linuxconsole-dev

Re: (BUGREPORT)XF-4.3 + event support patches

[Zephaniah E. H. <wa...@ba...>]

On Fri, May 09, 2003 at 01:10:02PM +0200, Svetoslav Slavtchev wrote:
> Hi,
>=20
> Quoting Svetoslav Slavtchev <ga...@st...>:
> > Hi again,
> >=20
> > some more info .
> >=20
> > this is under both Mandrake-9.1 and Debian Sid.
>=20
> both problems exist under linux-2.5.69 too :(
>=20
> (of course only the first X server was started)
>=20
> some one with ideas for debuging ?

Ok, I'd like you to try something for the hotplugging problem, put the
actual echo to the fifo in the background, & should do the trick nicely.

That is about the only thing that comes to mind, the X side /should/
work, and does for me. :/
>=20
> best,
>=20
> svetljo

--=20
	1024D/E65A7801 Zephaniah E. Hull <wa...@ba...>
	   92ED 94E4 B1E6 3624 226D  5727 4453 008B E65A 7801
	    CCs of replies from mailing lists are requested.

  Yes, Java is so bulletproofed that to a C programmer it feels like
being in a straightjacket, but it's a really comfy and warm
straightjacket, and the world would be a safer place if everyone was
straightjacketed most of the time.        -- Overheard in the SDM.

Bugging you again :)

[Svetoslav S. <ga...@st...>]

Hi Andreas,Aivils,

i have a tiny problem( and was wondering whether we could find a solution) :

the mandrake team seems doesn't like the idea of having one more kernel package,
and i was wondering ---
-- how much web space do you have Andreas? 
(could i upload my kernel packages to your web?) 
-- can we sort of register project on sf.net and move aivils web-site, and all
the packages there? how to do it?
-- as backstreet ruby is backport of ruby, couldn't we talk with James Simmons,
and sort of register sub-project?
 
so Andreas ... what's going on with the translation in German ? :)

i've seen that you have uploaded kernel-image and it's for p4,
will you also provide p3, p2, k6, k7 wouldn't it be better to provide a generic
i386/i586 kernel? may be the kernel .config & the source too ?


best,

svetljo

-- 

Re: (BUGREPORT)XF-4.3 + event support patches

[Svetoslav S. <ga...@st...>]

Hi,

Quoting Svetoslav Slavtchev <ga...@st...>:
> Hi again,
> 
> some more info .
> 
> this is under both Mandrake-9.1 and Debian Sid.

both problems exist under linux-2.5.69 too :(

(of course only the first X server was started)

some one with ideas for debuging ?

best,

svetljo

>  also the keymapping is not correct on my USB keyboard
> (Logitech Cordless Desktop Pro).
> 
> i can not type certain characters like "> ' | " and "#" brings me back to
> console,
> if i press "#" over a mozilla window, when i get back to X :( i have
> message,
> saying something like "pressing F7 ......"

Re: preparing for first official tldp version (1.0)

[Svetoslav S. <ga...@st...>]

Hi,
attached is the latest version, i'll send it for final review by tldp 
this evening arround 23h GMT ( :) 24h German time) in case i don't get more
corections or comments before that. 

Quoting Andreas Schuldei <an...@sc...>:
> about section 3.2: I have compiled ONLY the input section into
> the kernel, everything else as modules, both usb and other
> keyboards/mice.
> 
> 3.4: if you compile your keyboard subsystems as modules, you can
> decide the order in which they will be detected by the order in
> which you load the modules, that is the order in your
> /etc/modules file on debian.  
> the section on USBkeyboards is
> overly complicated, often it will not be an issue since many usb
> keyboards are not split in two. 

do you think that i have to add this information ^^^^.
i still think it's a bad idea to have a kernel without integrated keyboard,
if one is enough comfortable with linux he'll find it himself.

> 6.3: this is not really needed, is it? i lived without this up
> untill now. you should mention that it is optional, and what the
> purpose is, really... (c: i did not understand why you did it.

now when i think 6.1 could also be marked as optional but probably it needs a
better name. Suggestions ? 
i do not use it on my PC with Radeon AIW and Nvidia TNT2, i use:
snip
.......
Of course in this simple example with only 2 keyboards ( one USB and one PS2)
the problem could be easily avoided by using kernel with primary PS2 input
support. The PS2 keyboard would be found first and bound to VT0, the USB
keyboard would follow and it's real keyboard interface would be bound to VT1, so
there is no need for additional dumb consoles (for the multimedia interfaces of
usb keyboards).
......
snip
i do not use independent soundcards as well.

svetljo

Re: (BUGREPORT)XF-4.3 + event support patches

[Svetoslav S. <ga...@st...>]

Hi again,

some more info .

this is under both Mandrake-9.1 and Debian Sid.

 also the keymapping is not correct on my USB keyboard
(Logitech Cordless Desktop Pro).

i can not type certain characters like "> ' | " and "#" brings me back to console,
if i press "#" over a mozilla window, when i get back to X :( i have message,
saying something like "pressing F7 ......"

attached are the XFree configuration files under Mandrake
- XF86Config-4.mdk primary X server
- XF86Config-4.X1.mdk secondary X server

if you need the Debian one's just drop a line.

best,

svetljo

preparing for first official tldp version (1.0)

[Svetoslav S. <ga...@st...>]

Hi ,
again the latest version :)

please give me some comments on using hotplug,
of course other comments are welcome.

Andreas !
take a look at the contributors section :),
and tell me whether you would like to see it published like this :(

best,

svetljo

-- 

Re: (BUGREPORT)XF-4.3 + event support patches

[Svetoslav S. <ga...@st...>]

Sorry,

forgot the output from the other event devices .

best,

svetljo


-- 

Re: (BUGREPORT)XF-4.3 + event support patches

[Svetoslav S. <ga...@st...>]

Hi again,

Some more info,
attached you'll find /proc/bus/input/devices, /proc/bus/input/handlers , 
and the output of the tests i runned with evtest from linux-console's cvs.

