Am Mittwoch, 6. Dezember 2006 17:07 schrieb Duncan Sands:
> On Tuesday 5 December 2006 18:02, Oliver Neukum wrote:
> > Am Dienstag, 5. Dezember 2006 16:54 schrieb Duncan Sands:
> >
> > > it is helpful for devices that want to upload firmware in their probe
> > > method. Getting the firmware from the firmware subsystem can take a long
> > > time (due to timeouts for example, which may be infinite, or because it is
> > > waiting for a filesystem to be mounted etc). Parallel probing keeps the
> > > rest
> > > of the USB subsystem running while that device is waiting, which is
> > > particularly
> > > important if the filesystem you are waiting for is on a USB device, since
> > > otherwise it could be a deadlock!
> >
> > No, the number of times you can spawn of a new thread is limited.
> > You must be able to cope with that. The configuration you described
> > is buggy.
>
> Hi Oliver, I don't see anything inherently wrong with someone wanting to use
> a USB device which obtains firmware via another USB device. Are you simply
> pointing
> out that loading firmware in the probe method will still be buggy even with
> Greg's
> new scheme, or are you arguing that the above configuration is inherently
> wrong?
The latter. To have a robust system, you must survive probe or resume in
any order and deal with some failure modes. Firmware and loaders are to
be placed on / fs.
Regards
Oliver
-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
[email protected]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
