EasyCat Example

[kapil13j]

Hello,

I came across this library and I am testing it on my BeckHoff EtherCat devices. When I executed the easycat example, it discovers the 3 slaves on the bus but when it tries to call bus.createMapping(io_buffer), it gives an segmentation fault error. The error seems at line 562 in bus.cc file. Has anyone came across this issue? Any help appreciated!!

[leducp]

Hello,

KickCAT rely on the eeprom configuration on each slaves to determine which ressources (SyncManager, FMMU) to use. The crash you encountered seems to be due to a eeprom that do not have a sync manager section fully described.

You can check this with the printInfo() after the bus initialization.

For instance this is what I get with a working device (CANOpen):

 -*-*-*-*- slave 1001 -*-*-*-*-
Vendor ID:       0x00001337
Product code:    0x12783456
Revision number: 0x00000001
Serial number:   0x00000000
mailbox in:  size 128 - offset 0x1000
mailbox out: size 128 - offset 0x1080
supported mailbox protocol: 0x04
EEPROM: size: 3968 - version 0x01

SII size: 164
FMMU[0] Outputs (Slave to Master)
FMMU[1] Inputs  (Master to Slave)
SM[0] config
     physical address: 0x1000
     length:           128
     type:             MailboxOut
SM[1] config
     physical address: 0x1080
     length:           128
     type:             MailboxInt
SM[2] config
     physical address: 0x1100
     length:           0
     type:             Output (Slave to Master)
SM[3] config
     physical address: 0x1180
     length:           0
     type:             Input  (Master to Slave)

Be sure to have SM and FFMU entries for logical accesses (Input/Output, not mailbox).

[leducp]

You are calling printInfo() function on a slave that does not come from the bus: this is why the data are outlier.
Get the slave info like that:

Slave& slave_0 = bus.slaves().at(0);
printInfo(slave_0);

[kapil13j]

I called the print info, it is giving the proper information about slaves. Please look at the screenshots. But still I see segmentation fault after bus initialization.

[leducp]

OK so the eeproms of your slaves are coherent yet strangely configured. However it should not lead to a crash on our side so it seems that there is a bug on an edge case here.
I need you to dump the eeprom files of your slaves in order to investigate further. To do so you have to rebase on latest master, rebuild and then use the new tool eeprom like that:

cd build
make
sudo ./tools/eeprom 0 dump slave_0.bin /network_interface/
sudo ./tools/eeprom 1 dump slave_1.bin /network_interface/
sudo ./tools/eeprom 2 dump slave_2.bin /network_interface/

[kapil13j]

Sure, I have saved the eeprom files as you suggested. please have a look and let me know what could we do next.

Thanks,
Kapil

slave_eeprom.zip

[leducp]

Hello Kapil,

I managed to find the bug in KickCAT thanks to your eeprom dump. A patch is waiting for test and approval (#97)

[leducp]

Hello Kapil,

The patch was merged today so you should be able to start the bus now. However you still need to adapt the easycat_example to suit your slaves configuration (slave 0 and 3 have no inputs nor outputs and 1 has only one byte input). If you don't adapt the example, the code will crash on output access (177 and 181).