exporting CARD_INFO_GET to user space
rkagan at mail.ru
Sat Feb 26 16:24:23 EST 2005
On Thu, Feb 24, 2005 at 11:23:41PM +0100, Marco d'Itri wrote:
> On Feb 24, Roman Kagan <rkagan at mail.ru> wrote:
> > which makes me think that they might be also available in PCI xDSL
> > devices, and the like.
> Not that this is much relevant, because I do not know any PCI DSL modem
> driver which works without at least a proprietary binary DSP library.
Still, running "strings" on the firmware for SpeedTouch and ZyXEL 630-11
reveals that they know most, if not all, of the stats known to
AccessRunner; according to Gilles, so does ADI Eagle. I'm by no means
saying I have any idea on how to get them via USB, but it may be
possible; the generic interface for publishing the stats will prove
useful then :)
Besides, sysfs imposes certain lifetime management constraints. For
usbatm subdrivers, cxacru in particular, lifetime is controlled by the
usbatm core. Thus having per subdriver sysfs data will need extra
interactions with the usbatm core; maintaining the stats in the core
itself will make it much easier.
> > Unfortunately I still don't have a clear perception of Chas' attitude to
> > the proposed several weeks ago class/atm stuff. I'll reiterate on that once
> > 2.6.11 is out.
> Please send a reminder now, hotplug support for ATM interfaces is very
> important for distributions trying to cleanly integrate support for USB
OK, I'll try to.
> BTW, are you going to release the program which generates the parameter
I'm reluctant to spend time on it until we see it's needed by anybody.
So far people seemed to be able to establish ADSL connections without
it, and adding an extra configuration possibility will most likely
result in even more confusion, given the lack of documentation.
BTW the reason I made it optional was the observation that the parameter
file mistakenly created in the big-endian format made no apparent
difference compared to the correct little-endian one :)
More information about the Usbatm