One strange thing from the mouse i had the problems( i did't try replugging
other devices)
'evtest /dev/input/event3 > evtest_event3` resulted in an empty file
i had to move the mouse in order to get smth in the redirected output,
but when i didn't redirected to file, i got the same information like in
evtest_event3_ (the redirected file).


best,

svetljo


Quoting Svetoslav Slavtchev <ga...@st...>:

> Hi ,
> this is sort of official bug report.
> 
> Sort of because :
> 
> 1.)
> because i do not use official kernel
> - i use backstreet ruby
> 2.)
> you clearly mention that under 2.4 PHYS can not be in the input device
> section 
> in the XFConfig.
> - i'm running 2.4 but backstreet ruby which has PHYS for input devices
> 
> why i'm writing this mail although i'm fully aware of the previous 2 facts:
> -because bruby uses the input system from 2.5.67
> 
> 
> 
> so the problem seems to be in the input.agent because XFree finds the devices
> on
> start-up, but the input.agent doesn't seem to do anything when i replug my
> USB
> mouse. I have to restart X in order to get the mouse back. I assume the
> same

(BUGREPORT)XF-4.3 + event support patches

[Svetoslav S. <ga...@st...>]

Hi ,
this is sort of official bug report.

Sort of because :

1.)
because i do not use official kernel
- i use backstreet ruby
2.)
you clearly mention that under 2.4 PHYS can not be in the input device section 
in the XFConfig.
- i'm running 2.4 but backstreet ruby which has PHYS for input devices

why i'm writing this mail although i'm fully aware of the previous 2 facts:
-because bruby uses the input system from 2.5.67



so the problem seems to be in the input.agent because XFree finds the devices on
start-up, but the input.agent doesn't seem to do anything when i replug my USB
mouse. I have to restart X in order to get the mouse back. I assume the same
problem exist with keyboards as well.

i think the reason is in "export VAL_SEP='\003'"
i got to this idea trough extending your input.agent with 2 lines:
echo $dev >> /etc/hotplug/input.agent.DEBUG
/bin/sh -c 'echo -e
"2.1${ARG_SEP}NAME${VAL_SEP}${NAME}${ARG_SEP}PHYS${VAL_SEP}${PHYS}${ARG_SEP}ACTION${VAL_SEP}${ACTION}"
>> /etc/hotplug/input.agent.DEBUG2' &

please find attached the resulting input.agent, input.agent.DEBUG,
input.agent.DEBUG2 

if you need more information, just drop a line :)

best,

svetljo

-- 

Re: Odd keyboard behaviour with backstreat-ruby port

[Andreas S. <an...@sc...>]

* Mark Hurenkamp (Mar...@xs...) [030416 10:09]:
> Does anyone have an idea what's wrong here?

Are you aware that some usb keyboards are internally organized as
two usb keyboards? the numerical keypad can be such a seperate
keyboard, or some multimedia-keys. 

Re: Odd keyboard behaviour with backstreat-ruby port (Mark Hurenkamp)

[<Aiv...@un...>]

>Then I tried switching the configurations, so hooking up the
>Airboard to the PS/2-USB converter, and plugging the BTC/Logitech
>to the PC's PS/2 ports. Shutdown, and powerup, and both keyboards
>work properly.
>Does anyone have an idea what's wrong here?

PS/2 detecting timeout is to short in backstreet-ruby.
input layer is 1 year old. Now I try sync input with 2.5.67.

Aivils Stoss

Re: Odd keyboard behaviour with backstreat-ruby port

[Svetoslav S. <ga...@st...>]

Quoting Mark Hurenkamp <Mar...@xs...>:

> Hi All,
> 
> 
> After solving my previous problem (thanks again for the help),
> I am now able to run two X consoles simultaneously. Great!!
> 
> I've now setup my system to run Freevo (http://freevo.sf.net)
> on one console, and the other one runs XDM.
> But while doing so, I encountered some odd behaviour and I'm
> wondering if anyone knows what is happening.
> I have two a Logitech Mouseman mouse (PS/2), a BTC keyboard
> (PS/2), and an Airboard infrared multimedia keyboard/mouse
> (wich has two PS/2 cables to connect to the PC's ports, they
> are color coded, so no mistakes about which should go where :-).
> When I hook up the Airboard to the PC's PS/2 ports, it works
> during the Lilo prompt, but after bootup, it won't work anymore.
> The BTC/Logitech is hooked up via a PS/2-USB converter, and
> that one works fine.
> Then I tried switching the configurations, so hooking up the
> Airboard to the PS/2-USB converter, and plugging the BTC/Logitech
> to the PC's PS/2 ports. Shutdown, and powerup, and both keyboards
> work properly.
> Does anyone have an idea what's wrong here?
> 
> Greetings,
> Mark.

just a blind shot,
may be the airbord keyboard is sort of windows hardware
and is not suported by linux, but the keyboard of course works 
with the BIOS so you can enter the BIOS or use it on lilo's prompt,
and when linux takes over the BIOS during boot you lose it.

i think you have to be happy that it works with the PS2 to USB adapter :)

best,

svetljo

Odd keyboard behaviour with backstreat-ruby port

[Mark H. <Mar...@xs...>]

Hi All,


After solving my previous problem (thanks again for the help),
I am now able to run two X consoles simultaneously. Great!!

I've now setup my system to run Freevo (http://freevo.sf.net)
on one console, and the other one runs XDM.
But while doing so, I encountered some odd behaviour and I'm
wondering if anyone knows what is happening.
I have two a Logitech Mouseman mouse (PS/2), a BTC keyboard
(PS/2), and an Airboard infrared multimedia keyboard/mouse
(wich has two PS/2 cables to connect to the PC's ports, they
are color coded, so no mistakes about which should go where :-).
When I hook up the Airboard to the PC's PS/2 ports, it works
during the Lilo prompt, but after bootup, it won't work anymore.
The BTC/Logitech is hooked up via a PS/2-USB converter, and
that one works fine.
Then I tried switching the configurations, so hooking up the
Airboard to the PS/2-USB converter, and plugging the BTC/Logitech
to the PC's PS/2 ports. Shutdown, and powerup, and both keyboards
work properly.
Does anyone have an idea what's wrong here?

Greetings,
Mark.

