MPC5748G lwip ping protocol package delay

cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

MPC5748G lwip ping protocol package delay

2,107 Views
hansonhe
NXP Employee
NXP Employee

Hello,

We use LWIP protocol for network communication at present, we find that the delay of our PC ping MPC5748 is higher than 1ms, and the maximum value is 8ms. As a result, the speed of TCP transmission through Ethernet can only be limited to 500KB/s. PHY chip is KSZ8081, the circuit is from the DEVKIT-MPC5748G development board.

0 Kudos
7 Replies

1,592 Views
zhangah
Contributor II

Hi,we met the same problem now.Have you solve this problem? Can you give us some suggestion?

0 Kudos

1,592 Views
martin_kovar
NXP Employee
NXP Employee

Hello,

could you please show me the screen of the ping command? What values can you see?

Regards,

Martin

0 Kudos

1,592 Views
zhangah
Contributor II

Hi,we use the example project(name : lwip_mpc5748g) to run with the modification like flowing:

 

our lwipopts.h :

 

#ifndef LWIP_LWIPOPTS_H
#define LWIP_LWIPOPTS_H

#define TCPIP_MBOX_SIZE 10

#define DEFAULT_UDP_RECVMBOX_SIZE 10
#define DEFAULT_TCP_RECVMBOX_SIZE 10
#define DEFAULT_RAW_RECVMBOX_SIZE 10
#define DEFAULT_ACCEPTMBOX_SIZE 10

#define LWIP_NETIF_TX_SINGLE_PBUF 1
#define LWIP_SUPPORT_CUSTOM_PBUF 1
#define MEMP_USE_CUSTOM_POOLS 0
#define MEM_USE_POOLS 0

#define LWIP_IPV4 1
#define LWIP_IPV6 0

#define NO_SYS 0
#define LWIP_SOCKET 1//(NO_SYS==0)
#define LWIP_NETCONN 1//(NO_SYS==0)

#define LWIP_IGMP LWIP_IPV4
#define LWIP_ICMP LWIP_IPV4

#define LWIP_SNMP LWIP_UDP
#define MIB2_STATS LWIP_SNMP

#define LWIP_DNS LWIP_UDP
#define LWIP_MDNS_RESPONDER LWIP_UDP

#define LWIP_NUM_NETIF_CLIENT_DATA (LWIP_MDNS_RESPONDER)

#define LWIP_HAVE_LOOPIF 0
#define LWIP_NETIF_LOOPBACK 0

#define TCP_LISTEN_BACKLOG 1

#define LWIP_COMPAT_SOCKETS 1
#define LWIP_SO_RCVTIMEO 1
#define LWIP_SO_RCVBUF 1

