Ethernet fails in Colibri IMX6 WCE7

I have 1.2b6 bootloader and 1.2b6 BSP based WEC7 image in Colibri IMX6 module and there is a strange behaviour with ethernet.
If I have dhcp enabled it works but when I define static IP address it works only if ethernet cable is connected during boot time.
If I leave cable off during boot and connect it a couple of seconds after boot is fnished (Done NETUI is displayed in serial port debug window) following error message is displayed.

Ethernet: Connected at: 100Mbps full duplex

Exception ‘Data Abort’ (0x4): Thread-Id=057b0072(pth=af8e32e0), Proc-Id=00400002(pprc=82f185e0) ‘NK.EXE’, VM-active=04eb004e(pprc=af8b3940) ‘servicesd.exe’
PC=ef9d7b18(afd.dll+0x00017b18) RA=00000001(???+0x00000001) SP=c0e3f190, BVA=00000001

When I give “services list” command on local display before I connect ethernet cable it shows that ftp and telnet services are in state “running”.
Same command after connecting the cable gives state “unknown” for both services and they do not work.
Ping is still working.

Here are my tcp registry settings

[HKEY_LOCAL_MACHINE\Comm\ENET1\Parms\TcpIp]

; This should be MULTI_SZ

“DefaultGateway”=“192.168.10.1”

; This should be SZ… If null it means use LAN, else WAN and Interface.

“LLInterface”=""

; Use zero for broadcast address? (or 255.255.255.255)

“UseZeroBroadcast”=dword:0

; Thus should be MULTI_SZ, the IP address list

“IpAddress”=“192.168.10.212”

; This should be MULTI_SZ, the subnet masks for the above IP addresses

“Subnetmask”=“255.255.255.0”

; If we set this the above entries are ignore, but we leave them as a sample

“EnableDHCP”=dword:0

;DHCP will keep retrying

“DhcpMaxRetry”=dword:ffffffff

;INIT delay, RFC2131 recommends 10 seconds

“DhcpInitDelayInterval”=dword:2710

;Number of retries before the dialogue box comes up

“DhcpRetryDialogue”=dword:ffffffff

“InterfaceMetric”=dword:14

“DNS”=“8.8.8.8”

I have more information about the problem.
Also WEC7 1.2b6 image downloaded from Toradex support has same problem.
Static address itself was not enough to bring this bug visible.
After adding “DNS”=“8.8.8.8” also Toradex image breaks same way as mine.

You were using same configuration with previous releases of the BSP?

I had similar ethernet problem with 1.2b4 but I thought then incorrectly that it was related to enabling USB master mode.
With 1.2b6 my latest test was to install Toradex 1.2b6 image and change IP to static. It did not trigger the problem but when I added DNS the bug was there. If I remove DNS it works again.

Problem solved for me.
When I gave command “services stop ntp0:” before connecting the ethernet cable it did not crash.

Next I created my own 1.2b6 image using Toradex workspace but removed catalog item “SNTP Client Service”.
NTP service is no longer reported by “services list” command and ethernet works.

It is quite obvious that SNTP client service is crashing and it crashes also services.exe.
I don’t need builtin SNTP client so this no longer a problem for me.

Thank you for reporting your discovery. The NTP service tries to contact the servers as soon as the connection is established, this may be what triggers the issue. The time also depends on the other side (router/hub). We did not change anything in the SNTP component, but we will check recent updates to understand if something has been changed there.