On Thu, Sep 12, 2024 at 08:26:58AM GMT, Janne Johansson wrote:
> > There is the beginnings of such an implementation in i2c_io.h
> > (JI2C_IOCTL_EXEC behind a ‘#ifdef notyet’ ) but this code has been around
> > since 2004. Is this something desirable (notyet == tobedone) or is
> > the OpenBSD philosophy to keep access to i2c devices strictly in kernel
> > space?
> There is some information in this thread
> https://marc.info/?l=openbsd-tech&m=164510604215263&w=2
> about the feelings from the project to just randomly allow access to
> all kinds of gpio pins a board can expose. I don't know if i2c devices
> generally control equally "dangerous" parts on motherboards and sbcs,
> but if they do, then some restraint might be in order.

Some i2c devices are Hardware Security Modules (like the ATECC608A) and I
think it's also possible to do bad things by changing the behaviour or
measurements made by a temperature sensor for example.

> Now, this i2c behind usb would presumably not be connected to
> something the computer owner would not want to allow programmatic
> access to, but the previous non-detachable i2c buses might have been.

This means making the difference between an internal i2c bus (pci or fdt)
and a detachable bus. 

In this case, what FreeBSD (and NetBSD I think) does for this I2C-Tiny-USB
(and CP2112) driver by exposing /dev/iic* can also be done in the form of
a library using LibUSB and /dev/ugenx.yy (and therefore 100% user space).

Either way is fine, I can finish my driver or rather make a lib and create
a port for it.

-- 
Denis