Benjamin Tissoires

adding @whot to the thread, as he is the main user of the mt protocol. The problem here is that the doc is not always accurate, and when userspace starts...

> > For hidraw what we will likely need is a `LD_PRELOAD`-capable wrapper that routes `open("/dev/hidraw")` through whichever portal we want to use. > > Doing blocking D-Bus'y things in...

> > If a file is opened in the session, we know it has access to it (with `uaccess`). > > The problem here is that we also want to...

> That sounds racy, and that it might cause the problems mentioned in https://gitlab.freedesktop.org/bentiss/logind-hidraw/-/issues/1: > > ``` > Plus, the ioctl chain does not actually open the device until everyone...

> > This is racy, yes, but do we really care that the device is readable for a few (milli)seconds? > > Yes, absolutely. Not all "HID" attached devices are...

> So… what is the status here? Did the kernel patches get queued up? We got a small step back on the kernel side: when @hadess submitted his [USB patch](https://lore.kernel.org/linux-usb/[email protected]/T/#t),...

>> revoke ioctl at the VFS level, not per subsystem. > wasn't that attempted before and failed? Yes, that was one of the remark in the thread by Greg KH....

sorry, I haven't touch wekan in a while I I don't even recall having coded such a feature :( Trying to dig into that now would be quite some more...

> Also: I'd noticed early on in debugging that running rmmod wacom and then modprobe wacom would cause the device to disconnect (as expected) but never come back until I...

Sorry, I should have mentioned it. I used https://github.com/vpelletier/python-hidraw it's a one file module, entirely in python. So worse case, we can just add it as a submodule of openrazer.