06_dcmac on dev branch

[papeg]

We want to test the dcmac connections with the latest version of SLASH we are using (based on dev).

I was able to build the example with some minor changes in the CMakeLists.txt:

papeg@43f100f

We have connected port 0 and port 2 with a QSFPDD cable (which should match the configuration in the example).

Using dmac_init.py is reported to be up:

python ../../../linker/resources/dcmac/driver/dcmac_init.py -d 61 -k
Keep ALIVE path
DCMAC 0 link still up after 20.1 s

network_end2end_test.py runs through, but shows 0 values for all counters.

Using the cpp host code (06_dmac.cpp), the FPGA crashes after starting the kernels with invalid PCI data read:

[Wed May  6 09:06:44 2026] ami 0000:61:00.0: ERROR           : Logging thread error - invalid PCI data read!
[Wed May  6 09:06:44 2026] eqdma_cpm5_hw_error_process: Global Err Reg(0x248) = 0xffffffff
[Wed May  6 09:06:44 2026] ami: CRITICAL WARNING: cmd id: 74 timed out(timeout), hot reset is required

[mariodruiz]

Is the 'crash' you are report triggering a reboot or is just this error you provided?

[papeg]

I need to check with my colleague if the reboot was triggered, but it definitely required a reboot to recover from it.

[papeg]

We were able to pin down the error a bit more. It is caused by executing the cpp host code a second time. The first run deadlocks and is then cancelled manually. The second run then triggers the reboot. Attached is the full dmesg output. The runs of the python scripts yesterday must have been in between these two runs. I hope this helps in understanding the error better, thanks a lot.

dmesg.txt

[mariodruiz]

Hi @papeg,
Thank you for the details, we are looking into it.

Can you do a PR with the changes you made to the cmake to dev?

[papeg]

Great, thanks for looking into it!

Here is the PR with the changes: #81