Using IOCTL for GPIO control

[falcon35180]

I have built modbusd for an embedded NVR with an ARM CPU (HiSilicon Hi3520D V300) and am using it to communicate with a PV inverter.

The kernel module for GPIO control exposes an IOCTL interface to set pin state and direction - no sysfs file interface, unfortunately.

I figured out the correct commands to control the appropriate pin through reverse engineering (disassembling, decompiling, writing test programs).

This particular driver takes a structure pointer as an IOCTL argument; there may be others out there that take int/long arguments.

I made a slight modification to tty.c for this purpose and it works well for my case. It would be great to find a way to implement this in a more generic way.

I haven't forked the repo to publish the code since it's not really ready yet. I may work on it some more when I get a chance, but I'm mentioning it in case someone else wants to take it on.

Happy to assist in any way I can in the meantime.

[3cky]

Thanks for the information. Still not sure how to incorporate IOCTL GPIO controls in mbusd in a generic way, but this functionality indeed can be helpful for someone. A good first step will probably be to create a pull request with your changes to make them accessible to anyone.

[falcon35180]

I have some ideas for the design of the feature to make it versatile. An example configuration could be:

trx_control = ioctl trx_sysfile = /dev/gpio_ctl trx_ioctl_req = 0x4 trx_ioctl_arg_type = struct

# 1 when RTS set, 0 when clear # rts_inv for inverse trx_ioctl_val_1 = rts

I think this could be a good starting point, allowing for future extensions without breaking compatibility. More configuration values could be added, such as:

trx_ioctl_set_req trx_ioctl_clr_req

trx_ioctl_set_val_<n> trx_ioctl_clr_val_<n>

As I said, I'll do some more work on the code when I get a chance. What I have so far is just a quick hack to adapt the program to my needs.

[falcon35180]

Update, in case people are wondering:

I have implemented as per the first part of my previous comment. I'll do some more refinement regarding memory allocation/freeing and safety testing (I really want to minimise chances of issuing random IOCTLs to drivers!) before uploading the code and submitting a pull request.

Rest assured that work is in progress!

[falcon35180]

Update - I have done a bit more work on my code. It's been pushed to my own fork and a PR has been submitted.

It seems to work well and is in use on my own system.

I even managed to fix an unrelated bug while I was at it.

I would encourage everyone to review it themselves before putting it to work, but it looks OK so far.