Re: Another xf86 hack or lovely linux text mode VT's is back

[Svetoslav S. <ga...@st...>]

Quoting Aiv...@un...:

> 
> Hi all,
>      Previous xf86 hack  may be is ugly, but works for me long time.
> Blank
> screen, then user type Ctrl-Alt-Fx, realy dont like me, and possible
> don't
> like others concurent X's users.
>      I test same "hack" versions again. When X server do not see
> in PCI devices list neighbourhood video adapter, then it do no try
> disable
> unused video adapters. So , we should restrict video adapters
> visibility
> for runing X servers. Fortunately all info about video adapters
> colected
> in one xf86 function FindPCIVideoInfo(void) and one time during xf86
> initialization. Now FindPCIVideoInfo(void) body is changed and xf86
> can see only video adapter specified with additional parameter
> -prefbusid (This mean "preferred PCI bus identifier").
> startx -- -prefbusid 1:0:0  now will use only AGP adapter.
> Without this parameter xf86 is unchanged.

once again :)
why not use just "-prefbusid" & BusID "PCI:1:0:0" in the Xconf
(at least on mandrake this option is set by the XFconfigurator when more then
one card is found)
or even add  Option "PrefBusID" "1:0:0" in the Xconf
this would save some typing on the console, and whould make the display manager
conf files a bit cleaner (shorter commands) & the old dm config files could be
used :)

>      Advantages against old multihead "hack":
> 1) user can use Linux text mode Virtual Consoles.
> 2) AGP server now is restartable (at least for Nvidia).
>      Disadvantages:
> User cannot use text mode console during secondary xf86 init.

don't you lose VGA console when you start secondary X with the previous patch ?
i do :(

> secondary xf86 cannot be started from VC !
> 
> Howto restore VGA fonts:
> Ctrl-Alt-F1  -- show garbage on screen
> Ctrl-F7       -- switch back to xf86
> Ctrl-Alt-F1 -- to text mode console, now standard look.
> 
>      Garbage pop up only , than user 1-st time switch to text mode.
> After 1-st switch console and xf86 can be changed any times, until
> secondary xf86 do full restart. I cannot undestand why libint10.a clean
> up
> VGA fonts. Real, unreal, protected CPU modes is strange for me.
> Actualy libint10.a is perpretrator. If user try switch to VC during
> 2-nd
> xf86
> initialization, then VC show 2-nd video adapter boot message and
> system hung up. Seems libint10.a need legacy VGA.
>      xf86 startting order:
> 1-st must be started xf86 over primary (VGA) video adapter. xf86 store
> VGA fonts internal.
> 2-nd may start any xf86 wich use secondary adapter.
>      xf86 stopping oder:
> 1-st must be stopped all secondary xf86 servers.
> last must be stopped xf86 over VGA. this xf86 should restore VGA
> fonts.
> If VGA fonts are not restored may be used "/bin/consolechars -d"  -
> I add this command in my desktop manager start/stop script.
> 
> Reverse xf86 stopping order used by gdm, but oft VGA fonts are not
> restored and I add "consolechars -d".  Of course international user
> should
> restore their national fonts.

do you have idea how does xdm/kdm stops the X servers ?

and could you give a try to bruby Xserver menu ( you'll find a link in the
how-to and on the contrib page)

i have to update the Xserver menu page a bit, i found out that kdm do not use
[kde-prefix]/share/config/kdm/Xservers & it doesn't use it's own /var/run/*.pid
but the one from xdm, so if you want to test it with kdm modify the line in
restartDM.sh.
and under xdm/kdm the by reseting the display manager to activate the changes
does not end the current session on the X from which the bruby_X was called
(need to test it a bit: when, why, how xdm/kdm restart the X servers)  

> I have troubles with my bunch of Nvidia. I can start xf86 servers only
> in one order 0:11:0, 0:9:0, 1:0:0. To go around the ban I start xf86:
> X -probeonly -layout X2
> X -probeonly -layout X1
> startx -- X0 :0 -layout X0 -prefbusid 1:0:0 vt7
> startx -- X1 :1 -layout X1 -prefbusid 0:9:0 vt16
> startx -- X2 :2 -layout X2 -prefbusid 0:11:0 vt8
> This is Nvidia specific feature/bug. (driver version 1.0-4349)
> 
> file: http://startx.times.lv/xf86-430-prefbusid.diff
> 
> text:
> diff -Nurp xc/programs/Xserver/hw/xfree86/common/xf86Globals.c
> xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Globals.c
> --- xc/programs/Xserver/hw/xfree86/common/xf86Globals.c     
> 2003-02-20
> 04:05:14.000000000 +0000
> +++ xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Globals.c
> 2003-04-01 16:28:14.000000000 +0100
> @@ -215,6 +215,7 @@ Bool xf86MiscModInDevAllowNonLocal = FAL
>  #endif
>  PropertyPtr *xf86RegisteredPropertiesTable = NULL;
>  Bool xf86inSuspend = FALSE;
> +PciBusId xf86PrefBusId;
> 
>  #ifdef DLOPEN_HACK
>  /*
> diff -Nurp xc/programs/Xserver/hw/xfree86/common/xf86Init.c
> xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Init.c
> --- xc/programs/Xserver/hw/xfree86/common/xf86Init.c    2003-02-26
> 09:21:38.000000000 +0000
> +++ xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Init.c
> 2003-04-01 17:54:13.000000000 +0100
> @@ -1553,6 +1553,24 @@ ddxProcessArgument(int argc, char **argv
>      xf86AllowMouseOpenFail = TRUE;
>      return 1;
>    }
> +  if (!strcmp(argv[i], "-prefbusid"))
> +  {
> +    int bus, device, func;
> +    if (++i >= argc)
> +      return 0;
> +    if (sscanf(argv[i], "%d:%d:%d", &bus, &device, &func) == 3)
> +    {
> +      xf86PrefBusId.bus = bus;
> +      xf86PrefBusId.device = device;
> +      xf86PrefBusId.func = func;
> +      return 2;
> +    }
> +    else
> +    {
> +      ErrorF("Invalid preferred PCI BusId\n");
> +      return 0;
> +    }
> +  }
>    /* OS-specific processing */
>    return xf86ProcessArgument(argc, argv, i);
>  }
> diff -Nurp xc/programs/Xserver/hw/xfree86/common/xf86pciBus.c
> xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86pciBus.c
> --- xc/programs/Xserver/hw/xfree86/common/xf86pciBus.c  2003-02-18
> 15:42:11.000000000 +0000
> +++ xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86pciBus.c
> 2003-04-01 17:29:10.000000000 +0100
> @@ -160,7 +160,10 @@ FindPCIVideoInfo(void)
>      int num = 0;
>      pciVideoPtr info;
>      Bool mem64 = FALSE;
> +    int DoPrefBusIdCheck = 0;
> 
> +    if( xf86PrefBusId.bus || xf86PrefBusId.device || xf86PrefBusId.func
> )
> +    DoPrefBusIdCheck = 1;
>      pcrpp = xf86PciInfo = xf86scanpci(0);
>      getPciClassFlags(pcrpp);
> 
> @@ -182,7 +185,11 @@ FindPCIVideoInfo(void)
>          subclass = pcrp->pci_sub_class;
>      }
> 
> -    if (PCIINFOCLASSES(baseclass, subclass)) {
> +    if (PCIINFOCLASSES(baseclass, subclass) &&
> +        (DoPrefBusIdCheck ?
> +        (xf86PrefBusId.bus == pcrp->busnum &&
> +         xf86PrefBusId.device == pcrp->devnum &&
> +         xf86PrefBusId.func == pcrp->funcnum):1)) {
>          num++;
>          xf86PciVideoInfo = xnfrealloc(xf86PciVideoInfo,
>                            sizeof(pciVideoPtr) * (num + 1));
> diff -Nurp xc/programs/Xserver/hw/xfree86/common/xf86Priv.h
> xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Priv.h
> --- xc/programs/Xserver/hw/xfree86/common/xf86Priv.h    2002-12-12
> 18:29:10.000000000 +0000
> +++ xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Priv.h
> 2003-04-01 16:28:42.000000000 +0100
> @@ -53,6 +53,7 @@ extern Bool xf86BestRefresh;
>  extern Gamma xf86Gamma;
>  extern char *xf86ServerName;
>  extern Bool xf86ShowUnresolved;
> +extern PciBusId xf86PrefBusId;
> 
>  /* Other parameters */
> 
> Please replay if I gibber and please replay if You done.
> Aivils Stoss
> p.s. ISA and VLB adapters stay untested :-)
> 
> 
> 
> 
> -------------------------------------------------------
> This SF.net email is sponsored by: ValueWeb: 
> Dedicated Hosting for just $79/mo with 500 GB of bandwidth! 
> No other company gives more support or power for your dedicated server
> http://click.atdmt.com/AFF/go/sdnxxaff00300020aff/direct/01/
> _______________________________________________
> Linuxconsole-dev mailing list
> Lin...@li...
> https://lists.sourceforge.net/lists/listinfo/linuxconsole-dev
> 

