Trouble in porting tc1126/tcm1680/cn1100 ts driver, no btn_touch event?

[Peter Hutterer]

On Tue, Dec 19, 2017 at 03:35:29PM +0100, far5893 wrote:
> 
> > In the out of order list of things that are simply wrong and tell me
> > that you should not continue on this code base:
> > - there is no license attached to this repo, nor to the one you forked
> > (seems to be https://github.com/xchetah/tc1126-sunxi), nor to the
> > original one (https://github.com/yangsong2512/TCMCode)
> There is generic GPL license.

unless a license is listed in every file or some other indication that the
license applies to all files in the repo, the default is that the files are
copyrighted by the author with all rights reserved. The README doesn't hint
at GPL and there is no COPYING or LICENSE file.

working on code like this without FOSS-compatible licenses can open you up
to a world of hurt.

Cheers,
   Peter

> >    -> this is plain wrong and tell me you shouldn't use this as a base
> > code as you have no idea what can legally happen next
> > - the driver is completely wrong: it says it's SPI but I see
> > everywhere that it is a I2C driver
> Yes,only a naming issue.
> I suppose that original touch screen was spi capable.
> > - the driver obviously doesn't know how to use the Linux Multitouch
> > protocol (which explains why your device doesn't work). The slots are
> > not used, not even initialized and the single touch emulation that
> > libinput/Xorg expect simply can't work
> I add slot initialization and use,after this mail.
> > - there is no git history shared between the 3 repos (yours, the one
> > you forked, the original), so you can not basically fix something in
> > one repo and fix it in the other ones
> the original seems very different and don't know how much is "original".
> > - the driver is not following the usual device model (it will  support
> > only one touchscreen), and the init process is just garbage (it
> > registers the device while in the init, where it should just register
> > the driver)
> It register driver  in init  , and in *probe* , initialize input dev  and  reset TS.
> >   -> actually this might not be entirely the case, but it's clear that
> > it was written until it worked and then people stopped touching it in
> > fear of breaking it
> > - full of commented out parts, we don't know if they have been
> > removed/written by mistake or not.
> I'm tring to clean it.
> > 
> > Anyway, my advice would be to turn over to the sunxi community (if
> > this is a touchscreen usually connected to AllWinner SoCs) and require
> > help from them. Maybe someone will help you there, but IMO, the effort
> > needed to have a working driver doesn't even match the price of the
> > device, so I'd just buy something new and supported.
> > 
> Thanks , but i rewrite Report function following your suggestions and seems that work well
> i have  tablet with silead  and i work to obtain same events as silead_ts.
> Now pointer is fully working, and driver compile well.
> 
> I know that none will never sell a product with this driver , but for hobbyist is sufficient,
> and if a skilled kernel driver writer want try to write from scratch it is a *working* starting point.
> 
> Miky