USB Ethernet (aue) panics on 4.10 (kern/69319)

[Barry Bouwsma]

On Thu, 2 Dec 2004 00:08:31 +0000, Scott Mitchell wrote:

> > I've put what I *think* I've compiled into a module up at
> > https://NOSPAM.dydnds.dk/hacks/

> Cool.  I'll take a look at this over the weekend.  if_axe looks similar
> enough to the rest of the USB Ethernet drivers (they were all wpaul
> creations to begin with) so it shouldn't be too hard to put together
> something that's mostly right...

I'm overjoyed to report that your patches to if_aue and its ilk
(and uhbdi-whatnot) are what I needed to get if_axe working, whee

And that's what I'm sending this mail out on, no less.  Though I
haven't done a Real Test -- I've only made a 10Mbit/sec connection
to a Cabal Modem.  I'll probably try a 100Mbit/sec switch sometime
Real Soon Now.

There are two more if_axe.c files in the location given above to
be downloaded now -- -NEWTEST, and -THIS_REALLY_WORKS!!@!!
The latter is what I'm using.  The only real difference between
it and -NEWTEST is that the former has splx(s) and the like, taken
from either NetBSD or DragonFly's if_axe code.

I don't think that locking was what caused me to have panics in
nd6_timer; I think it's one of my homebrewed scripts I use to
assign IPv6 addresses, that is sub-optimal when more than one
dhclient'ed interface is present.  (Not that I should be able to
induce an IPv6-related panic by ifconfig'ing interfaces, but it's
been that way for years)

Anyway, I still plan to take the latest -current code, and merge
in what's needed from my ugly-hacked versions as neatly as possible,
and put that up eventually as if_axe.c with no suffix.

Note that both NetBSD-current and DragonFlyBSD have if_axe.c in
their source trees nowadays, so someone knowledgeable may well
want to review what they've done and if it's applicable.  I took
the splx() and such from one of them in a vain attempt to get a
non-crashing module months ago, without having the faintest idea
what I'm doing, so I'd welcome comments from anyone who can set
me straight.


currently being tested connected to OHCI.  Haven't tried it with
UHCI or EHCI to see if the eternal-probe problem is present with
the former, and to verify the latter works.  Only been online
for some minutes.  Problems will be reported.  OHCI is mostly
that from the latest 4.x at present.  Device is a Linksys USB200M.
`dmesg' output relevant to it is
whoops, not in dmesg.boot as I came out of single-user, hmmm, try this
 axe0: Linksys USB 2.0 10/100 ethernet controller, rev 2.00/0.01, addr 3
 axe0: Ethernet address: 00:10:60:25:f1:e5
 miibus0: <MII bus> on axe0
 rlphy0: <RTL8201L 10/100 media interface> on miibus0
 rlphy0:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto
 bpf: axe0 attached.
`ifconfig' looks like
axe0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
        inet6 fe80::210:60ff:fe25:f1e5%axe0 prefixlen 64 scopeid 0x1
        inet 84.72.24.80 netmask 0xfffff000 broadcast 255.255.255.255
        inet6 2002:5448:1850:0:210:60ff:fe25:f1e5 prefixlen 64
        inet6 2002:5448:1850:feed:babe:dead:c0de:d00d prefixlen 48
        inet 172.27.72.27 netmask 0xffffff00 broadcast 172.27.72.255
        ether 00:10:60:25:f1:e5
        media: Ethernet autoselect (10baseT/UTP)
        status: active
guess I'll let it run for a while with OHCI and see if there are any
showstoppers -- but I should probably boot cleanly with it too...
I loaded it as a kernel module


thkans
barry bouwsma