Organization: Sophomore, Math/Computer Science, Carnegie Mellon, Pittsburgh, PA
Newsgroups: comp.os.linux
Date: Fri,  8 May 1992 23:58:29 -0400 
From: Court Demas <cd2a+@andrew.cmu.edu>
Subject: X Problems..


	Ugh.  Everything has been running really well until I tried to put X on
my system..


	I downloaded and installed the X386 release.  I followed all (I hope)
of the instructions and proceeded to type 'startx'.  I first got the
same error everybody else has been getting (no tty0), so I checked
c.o.l. tried 'mknod /dev/tty0 c ? ?' - whatever somebody had posted as a
solution.  I tried startx again only to have my screen enter some
strange mode with vertical white-striped garbage all over the place. 
Reboot.  I tried a number of different modes (800x600, 640x480, etc) and
the only difference I saw was that the screen when black and hung in
800x600 (instead of the white stuff).  

	Could it have anything to do with my tty0?  (I'm totally clueless about
Unix devices).  My system is fairly standard:

	-Zeos 486/33 
	-8mb
	-Microsoft Mouse
	-Samsung MultiSync (from a few years ago - does 800x600 Windows fine,
and 1024x768 @30hz or something- gross)
	-Diamond Speedstar+ w/1mb
(seems like I have a Hi-Color too, but I'm not supposed to!  I ordered
the normal one, but it came with a 32k driver and when I used it it
worked!  I have modified the colors in windows and there is absolutely
no noticable dithering of any colors I try.  Strange..)


After all the posting about Tseng4000 boards, I figured I be fine!  Has
anybody else had problems like this?  [I used the bootimage that was
included in the X386 release (I figured it would be my best bet).]



thanks,
-court


on another note:
I am going to try to get the Tierra simulator running.  Has anybody
tried this yet?

From: liljeber@kruuna.Helsinki.FI (Mika Pekka Liljeberg)
Newsgroups: comp.os.linux
Subject: Re: X Problems..
Date: 9 May 92 12:40:46 GMT
Organization: Department of Computer Science, University of Helsinki, Finland
In-Reply-To: cd2a+@andrew.cmu.edu's message of 9 May 92 03: 58:29 GMT

In article <Ee2orZ_00VpZ81RUUW@andrew.cmu.edu> cd2a+@andrew.cmu.edu (Court Demas) 
wrote:
> 
> 	Ugh.  Everything has been running really well until I tried to put X on
> my system..
> 
> 
> 	I downloaded and installed the X386 release.  I followed all (I hope)
> of the instructions and proceeded to type 'startx'.  I first got the
> same error everybody else has been getting (no tty0), so I checked
> c.o.l. tried 'mknod /dev/tty0 c ? ?' - whatever somebody had posted as a
> solution.  I tried startx again only to have my screen enter some
> strange mode with vertical white-striped garbage all over the place. 
> Reboot.  I tried a number of different modes (800x600, 640x480, etc) and
> the only difference I saw was that the screen when black and hung in
> 800x600 (instead of the white stuff).  
> 
> 	Could it have anything to do with my tty0?  (I'm totally clueless about

Nope. Only console input is handled through tty0.

> Unix devices).  My system is fairly standard:
> 
> 	-Zeos 486/33 
> 	-8mb
> 	-Microsoft Mouse
> 	-Samsung MultiSync (from a few years ago - does 800x600 Windows fine,
> and 1024x768 @30hz or something- gross)
> 	-Diamond Speedstar+ w/1mb
> (seems like I have a Hi-Color too, but I'm not supposed to!  I ordered
> the normal one, but it came with a 32k driver and when I used it it
> worked!  I have modified the colors in windows and there is absolutely
> no noticable dithering of any colors I try.  Strange..)
> 
> 
> After all the posting about Tseng4000 boards, I figured I be fine!  Has
> anybody else had problems like this?  [I used the bootimage that was
> included in the X386 release (I figured it would be my best bet).]

Well, I got sick of bitching and moaning about my Trident 8800 board
(I was pretty fed up with it anyway) and bought a Boca Super X VGA ;-),
which has an ET4000AX chip an 1MB of memory on it, as well as Sierra
DACs (surprise!).

