[mythtv-users] OT. WOL for my myth box.

[David Watkins]

First of all thanks for the reply.

> The nVidia-based MSI mobos I have seem to require that the WOL packet
> go to the -broadcast- address, e.g., even though the packet has the
> machine's MAC inside it, the network interface probably isn't even
> -listening- for that packet unless it's being broadcast.

I've been sending to 192.168.1.255 (interface is 192.168.1.10).  I
think that's the broadcast address?

> The VIA-based MSI mobo I have works if the packet is directed either
> straight to that machine's MAC or to the broadcast address.

I'll try sending to the MAC address directly.
>
> On the other hand, those nVidia-based MSI mobos (I've got 3 of them,
> all identical) have some sort of bug under at least Ubuntu Breezy
> (haven't tried them under either Hoary or Dapper), in that rebooting
> the machine leaves the network interface in some peculiar state such
> that sending it a WOL packet won't actually wake up the machine.
> (I -believe- that this might have been fixed upstream in some later
> kernel [ISTR it was some debugging statement that was inadvertently
> left in or something like that], but I haven't tested that yet.)

That seems to be the sort of problem I've got, but I'd like to try and
prove whether it's the motherboard or the NIC.  I'm starting to read
problems about some of the newer,  budget 8139 cards which aren't
necessarily compatible with the older drivers.  The 8139cp driver
won't load for me, which I believe it should if the card is genuinely
an 8139C+.  That's why I'm trying to understand some of the PME flags
that I'm seeing.  Most of them seem to be PME- which I presume means
'OFF'?

> However, if one of those nVidia-based MSI's is turned off cold, then
> booted directly into the BIOS (e.g., hold down DEL) or otherwise
> prevented from booting Ubuntu, and then turned off manually again,
> the WOL packet -will- wake it up.  (I discovered that by trial and
> error while I was getting WOL to work at all.)

Well I'll try that.  It would be success of a sort, though not a solution.

> So you might have one or both of these sorts of things going on;
> I'd recommend trying various combinations.  Note that I've been
> using the ether-wake.c program by Donald Becker (v1.09 11/12/2003,
> www.scyld.com), which I found via a web search.  You could try that,
> or use whatever you're using now if it can also broadcast.

Yeah I've used that, and several other programmes to try and send the
wakeup.  I think the problems with the listening end.

Thanks again.  If I do get anywhere I'll post back.

D