RTNET i.mx6 raw packets

RTNET i.mx6 raw packets

[Paal Tamas]

Dear All, I would like to use RTNET with raw packets on the fec interface of my i.mx6 CPU. The purpose of the whole thing is to use the Soem ECAT master.
I am lost. I need your help. Xenomai_3.1, kernel 4.14.78.I am using this rtnet fec patch:
https://xenomai.org/pipermail/xenomai/2019-March/040580.html The fec interface is eth0.
I am initializing RTnet the following way: ifconfig eth0 down
modprobe rtnet_fec
modprobe rtpacket
echo 2188000.ethernet >  /sys/bus/platform/drivers/fec/unbind
echo 2188000.ethernet >  /sys/bus/platform/drivers/rtnet_fec/bind
/usr/xenomai/sbin/rtifconfig rteth0 up
   >RTnet: registered rteth0
   >rtnet_fec 2188000.ethernet: 2188000.ethernet supply phy not found, using dummy regulator
   >pps pps0: new PPS source ptp0
   >libphy: fec_enet_mii_bus: probed
   >rtnet_fec 2188000.ethernet rteth0: RTnet mode enabled
   >rtnet_fec 2188000.ethernet rteth0: registered PHC device 0
   >Micrel KSZ9031 Gigabit PHY 2188000.ethernet-1:07: attached PHY driver [Micrel KSZ9031 Gigabit PHY] (mii_bus:phy_addr=2188000.ethernet-1:07, irq=POLL)
/usr/xenomai/sbin/rtifconfig
   >rteth0    Medium: Ethernet  Hardware address: F8:DC:7A:1B:88:F1
   >          UP BROADCAST  MTU: 1500
   >          RX packets:0 errors:0 dropped:0 overruns:0 frame:0
   >          TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
   >          collisions:0
   >          RX bytes:0 (0.0 b)  TX bytes:0 (0.0 b) If I load the ipv4 module and add the routing info: modprobe rtipv4
/usr/xenomai/sbin/rtifconfig rteth0 down
/usr/xenomai/sbin/rtifconfig rteth0 up 192.168.100.50
   >rtnet_fec 2188000.ethernet rteth0: carrier lost
   >rtnet_fec 2188000.ethernet rteth0: Link is Up - 100Mbps/Full - flow control rx/tx
   >rtnet_fec 2188000.ethernet rteth0: carrier detected
/usr/xenomai/sbin/rtifconfig
   >rteth0    Medium: Ethernet  Hardware address: F8:DC:7A:1B:88:F1
   >          IP address: 192.168.100.50  Broadcast address: 192.168.100.255
   >          UP BROADCAST RUNNING  MTU: 1500
   >          RX packets:65 errors:0 dropped:0 overruns:0 frame:0
   >          TX packets:7 errors:0 dropped:0 overruns:0 carrier:0
   >          collisions:0
   >          RX bytes:7605 (7.4 Kb)  TX bytes:686 (686.0 b)
/usr/xenomai/sbin/rtroute add 192.168.100.100 00:13:3b:93:07:20 dev rteth0
/usr/xenomai/sbin/rtping 192.168.100.100
   >Real-time PING 192.168.100.100 56(84) bytes of data.
   >64 bytes from 192.168.100.100: icmp_seq=1 time=499.7 us
   >64 bytes from 192.168.100.100: icmp_seq=2 time=424.0 us
   >64 bytes from 192.168.100.100: icmp_seq=3 time=531.0 us
   >64 bytes from 192.168.100.100: icmp_seq=4 time=750.3 us
   >^C
   >--- 192.168.100.100 rtping statistics ---
   >4 packets transmitted, 4 received, 0% packet loss
   >worst case rtt = 750.3 us
   
As you can see ping works. Now comes the difficulties.
I try to send raw packet this was (pseudo code coming): #define ETH_P_ECAT 0x88A4
const unsigned char txframe[]={0xff,0xff,0xff,0xff,0xff,0xff,0x1,0x1,0x1,0x1,0x1,0x1,0x88,0xa4,0xe,0x10,0x7,0x1,0x0,0x0,0x0,0x0,0x2,0x0,0x0,0x0,0x0,0x0,0x0,0x0};
int sock = rt_dev_socket(PF_PACKET, SOCK_RAW, htons(ETH_P_ECAT));    
struct ifreq ifr;
strcpy(ifr.ifr_name, "rteth0");
rt_dev_ioctl(sock, SIOCGIFINDEX, &ifr);
struct sockaddr_ll sll;
sll.sll_family = AF_PACKET;
sll.sll_ifindex = ifr.ifr_ifindex;
sll.sll_protocol = htons(ETH_P_ECAT);
rt_dev_bind(sock, (struct sockaddr *)&sll, sizeof(sll));
unsigned char buf[4096];
memset(&buf,0,4096);
for(i=0;i<30;i++)buf[i]=txframe[i];
rt_dev_send(sock, buf, 64, 0);
rt_dev_close(sock); Almost every function call returns positive value (sock is 3, ifindex is 1) only rt_dev_send() failes. It returns EFAULT (-14) that means "Bad Address". Memory fault? I do not see why!
I am testing it since a few days. I tried several things (allocate buffer differently, use different MAC addresses, data size in the data and so on), but nothing worked. I get the same error all the time.
I have already enabled CONFIG_XENO_DRIVERS_RTNET_CHECKED for kernel configuration, but I do not see any error appearing. Note:
If I compile and run the same code with the socket(),ioctl(),bind(),send(),close() function on my other ethernet interface "eth1" (that is a USB to ethernet adapter and has no RTNET support) it works
fine till RTnet is not initialized. If I initialize RTNET the way I desribed above the ioctl() (to read the interface index) and send() functions return EPERM (-1) error (Operation not permitted). (I am working as root.) This is very strange too.
I would assume that RTNET has no effect on the "standard" networking interface.
What am I doing wrong? I checked the mailing list for more than a year back, but I found no answer.
Any help is appreciated.
Thank you in advance,
Tamas 

