip.hdr.ip_src, &b->ip.hdr.ip_dst, pkt_len); > + > + debug_cond(DEBUG_DEV_PKT, > + "In__________________________________________\n"); > + > + b->ip.hdr.ip_src = net_server_ip; > + b->ip.hdr.ip_dst = net_ip; > + b->ip.hdr.ip_sum = 0x0000; > + if (tcp_rx_xsum != compute_ip_checksum(b, IP_HDR_SIZE)) { > + debug_cond(DEBUG_DEV_PKT, > + "TCP RX IP xSum Error (%pI4, =%pI4, len=%d)\n", > + &net_ip, &net_server_ip, pkt_len); > + return; > + } > + > + /* > + * Build Pseudo header and Verify TCP header > + */ > + tcp_rx_xsum = b->ip.hdr.tcp_xsum; > + b->ip.hdr.tcp_xsum = 0x0000; > + if (tcp_rx_xsum != tcp_set_pseudo_header((uchar *)b, b->ip.hdr.ip_src, > + b->ip.hdr.ip_dst, tcp_len, > + pkt_len)) { > + debug_cond(DEBUG_DEV_PKT, > + "TCP RX TCP xSum Error (%pI4, %pI4, len=%d)\n", > + &net_ip, &net_server_ip, tcp_len); > + return; > + } > + > + tcp_hdr_len = (b->ip.hdr.tcp_hlen >> 2); > + payload_len = tcp_len - tcp_hdr_len; > + > + if (tcp_hdr_len > TCP_HDR_SIZE) > + tcp_parse_options((uchar *)b + IP_TCP_HDR_SIZE, > + tcp_hdr_len - TCP_HDR_SIZE); > + /* > + * Incoming sequence and ack numbers are server's view of the numbers. > + * The app must swap the numbers when responding. > + */ > + > + tcp_seq_num = ntohl(b->ip.hdr.tcp_seq); > + tcp_ack_num = ntohl(b->ip.hdr.tcp_ack); > + > + /* > + * Sent first packet send from server as first to app > + * Other packets could be ordered, but are currently not ordered. > + */ > + > + tcp_action = tcp_state_machine(b->ip.hdr.tcp_flags, > + &tcp_seq_num, payload_len); > + > + /* > + * State altering command to be sent. > + * The packet sequence and ack numbers are in the tcp_seq_num > + * and tcp_ack_num variables. The current packet, its position > + * in the data stream, is the in the range of those variables. > + * > + * In the "application push" invocation, the TCP header with all > + * its information is pointed to by the packet pointer. > + * > + * in the typedef > + * void rxhand_tcp(uchar *pkt, unsigned int dport, > + * struct in_addr sip, unsigned int sport, > + * unsigned int len); > + * *pkt is the pointer to the payload > + * dport is used for tcp_seg_num > + * action_and_state.s_addr is used for TCP state > + * sport is used for tcp_ack_num (which is unused by the app) > + * pkt_ length is the payload length. > + * > + * TCP_PUSH from the state machine with a payload length of 0 is a > + * connect or disconnect event > + */ > + > + tcp_activity_count++; > + if (tcp_activity_count > TCP_ACTIVITY) { > + puts("| "); > + tcp_activity_count = 0; > + } > + > + if ((tcp_action & TCP_PUSH) || (payload_len > 0)) { > + debug_cond(DEBUG_DEV_PKT, > + "TCP Notify (action=%x, Seq=%d,Ack=%d,Pay%d)\n", > + tcp_action, tcp_seq_num, tcp_ack_num, payload_len); > + > + action_and_state.s_addr = tcp_action; > + (*tcp_packet_handler) ((uchar *)b + pkt_len - payload_len, > + tcp_seq_num, action_and_state, > + tcp_ack_num, payload_len); > + > + } else if (tcp_action != TCP_DATA) { > + debug_cond(DEBUG_DEV_PKT, > + "TCP Action (action=%x,Seq=%d,Ack=%d,Pay=%d)\n", > + tcp_action, tcp_seq_num, tcp_ack_num, payload_len); > + > + /* > + * Warning: Incoming Seq & Ack sequence numbers are transposed here > + * to outgoing Seq & Ack sequence numbers > + */ > + net_send_tcp_packet(0, ntohs(b->ip.hdr.tcp_src), > + ntohs(b->ip.hdr.tcp_dst), > + (tcp_action & (~TCP_PUSH)), > + tcp_seq_num, tcp_ack_num); > + } > +} > -- > 2.11.0 > > _______________________________________________ > U-Boot mailing list > U-Boot at lists.denx.de > https://lists.denx.de/listinfo/u-boot