Re: [U-Boot] [PATCH] TCP and wget implementation v4
A note on this TCP implementation. In TCP the transmitting TCP guarantees delivery of a stream, and the receiving TCP guarantees ordered of delivery of the stream. In this implementation The kernel memory buffer and the TCP sequence number is used to order the stream. for the application, and the application is the kernel itself. wget is not considered the application, and does receive packets "out of order."
It seems like it would be possible to just store off a packet that is ahead of its neighbor and not call any upper handler until the needed packet arrives. Then all upper layers wouldn't need to know about the reordering.
There is, for u-boot a large number of buffers used on RX, in order to have a long work queue of kernel data.
CONFIG_SYS_RX_ETH_BUFFER 50
The TCP transmit window is approximately 25 such packets, so overruns are eliminated.
There are about 3300 kernel data packets in this test kernel, so missing a few, 3 to 5, has little impact on elapsed time, and they remain in the input buffer pool ahead of the HTTP header,
The only critical order for this implementation is the TCP header. Without processing the TCP header we do not know the stream address of the first byte of kernel data. It is easy when we know this to map TCP stream address to kernel data offset.
Advice on the reset buffer approach are invited. It requires an interface between the wget application to reset the buffer index.
Between wget and what? The TCP implementation? It seems like something that should be abstracted from wget.
wget receives packets without intermediate ordering. Ordering data is a function of wget placing the kernel data at the correct offset, based of TCP stream location, at the correct memory address.
wget is the agent which correctly orders data. There is no socket analogue, the abstraction, and a socket's linked list buffer reordering, which is the generally recognized reordering point for TCP data.
Correct ordering is a primary task of this wget implementation, becuse the destination is a kernel image, and this choice very greatly simplifies the TCP stack, while reducing code complexity, and limits code size.
Reordering would require a new piece of code similar to the fragment assemble piece of code in net.h, and that's an amazing, but complex, piece of code. My objective was to produce something as simple as possible.
The TCP stack delivers packets in the order they come off the wire. Wget puts them in the correct order based on TCP sequence number to build the kernel image, and the TCP sequence number/ack protocol ensures complete delivery of a stream.
On Fri, Jan 5, 2018 at 4:10 PM, Duncan Hare dh@synoia.com wrote:
A note on this TCP implementation. In TCP the transmitting TCP guarantees delivery of a stream, and the receiving TCP guarantees ordered of delivery of the stream. In this implementation The kernel memory buffer and the TCP sequence number is used to order the stream. for the application, and the application is the kernel itself. wget is not considered the application, and does receive packets "out of order."
It seems like it would be possible to just store off a packet that is ahead of its neighbor and not call any upper handler until the needed packet arrives. Then all upper layers wouldn't need to know about the reordering.
There is, for u-boot a large number of buffers used on RX, in order to have a long work queue of kernel data.
CONFIG_SYS_RX_ETH_BUFFER 50
The TCP transmit window is approximately 25 such packets, so overruns are eliminated.
There are about 3300 kernel data packets in this test kernel, so missing a few, 3 to 5, has little impact on elapsed time, and they remain in the input buffer pool ahead of the HTTP header,
The only critical order for this implementation is the TCP header. Without processing the TCP header we do not know the stream address of the first byte of kernel data. It is easy when we know this to map TCP stream address to kernel data offset.
Advice on the reset buffer approach are invited. It requires an interface between the wget application to reset the buffer index.
Between wget and what? The TCP implementation? It seems like something that should be abstracted from wget.
wget receives packets without intermediate ordering. Ordering data is a function of wget placing the kernel data at the correct offset, based of TCP stream location, at the correct memory address.
wget is the agent which correctly orders data. There is no socket analogue, the abstraction, and a socket's linked list buffer reordering, which is the generally recognized reordering point for TCP data.
Correct ordering is a primary task of this wget implementation, becuse the destination is a kernel image, and this choice very greatly simplifies the TCP stack, while reducing code complexity, and limits code size.
Reordering would require a new piece of code similar to the fragment assemble piece of code in net.h, and that's an amazing, but complex, piece of code. My objective was to produce something as simple as possible.
The TCP stack delivers packets in the order they come off the wire. Wget puts them in the correct order based on TCP sequence number to build the kernel image, and the TCP sequence number/ack protocol ensures complete delivery of a stream.
OK, it sounds like you've got a solution and a preference, so if it makes more sense for this embedded implementation to maintain simplicity, then so be it.
-Joe
Date: Wed, 10 Jan 2018 17:54:07 -0800 Subject: [PATCH] git_msg_1
TCP and wget implementation.
