Thu, Jul 21, 2005 at 08:20:11PM -0400, Miles Nordin:
> >>>>> "mn" =3D=3D Miles Nordin <carton@Ivy.NET> writes:
>=20
>     mn> I found the problem also occurs with an hme card.
>=20
> however, I get these messages in dmesg that i didn't get from gem:
>=20
> hme0: status=3D400<MAXPKTERR>
> hme0: status=3D400<MAXPKTERR>
> hme0: status=3D400<MAXPKTERR>
> hme0: status=3D400<MAXPKTERR>
> hme0: status=3D500<SENTFRAME,MAXPKTERR>
> hme0: status=3D400<MAXPKTERR>
>=20
> There are ~30 of these over a day or so, many individual stalls.  I'm
> not sure that every stall prints a console message, though.

These are from the DMA ring filling.  See _HME_NDESC in /sys/dev/ic/hme.c.
IIRC, that can be as much as 256.

These messages were disabled at some point, before 3.0.  hme.c 1.39, I thin=
k.

> 3.0_BETA is still building.  There is no ALTQ patch for 3.0_BETA, and
> PF/ALTQ still isn't integrated into NetBSD, so I'm not sure that's
> ultimately going to work out for me, but I can at least report whether
> it fixes these hme/gem problems before trying something else.