Any dependency of sysrq for labgrid console?

[atline]

It's maybe not the problem of labgrid itself, I just want to see if you can give me some inspire.

My problem is: on most of i.mx platforms, it's ok to send break to use sysrq, but for some platform, it can't. E.g. imx7ulp as next:

root@imx7ulpevk:~# echo 1 > /proc/sys/kernel/sysrq
root@imx7ulpevk:~#
Enter command. Try 'help' for a list of builtin commands
-> break

----------------------
h
-sh: h: command not found

What I have tried?

1. I tried to directly using microcom to open it with microcom -s 115200 -p /dev/ttyUSB35, it works with sysrq
2. But if I directly use telnet as next microcom -s 115200 -t 192.168.100.122:10014, it doesn't work with sysrq, same error with when using labgrid-console

So, it seems looks like it only can't work for ser2net usage, which labgrid utilize, and just for some i.mx platforms. Do you have any experience what other dependency to use break command in microcom to trigger sysrq? E.g. any special rootfs/kernel settings? Thanks.

[Emantor]

This could be a problem on multiple layers, I don't know whether microcom correctly sends the break signal over rfc2217, it might get modified by ser2net (which has multiple configuration options for breaks) or it might not propagate through ser2net to the serial port.

[jluebbe]

Ah, you already asked at cminyard/ser2net#145. Has this behavior changed over versions, so that we need to handle them differently?

[atline]

I think sending break is still there even we use nobreak if we're talking about the same concept.

I send a question here: cminyard/ser2net#145, in fact I conclude: for the usb-to-serial chip which use CP210X from Silicon Labs, I have to configure nobreak to allow send break after linux up.
Should you have any other concern, could you also join that discussion? Thanks.

[atline]

@jluebbe , YES, I asked the question. This is the first time our engineer ask to use sysrq on some platform which use CP210x chip. I check all i.mx platforms, it just affect imx6ul, imx6ull, imx7ulp, imx6ulz, imx6ulz-lite. I'm still waiting ser2net maintainer's reply. I don't have the background related to this topic, do you mean it will related to our driver code change on these boards? If that, I can ask the driver owner.

[atline]

@jluebbe, @Emantor

1. I rollback to NXP release 3 years with old kernel 5.10, the issue still there. As no hardware change for this board recent years, so I believe it's not a regression issue. Just our engineers didn't use sysrq for these old boards which chip comes from Silicon Labs.

2. And the ser2net maintainer replied more, please take a look from About `nobreak` in ser2net configure cminyard/ser2net#145 (reply in thread)
   From that, it looks it's even safe for non-CP210X chip, the nobreak just turn break off when start, nothing more. Maybe you want to join that thread if you have more concerns.

3. Here, we use NetworkSerialPort, so not rely on labgrid exporter to start the ser2net. So I can separately add nobreak to all CP210x boards. But for labgrid code base, I understand you may want to add that option to exporter as many people use other SerialPort resource based on udev. It looks nobreak also ok for i.mx series chip without CP210x, but I don't know if any impact for other company's chip, you may can evaluate it with your lab.

[Emantor]

Thanks for confirming with the ser2net maintainer, this confirms my suspicion that nobreak does not disable sending breaks which I gleaned from a cursory source code read. @jluebbe can you implement nobreak and test this in your lab environment to confirm there are no issues? Alternatively I can also write the PR.