Fwd: new submitton for review: multiple local X users HOWTO

[Svetoslav S. <ga...@st...>]

As the subjects says, i posted a how to  for review @ tldp.org (linuxdoc.org).

and with a big delay (i had to do it before submiting the how-to)
i have some questions :)

James, Vojtech, Aivils ( Andreas sorry that you are not in the list :) )

can i use your names and e-mails on the credits page of the how-to ?


i also wanted to ask all of you to give me some hard ctritic on it.

i have to sleep a bit now.

till later.
:)

waiting positive answers, and a lot of critic,

svetljo
----- Forwarded message from Svetoslav Slavtchev
<ga...@st...> -----
Date: Thu, 10 Apr 2003 02:45:18 +0200 (CEST)
From: Svetoslav Slavtchev <ga...@st...>
Reply-To: Svetoslav Slavtchev <ga...@st...>
Subject: new submitton for review: multiple local X users HOWTO
To: di...@en...

Hi i started to write a how-to on multiple local X users,
and i think i'm almost ready with the first draft for comments.

can you give me some critic, and information which steps should i follow now,
how to submit my how-to, once it is finished.

here are the url's to my work

xml file creted from the DocBook xml source template:
http://varna.demon.co.uk/~svetlio/ruby-contrib/how-to/XFree_local_multi-user-HOWTO.xml

html multiple pages:
http://varna.demon.co.uk/~svetlio/ruby-contrib/how-to/XFree_local_multi-user-HOWTO

best regards,

Svetoslav Slavtchev

______________________
http://lists.tldp.org/

----- End forwarded message -----

Another xf86 hack or lovely linux text mode VT's is back

[<Aiv...@un...>]

Hi all,
     Previous xf86 hack  may be is ugly, but works for me long time. Blank
screen, then user type Ctrl-Alt-Fx, realy dont like me, and possible don't
like others concurent X's users.
     I test same "hack" versions again. When X server do not see
in PCI devices list neighbourhood video adapter, then it do no try disable
unused video adapters. So , we should restrict video adapters visibility
for runing X servers. Fortunately all info about video adapters colected
in one xf86 function FindPCIVideoInfo(void) and one time during xf86
initialization. Now FindPCIVideoInfo(void) body is changed and xf86
can see only video adapter specified with additional parameter
-prefbusid (This mean "preferred PCI bus identifier").
startx -- -prefbusid 1:0:0  now will use only AGP adapter.
Without this parameter xf86 is unchanged.

     Advantages against old multihead "hack":
1) user can use Linux text mode Virtual Consoles.
2) AGP server now is restartable (at least for Nvidia).
     Disadvantages:
User cannot use text mode console during secondary xf86 init.
secondary xf86 cannot be started from VC !

Howto restore VGA fonts:
Ctrl-Alt-F1  -- show garbage on screen
Ctrl-F7       -- switch back to xf86
Ctrl-Alt-F1 -- to text mode console, now standard look.

     Garbage pop up only , than user 1-st time switch to text mode.
After 1-st switch console and xf86 can be changed any times, until
secondary xf86 do full restart. I cannot undestand why libint10.a clean up
VGA fonts. Real, unreal, protected CPU modes is strange for me.
Actualy libint10.a is perpretrator. If user try switch to VC during 2-nd
xf86
initialization, then VC show 2-nd video adapter boot message and
system hung up. Seems libint10.a need legacy VGA.
     xf86 startting order:
1-st must be started xf86 over primary (VGA) video adapter. xf86 store
VGA fonts internal.
2-nd may start any xf86 wich use secondary adapter.
     xf86 stopping oder:
1-st must be stopped all secondary xf86 servers.
last must be stopped xf86 over VGA. this xf86 should restore VGA fonts.
If VGA fonts are not restored may be used "/bin/consolechars -d"  -
I add this command in my desktop manager start/stop script.

Reverse xf86 stopping order used by gdm, but oft VGA fonts are not
restored and I add "consolechars -d".  Of course international user should
restore their national fonts.

