On Wednesday 8 November 2006 21:17, David Brownell wrote:
> > I don't know what fxload does exactly,
> 
> You can then read its manpage ... :)
> 
> Or see the section on the topic at
> 
>   http://linux-hotplug.sourceforge.net/?selected=usb
> 
> which gives complete examples of:
> 
> >             but couldn't there be a simple
> > usb driver that takes care of fxloading: it would attach itself to
> > any usb device that matches the ids for a device known to need fxloading;
> > it would then get the firmware using the firmware loading infrastructure
> > (the firmware file name would need to be calculated from the device ids
> > somehow), and upload it to the device.
> 
> Exactly what's described in the 'Downloading Firmware with "fxload"'
> seection of the webpage referenced above.  No new kernel code needed;
> it works out-of-the-box on some Linux distros, others may need grab
> the fxload package.

Yes, but why use a custom userspace mechanism (fxload) when you could
use a standard userspace mechanism (firmware loading subsystem)?
And why require people to install an extra program (fxload) when it
could work out of the box due to in kernel support?

The speedtouch driver used to be in a similar situation: there was a userspace
tool (modem_run) to take care of firmware loading.  The kernel driver took care
of the rest, once the firmware was loaded.  There didn't seem much point in
implementing firmware loading in the kernel driver; yet one happy day David
Woodhouse implemented it, and it turned out to be a big improvement.  Why?
Because with the new scheme all you have to do was dump firmware files in
the right directory, while with the old scheme you needed to also install
modem_run, create a hotplug script etc (hassles); and with the old scheme
there were more ways things could go wrong: I used to spend a lot of time
debugging users' hotplug problems and usbfs problems.  Not all of these are
relevant to fxload, yet my speedtouch experience makes me wonder whether
things wouldn't be better with an in-kernel fxload.

> >             At that point the device
> > transmogrifies itself into a device with different ids, the real driver
> > attaches itself and away you go.  Laurent wouldn't need to add stuff to his
> > driver, because this would be a separate (and universal) driver.

Duncan.

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
linux-usb-devel@lists.sourceforge.net
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel

Reply via email to