On 2012-12-08 2:34 PM, Daniel Gimpelevich wrote:
> On Sat, 2012-12-08 at 10:49 +0000, David Woodhouse wrote: 
>> On Fri, 2012-12-07 at 18:58 -0800, Daniel Gimpelevich wrote:
>> > > > OK, but hotplug never generates any atm events, so Felix suggested this
>> > > > instead.
>> > > 
>> > > Hm, it should do. We have other scripts in /etc/hotplug.d/atm, and
>> > > hotplug2.rules definitely *looks* like it should be invoking them.
>> > > 
>> > 
>> > There is only one such script, and you should make sure it still
>> > works, because I put one in to test, and it was never executed.
>> 
>> That needs fixing then. Don't just do things wrongly to work around a
>> bug elsewhere — *fix* the bug. I may be able to take a look at this in a
>> few days' time, once my Geos is up and running again.
>> 
>> Bad Felix, if he really advised you to use an initscript and he meant it
>> as a long-term thing rather than a temporary hack for testing the
>> concept.
>> 
> 
> Does solos-pci add the sysfs entri(es) when the module is inserted, or
> later? Modules are inserted before hotplug is running, so if solos-pci
> does not add at that time, that could account for the discrepancy. I
> would have assumed Felix knew hotplug wasn't running yet when he
> suggested this.
As for the time when hotplug is running wrt. loaded modules - normally
udevtrigger should take care of generating the right events.

I was suggesting this as a hack rather than a proper fix. I don't see
any of the script based MAC address override approaches as a long term
solution.

The proper long term solution is to make sure that the kernel does the
right thing and doesn't require userspace fixup for these devices.

- Felix
_______________________________________________
openwrt-devel mailing list
[email protected]
https://lists.openwrt.org/mailman/listinfo/openwrt-devel

Reply via email to