#define LWIP_TCPIP_CORE_LOCKING 1

#define LWIP_NETIF_LINK_CALLBACK 0
#define LWIP_NETIF_STATUS_CALLBACK 0

#define LWIP_NETCONN_SEM_PER_THREAD 1

#define LWIP_SOCKET_SET_ERRNO 1

#ifdef LWIP_DEBUG

#define LWIP_DBG_MIN_LEVEL 0
#define PPP_DEBUG LWIP_DBG_OFF
#define MEM_DEBUG LWIP_DBG_OFF
#define MEMP_DEBUG LWIP_DBG_OFF
#define PBUF_DEBUG LWIP_DBG_OFF
#define API_LIB_DEBUG LWIP_DBG_OFF
#define API_MSG_DEBUG LWIP_DBG_OFF
#define TCPIP_DEBUG LWIP_DBG_OFF
#define NETIF_DEBUG LWIP_DBG_OFF
#define SOCKETS_DEBUG LWIP_DBG_OFF
#define DNS_DEBUG LWIP_DBG_OFF
#define AUTOIP_DEBUG LWIP_DBG_OFF
#define DHCP_DEBUG LWIP_DBG_OFF
#define IP_DEBUG LWIP_DBG_OFF
#define IP_REASS_DEBUG LWIP_DBG_OFF
#define ICMP_DEBUG LWIP_DBG_OFF
#define IGMP_DEBUG LWIP_DBG_OFF
#define UDP_DEBUG LWIP_DBG_OFF
#define TCP_DEBUG LWIP_DBG_OFF
#define TCP_INPUT_DEBUG LWIP_DBG_OFF
#define TCP_OUTPUT_DEBUG LWIP_DBG_OFF
#define TCP_RTO_DEBUG LWIP_DBG_OFF
#define TCP_CWND_DEBUG LWIP_DBG_OFF
#define TCP_WND_DEBUG LWIP_DBG_OFF
#define TCP_FR_DEBUG LWIP_DBG_OFF
#define TCP_QLEN_DEBUG LWIP_DBG_OFF
#define TCP_RST_DEBUG LWIP_DBG_OFF
#endif

#define LWIP_DBG_TYPES_ON (LWIP_DBG_ON|LWIP_DBG_TRACE|LWIP_DBG_STATE|LWIP_DBG_FRESH|LWIP_DBG_HALT)


/* ---------- Memory options ---------- */
/* MEM_ALIGNMENT: should be set to the alignment of the CPU for which
lwIP is compiled. 4 byte alignment -> define MEM_ALIGNMENT to 4, 2
byte alignment -> define MEM_ALIGNMENT to 2. */
/* MSVC port: intel processors don't need 4-byte alignment,
but are faster that way! */

#define MEM_ALIGNMENT 4

/* MEM_SIZE: the size of the heap memory. If the application will send
a lot of data that needs to be copied, this should be set high. */
#define MEM_SIZE 32768

/* MEMP_NUM_PBUF: the number of memp struct pbufs. If the application
sends a lot of data out of ROM (or other static memory), this
should be set high. */
#define MEMP_NUM_PBUF 16
/* MEMP_NUM_RAW_PCB: the number of UDP protocol control blocks. One
per active RAW "connection". */
#define MEMP_NUM_RAW_PCB 4
/* MEMP_NUM_UDP_PCB: the number of UDP protocol control blocks. One
per active UDP "connection". */
#define MEMP_NUM_UDP_PCB 8
/* MEMP_NUM_TCP_PCB: the number of simulatenously active TCP
connections. */
#define MEMP_NUM_TCP_PCB 8
/* MEMP_NUM_TCP_PCB_LISTEN: the number of listening TCP
connections. */
#define MEMP_NUM_TCP_PCB_LISTEN 8
/* MEMP_NUM_TCP_SEG: the number of simultaneously queued TCP
segments. */
#define MEMP_NUM_TCP_SEG 16
/* MEMP_NUM_SYS_TIMEOUT: the number of simulateously active
timeouts. */
#define MEMP_NUM_SYS_TIMEOUT 15

/* The following four are used only with the sequential API and can be
set to 0 if the application only will use the raw API. */
/* MEMP_NUM_NETBUF: the number of struct netbufs. */
#define MEMP_NUM_NETBUF 16
/* MEMP_NUM_NETCONN: the number of struct netconns. */
#define MEMP_NUM_NETCONN 32
/* MEMP_NUM_TCPIP_MSG_*: the number of struct tcpip_msg, which is used
for sequential API communication and incoming packets. Used in
src/api/tcpip.c. */
#define MEMP_NUM_TCPIP_MSG_API 32
#define MEMP_NUM_TCPIP_MSG_INPKT 32


/* ---------- Pbuf options ---------- */
/* PBUF_POOL_SIZE: the number of buffers in the pbuf pool. */
#define PBUF_POOL_SIZE 128

/** SYS_LIGHTWEIGHT_PROT
* define SYS_LIGHTWEIGHT_PROT in lwipopts.h if you want inter-task protection
* for certain critical regions during buffer allocation, deallocation and memory
* allocation and deallocation.
*/
#define SYS_LIGHTWEIGHT_PROT (NO_SYS==0)


