Julian Coleman wrote:
> Matthias Pfaller wrote:
> > I'm curious if this really was causing the problems. I compiled a more =
recent
> > -current kernel for my test machine last night. After work I'll try if =
my
> > EA412 is still working with this kernel (just in case it's a mi bug).
>=20
> I don't think so.  Changing this didn't make any difference.  BTW, the ke=
rnel
> I'm running on is about 2 weeks old.

In the meantime I tried a newer kernel on my spare pc532 (the one with the
EA412) and it still is working ok. But the tree I was building from isn't
really -current, it is about 4 weeks old.

> I don't know if this is an MI bug or
> not.  One thing is that SCSI_NOSLEEP isn't used much in other SCSI code.
> Perhaps on the TT, it does tickle a race condition in the SCSI code (as L=
eo
> pointed out).  I don't suppose it's possible for you to attach your EA412=
 to
> an Atari is it?

Sorry, but the only Atari I have is an old 260ST and I haven't turned it
on for years... A TT is missing in my collection.

> Or try if_se.c without SCSI_NOSLEEP on the PC532?

Sorry, no. You have to use SCSI_NOSLEEP. Both, se_ifstart and se_recv
can be called from an interrupt context. sleeps are not allowed here.

Matthias
--=20
Matthias Pfaller                            Software Entwicklung
marco Systemanalyse und Entwicklung GmbH    Tel   +49 8131 5161-41
Hans-B=F6ckler-Str. 2, D 85221 Dachau         Fax   +49 8131 5161-66
http://www.marco.de/                        Email leo@dachau.marco.de