Hi Ivan,
On Sunday 11 November 2007, Ivan N. Zlatev wrote:
> On 11/10/07, Laurent Pinchart <[EMAIL PROTECTED]> wrote:
> > Hi Ivan, Tom, Felix and all other iSight users.
> >
> > On Thursday 08 November 2007, Felix Möller wrote:
> > > Hi,
> > >
> > > >> But as we currently stand everything works so nicely as it is. Why
> > > >> should we have to spend precious time to port all of it to user
> > > >> land, read on how to write udev rules, bash scripts and what the
> > > >> best solution for distributions will be in terms of all of that.
> > > >>
> > > >> :-(
> > > >
> > > > I think usability is also an argument. For a user it is already
> > > > complex enough to install the driver without firmware, scripts etc. I
> > > > think it is better to have a solution, that "just works" and can be
> > > > handled by users easily in contrast to an very clean solutions noone
> > > > understands.
> > >
> > > yes I think so too.
> > >
> > > And I think firmware loading from the kernel is not that uncommon the
> > > new intel wlan driver iwlwifi loads its firmware too.
> >
> > EZ-USB firmware loading have traditionally been done in userspace. The
> > fxload utility has been designed for that purpose. I'm also not keen on
> > adding a dependency on SHA1 to the Linux UVC driver, especially if only
> > one buggy device requires it.
>
> I see. Kernel space fxload firmware loader sounds a bit off scope.
It's a bit out of scope for the Linux UVC driver, but it might still be
useful.
Basically, my position is that I don't want to decide if firmware loading
should be done in userspace or kernelspace. EZ-USB firmware loading is
currently done in userspace by the fxload utility. I'm not against moving
this to the kernel if USB gurus (I'm thinking about the linux-usb-devel
mailing list) think it would be a better approach. In that case, I would
however insist on implementing this in a generic way, not specific to the
Linux UVC driver.
To summarise the issue, I'm ok with a userspace implementation (which already
exists). If you feel the task would be better handled in the kernel, I would
be ok with a generic kernelspace approved the linux-usb-devel.
> Do you know off the top of your head a driver I can look into for their
> fxload based userspace hotplug(?) integration (scripts?)? I am willing
> to look into a fxload based setup as long as it's convenient for
> distributions.
The Xilinx ISE toolchain uses fxload. The software can be downloaded at no
cost. As the package is a bit big (around 1GB), here's the udev rule:
BUS=="usb", ACTION=="add", SYSFS{idVendor}=="03fd", SYSFS{idProduct}=="0009",
RUN+="/sbin/fxload -v -t fx2 -I /usr/share/xusbdfwu.hex -d $TEMPNODE"
Cheers,
Laurent Pinchart
_______________________________________________
Linux-uvc-devel mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/linux-uvc-devel