This is the interface and Kconfig files for introducing TCP and wget into u-boot.
Interfaces are in net.c and net.h, ping.c is modified to the new ip send interface, and UDP and TCP have shim procedures call map the protocol interface to the ip interface.
The UDP interface is unchanged, and the existing UDP programs need no changes.
All the code is new, and not copied from any source.
This code should compile.
Next step? git push? And I do wish I could get git send-mail configureg with yahoo.
Signed-off-by: Duncan Hare DH@synoia.com
From: Duncan Hare DH@synoia.com Date: Wed, 10 Jan 2018 17:54:07 -0800 Subject: [PATCH] git_msg_1
Signed-off-by: Duncan Hare DH@synoia.com ---
include/net.h | 32 ++++++++++++++---- net/Kconfig | 5 +++ net/net.c | 102 +++++++++++++++++++++++++++++++++++++++++++++++----------- net/ping.c | 9 ++---- 4 files changed, 115 insertions(+), 33 deletions(-)
diff --git a/include/net.h b/include/net.h index 455b48f6c7..d231987e6e 100644 --- a/include/net.h +++ b/include/net.h @@ -15,17 +15,26 @@ #include <asm/cache.h> #include <asm/byteorder.h> /* for nton* / ntoh* stuff */
-#define DEBUG_LL_STATE 0 /* Link local state machine changes */ -#define DEBUG_DEV_PKT 0 /* Packets or info directed to the device */ -#define DEBUG_NET_PKT 0 /* Packets on info on the network at large */ +#define DEBUG_LL_STATE 0 /* Link local state machine changes */ +#define DEBUG_DEV_PKT 0 /* Packets or info directed to the device */ +#define DEBUG_NET_PKT 0 /* Packets on info on the network at large */ #define DEBUG_INT_STATE 0 /* Internal network state changes */
/* * The number of receive packet buffers, and the required packet buffer * alignment in memory. * + * The nuber of buffers for TCP is used to calculate a static TCP window + * size, becuse TCP window size is a promise to the sending TCP to be able + * to buffer up to the window size of data. + * When the sending TCP has a window size of outstanding unacknowledged + * data, the sending TCP will stop sending. */
+#if defined(CONFIG_TCP) +#define CONFIG_SYS_RX_ETH_BUFFER 50 /* For TCP */ +#endif + #ifdef CONFIG_SYS_RX_ETH_BUFFER # define PKTBUFSRX CONFIG_SYS_RX_ETH_BUFFER #else @@ -354,6 +363,8 @@ struct vlan_ethernet_hdr {
#define IPPROTO_ICMP 1 /* Internet Control Message Protocol */ #define IPPROTO_UDP 17 /* User Datagram Protocol */ +#define IPPROTO_TCP 6 /* Transmission Control Protocol */ +
/* * Internet Protocol (IP) header. @@ -538,7 +549,7 @@ 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 */ @@ -596,10 +607,10 @@ int net_set_ether(uchar *xet, const uchar *dest_ethaddr, uint prot); int net_update_ether(struct ethernet_hdr *et, uchar *addr, uint prot);
/* Set IP header */ -void net_set_ip_header(uchar *pkt, struct in_addr dest, struct in_addr source); +void net_set_ip_header(uchar *pkt, struct in_addr dest, struct in_addr source, + u16 pkt_len, u8 prot); void net_set_udp_header(uchar *pkt, struct in_addr dest, int dport, - int sport, int len); - + int sport, int len); /** * compute_ip_checksum() - Compute IP checksum * @@ -670,9 +681,16 @@ static inline void net_send_packet(uchar *pkt, int len) * @param sport Source UDP port * @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, 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/net/Kconfig b/net/Kconfig index 414c5497c7..625ad291bb 100644 --- a/net/Kconfig +++ b/net/Kconfig @@ -45,4 +45,9 @@ config BOOTP_VCI_STRING default "U-Boot.arm" if ARM default "U-Boot"
+config TCP + bool "Include Subset TCP stack for wget" + help + TCP protocol support for wget. + endif # if NET diff --git a/net/net.c b/net/net.c index 4259c9e321..79255c11eb 100644 --- a/net/net.c +++ b/net/net.c @@ -107,6 +107,12 @@ #if defined(CONFIG_CMD_SNTP) #include "sntp.h" #endif +#if defined(CONFIG_TCP) +#include "tcp.h" +#endif +#if defined(CONFIG_CMD_WGET) +#include "wget.h" +#endif
