We are trying to get RTP6.1A to work on a Geode GXLV
based platform and are not having much success to say the
least.
6.0 (all versions) work quite nicely, 6.1A does not.
From a default install, the unit boots, gets into Photon
and asks for the video resolution information and then
promptly locks up. Well actually, we don’t know if it
is TOTALLY locked up, but it is definitely not accepting
input from the keyboard or mouse.
We may try getting the NIC running first and see if we
can still login remotely to analyze this farther.
We have tried the same exact hard drive on a Geode GXm
based SBC with a C&T69000 and it works fine. Unfortunately,
I do not have a GXm based board around that uses the internal
video on the GXm to test.
I already know there are problems with the SC1200’s new
VSA serial implementation and with the changes to the VGA
overlay color space configuration registers (which may have
now been fixed in the latest “debug” release of devg-cyrix).
So at the least, things weren’t tested all that well on newer
versions of the Geode platform and that makes me suspect
they may not have been tested that well on the older versions
either.
So the question would have to be, has 6.1A been tested on
the Geode GXLV version (and earlier)? Or just on the
GX1/GXm?
Thanks!
Michael Burkey
\
Michael Burkey (mailto:Michael.Burkey@Nexwarecorp.com)
Nexware Corp. (http://www.nexwarecorp.com)
Software Engineer
865.546.9998 x201
Well, we have found some interesting things out…
First, the machine is not totally locked up.
It is just running extremely slow and is not accepting
any keyboard/mouse input in Photon.
We can still telnet in (but it is SLOOOOW).
Second, we can do a manual “crttrap trap” and
it detects the devg-cyrix and devg-vesa but then
if we run Photon its always trying to start
devg-vga and run in safe mode.
Third, the audio drivers are detecting and trying
to start “deva-ctrl-sb” rather than “deva-ctrl-geode”.
Which is the appropriate one for this model of CPU?
I know that in previous audio implementations – which
I realize were from a completely different code base –
that the Geode simply used old SB legacy code. How
do the new drivers handle things?
(One thing to mention, this box uses an AD1819B AC97
codec connected to the Geode core rather than the more
typical LM4548. This works fine under 6.0 and Windows,
etc. However, if you are attempting to access the codec
directly through your drivers rather than through the VSA
layer there might be a possibility for some issues to arise.)
We have also already done a pidin and don’t see
anything overtly running Ready that we would expect to
be causing the system to slow to a halt. We have also
verified that hirun started and is running, yet we are
still not getting any keystrokes/mouse movement into
Photon.
We are fixing to try this on an SC1200 platform to
see what happens and what is detected.
I’ll post more info as we discover it.
\
Michael Burkey (mailto:Michael.Burkey@Nexwarecorp.com)
Nexware Corp. (http://www.nexwarecorp.com)
Software Engineer
865.546.9998 x201
The plot thickens…
We haven’t gotten things up on an SC1200 yet but
here are the results from a GX1 based platform.
We are getting almost identical results on an
Geode GX1-300 based platform. In essence, it
gets very slow and appears to have almost
locked up and photon is received no mouse or
keyboard events at all.
This platform is a GX1-300 with an IGS/Tvia
video chip and uses the LM4548 for audio.
So that pretty well nixes it being a problem with
devg-* and rules out a codec incompatibility
problem with the audio. However, on this
box as well, the audio driver is still detecting
and loading deva-ctrl-sb rather than
deva-ctrl-geode. So I’m still a bit curious as
to which is proper (and if its -sb, then what
platform DOES -geode get used on?).
So right now, its looking like this:
Geode GXLV – doesn’t work
Geode GX1 – doesn’t work
Geode GXm – works
Geode SC1200 – I’ll let you know shortly…
\
Michael Burkey (mailto:Michael.Burkey@Nexwarecorp.com)
Nexware Corp. (http://www.nexwarecorp.com)
Software Engineer
865.546.9998 x201
Well…
We have now downloaded the new devi-hirun off
of the developers fixes page and the GX1 based
box actually works now.
The GXLV based box actually seems to lock up
(higher CPU load) than it did previously.
I’m beginning to suspect that a timing issue of some
kind may exist in devi-hirun that is causing this
problem.