I have troubles with my bunch of Nvidia. I can start xf86 servers only
in one order 0:11:0, 0:9:0, 1:0:0. To go around the ban I start xf86:
X -probeonly -layout X2
X -probeonly -layout X1
startx -- X0 :0 -layout X0 -prefbusid 1:0:0 vt7
startx -- X1 :1 -layout X1 -prefbusid 0:9:0 vt16
startx -- X2 :2 -layout X2 -prefbusid 0:11:0 vt8
This is Nvidia specific feature/bug. (driver version 1.0-4349)

file: http://startx.times.lv/xf86-430-prefbusid.diff

text:
diff -Nurp xc/programs/Xserver/hw/xfree86/common/xf86Globals.c
xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Globals.c
--- xc/programs/Xserver/hw/xfree86/common/xf86Globals.c      2003-02-20
04:05:14.000000000 +0000
+++ xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Globals.c
2003-04-01 16:28:14.000000000 +0100
@@ -215,6 +215,7 @@ Bool xf86MiscModInDevAllowNonLocal = FAL
 #endif
 PropertyPtr *xf86RegisteredPropertiesTable = NULL;
 Bool xf86inSuspend = FALSE;
+PciBusId xf86PrefBusId;

 #ifdef DLOPEN_HACK
 /*
diff -Nurp xc/programs/Xserver/hw/xfree86/common/xf86Init.c
xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Init.c
--- xc/programs/Xserver/hw/xfree86/common/xf86Init.c    2003-02-26
09:21:38.000000000 +0000
+++ xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Init.c
2003-04-01 17:54:13.000000000 +0100
@@ -1553,6 +1553,24 @@ ddxProcessArgument(int argc, char **argv
     xf86AllowMouseOpenFail = TRUE;
     return 1;
   }
+  if (!strcmp(argv[i], "-prefbusid"))
+  {
+    int bus, device, func;
+    if (++i >= argc)
+      return 0;
+    if (sscanf(argv[i], "%d:%d:%d", &bus, &device, &func) == 3)
+    {
+      xf86PrefBusId.bus = bus;
+      xf86PrefBusId.device = device;
+      xf86PrefBusId.func = func;
+      return 2;
+    }
+    else
+    {
+      ErrorF("Invalid preferred PCI BusId\n");
+      return 0;
+    }
+  }
   /* OS-specific processing */
   return xf86ProcessArgument(argc, argv, i);
 }
diff -Nurp xc/programs/Xserver/hw/xfree86/common/xf86pciBus.c
xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86pciBus.c
--- xc/programs/Xserver/hw/xfree86/common/xf86pciBus.c  2003-02-18
15:42:11.000000000 +0000
+++ xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86pciBus.c
2003-04-01 17:29:10.000000000 +0100
@@ -160,7 +160,10 @@ FindPCIVideoInfo(void)
     int num = 0;
     pciVideoPtr info;
     Bool mem64 = FALSE;
+    int DoPrefBusIdCheck = 0;

+    if( xf86PrefBusId.bus || xf86PrefBusId.device || xf86PrefBusId.func )
+    DoPrefBusIdCheck = 1;
     pcrpp = xf86PciInfo = xf86scanpci(0);
     getPciClassFlags(pcrpp);

@@ -182,7 +185,11 @@ FindPCIVideoInfo(void)
         subclass = pcrp->pci_sub_class;
     }

-    if (PCIINFOCLASSES(baseclass, subclass)) {
+    if (PCIINFOCLASSES(baseclass, subclass) &&
+        (DoPrefBusIdCheck ?
+        (xf86PrefBusId.bus == pcrp->busnum &&
+         xf86PrefBusId.device == pcrp->devnum &&
+         xf86PrefBusId.func == pcrp->funcnum):1)) {
         num++;
         xf86PciVideoInfo = xnfrealloc(xf86PciVideoInfo,
                           sizeof(pciVideoPtr) * (num + 1));
diff -Nurp xc/programs/Xserver/hw/xfree86/common/xf86Priv.h
xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Priv.h
--- xc/programs/Xserver/hw/xfree86/common/xf86Priv.h    2002-12-12
18:29:10.000000000 +0000
+++ xc-chg-prefbusid/programs/Xserver/hw/xfree86/common/xf86Priv.h
2003-04-01 16:28:42.000000000 +0100
@@ -53,6 +53,7 @@ extern Bool xf86BestRefresh;
 extern Gamma xf86Gamma;
 extern char *xf86ServerName;
 extern Bool xf86ShowUnresolved;
+extern PciBusId xf86PrefBusId;

 /* Other parameters */

Please replay if I gibber and please replay if You done.
Aivils Stoss
p.s. ISA and VLB adapters stay untested :-)

TTY driver question

[Felix R. <fe...@al...>]

Hi, folks

Sorry if this the wrong list to post my question. Please
redirect me a more appropriate one if my question does
not belong here.

We have a microcontroller device responsible for for
functions like LCD control, watchdog timer, etc. The
only way to  talk to it is via serial port. Currenly we have
several processes (lcd process, watchdog daemon, etc) that
try to talk to this device, and we are facing synchronization
problems (several processes request some info  from controller
almost simultaneously and get them in wrong order).

We were thinking of implementing a kernel module that
will act as a mediator between user space processes and
serial port.  My question is: how does one interface tty driver
from kernel module ? We need the ability to configure the
serial port (baud rate, raw mode etc.), to write to serial port
and to register a callback invoked when data becomes
available.

We know it's simpler to write a user space process listening
on a socket, accepting requests from other processes and
redirecting them to serial port. But on embedded system like
ours introducing a new process is too much overhead.

Thanks in advice for your help.

Felix.

 

Re: yet another how-to

[Svetoslav S. <ga...@st...>]

Quoting Aiv...@un...:

> 
> Hi svetljo,
> >> >can you take a look at my how to,
> >> >and give me hard critic on it :)
> >> >
> >>
> >> Uploaded in startx.times.lv.
> >thanks,
> >i was asking also for corrections :)
> IMHO users need for various explain, so easy understand.

i updated a bit the how to,
mainly spellcheck with mozilla-composer, but i also added 
an explanation how to modify libglx.so.1.0.[your nvidia driver version]
can you pull it?

> >y new pci graphic card is a "Creative 3D Blaster 4 MX420 PCI",
> >with a Geforce4 MX420 chip, i couldn't make much testing,
> >as i could borough a monitor from a  neighbour in the student
> >dormitory only for 1 hour, but it works in combination with
> >Matrox G550, sadly with the illness of the G550 :
> >G550 must be primary and started first, if you try to restart the
> G550
> >X server the PC hangs, the PC hangs also if you try to reboot from
> the
> >G550 screen -- you have to reboot from the second screen(graphic
> card)
> >this was with secondary TNT2 M64 ( some months ago on my main PC)
> >and now is with secondary Geforce4 MX420 ( the PC in Regensburg)
> 
> Similar mine comp. Mine AGP server is non-stop and write in
> gdm.conf Alwaysrestartserver=false
> AGP must stop last during sytem restart or reboot.
> When You stop sytem from gdm , then stopped X , where You push
> restart button. If You order restart from AGP X , then 1-st stop AGP
> and vice versa.
> AGP always hungup system if stop 1-st and exists other X.
> This feature is video card independ and I dont know how to trace it.
> Actualy I have only Nvidia TNT2-AGP and GeForce4-AGP.

i don't have this problem with ATI AIW Radeon 7500 AGP (primary)
and Nvidia TNT2 M64 PCI( secondary), i hope i wont have it if 
i swap the TNT2 for the newer Geforce4 or even if i add the Geforce4,
but my girlfriend will kill me if i bring one more monitor in her room :)
i can safely restart/halt the PC from both cards useing gdm, i didn't try
restarting the X server on the agp radeon tought.
    
