On Fri, Jun 14, 2024, at 12:03, Wolfram Sang wrote: > Hi Arnd, everyone, > >> I could imagine treating both gpio-virtuser and this code as >> a gpiolib extension rather than a consumer (which is usually >> part of some other subsystem's driver). > > I have difficulties seeing this. For the analyzer, at least. It does not > extend gpiolib in a way another consumer could make use of it? That's the same as the chardev support in gpiolib: it doesn't provide functionality to in-kernel consumers but does give an interface to userspace. Arguably, integrating the logic analyzer into the chardev device itself would make sense from an interface, as you could define it as ioctls and mmap instead of the debugfs interface. Of course we don't want to make it a first-class interface because of the reasons you explain in the cover letter, but it would totally make sense to me to call it a debugging feature of libgpio instead of calling it a standalone driver. Arnd