[vox-tech] Losing keyboard and mouse between boots

[Rick Moen]

Quoting Peter Jay Salzman (p at dirac.org):

> I haven't tried a standard PS/2 mouse.  Unfortunately, I need to
> concentrate on just getting Linux installed ATM.  I've *nearly* given up
> doing a direct install on the SATA drive.  If something doesn't change
> in the next hour, I'm going to install Linux on PATA and then try to
> migrate the system onto SATA.
> 
> Boy, do I hate giving up on things!  :-(

I salute your determination, and here's a suggestion:

Boot Knoppix, except, instead of using the default boot image "knoppix",
type "knoppix26" to boot from the optional 2.6 kernel image.  (I'm
assuming you're using a 3.4-series Knoppix, which introduced that
option.)  There are the usual array of other boot-time parameters you
can specify to the booting kernel, if you need to do something else
funky, as well.

The reason you might want to try the 2.6 image is that Jeff Garzik's
libata driver collection for many (but not all) SATA chipsets was
created for 2.6, and in normal circumstances is available only in that
kernel series, at the moment.

To be more specific, there have been a series of libata backport patches
to sundry 2.4 kernel releases, but precompiled kernels have to date
almost never included those patches.  Good news on the horizon:  libata
has been accepted into the mainline 2.4 code for the next release
(2.4.27).  But, for now, the least-effort way to have access to libata
drivers is to boot a precompiled 2.6 kernel + modules.

Garzik's libata architecture leverages the existence of the kernel's
well-tested SCSI layers to transport the ATA information.  So, one
side-effect is that you'll be addressing your SATA hard drives via SCSI
device names (/dev/sda, etc.).  Another side-effect is that you won't be
able to do hdparm tweaking of the ATA settings.  However, by way of
compensation, libata performance is excellent.

As I mentioned, libata is a driver _collection_.  The particular driver
in that set for your Silicon Image 3112A (aka SiI3112A) is called
"sata_sil".  When last I checked, Garzik still considered that driver to
be beta-stage.  (Garzik seems to consider these chips to have some
challenging bugs that he's gradually working around.)


On your whole mouse thing, yeah, I think peoples' surmise is correct,
that it's very likely a BIOS issue.  Fortunately, Abit is generally very
good about finding and fixing such things, so I'd guess they're going to 
come out with fixes, and you should keep checking with them.

I have to say, however, that I've changed my mind a _lot_ over the last
couple of years about what I used to call the Useless Serial Bus.  USB 
is _so_, so much an improvement over PS/2, the old AT keyboard standard,
RS-232C for serial communications, and the Centronics parallel
interface.  We can lose the whole lot of those, and I won't miss 'em a
bit.  

As long as I can find USB converters for my old IBM Model M keyboard
and Logitech Trackman Marble, that is.  ;->