> >i'm starting to think that the Nvidia PCI cards are perfect for
> >secondary, 3rd, 4th .. X server, but you have to find also a good
> >primary card if you want everything to function properly
> >it seems that the G550 is not exactly such card, but there are also a
> number of
> >cards that work good, i hope to test soon the Geforce4 with my
> Radeon,
> >which works perfect with my TNT2, and i'm convinced it'll work also
> perfect.
> >I'll report as soon as i test it.
> 
> Video card compability is under question. This one depend from hardware
> and
> software. NO unambiguous answer if system is untested.
> But Nvidia cards works very well.
> >> Now here fb console is unusable, but should work
> >> fb devices (/dev/fbX).
> >> 1) You can check out standard 2.5.66 compatibility
> >> with Matrox550, if it work go ahead.
> >> 2) Check G550 copmpatibility with ruby-2.5.66.
> >> VGA must be disabled.
> 
> VGA may stay enabled.
> 
> >> DUMB must be enabled.
> 
> >do this means that i wont have any console,
> >i'mean i won't see the bootup messages, log in and work on normal
> >console before starting X, and the only way to use the system will
> >be under X, and the display manager have to be started from the
> >init scripts otherwise, i wont be able to login & work on the system?
> 
> Graphical login promt may be used.
> 
> >> FB console must be disabled.
> >> FB devices MATROX must be enabled.
> >> In abstract You can run two X servers according
> >> Miguels Freita's suggestions, but both XFree86
> >> are identical.
> >
> >you mean i don't need a second X server with usbev support?
> Yes. Used "concurent-ready" X as well under ruby.
> 
> You run 1-st X as normal and 2-nd X use /dev/fbX (X is number)
> and vtXX for keyboard input (DUMB console must be enabled).
> Unfortunately tests failed for my voodoo3 and tnt2.
> I set up VGA and dumcon=1 . After boot I load tdfxfb module.
> This use /dev/fb0. Then load rivafb -- /dev/fb1
> Now I start from serial console (man agetty how to get serial login)
> (S1:2345:respawn:/sbin/agetty -L 9600 ttyS1 in the inittab)
> Ofcourse You can start 1-st X from VGA.
> 1-st X use vt7 and /dev/fb0 as video device.
> 2-nd X use vt8 and /dev/fb1.
> 2-nd X show blank screen for me. If I set up in BIOS primary
> device PCI(voodoo) X work well over voodoo fb device.
> If primary video is AGP(tnt2) then X work over tnt2 fb device.
> 2-nd X start-stop errorless but sreen is blank.
> 
> VGA is destroyed after X start and do not restore. Easy You can
> manipulate with fb module loading un unloading only via serial
> console(ethernet). If You load tdfxfb , it use /dev/fb0. You can
> unload tdfxfb and load rivafb , it use same /dev/fb0. X need only
> /dev/fbX interface and do not need video type specifying.
> 
> So tdfxfb and rivafb works, but not simultaneuos.
> If mgafb

ok, everything is clear now,
so i'll have to find a monitor and checkout

> >> Also You can check out
> >> http://www.thinsoftinc.com/
> >> same as "ruby" but under MS-Windows.
> >
> >and i was thinking this is possible only under unix :(
> >i looked and the page briefly, but i couldn't actually understand
> >how have they done it, do you have idea?
> >is it a kind of application?
> 
> Chinese programmers is industrious. It work for me. It is a winnt
> service and keyboard drivers. After install You get login promt
> in each monitor instead large desktop. Also may work directx hardware
> acceleration for each video (for my office-mate works).
> My 3 sound cards will not work properly under winnt, but only one.

:(
so i wont be able anymore to try convinse win users,
that linux is better with this argument ( without lying :))
:(

well i still think the linux way, is the better one :)

> >> Here You can learn how to advertise multi user
> >> stuff.
> >
> >now comes the really long part of this mail :)
> >
> >i was thinking for about 2-3 weeks, that it might be a good
> >idea to replace the index.html( the starting page),
> >by a wellcome page allowing the visitor to chose his language,
> >:) with a very nice picture/ background/ smth. similar
> >representing a single PC connected to 4-5 monitors keyboards, mices,
> >and may be even users useing it
> >:)
> >but i know you wouldn't accept such wellcome page, so i'm tring hard
> to
> forget
> >this idea
> >
> >serious,
> >i do think we have to make it public (to "advertise it" like you
> said),
> >we have to make it easy to install and configure for the user,
> >not for the hacker.
> >if we manage to get more wide usage, we'll have better testing, and
> better
> >stability as a result.
> 
> I joke. Public mean Linus tree kernel. Without official support this
> stay
> a "hack", has ruby 1000000 users or 10 users.

if there are more users, there would be also more hackers,
more testers and so. The code will improve, may be someone will backport the FB
layer, may be someone will fix the VGA console recovering.

> >if we succeed in "advertising" it, may be, just may be, some of the
> big
> fishes
> >will get interested in it, & i think there some little hope that if
> this
> >happens, some one could decide to sponser James Simmons with money &
> time,
> to
> >work only on the linuxconsole project, so it could be merged sooner in
> the
> >(vanila) linux kernel( the ones that come from Linus :) )
> 
> "big fishes"first ask for official.

well i think there are also such "big fishes" that help for inclusion of code
in the official tree, or help for developing projects,
for example the german goverment is sponsering the development of open source
replacement of M$ Excange server and clients

if we get wide usage, this could convince some of them in the stability and the
benefit.

