Mity CPU Platforms » MitySOM-5CSX Altera Cyclone V

Hi Atef,

The image you downloaded should have an SSH server (OpenSSH) running by default, yes. ps by default doesn't show all processes, which is why the grep in your screenshot didn't find it. Running ps -A or ps aux should confirm that sshd is running.

We don't have any known issues with SSH on this image, but since you're having trouble pinging as well, I suspect this is a more general Ethernet issue. One change that our more recent kernels (including the image you downloaded but probably not the one that came preinstalled on the original dev kit) have is that the default PHY skew adjustments in our device tree require the PHY to be detected at address 3. If it's not, the skews will most likely be missing, which can cause quite bad packet loss depending on the carrier.

You can confirm what PHY address was detected by running ethtool eth0 | grep PHYAD. Note that previous versions of our dev kit used address 2 and also had missing level translation that could cause unreliable address latching; if you referenced revision 10 or earlier, there's a good chance this is the issue.

If the PHY is reliably being detected at an address other than 3, you can try changing the reg value in the phy0 device tree node to match; or, if you let us know the detected address, we can provide an updated device tree with the change. Long term, if you're planning to use a custom device tree anyway, there's not really any reason to match the dev kit's PHY address if you used 2 or another value in your design (although we still recommend the level translation discussed in the link above, if possible). If you do want your carrier to work better with our stock device tree though you'd need to change the PHYAD strapping resistors to select address 3.

As an aside, in case you have a local Yocto build you'd like to install packages from, our MitySOM-5CSx images are configured with rpm as the package manager rather than opkg. We don't host a package server though, so you'd need to build any desired packages yourself.

Zach

Hi Zach,
Thank you for your reply.
I ran ethtool eth0 | grep PHYAD on my system, and it confirms that the PHY is detected at address 3, as shown in the attached screenshot.
Could you please let me know if the Ethernet hardware section of my carrier board (PHY and RGMII connections) looks correct (R525 is not connected) ?
clipboard-202512090911-npx5h.png
Also, how can I verify that there are no RGMII skew/timing issues in my hardware design? Do you have any specific checks or measurement methods you recommend for this?
clipboard-202512090915-aewdr.png
Thanks

Hi Zach,

Just a quick update on my side.
I am now able to communicate over Ethernet with our custom carrier board. However, in some cases I still observe data loss that results in dropped connections. My current suspicion is that this may be related to the RGMII traces on our Ethernet interface at 1 Gb/s (signal integrity / skew issues).
If you have additional recommendations for checking RGMII layout or timing on the MitySOM-5CSx, I would be very interested.
Thanks,

Hello,

For layout, the two groups you have outlined in your schematic snippet should be matched in length and ideally be on the same layer. The clocks should be separated from the data lanes by at least 5x trace width. We usually use a 5mil matching rule for RGMII trace groups, with 50 Ohm impedance.

The PHY datasheet also has some routing rules regarding the ground plane between the PHY and the magnetics and Jack.

With regards,
Mike

Hi Zach,

Could you please generate a Linux image for the 5CSX-H6-4YA based on the Linux image available at the following link (https://support.criticallink.com/redmine/projects/mityarm-5cs/wiki), but configured so that the Ethernet link is limited to 100 Mbps instead of 1 Gbps?

Regards

Hi Atef,

You should be able to do this via ethtool:

ethtool -s eth0 speed 100 duplex full

My appologizes if you already tried this but thought I'd give a quicker approach if you're trying to test out different speeds.

Regards,
Dan