/* ---------- TCP options ---------- */
#define LWIP_TCP 1
#define TCP_TTL 255

/* Controls if TCP should queue segments that arrive out of
order. Define to 0 if your device is low on memory. */
#define TCP_QUEUE_OOSEQ 1

/* TCP Maximum segment size. */
#define TCP_MSS 1460

/* TCP sender buffer space (bytes). */
#define TCP_SND_BUF 2920

/* TCP sender buffer space (pbufs). This must be at least = 2 *
TCP_SND_BUF/TCP_MSS for things to work. */
#define TCP_SND_QUEUELEN (4 * TCP_SND_BUF/TCP_MSS)

/* TCP writable space (bytes). This must be less than or equal
to TCP_SND_BUF. It is the amount of space which must be
available in the tcp snd_buf for select to return writable */
#define TCP_SNDLOWAT (TCP_SND_BUF/2)

/* TCP receive window. */
#define TCP_WND 65535

/* Maximum number of retransmissions of data segments. */
#define TCP_MAXRTX 12

/* Maximum number of retransmissions of SYN segments. */
#define TCP_SYNMAXRTX 4


/* ---------- ARP options ---------- */
#define LWIP_ARP 1
#define ARP_TABLE_SIZE 10
#define ARP_QUEUEING 1


/* ---------- IP options ---------- */
/* Define IP_FORWARD to 1 if you wish to have the ability to forward
IP packets across network interfaces. If you are going to run lwIP
on a device with only one network interface, define this to 0. */
#define IP_FORWARD 0

/* IP reassembly and segmentation.These are orthogonal even
* if they both deal with IP fragments */
#define IP_REASSEMBLY 1
#define IP_REASS_MAX_PBUFS 10
#define MEMP_NUM_REASSDATA 10
#define IP_FRAG 1


/* ---------- ICMP options ---------- */
#define ICMP_TTL 255


/* ---------- DHCP options ---------- */
/* Define LWIP_DHCP to 1 if you want DHCP configuration of
interfaces. */

#define LWIP_DHCP 0


/* 1 if you want to do an ARP check on the offered address
(recommended). */
#define DHCP_DOES_ARP_CHECK (LWIP_DHCP)


/* ---------- AUTOIP options ------- */
#define LWIP_AUTOIP (LWIP_DHCP)
#define LWIP_DHCP_AUTOIP_COOP (LWIP_DHCP && LWIP_AUTOIP)


/* ---------- UDP options ---------- */
#define LWIP_UDP 1
#define LWIP_UDPLITE LWIP_UDP
#define UDP_TTL 255


/* ---------- RAW options ---------- */
#define LWIP_RAW 1


/* ---------- Statistics options ---------- */

#define LWIP_STATS 1
#define LWIP_STATS_DISPLAY 0

#if LWIP_STATS
#define LINK_STATS 1
#define IP_STATS 1
#define ICMP_STATS 1
#define IGMP_STATS 1
#define IPFRAG_STATS 1
#define UDP_STATS 1
#define TCP_STATS 1
#define MEM_STATS 1
#define MEMP_STATS 1
#define PBUF_STATS 1
#define SYS_STATS 1
#endif /* LWIP_STATS */


/* ---------- PPP options ---------- */

#define PPP_SUPPORT 0

#define LWIP_CHECKSUM_CTRL_PER_NETIF 0
#define CHECKSUM_GEN_IP 0
#define CHECKSUM_GEN_UDP 0
#define CHECKSUM_GEN_TCP 0
#define CHECKSUM_GEN_ICMP 0
#define CHECKSUM_GEN_ICMP6 LWIP_IPV6
#define CHECKSUM_CHECK_IP 0
#define CHECKSUM_CHECK_UDP 0
#define CHECKSUM_CHECK_TCP 0
#define CHECKSUM_CHECK_ICMP 0
#define CHECKSUM_CHECK_ICMP6 LWIP_IPV6
#define LWIP_CHECKSUM_ON_COPY 0