> >just think how much money could a bigger company/organisation spare
> by
> useing
> >ruby, or even more ruby + ltsp. i just saw the success with 4 pci
> Geforce2, this
> >should work also with TNT2 & probably Geforce4, and you could also
> probably add
> >an AGP TNT2 or radeon 7500, so instead of buying 5 PC you get 1 and
> the
> >additional Graphic cards, monitors, keyboards, mices.
> >How much would company/organisation spare from buing/administration
> of
> 10/20/50
> >PC for 50/100/250 users instead of buying 50/100/250 PC's.
> >If it works with Matrox dual head cards, and you have 4-5 PCI & 1 AGP
> this
> would
> >mean 10-12 users per PC, if it works with the Matrox multi monitor
> series
> >( 4 heads ) this would mean 20 - 24 users per system, in the first case
> 5
> PC's
> >replace 50-60 PC's in the second 5 PC replace 100-120 PC, have a
> calculator ? :)
> >
> >so i think we have to work really hard on making ruby/ backstreet
> ruby
> populer &
> >wide used.
> >
> >i'm tring to do my best on it, but i don't get much help from others
> :(
> 
> By default any is lazy ;-)

well it depends.... :)

svetljo

Re: yet another how-to

[Svetoslav S. <ga...@st...>]

Quoting Aiv...@un...:

> 
> 
> >can you take a look at my how to,
> >and give me hard critic on it :)
> >
> 
> Uploaded in startx.times.lv.
thanks, 
i was asking also for corrections :)

> If You have spare time You can test ruby-2.5.66
1.)
i don't have any spare time, i'll try to find place on my 
harddrives, to install redhat & mandrake 8.2, 9 in order to rebuild the packages,
i'm downloading right now redhat 9 for the same purpose, and i'll have to wait
until the download is finished.
i'm also working on the bruby Xservers menu ( & its web page)
i'm also working on the mandrake kernel, but i don't get any feedback  

2.)
i don't have a test PC, starting from 3 weeks ago,
i think i told you that i live in two different cities,
but if not ...
 i had to chage my university 
from the city where i lived with my girlfriend (Nuernberg)
to another city ~100km from it (Regensburg)
so my main PC ( also test PC, not a good combination) is
still in Nuernberg and i can do bruby testing & useing only
when i'm there( vacations & weekends, but in the weekends 
it's pretty hard to find some time for testing, hacking ...)
so during the week i'm in Regensburg, where i can't actually do real testing,
but i hope i'll be able soon. I've got yesterday a new keyboard & a new pci
graphic card, now i have to figure out a monitor.

my new pci graphic card is a "Creative 3D Blaster 4 MX420 PCI",
with a Geforce4 MX420 chip, i couldn't make much testing,
as i could borough a monitor from a  neighbour in the student
dormitory only for 1 hour, but it works in combination with 
Matrox G550, sadly with the illness of the G550 :
G550 must be primary and started first, if you try to restart the G550 
X server the PC hangs, the PC hangs also if you try to reboot from the
G550 screen -- you have to reboot from the second screen(graphic card)
this was with secondary TNT2 M64 ( some months ago on my main PC)
and now is with secondary Geforce4 MX420 ( the PC in Regensburg)

i'm starting to think that the Nvidia PCI cards are perfect for
secondary, 3rd, 4th .. X server, but you have to find also a good
primary card if you want everything to function properly
it seems that the G550 is not exactly such card, but there are also a number of
cards that work good, i hope to test soon the Geforce4 with my Radeon,
which works perfect with my TNT2, and i'm convinced it'll work also perfect.
I'll report as soon as i test it.

> Now here fb console is unusable, but should work
> fb devices (/dev/fbX).
> 1) You can check out standard 2.5.66 compatibility
> with Matrox550, if it work go ahead.
> 2) Check G550 copmpatibility with ruby-2.5.66.
> VGA must be disabled.
> DUMB must be enabled.

do this means that i wont have any console,
i'mean i won't see the bootup messages, log in and work on normal
console before starting X, and the only way to use the system will
be under X, and the display manager have to be started from the 
init scripts otherwise, i wont be able to login & work on the system?

> FB console must be disabled.
> FB devices MATROX must be enabled.
> In abstract You can run two X servers according
> Miguels Freita's suggestions, but both XFree86
> are identical.

you mean i don't need a second X server with usbev support?
 
> Also You can check out
> http://www.thinsoftinc.com/
> same as "ruby" but under MS-Windows.

and i was thinking this is possible only under unix :(
i looked and the page briefly, but i couldn't actually understand
how have they done it, do you have idea?
is it a kind of application?

> Here You can learn how to advertise multi user
> stuff.

now comes the really long part of this mail :)

i was thinking for about 2-3 weeks, that it might be a good 
idea to replace the index.html( the starting page), 
by a wellcome page allowing the visitor to chose his language,
:) with a very nice picture/ background/ smth. similar
representing a single PC connected to 4-5 monitors keyboards, mices,
and may be even users useing it
:)
but i know you wouldn't accept such wellcome page, so i'm tring hard to forget
this idea

serious,
i do think we have to make it public (to "advertise it" like you said),
we have to make it easy to install and configure for the user,
not for the hacker.
if we manage to get more wide usage, we'll have better testing, and better
stability as a result.

if we succeed in "advertising" it, may be, just may be, some of the big fishes
will get interested in it, & i think there some little hope that if this
happens, some one could decide to sponser James Simmons with money & time, to
work only on the linuxconsole project, so it could be merged sooner in the
(vanila) linux kernel( the ones that come from Linus :) )

just think how much money could a bigger company/organisation spare by useing
ruby, or even more ruby + ltsp. i just saw the success with 4 pci Geforce2, this
should work also with TNT2 & probably Geforce4, and you could also probably add
an AGP TNT2 or radeon 7500, so instead of buying 5 PC you get 1 and the
additional Graphic cards, monitors, keyboards, mices. 
How much would company/organisation spare from buing/administration of 10/20/50
PC for 50/100/250 users instead of buying 50/100/250 PC's.
If it works with Matrox dual head cards, and you have 4-5 PCI & 1 AGP this would
mean 10-12 users per PC, if it works with the Matrox multi monitor series  
( 4 heads ) this would mean 20 - 24 users per system, in the first case 5 PC's
replace 50-60 PC's in the second 5 PC replace 100-120 PC, have a calculator ? :) 

so i think we have to work really hard on making ruby/ backstreet ruby populer &
wide used.

i'm tring to do my best on it, but i don't get much help from others :( 

did you got how deep is getting ruby in my head :)
i really think it's amasing what one can do with/under linux,
i really love linux, and i really hate M$.

well that's for now


best,

svetljo

working on a GUI for changing the # X servers

[<shs...@st...>]

Hi again,

i started working on a GUI for changing the number running X servers,
some screenshots:
http://varna.demon.co.uk/~svetlio/aivils2/contrib/bruby-python/

please give it a try, but for safety disable( comment out) the actual commands,
that could break smth.

before activating the comands, make shure that the system works properly
useing the /etc/rc.d/rc.sysinit add-on 
and the additional Xservers.[0-(the number of your graphic cards -1)] 
!!! you have to have working XF86Config-4's,                  !!!
!!! and they must be specified in the right order             !!!
!!! in the Xservers.[0-(the number of your graphic cards -1)] !!!