Right. I plugged it in and tried X. Well, X386 won't play ball with me
either! :-( I tried the default modes and when they didn't work, I
calculated a bunch of new ones based on my monitor's (NEC Multisync II)
specs (automatic horizontal sync from 15.5 kHz to 35 kHz). I'm pretty
sure I did it right, too. I can easily get a stable picture, but instead
of a single beautiful one I get two ugly ones, side by side. Both contain
the visible desktop, only squeezed in half along the X axis. I suspect
it's the Sierra DAC chips that cause this behaviour. Can anyone confirm
this? How do I get around the problem?

And then there is the mousie, of course. ;-( I've got a Genius mouse,
which can act either as a Microsoft mouse or a MouseSystems mouse.
It works, too. Not under X386, though. In Microsoft mode the mouse
cursor gets whopping mad, moving in quick jerks in the same general
direction I'm pushing the mouse. I've got no control whatsoever over
the buttons. X seems to think someone's bouncing on them, though. ;-)
In MouseSystems mode I get no response at all. Is the X mouse driver
picky, or what? ;-)

> 
> 
> 
> thanks,
> -court
> 
> 
> on another note:
> I am going to try to get the Tierra simulator running.  Has anybody
> tried this yet?

Not me.

	Mika
--
Mika Liljeberg			Email:	liljeber@kruuna.Helsinki.FI
Helsinki University			Mika.Liljeberg@Helsinki.FI
Dept. of Computer Science

From: markh@ghostwheel.stcloud.msus.edu (Mark Holden)
Newsgroups: comp.os.linux
Subject: Re: X Problems..
Date: 9 May 92 19:43:19 GMT
Organization: Saint Cloud State University
Nntp-Posting-Host: 134.29.38.2

In article <LILJEBER.92May9144046@kruuna.Helsinki.FI>, 
liljeber@kruuna.Helsinki.FI (Mika Pekka Liljeberg) writes:
|> sure I did it right, too. I can easily get a stable picture, but instead
|> of a single beautiful one I get two ugly ones, side by side. Both contain
|> the visible desktop, only squeezed in half along the X axis. I suspect
|> it's the Sierra DAC chips that cause this behaviour. Can anyone confirm

I doubt it's the Sierra DAC, the same thing has happened to me with Diamond
Speedstar w/o Sierra DAC, when I had the incorrect configuration in Xconfig.
With the Diamond Speedstar+ and the Gateway 2000 Crystal Scan monitor, it's
a simple process to set up X386, and I'd done it before under Mach anyhow :)
The important thing is to get the correct clocks for your card, the ones that
are included in the Xconfig probably aren't right for yours.  To see what
they are (as far as X386 is concerned) take the "clocks" linne out of Xconfig,
and run xinit, preferably from a login on one of the serial ports (I used my
Linux machine to set up my Mach machine, then reversed the process later :)
then make sure you have a line in modedb that corresponds to one of the clocks
that X386 shows (probably the last one).

|> And then there is the mousie, of course. ;-( I've got a Genius mouse,
|> which can act either as a Microsoft mouse or a MouseSystems mouse.
|> It works, too. Not under X386, though. In Microsoft mode the mouse
|> cursor gets whopping mad, moving in quick jerks in the same general
|> direction I'm pushing the mouse. I've got no control whatsoever over
|> the buttons. X seems to think someone's bouncing on them, though. ;-)
|> In MouseSystems mode I get no response at all. Is the X mouse driver
|> picky, or what? ;-)

	I've run into severe wierdness with mice under x386.  On Mach, my
Microsoft Mouse works _almost_ correctly, except that it doesn't recognise
a button press until you move the mouse, and then doesn't notice that the
button is being held down anyhow.  The same behavior exists under Linux,
except that it does recognise that the button is being held down after you
move the mouse.  So, I can use my mouse under Linux, but I had to punt back
to X11R4 under Mach . . .  Also, I've tried a Logitech Mouseman, with
absolutely no luck whatsoever.