#endif /* LWIP_LWIPOPTS_H */

 

 

 

our lwipcfg.h :

/**
* Additional settings for the particular port.
*/

#define LWIP_PORT_INIT_IPADDR(addr) IP4_ADDR((addr), 192,168,8,1)
#define LWIP_PORT_INIT_GW(addr) IP4_ADDR((addr), 192,168,8,1)
#define LWIP_PORT_INIT_NETMASK(addr) IP4_ADDR((addr), 255,255,0,0)

/* MAC address is passed as compilation option */

/* configuration for applications */

#define LWIP_CHARGEN_APP 1
#define LWIP_DNS_APP 0
#define LWIP_HTTPD_APP 1
/* Set this to 1 to use the netconn http server,
* otherwise the raw api server will be used. */
/*#define LWIP_HTTPD_APP_NETCONN */
#define LWIP_HTTPD_APP_NETCONN 1
#define LWIP_NETBIOS_APP 0
#define LWIP_NETIO_APP 0
#define LWIP_RTP_APP 0
#define LWIP_SHELL_APP 0
#define LWIP_SNTP_APP 0
#define LWIP_SOCKET_EXAMPLES_APP 1
#define LWIP_TCPECHO_APP 1
/* Set this to 1 to use the netconn tcpecho server,
* otherwise the raw api server will be used. */
/*#define LWIP_TCPECHO_APP_NETCONN */
#define LWIP_TCPECHO_APP_NETCONN 1
#define LWIP_UDPECHO_APP 1
#define LWIP_LWIPERF_APP 1

 

 

our test task: this is the only taskjust sending several Ascii in cycle

static void
sockex_nonblocking_connect(void *arg)
{
int s;
int ret;
u32_t opt;
#if LWIP_IPV6
struct sockaddr_in6 addr;
#else /* LWIP_IPV6 */
struct sockaddr_in addr;
#endif /* LWIP_IPV6 */
fdsets sets;
struct timeval tv;
u32_t ticks_a, ticks_b;
int err;
const ip_addr_t *ipaddr = (const ip_addr_t*)arg;
INIT_FDSETS(&sets);

/* set up address to connect to */
memset(&addr, 0, sizeof(addr));
#if LWIP_IPV6
addr.sin6_len = sizeof(addr);
addr.sin6_family = AF_INET6;
addr.sin6_port = PP_HTONS(SOCK_TARGET_PORT);
inet6_addr_from_ip6addr(&addr.sin6_addr, ip_2_ip6(ipaddr));
#else /* LWIP_IPV6 */
addr.sin_len = sizeof(addr);
addr.sin_family = AF_INET;
addr.sin_port = PP_HTONS(SOCK_TARGET_PORT);
inet_addr_from_ip4addr(&addr.sin_addr, ip_2_ip4(ipaddr));
#endif /* LWIP_IPV6 */

/* first try blocking: */

/* create the socket */
#if LWIP_IPV6
s = lwip_socket(AF_INET6, SOCK_STREAM, 0);
#else /* LWIP_IPV6 */
s = lwip_socket(AF_INET, SOCK_STREAM, 0);
#endif /* LWIP_IPV6 */
LWIP_ASSERT("s >= 0", s >= 0);

/* connect */
ret = lwip_connect(s, (struct sockaddr*)&addr, sizeof(addr));
/* should succeed */
LWIP_ASSERT("ret == 0", ret == 0);

/* write something */ 
while(1)
{

ret = lwip_write(s, "test", 4);
LWIP_ASSERT("ret == 4", ret == 4);
}

/* close */
ret = lwip_close(s);
LWIP_ASSERT("ret == 0", ret == 0);

return; 
}

 

 

the result is that:just 129kB/s.

pastedImage_1.png

ps:

1.our phy is 8081,and cpu is 5748G with model of RMII.