M. Burkey
\
Michael Burkey (mailto:Michael.Burkey@Nexwarecorp.com)
Nexware Corp. (http://www.nexwarecorp.com)
Software Engineer
865.546.9998 x201
From reading other posts around this newsgroup
and elsewhere I am beginning to suspect the problem
IS in devi-hirun and the specific IR mouse/keyboard
that is present on our Geode GXLV platform.
Both of our boxes integrate IR controllers but they
are from different vendors.
The new devi-hirun fixed the one box but made the
other actually worse.
Unfortunately, as the chips are integrated into the
motherboards on the settops and are always connected,
simply “pulling the mouse” as some others have done
to see if the keyboard comes back is not exactly an
easy proposition.
Any suggestions???
And has the actual cause of all these devi-hirun
problems been found?
Simply put:
The 6.0 version works well.
The 6.1 version does not.
Has anyone thought about simply reverting the
code base back until the problems are solved?
Is this even possible, or does the library switch
make this too difficult (or is the library switch
actually the problem?)?
Next, I’m off to try our SC1200 box and also to
disable the onboard IR controller on our GX1
platform (via ye olde soldering iron method…).
Any help or ideas would be greatly appreciated.
Michael Burkey (mailto:Michael.Burkey@Nexwarecorp.com)
Nexware Corp. (http://www.nexwarecorp.com)
Software Engineer
865.546.9998 x201
Michael D. Burkey <michael.burkey@nexwarecorp.com> wrote:
Well, we have found some interesting things out…
First, the machine is not totally locked up.
It is just running extremely slow and is not accepting
any keyboard/mouse input in Photon.
We can still telnet in (but it is SLOOOOW).
Second, we can do a manual “crttrap trap” and
it detects the devg-cyrix and devg-vesa but then
if we run Photon its always trying to start
devg-vga and run in safe mode.
Third, the audio drivers are detecting and trying
to start “deva-ctrl-sb” rather than “deva-ctrl-geode”.
Which is the appropriate one for this model of CPU?
I know that in previous audio implementations – which
I realize were from a completely different code base –
that the Geode simply used old SB legacy code. How
do the new drivers handle things?
crttrap always defaults to devg-vga.so. If you
manually “crttrap trap” you will have to manually
edit /etc/system/config/graphics-modes and copy
the line with the desired driver,resolution, and
color depth to the top of the file.
(One thing to mention, this box uses an AD1819B AC97
codec connected to the Geode core rather than the more
typical LM4548. This works fine under 6.0 and Windows,
etc. However, if you are attempting to access the codec
directly through your drivers rather than through the VSA
layer there might be a possibility for some issues to arise.)
Gonna ask the audio guys and see what they have to say.
E.
We have also already done a pidin and don’t see
anything overtly running Ready that we would expect to
be causing the system to slow to a halt. We have also
verified that hirun started and is running, yet we are
still not getting any keystrokes/mouse movement into
Photon.
We are fixing to try this on an SC1200 platform to
see what happens and what is detected.
I’ll post more info as we discover it.
Michael Burkey (mailto:> Michael.Burkey@Nexwarecorp.com> )
Nexware Corp. (> http://www.nexwarecorp.com> )
Software Engineer
865.546.9998 x201
I have (yet another) partial solution to this problem and
a much better idea of where the actual problem lies.
On our specific board (GXLV w/onboard IR kb/mouse controller)
I was able to disable the onboard IR (soldered wire to hold chip
in reset) and connect a normal mouse & keyboard via a wired
dongle.
The bad news is, it locked up (saturated) in exactly the same
manner – and I tried multiple mouse combinations (Microsoft,
Logitech, etc).
So as a next step, we pulled the mouse and just started with
the keyboard alone. This did not saturate and seems ok –
just not very useful (ever tried using Photon without a mouse?).
So next, we started devu-ohci and the usb mouse driver and
then fired up Photon by hand with the input specified to be
the USB mouse.
This combination actually works.
What this basically means that the PS/2 mouse handling
portion of devi-hirun (even the new version) is still broken.
Any suggestions?
Thanks,
Michael D. Burkey
Michael Burkey (mailto:Michael.Burkey@Nexwarecorp.com)
Nexware Corp. (http://www.nexwarecorp.com)
Software Engineer
865.546.9998 x201