Re: IPL_NET on if_le_vme.c

[Tetsuya Isaki <isaki%pastel-flower.jp@localhost>]

At Wed, 25 Jun 2008 00:40:54 +0900,
Izumi Tsutsui wrote:
> (please keep Cc: to me)
:
> > While reading source, I have a question.
> > 
> > Here, IPL_NET is 5 and PSL_IPL is 0x0700,
> > therefore, the condition always false?
> 
> Well, maybe it's my botch on MI softintr(9) change.
> IPL_NET was (PSL_S|PSL_IPL3) before that.
> (yes, using ipl2psl_table[] on atari might cause some extra overhead.
>  now I wonder if it's worth to make all m68k ports consistent..)

I think that it isn't necessary to unite it,  if there are
some reasons.

> > I'm not sure but how about this?
>  :
> > -   if ((sr & PSL_IPL) < (IPL_NET & PSL_IPL))
> > +   if ((sr & PSL_IPL) < PSL_IPL5)
> 
> (ipl2psl[IPL_NET] & PSL_IPL) is more reasonable than PSL_IPL5, I think.

Ah, that's right.
I'll post the patch at next mail.
---
Tetsuya Isaki <isaki%pastel-flower.jp@localhost / isaki%NetBSD.org@localhost>