2.the dhcp and auto ip model is disable.

3.the tcp server run in the pc with the configuration of Ip:192.168.8.2 and port:21.

4.the tcp client run in the 5748G with the configuration of Ip:192.168.8.1.

 

0 Kudos

1,592 Views
martin_kovar
NXP Employee
NXP Employee

Hello,

could you please describe the application you use for bandwidth measurement? From the figure I do not understand, what data are sent to microcontroller. If you want to measure bandwidth, I recommend you to use packets with maximal length 1500 bytes with UDP protocol on transport layer.

If you use TCP instead of UDP, there is some overhead because of Three-way handshake and acknowledgement.

But original question was related to ICMP protocol, which by default uses ICMP packets with 32 bytes data length, which are not suitable for bandwidth measurement.

So, could you please let me know, what the application measures and how can I interpret the results you posted?

Regards,

Martin

0 Kudos

1,592 Views
zhangah
Contributor II

Hello,

     there's another strange phenomena that the bandwidth is affected by the Cpu core clock.

     The explicit test is implement like following:

    1. our  Oscillators  circurit is :

pastedImage_1.png

    2. our program is the same like:

  

while(1)
{

ret = lwip_write(s, buff,1024);
// LWIP_ASSERT("ret == 1024", ret == 1024);

if(ret < 0)
{
ret = lwip_close(s);
/* create the socket */
#if LWIP_IPV6
s = lwip_socket(AF_INET6, SOCK_STREAM, 0);
#else /* LWIP_IPV6 */
s = lwip_socket(AF_INET, SOCK_STREAM, 0);
#endif /* LWIP_IPV6 */
LWIP_ASSERT("s >= 0", s >= 0);
/* connect */
ret = lwip_connect(s, &addr, sizeof(addr));
}
//sys_msleep(1);
}

  3.Different CPU core clock setting in PE like:

A:

pastedImage_2.png     

the bandwidth result is:

pastedImage_4.png

B:

pastedImage_3.png

the bandwidth result is:

pastedImage_7.png

so,the result is that: A is much higher then B.Can someone explane the reason for us.Thanks!

0 Kudos

1,592 Views
zhangah
Contributor II

Hello,

     we use the only task of sending several bytes througth tcp protocol like this:

      *****************************************************************************

while(1)
{

ret = lwip_write(s, buff,1024);
// LWIP_ASSERT("ret == 1024", ret == 1024);

if(ret < 0)
{
ret = lwip_close(s);
/* create the socket */
#if LWIP_IPV6
s = lwip_socket(AF_INET6, SOCK_STREAM, 0);
#else /* LWIP_IPV6 */
s = lwip_socket(AF_INET, SOCK_STREAM, 0);
#endif /* LWIP_IPV6 */
LWIP_ASSERT("s >= 0", s >= 0);
/* connect */
ret = lwip_connect(s, &addr, sizeof(addr));
}
//sys_msleep(1);
}

*****************************************************************************

actually,now the bandwidth can run up to about 4M Bytes/s. but that seemed to the highest speed,no matter how I could change the parameters in lwipopts.h.

0 Kudos

1,592 Views
martin_kovar
NXP Employee
NXP Employee

Hello Hanson,

I checked the lwIP example project with my computer and I can see following values:

pastedImage_1.png

First time, there was 4ms delay, but in all other cases, there is average delay about 1-2ms, which seems to be correct. This delay is caused by the microcontroller, which must create and send ICMP reply packet and it is unrelated to transfer rate.

Delay caused by transfer the ICMP reply packet through the twisted pair cable is negligible. Main delay is cause by the microcontroller, during ICMP handling.

Also, this example is based on FreeRTOS and in case you run another tasks, 8ms delay is not high. From my point of view, example behavior is correct.

Please let me know, if you need any other information.

Regards,

Martin

0 Kudos