Messages in this thread |  | | From | Ondrej Zary <> | Subject | Re: [PATCH WIP] parport: add device model | Date | Fri, 10 Apr 2015 20:24:23 +0200 |
| |
On Friday 10 April 2015 16:30:38 Sudip Mukherjee wrote: > This is work-in-progree, not for applying to any tree. Posting now for > your comments so that I know if I am in the proper track. > > in parport_register_driver() driver is registered but i am not linking > anywhere the device with the driver, but yet when I am testing this > patch I am seeing in sys tree that parport0 is linked with > the lp driver. Is it done in the device core? I am missing this step > somewhere. > > In parport_claim() the attach is unchecked as of now, I think we will > need my initial patch series of monitoring the attach return value along > with it. > > while testing I am getting NULL dereference with daisy.c, and after > disabling PARPORT_1284 , I am getting some new errors. so if you are > testing this patch please keep in mind that still lots of work is > pending. > My main intention to post it now is to know if my approach is correct.
Many newer parallel port devices support plug&play (IEEE1284 device ID) but Linux never supported it properly. The ID is probed and even the class is printed in the kernel log (drivers/parport/probe.c) but there's no support for module autoloading based on that. This could be a good opportunity to add this support. I was thinking about this while playing with some parport webcams recently.
-- Ondrej Zary
|  |