[mythtv] Backend segfault backtrace, ??ERY_RECORDER

[Dennis Cartier]

> I'm using redhat 9, with 2.4.21, low latency & preemptive patches.  The
> board is a P2B-D, with MSI slotkey adapters. I would be suprised if it
> was a problem with the board, this sucker has been running for almost 6
> years, with hardly any downtime, it's by far the best board I've
> ever owned.

I tried removing the low latency and preemptive options as well. Did not
seem to help.

Yes, my board was also a rock for years, that is why I decided to use it as
the backend.

>
> I'm using a 3com 3c905 in the box now.

I had a Via Rhine based NIC, I also tried a DEC Tulip and am now using a
3c905B.

>
> I'll take one cpu out and see if I still get the error, did you also
> switch to a non SMP kernel?

No I just removed 1 CPU and put the terminating card in the second slot. I
did not recompile the kernel yet. An SMP kernel will boot fine with only 1
CPU.

>
> I've been trying to crash it again, it doesnt seem to want to crash if
> I'm watching on the 1st card, almost always crashes when I'm messing
> around with the 2nd card.  I may take the older one back to compusa for
> a new one, I've only had them a few weeks now.

I only had 1 card at the time, now I have 4 in 2 different backends. I found
that running mythbackend under gdb seemed to decrease the frequecy of the
problem, but not eliminate it completely. Pulling 1 CPU did the trick. No
crashes since.

>
> I just noticed this mesg, have not noticed this before, may be
> related too.
>
> ivtv: Not enough free buffers, stream 0
>

This was on the ivtv lists awhile ago. Can't remember what it was about, DMA
or low memory I think.

Dennis