More VK204-25-USB fun
: the device actually fails the USBCV 1.3.1 compliance test because it reports a zero-length read buffer. Oops! I'm hoping to come up with some kind of software workaround for linux, since windows obviously manages to do it.Update:
Success! Now when the device is detected, I overwrite the the kernel's copy of the wMaxPacketSize with its "real" value (i.e. NOT the one provided by USB). The serial port is now usable.
However, I'm not out of the woods yet. LCDd can't *quite* communicate with it properly. But I have tested a bunch of the commands by writing them directly to the serial port, and it all seems to work. So, now that I've fixed the USB driver, it's time to fix the serial protocol driver :)