In message <Pine.SGI.3.95.970117183343.6954A-100000@physig4.ph.kcl.ac.uk>
          Mark Brinicombe <amb@physig.ph.kcl.ac.uk> wrote:

> Hi,
>   Ok I have finally got to new test kernels built and on the ftp site
> They are voy-4922 and sa110-4922.

Any chance of the new ps, w, kvm_mkdb and others appearing prebuilt ?

> 
> These are test kernels as described below.
> 
> Note: These kernels do not have ICS IDE or ATAPI support.
> 
> Ok These kernels have new versions of the com and lpt drivers that use the
> new bus space implemention. (as commited to the source tree)

I wonder whether the lpt driver is causing some problems with the
keyboard not reseting properly ?

Since I've been building my own kernels, I've not suffered from this
problem at all and I don't include the lpt driver in my config file. I
tried SA-4922 and couldn't get it to get past the keyboard reset even
after five attempts. I left one of the attempts for about 3-5 minutes
and it didn't time out either which surprised me, since I thought it was
supposed to timeout after 100s. If it helps narrow this one down, I can
send you either my config file or the diffs between it and SA110,
because I'm becoming convinced the keyboard reset problem is related to
a difference between my kernel and yours. Though, whether it is lpt
related is probably a real longshot.

> They also have a new version of the floppy driver using the new bus space
> stuff and an updated wd driver using the new bus space stuff (have not
> done ATAPI patches for this yet).
> The com driver has the bug fix for the GT665 combo chip
> The new interrupt fixes are in so there should be no ibuf overflows.
> You may still get some SILO overflows but that is a different matter (have
> this problem with NetBSD/i386 as well)

Well, with my own built kernels, the ibuf overflows have completely
disappeared and the SILO overflows are now pretty rare. I used to get
hundreds and now I only get a couple. Certainly looks good to me!

regards,
Pete