On Tuesday 03 November 2009 06:54, Sebastian Haas wrote:
> Wolfgang Grandegger schrieb:
> > Sebastian Haas wrote:
> >> Regarding firmware update. As this process is always pretty manufacturer
> >> specific, I don't see any need to build a general interface for it. Each
> >> device may have it's own solution.
> >
> > IIRC, there is a generic interface for firmware downloads, which sjould
> > be used. I think the softing driver uses it.
> The Softing board needs a firmware to be uploaded to actually run. The
> CPC-USB for example doesn't need an extra firmware to be loaded but
> instead you can upgrade/flash the internal firmware if you need.
> 
> I don't know the generic interface for firmware downloads, but I don't
> want to upgrade the firmware automatically if I connect a devices. AFAIK
> the firmware download mechanism starts automatically.
> 
> Sebastian

You are both right. When a device needs a firmware download to actually
run, the kernel's firmware mechanism should be used. Other devices
have some firmware in thier flash. You don't want to update these
devices automatically. So a special device specific implemenation is needed.

For USB devices I could think of libusb for such operations. This brings the 
advantage, that a driver could even refuse to load until a specific firmware
is flashed through libusb.

But this does not answer the question for PCI devices. This requires some driver
support. I could think of the kernel's firmware mechanism for this,
but with a manual trigger. This trigger fits into the same sheme as the
"enable 5V on pin1" task mentioned elsewhere in this thread.

So when we find a way for the "enable 5V on pin1" issue, firmware updating
is somehow solved from point of view.

Matthias
_______________________________________________
Socketcan-core mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/socketcan-core

Reply via email to