Re: [Qemu-devel] [PATCH V2 2/3] usb-gotemp: new module emulating a USB thermometer

[Greg KH]

On Wed, Nov 11, 2009 at 01:15:45AM +0100, Alexander Graf wrote:
> 
> On 11.11.2009, at 01:09, Anthony Liguori wrote:
> 
> > Scott Tsai wrote:
> >> On Wed, Nov 11, 2009 at 1:06 AM, Luiz Capitulino <address@hidden 
> >> > wrote:
> >>
> >>>> I'd certainly like to make this code useful for something other  
> >>>> than
> >>>> developer training.

What code?  Where is it at?

> >>>> How about a new monitor command "thermometer_set" that works like  
> >>>> "mouse_move"?
> >>>> "thermometer_set" would just set the temperature of the "first"
> >>>> thermometer device it finds.
> >>>>
> >>> Couldn't the device be a parameter?
> >>>
> >>> And I'd suggest usb_therm_set for the name.
> >>>
> >>>
> >>
> >> Looking at the existing "mouse_set" and "mouse_move" monitor  
> >> commands,
> >> they work on USB, PS/2 and other kinds of mice with "mouse_set"  
> >> selecting
> >> the mouse device affected by  "mouse_move".
> >> So how about a new command "therm_set" which selects the thermometer
> >> affected by "therm_temp" ?
> >>
> >> On a separate note, I understand that if a piece of code is not  
> >> useful enough
> >> we don't want to merge it to add to the maintenance burden.
> >> I still propose 'usb-gotemp' for merging because the fact that gregkh
> >> could give his
> >> driver tutorial several years in a roll to sizable audiences shows
> >> that there are people out there
> >> interested in getting into Linux driver development.
> >> With this code merged, people could follow the video and slides of  
> >> his talk
> >> without special hardware and this potentially grows the Linux  
> >> developer pool.
> >>
> >
> > And if Greg decides to change the device he uses for the tutorial,  
> > then in a few years it's not so useful anymore?
> 
> Well, why don't we ask him?

I don't understand the context here.

There is already an in-kernel driver for the gotemp usb device, that has
been there for many many years.  Why would you want to write a different
one instead?

Yes, I do use a different one for my "write a driver" tutorial, but when
I give that class, I note the differences between the userspace
interface for that driver, and the one that is already in the kernel.

> > That said, if we position this as an example device, I think that  
> > makes sense.
> 
> I personally don't think it should be a requirement for inclusion in  
> qemu that it's useful for years on for everyone using it. If there's a  
> big enough group of people using a feature it seems worthwhile. And if  
> Scott went through the trouble of implementing it, I'm pretty sure  
> there is enough of an audience around.

Implementing something I wrote 4 years ago?  :)

thanks,

greg k-h