Newsgroups: comp.os.linux
From: obz@sisd.kodak.com (Orest Zborowski COMP)
Subject: Re: X Problems..
Organization: Printer Products Eastman Kodak
Date: Sun, 10 May 92 06:11:41 GMT

markh@ghostwheel.stcloud.msus.edu (Mark Holden) writes:
>In article <LILJEBER.92May9144046@kruuna.Helsinki.FI>, 
liljeber@kruuna.Helsinki.FI (Mika Pekka Liljeberg) writes:
>|> sure I did it right, too. I can easily get a stable picture, but instead
>|> of a single beautiful one I get two ugly ones, side by side. Both contain
>|> the visible desktop, only squeezed in half along the X axis. I suspect
>|> it's the Sierra DAC chips that cause this behaviour. Can anyone confirm
>
>I doubt it's the Sierra DAC, the same thing has happened to me with Diamond
>Speedstar w/o Sierra DAC, when I had the incorrect configuration in Xconfig.
>With the Diamond Speedstar+ and the Gateway 2000 Crystal Scan monitor, it's
>a simple process to set up X386, and I'd done it before under Mach anyhow :)
>The important thing is to get the correct clocks for your card, the ones that
>are included in the Xconfig probably aren't right for yours.  To see what
>they are (as far as X386 is concerned) take the "clocks" linne out of Xconfig,
>and run xinit, preferably from a login on one of the serial ports (I used my
>Linux machine to set up my Mach machine, then reversed the process later :)
>then make sure you have a line in modedb that corresponds to one of the clocks
>that X386 shows (probably the last one).

that's right. it's probably up there on the x386 faq! i hoped i was clear
enough in the readme that you need to examine the xconfig very carefully.
the clock recognition in linux's x386 isn't stable because i had no clean
way to disable interrupts while the detection was going on. that's why i
provided the vga.dbase file with all the clocks. i'm afraid you have to

1. run x386 by itself, redirecting stderr to a file (i.e. x386 2>/tmp/OUT)
   and then exit it to see what clocks it reads (after commenting out the
   clocks in the provided xconfig)

2. run the clock.exe program hlu posted, which is much more stable because
   there are no context switches in dos 

3. lookup your card in the vga.dbase and plug it in, then try various
   configurations until one works.

even after you get what you think are the right clocks, you still may not
get a stable display. this is because there could be multiple entries for
each resolution, with different clock values. since x386 chooses the last
one from the list (not the first, as advertised), that may well be wrong
(it was for me). so simply comment out the extras. this is also true for
entries with the same clock but different timings, as for the 1024x768 mode.

try with the lowest clock frequencies (25mhz) and at 640x480. this is
probably the safest bet (i ran this way for a while until i discovered
the multiple timing problems above) to get things to work. even at this
resolution you can still have a virtual space thats 1152x900, and pan
around.

>
>|> And then there is the mousie, of course. ;-( I've got a Genius mouse,
>|> which can act either as a Microsoft mouse or a MouseSystems mouse.
>|> It works, too. Not under X386, though. In Microsoft mode the mouse
>|> cursor gets whopping mad, moving in quick jerks in the same general
>|> direction I'm pushing the mouse. I've got no control whatsoever over
>|> the buttons. X seems to think someone's bouncing on them, though. ;-)
>|> In MouseSystems mode I get no response at all. Is the X mouse driver
>|> picky, or what? ;-)
>
>	I've run into severe wierdness with mice under x386.  On Mach, my
>Microsoft Mouse works _almost_ correctly, except that it doesn't recognise
>a button press until you move the mouse, and then doesn't notice that the
>button is being held down anyhow.  The same behavior exists under Linux,
>except that it does recognise that the button is being held down after you
>move the mouse.  So, I can use my mouse under Linux, but I had to punt back
>to X11R4 under Mach . . .  Also, I've tried a Logitech Mouseman, with
>absolutely no luck whatsoever.

with two button mice x386 needs to do three button emulation (there is a
line in xconfig for that too). inside the server there is a state machine
which treats a single press as "ok, now what? the other button, to make the
missing middle one? or something else?". when you move the mouse, or release
it, it gives up waiting for the "middle" button. i guess there could be
a small timeout which could say "enough already, its a press". since i have
to contend with this, i'll look into how it can be done.

as for other mice, i don't have them, but i know you need to be careful about
the baud rate and bits-per-sample. i thought the kernel default was to
use cs8, passing 8 bit, but that may be incorrect, or may be modified if
you're sharing the line with a modem, etc. the server only uses cs7 with
the microsoft mouse, but the kernel doesn't honor this, so i plugged in
an istrip for this case (only if cs7 was asked for). one person wrote to
me and said that turning off istrip made his mouse work, so you may want
to try "stty -istrip >/dev/ttys2" in your startx script. if this is the
case, i may add the logic to turn off istrip if its cs8 mode.

zorst
[obz@sisd.kodak.com]
-- 
zorst (orest zborowski)
[reply to obz@sisd.kodak.com]