DECLARE_GLOBAL_DATA_PTR;
@@ -181,6 +187,7 @@ int net_ntp_time_offset; static uchar net_pkt_buf[(PKTBUFSRX+1) * PKTSIZE_ALIGN + PKTALIGN]; /* Receive packets */ uchar *net_rx_packets[PKTBUFSRX]; + /* Current UDP RX packet handler */ static rxhand_f *udp_packet_handler; /* Current ARP RX packet handler */ @@ -381,6 +388,9 @@ 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(); @@ -484,6 +494,11 @@ restart: nfs_start(); break; #endif +#if defined(CONFIG_CMD_WGET) + case WGET: + wget_start(); + break; +#endif #if defined(CONFIG_CMD_CDP) case CDP: cdp_start(); @@ -777,11 +792,41 @@ void net_set_timeout_handler(ulong iv, thand_f *f) }
int net_send_udp_packet(uchar *ether, struct in_addr dest, int dport, int sport, - int payload_len) + int payload_len) +{ + return net_send_ip_packet(ether, dest, dport, sport, payload_len, + 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, u8 action, u32 tcp_seq_num, + u32 tcp_ack_num) { uchar *pkt; int eth_hdr_size; int pkt_hdr_size; + if (proto == IPPROTO_UDP) { + 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) */ assert(net_tx_packet != NULL); @@ -799,9 +844,21 @@ int net_send_udp_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); - pkt += eth_hdr_size; - net_set_udp_header(pkt, dest, dport, sport, payload_len); - pkt_hdr_size = eth_hdr_size + IP_UDP_HDR_SIZE; +#if defined(CONFIG_TCP) + if (proto == IPPROTO_UDP) { +#endif + net_set_udp_header(pkt + eth_hdr_size, dest, + dport, sport, payload_len); + pkt_hdr_size = eth_hdr_size + IP_UDP_HDR_SIZE; + +#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) { @@ -1157,9 +1214,6 @@ void net_process_received_packet(uchar *in_packet, int len) /* Can't deal with anything except IPv4 */ if ((ip->ip_hl_v & 0xf0) != 0x40) return; - /* Can't deal with IP options (headers != 20 bytes) */ - if ((ip->ip_hl_v & 0x0f) > 0x05) - return; /* Check the Checksum of the header */ if (!ip_checksum_ok((uchar *)ip, IP_HDR_SIZE)) { debug("checksum bad\n"); @@ -1205,9 +1259,19 @@ void net_process_received_packet(uchar *in_packet, int len) * we send a tftp packet to a dead connection, or when * there is no server at the other end. */ + 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; } @@ -1365,8 +1429,7 @@ common: } /**********************************************************************/
-int -net_eth_hdr_size(void) +int net_eth_hdr_size(void) { ushort myvlanid;
@@ -1426,25 +1489,28 @@ int net_update_ether(struct ethernet_hdr *et, uchar *addr, uint prot) } }
-void net_set_ip_header(uchar *pkt, struct in_addr dest, struct in_addr source) +void net_set_ip_header(uchar *pkt, struct in_addr dest, struct in_addr source, + u16 pkt_len, u8 prot) { struct ip_udp_hdr *ip = (struct ip_udp_hdr *)pkt;
/* - * Construct an IP header. + * Construct an IP header. */ /* IP_HDR_SIZE / 4 (not including UDP) */ ip->ip_hl_v = 0x45; ip->ip_tos = 0; - ip->ip_len = htons(IP_HDR_SIZE); + ip->ip_len = htons(pkt_len); ip->ip_id = htons(net_ip_id++); - ip->ip_off = htons(IP_FLAGS_DFRAG); /* Don't fragment */ + ip->ip_off = htons(IP_FLAGS_DFRAG); /* Don't fragment */ ip->ip_ttl = 255; + ip->ip_p = prot; ip->ip_sum = 0; /* already in network byte order */ net_copy_ip((void *)&ip->ip_src, &source); /* already in network byte order */ net_copy_ip((void *)&ip->ip_dst, &dest); + ip->ip_sum = compute_ip_checksum(ip, IP_HDR_SIZE); }
void net_set_udp_header(uchar *pkt, struct in_addr dest, int dport, int sport, @@ -1460,11 +1526,8 @@ void net_set_udp_header(uchar *pkt, struct in_addr dest, int dport, int sport, if (len & 1) pkt[IP_UDP_HDR_SIZE + len] = 0;
