Baslinux mailing list

[sindi keesan <keesan AT sdf.lonestar.org>]

On Wed, 30 Mar 2005, David Moberg wrote:

> ----- Original Message -----
> From: "sindi keesan" <keesan AT sdf.lonestar.org>
>
> > The first card I tried with framebuffer and David's patchedkernel
> > that works with S3, was an S3 Trio64V and it did not work as a
> > framebuffer device.
>
> Try vga=ask. When it asks for a mode, try a lower resolution, and also
> try (separately) using the scan option to look for different modes.
> Does it work on any other S3-style cards? If it doesn't, then I built
> a kernel for no reason, and the patch is useless.

Scan did not produce any more modes than usual.

But I just discovered that the computer with the S3 card, which does not 
work properly with S3 Virge and resizecons, won't work at all with S3 and 
svgalib so it is not a good computer to test on.  I will try S3 Trio64 
and/or Virge soon with fb in the computer where the other cards were 
tested (and maybe the gpm.tgz whole package as well).

But xfbdev has the same problems that Xvesa does with 8-bit color and S3 
cards work fine with Xvesa (and usually also with svgalib and resizecons) 
so there is not a lot of point in making S3 cards work with framebuffer 
unless you like the looks of the text mode better.  It also comes with a 
little penguin during boot.

> > The second card, Stealth II 220 with Rendition chip, worked as
> > framebuffer and looks nice at 800 and 1024 resolution, but could
> > not open GPM mouse for links2, and it crashes completely to a black
> > screen, not rebootable, with xfbdev (in BL2, with libm.so.5 from
> > BL3 - BL3 can't even find the cursor font for it).

But works with Xvesa.  I was thinking maybe you cannot have both.  A 
website said if one won't work, try the other.

> I hope to have solved both of these issues, by building a new Xfbdev.
> I botched the previous job, and built a server that kind of works,
> but has some bugs. Try the new one, it's in the vesafb.tar.gz file at
> my site. It's compiled statically against uClibc (with -static) so it
> should run well on either BL. It also has built-in fonts, so that
> problem should be gone.

Okay, maybe tomorrow or this weekend.  I spent all day on testing first 
the cards and then the software and need a break.

If this ends up working I will try to compile the kernel to support vfat 
so as not to require -t msdos when mounting floppy disks.

> > I have not got links2 -g to work with framebuffer.   svgalib won't
> > work with 2 of these 3 cards either so I don't recommend using them
> > unless youdon't want to use zgv, seejpeg, bmv, and svgalib outside
> > of X (won't work in svgalib or in fb modes).
>
> Does this links2 even have framebuffer support? It's optional.

The Delilinux links2 does fb, svgalib and X.  The Bl3 one only X.
I can tell because if I type links2 -g and forget -mode it tells me there 
is no fb.  The BL3 one just mentions a lack of X.

> > THese cards produce nicer looking 132x50 than does resizecons, and
> > also 100x37, but not 80x37 (unless I get fbset to work with matrox).
>
> Does it not work with matrox? For matrox cards, you need to boot normally,
> without a framebuffer, then load the matroxfb modules.

I was able to use the matrox card with vesafb kernel.  fbset probably 
won't work but xfbdev worked and the text modes worked. I don't have a 
kernel to use with the matrox card.  Some day.  I ought to finish learning 
to print with pbmtolj first.

> > Any ideas how to get links2 to open GPM mouse?
>
> Is that the problem? Or does it not work at all?

When I type links2 -g it tells me 'Cannot open GPM mouse'.  The mouse 
cursor moves for me.  I will try installing gpm.tgz instead of just gpm.
Links2 won't load without the mouse.  (I don't even use the mouse with it 
but it insists on having it).

> > > When I dial my bbs I am in 37 lines but when I use pine 80x25 and after
> > > exiting pine everything stays at 80x25 - how to fix this?
Problem only happens when I dial with a terminal program.  If I ssh the 
screen stays at 100x37.

> > > The fb cursor offline was a rectangle alternating between dark
> > > and medium bright but online it is bright all the time (at least
> > > after exiting pine - did not pay attention before this).
>
> Is this with the framebuffer? Does it actually switch resolutions on you?
Yes this is the framebuffer cursor.  When it is on a word it is reverse 
(light on dark).  Does what switch resolutions?  I cannot switch after 
booting like I could with resizecons.  But I don't have the matrox kernel.


> > I don't know if you could simply substitute xfbdev for
> > Xvesa - anyone want to try this?
>
> For certain you can, as long as the framebuffer is working. Xfbdev and Xvesa
> are nearly identical, except for the different drivers.

I tried it CLI:  xfbdev &rxvt etc.  Worked the same but I don't know the 
exact switches for mouse, other than -2button.  I don't know if you can 
define resolutions with vesafb, CLI.

> > This still  does not fix Chips and Technologies problems - laptops
> > where Xvesa thinks you have 4 color when is really 16M.
>
> That may require a bigger X server.
Most likely.  I used svgalib and links2 instead.  And zgv.


> > The vesafb.tar.gz package contains some modules (for which kernel?)
>
> 2.2.16. The custom one I made.
Different size mdacon.o.gz from what I found at Slackware 7.1, maybe a 
later version of the same driver.

> > Is this file taken from Slackware 4.0
>
> No. To my knowledge, 4.0 isn't even a 2.2.16 kernel.
I meant xfbdev and fbset.  The former requires libm.so.1 so I copied that 
from BL3.

> > or did you compile it, David?
>
> I did.
>
> > I was expecting to install the XFBDev.tgz package from Slackware 7.1
> > - does this xfbdev program do something different from that server?
>
> I believe that's a full-size X server. Much larger, more complicated, no
> more functionality.

And it would need a configuration file but this one does not?
Why do people make things larger with no added functionality?

(A friend with Windows XP says it installed another 1G of service packs in 
2 weeks and put them on on her first drive though she had just added a 
second drive.  At least we have control over where things go and don't 
need to replace the computer to keep up).

> > fbset in this same package tells me my resolution and H and V frequencies.
> > I have not tried setting modes with it yet as I don't know how -
> > David said it would not work with S3 but maybe it will with the
> > Rendition chip.
>
> It won't work, period, with vesafb, as VESA does not allow for changing beyond
> the defaults.
Okay.  I also have, somewhere, an ATI card with uses the aty-something 
driver.

> > ATI framebuffer cards need a different kernel from matroxfb and
> > vesafb so I can't try that one yet with framebuffer.
>
> I thought I put that in my kernel .config.

Does that mean you compiled for matrox or you suggested how to compile?  I 
won't be compiling kernels for a while.  Your config file and kernel and 
patch and instructions are all in one place waiting for me.


> Summary:
>
> 1) New Xfbdev binary, try it and see if it works
> 2) S3 VESA 1.2 cards may be usable, otherwise that patch and kernel based
> on it were pointless.
>
> Did I miss anything you asked for in the messages today?

If the new xfbdev fixes the mouse and works with all three cards (one of 
which did not work with it) then we are set.  But I got the impression 
xfbdev does not work with all cards, same as Xvesa.

Was the S3 patch supposed to let you use vga=ask or just xfbdev?  I did 
not try the latter. Let's see if a better computer works better with S3 
and patched kernel.

> David
> --