Re: hme goes down with SENTFRAME,RXTERR

[Julian Coleman <jdc%coris.org.uk@localhost>]

Hi,

> sound like this:
> 
>  http://mail-index.netbsd.org/port-sparc64/2005/07/19/0008.html
> 
> follow the thread---it happened on hme also.

I'd be interested to know if the stalls are still there with gem on -current.
As part of revision 1.68 of src/sys/dev/ic/gem.c, there is a fix to reset RX
DMA on some errors.  My V120 is running 4.0 with (nearly) current gem code
and has been fine as an NFS server.

I wonder if a similar fix needs to be applied in some cases to our hme
driver.  However, the hme is simpler, so only a full reset may be possible,
which will discard all pending packets (as per ifconfig down/up).  I also
wonder why the watchdog timer doesn't kick in and reset the interface.

Thanks,

J

-- 
  My other computer also runs NetBSD    /        Sailing at Newbiggin
        http://www.netbsd.org/        /   http://www.newbigginsailingclub.org/