Re: RTNET i.mx6 raw packets

[Jan Kiszka]

On 01.10.21 10:03, Paal Tamas via Xenomai wrote:
> Dear All, I would like to use RTNET with raw packets on the fec interface of my i.mx6 CPU. The purpose of the whole thing is to use the Soem ECAT master.
> I am lost. I need your help. Xenomai_3.1, kernel 4.14.78.I am using this rtnet fec patch:
> https://xenomai.org/pipermail/xenomai/2019-March/040580.html The fec interface is eth0.

There was a proposal to merge that change
(https://xenomai.org/pipermail/xenomai/2021-July/046000.html), and I
almost did but had to revert it again due to build issues
(https://xenomai.org/pipermail/xenomai/2021-August/046088.html). Any
news on this, Jean-Baptiste?

Jan

-- 
Siemens AG, T RDA IOT
Corporate Competence Center Embedded Linux

Re: RTNET i.mx6 raw packets

[Jean-Baptiste Tredez]

We are working on a new patch for mainline kernel 5.10 + xenomai 
(previous patch was based on nxp kernel 5.4 + xenomai 3.1).

This work is under test. I hope to send a new proposal by the end of the 
month.

Jean-Baptiste

Le 07/10/2021 à 10:12, Jan Kiszka a écrit :
> On 01.10.21 10:03, Paal Tamas via Xenomai wrote:
>> Dear All, I would like to use RTNET with raw packets on the fec interface of my i.mx6 CPU. The purpose of the whole thing is to use the Soem ECAT master.
>> I am lost. I need your help. Xenomai_3.1, kernel 4.14.78.I am using this rtnet fec patch:
>> https://xenomai.org/pipermail/xenomai/2019-March/040580.html The fec interface is eth0.
> There was a proposal to merge that change
> (https://xenomai.org/pipermail/xenomai/2021-July/046000.html), and I
> almost did but had to revert it again due to build issues
> (https://xenomai.org/pipermail/xenomai/2021-August/046088.html). Any
> news on this, Jean-Baptiste?
>
> Jan
>

Re: RTNET i.mx6 raw packets

[Paal Tamas]

Dear Jean-Baptiste and Jan,  Do you have any update on this topic?  Are you sure that the issue I detected and described in my original mail (trying to use RAW_SOCKET via the fec in my i.mx6 using the mentioned xenomai patch, while PING is working when the UDP layer is loaded) is the cause of the actual fec driver implemention?I would be satisfied with that implemention if we could make it work somehow.  Thank you vey much again,Tamas 
Jean-Baptiste Tredez <jean-baptiste.tredez@ba-healthcare.com> írta:
>We are working on a new patch for mainline kernel 5.10 + xenomai 
>(previous patch was based on nxp kernel 5.4 + xenomai 3.1).
>
>This work is under test. I hope to send a new proposal by the end of the 
>month.
>
>Jean-Baptiste
>
>Le 07/10/2021 à 10:12, Jan Kiszka a écrit :
>> On 01.10.21 10:03, Paal Tamas via Xenomai wrote:
>>> Dear All, I would like to use RTNET with raw packets on the fec interface of my i.mx6 CPU. The purpose of the whole thing is to use the Soem ECAT master.
>>> I am lost. I need your help. Xenomai_3.1, kernel 4.14.78.I am using this rtnet fec patch:
>>> https://xenomai.org/pipermail/xenomai/2019-March/040580.html The fec interface is eth0.
>> There was a proposal to merge that change
>> (https://xenomai.org/pipermail/xenomai/2021-July/046000.html), and I
>> almost did but had to revert it again due to build issues
>> (https://xenomai.org/pipermail/xenomai/2021-August/046088.html). Any
>> news on this, Jean-Baptiste?
>>
>> Jan
>