- net_set_ip_header(pkt, dest, net_ip); - ip->ip_len = htons(IP_UDP_HDR_SIZE + len); - ip->ip_p = IPPROTO_UDP; - ip->ip_sum = compute_ip_checksum(ip, IP_HDR_SIZE); - + net_set_ip_header(pkt, dest, net_ip, IP_UDP_HDR_SIZE + len, + IPPROTO_UDP); ip->udp_src = htons(sport); ip->udp_dst = htons(dport); ip->udp_len = htons(UDP_HDR_SIZE + len); @@ -1485,7 +1548,8 @@ void copy_filename(char *dst, const char *src, int size)
#if defined(CONFIG_CMD_NFS) || \ defined(CONFIG_CMD_SNTP) || \ - defined(CONFIG_CMD_DNS) + defined(CONFIG_CMD_DNS) || \ + defined(CONFIG_CMD_WGET) /* * make port a little random (1024-17407) * This keeps the math somewhat trivial to compute, and seems to work with diff --git a/net/ping.c b/net/ping.c index 9508cf1160..254b646193 100644 --- a/net/ping.c +++ b/net/ping.c @@ -20,16 +20,11 @@ struct in_addr net_ping_ip; static void set_icmp_header(uchar *pkt, struct in_addr dest) { /* - * Construct an IP and ICMP header. + * Construct an ICMP header. */ - struct ip_hdr *ip = (struct ip_hdr *)pkt; struct icmp_hdr *icmp = (struct icmp_hdr *)(pkt + IP_HDR_SIZE);
- net_set_ip_header(pkt, dest, net_ip); - - ip->ip_len = htons(IP_ICMP_HDR_SIZE); - ip->ip_p = IPPROTO_ICMP; - ip->ip_sum = compute_ip_checksum(ip, IP_HDR_SIZE); + net_set_ip_header(pkt, dest, net_ip, IP_ICMP_HDR_SIZE, IPPROTO_ICMP);
icmp->type = ICMP_ECHO_REQUEST; icmp->code = 0;
Hi Duncan,
On Wed, Jan 10, 2018 at 8:18 PM, Duncan Hare dh@synoia.com wrote:
Date: Wed, 10 Jan 2018 17:54:07 -0800 Subject: [PATCH] git_msg_1
TCP and wget implementation.
Sounds like this should be the title of the series cover letter. I thought we agreed that this should be sent as a series. That means using patman to send all 3 matches to the list as once.
This is the interface and Kconfig files for introducing TCP and wget into u-boot.
Interfaces are in net.c and net.h, ping.c is modified to the new ip send interface, and UDP and TCP have shim procedures call map the protocol interface to the ip interface.
The UDP interface is unchanged, and the existing UDP programs need no changes.
All the code is new, and not copied from any source.
This code should compile.
Next step? git push?
Next step is to get sendmail setup and resend as a series using patman.
And I do wish I could get git send-mail configureg with yahoo.
https://help.yahoo.com/kb/SLN4724.html
Signed-off-by: Duncan Hare DH@synoia.com
Hi Duncan,
On Wed, Jan 10, 2018 at 8:18 PM, Duncan Hare dh@synoia.com wrote:
Date: Wed, 10 Jan 2018 17:54:07 -0800 Subject: [PATCH] git_msg_1
TCP and wget implementation.
Sounds like this should be the title of the series cover letter. I thought we agreed that this should be sent as a series. That means using patman to send all 3 matches to the list as once.
Next step is to get sendmail setup and resend as a series using patman.
And I do wish I could get git send-mail configureg with yahoo.
https://help.yahoo.com/kb/SLN4724.html
Signed-off-by: Duncan Hare DH@synoia.com
Ok, our definition of series differed. I was not thinking concurrent.
Will send next week. Duncan
Hi Duncan,
On Mon, Jan 15, 2018 at 11:01 AM, Duncan Hare dh@synoia.com wrote:
Hi Duncan,
On Wed, Jan 10, 2018 at 8:18 PM, Duncan Hare dh@synoia.com wrote:
Date: Wed, 10 Jan 2018 17:54:07 -0800 Subject: [PATCH] git_msg_1
TCP and wget implementation.
Sounds like this should be the title of the series cover letter. I thought we agreed that this should be sent as a series. That means using patman to send all 3 matches to the list as once.
Next step is to get sendmail setup and resend as a series using patman.
And I do wish I could get git send-mail configureg with yahoo.
https://help.yahoo.com/kb/SLN4724.html
Signed-off-by: Duncan Hare DH@synoia.com
Ok, our definition of series differed. I was not thinking concurrent.
I'm using common patch terminology. Please review our patch process.
https://www.denx.de/wiki/U-Boot/Patches
Will send next week.
Duncan
participants (3)
-
Duncan Hare -
Joe Hershberger -
Joe Hershberger