Today I built a new SD card with the recent (11Jan) .27 files and
rootfs from the angstrom../demo/beagleboard web page. The system
boots up fine, but...
1) It no longer sees the ASIX USB Ethernet adapter or turns on the
green lights for it. There is no mention of it in the boot up log.
And there is no eth0 device.
2) The mplayer no longer works at all. It runs, prints out its first
dozen or so status lines, and then hangs without any sound or video
playing. And once it hangs, the sound on the system no longer works
for any other task.
Any ideas?
A couple of additional notes...
Re: 1) When I use the kernel/rootfs from the Sakoman feeds, the
ethernet works fine.
Re: 2) When I use the kernel/rootfs from the Sakoman feeds, the
mplayer also no longer works.
So the mplayer problem seems related to Angstrom in general. While
the ethernet problem seems related to Angstrom from the demo/
beagleboard folder.
The mplayer problem in my feeds is a kernel issue (alsa was broken
with a recent checkin) and should be fixed within the next few days
when 2.6.28-omap2 is tagged.
Sakoman: Thanks much. I'll wait for the updates. That makes sense.
Koen: More information... My Trendnet Adapter works fine. But my
GWC Adapter no longer works with the demo kernel. Both use the Asix
Chip. Has there been some change in the driver for usb ethernet that
might be affecting that? Both GWC and Trendnet work with the Sakoman
feeds. Where would I look?
No. I can see from lsusb that the USB bus sees the ASIX chip. But
there is some piece of init code that no longer runs. So the green
lights don't come on and there is no eth0 seen. Others have had the
same problem. The GWC uses the same chip as everyone else, but they
may have wired it up slightly differently.
Right now, I'm waiting for the sound / mplayer bugs to get fixed and
hopefully this will start working again.
I don't think so. I'm not using NFS. And it's not that I can't get
an address. It doesn't even bring up an eth0. So I can do lsusb and
it shows the ASIX. But if I do ifconfig -a for example, it only shows
lo. And it doesn't "activate" the hardware to the point that the
green lights come on.
If I use one of the sakoman feed images (or an older Angstrom demo --
the 9/25 one), then it works fine.
Here's a thought that might be crazy: The GWC unit has an ethernet
MAC address that is obviously either wrong or illegal -- i.e. it has
the first four or so digit couplets as zeros (00:00:00:00:ab:cd) or
something like that. Perhaps the new code is checking for "valid" MAC
addresses and deciding that this is not valid.
Is that possible? Where would I look for an answer like this.
I don't think so. I'm not using NFS. And it's not that I can't get
an address. It doesn't even bring up an eth0. So I can do lsusb and
it shows the ASIX. But if I do ifconfig -a for example, it only shows
lo. And it doesn't "activate" the hardware to the point that the
green lights come on.
I can't check right now, but it seems like the fix for the NFS root
issue may have broken operation of usb network interfaces. I know mine
has been broken for a a couple of weeks, I'm not there right now, so I
can't confirm the problem.
Here's a thought that might be crazy: The GWC unit has an ethernet
MAC address that is obviously either wrong or illegal -- i.e. it has
the first four or so digit couplets as zeros (00:00:00:00:ab:cd) or
With the trendnet adapter, I often have to 'lukewarm boot' (turn off the power strip for 0.5-1 seconds) a time or two. The first time it comes up with all 00's. When I briefly tried the .28 kernel, it would sometimes not detect the adapter at all the first time through.
No, this isn't an intermittent problem for me. There is something in
the driver for ASIX that has decided that the GWC version is not OK to
start up. So it's not that I get an eth0 configured to 0.0.0.0 (as in
the patch that Philip referenced) -- it's that I don't get an eth0 at
all (no such device).
How would I track down if there has been any change in the ASIX driver
in the last month or two?
No, this isn't an intermittent problem for me. There is something in
the driver for ASIX that has decided that the GWC version is not OK to
start up. So it's not that I get an eth0 configured to 0.0.0.0 (as in
the patch that Philip referenced) -- it's that I don't get an eth0 at
all (no such device).
How would I track down if there has been any change in the ASIX driver
in the last month or two?
This was super helpful. Thanks. Here's what I have discovered -- and
I need more help...
The GWC ethernet adapter that I have checks in with USB id:
0xb95:772a.
There was a change on Nov 11 by Jason Cooper: (http://git.kernel.org/?
p=linux/kernel/git/tmlind/linux-
omap-2.6.git;a=commitdiff;h=ccf95402d0ae6f433f29ce88cfd589cec8fc81ad)
that added a table entry for that number.
I'm not sure how that might have broken the GWC. But the coincidence
seems too good to ignore.