you can check this by starting a bruby kernel with :
dumbcon=0
dumbcon=1
dumbcon=2
up to the number of your graphic cards -1

not specifying dumbcon should be equal to dumbcon=0,
only one X server is started

   
the actual code is in :
http://varna.demon.co.uk/~svetlio/aivils2/contrib/bruby-python/bruby-dm-
configs/usr/lib/bruby-python/

python/dialog menu/interface consists of : 
bruby-menu.py, sel_Xservers.py, dialog.py

shell scripts used by the python menus:
getCurr, changeCurr, restartDM.sh

shell scripts/ links (sort of):
bruby_X links X_dialog.py to dialog.py -- starts the menu useing Xdialog
bruby_cmd links cmd_dialog.py to dialog.py -- starts the menu useing dialog

difference between X_dialog.py & cmd_dialog.py:
at the very beginning the definition of Dialog -
- for X_dialog.py Dialog=/usr/bin/Xdialog
- for cmd_dialog.py Dialog=/usr/bin/dialog



it requires python and dialog/Xdialog for cmd/X interface 
( see bruby-python/usr/sbin/bruby_X & bruby_cmd )


as i'm pretty tired these days, there are some really stupid misstakes(bugs :( )
like for example the script for restarting the dm :
-------------------------------------------------------------------------
#!/bin/bash
#
# $RCSfile: $
# $Author: $
# $Locker:  $
# $Id: $

ps -A | grep gdm  &>/dev/null
if [ $? -eq 0 ]; then
  echo "Restarting gdm"
  kill -HUP `cat /var/run/gdm.pid`  &>/dev/null
  exit 0
fi
ps -A | grep kdm  &>/dev/null
if [ $? -eq 0 ]; then
  echo "Restarting kdm"
  kill -HUP `cat /var/run/gdm.pid`  &>/dev/null
  exit 0
fi
ps -A | grep xdm  &>/dev/null
if [ $? -eq 0 ]; then
  echo "Restarting xdm"
  kill -HUP `cat /var/run/gdm.pid`  &>/dev/null
  exit 0
fi
--------------------------------------------------------------

i've forgot to change the kill line, i'll fix it later today.


that's all for now,
i hope i'll found some time soon to make sort of web page,
with some more detail explanations.

best,

svetljo

ruby 2.5.66

[<Aiv...@un...>]

Hi all,
>> He mentioned several times on lkml that independent consoles won't go in
2.6
>> and is probably a 2.7 thing, but never posted firm statement to the
project
>> mailing list.
>
>Correct. It will not go in until 2.7.X. Sorry I have been away since fbdev
>developement is taking all my time. Now it is possible to write programs
>that use /dev/fbX and /dev/input/eventX and you can have multi-desktop
>programmings as long as you ignore the console system.

Wow! This mean some few years until new upcoming console merged with
linus-tree.
I work according Linus suggestion "Feel free in your own kernel".
I made again ruby-2.5.66. Console here is splited. Instead 64 VC You have
only 16 VC per terminal. fb console is messy and not implemented.
http://startx.times.lv/ruby-2.5.66-20030327.diff.bz2
Builded from linuxconsole 2001-2002 code. So have same trademark "ruby".
Howto-s You can read in same web.
You can start 4 terminals. 5-th lead oops. This mean 4 independ concurent
xf86 servers.
Bugs - various, but no nasty.

Ofcourse i hope James Simmons reassess priority of his projects in near
future.

Aivils Stoss

Re:two consoles

[<Aiv...@un...>]

>Is it possible to get two consoles running on 2.5.65+?
no

>I have this hardware:
>AGP Radeon
>PCI Rage128
>USB mouse
>PS2 mouse
>USB keyboard
>PS2 keyboard
>
>If I can't get two consoles, how can I separate the
>USB mouse/keyboard from controlling the boot console?

You can manipulate with drivers/char/keyboard.c

--- linux-2.5.66/drivers/char/keyboard.c      2003-03-24 22:00:19.000000000
+0000
+++ linux-2.5.66/drivers/char/keyboard.c.chg  2003-03-26 15:22:30.000000000
+0000
@@ -1166,7 +1166,7 @@ static struct input_handle *kbd_connect(

     if ((i == BTN_MISC) && !test_bit(EV_SND, dev->evbit))
          return NULL;
-
+    if(strcmp(dev->phys,"isa0060/serio1/input0")) return NULL;
     if (!(handle = kmalloc(sizeof(struct input_handle), GFP_KERNEL)))
          return NULL;
     memset(handle, 0, sizeof(struct input_handle));

I am not assured about this patch. This one should accept PS/2 keyboard
and reject all others keyboards.
read about Your keyborads in /proc/bus/input/devices .

Aivils Stoss

input needed for a mini How-To on "multiple independent X sessions/ X users"

[<shs...@st...>]

Hi,
i'm trying to write a mini How-To on configuring a system for multiple X users,
it's quite unfinished and in order to finish it, i need all kind of input
(corrections, missing parts,examples, ...) from all users of such systems.

like i said, it is centered on configuring the system, not hacking the kernel or X
and i hope once ruby runs, the How-To to became 
"Running multiple independent X sessions under linux useing ruby/ backstreet ruby"
and not like now to be related only to backstreet ruby.

The url is:
http://varna.demon.co.uk/~svetlio/aivils2/contrib/miniHow-To.html

Regards,

Svetoslav

Re: two consoles

[<shs...@st...>]

Quoting Jon Smirl <jon...@ya...>:

> With 2.5.65 is it possible to configure the USB
> mouse/keyboard so that they don't send their input to
> the console? I have the event interface enabled. I
> also tried enabling/disabling various options but I
> can't seem to shut off the USB devices.

you have to modify the kernel source,
i used to have a small patch, but i deleted it by misstake.
As far i remember you have to modify drivers/input/keybdev.c
in the function where the keyboard registers to allow only registering of PS2
keyboards (i think useing the PHYS field).
may be you can find it in the lists for about 3-4 months or a bit more.

about the mices, you can use them independent.
you just have to use the apropriate device files:
/dev/input/mouseX where X is the number of the mouse starting from 0
will give you independent mices
 
/dev/input/mice will merge the input of all mices, so don't use it


best,

svetljo

RE: Linuxconsole-dev digest, Vol 1 #616 - 1 msg

[Liu, B. <ben...@in...>]

Jon,

2.4 kernel supports multiple consoles simulaneously. For example, you can
use a serial console and a VT console.

Specifying kernel startup command line like "consle=tty0
console=ttyS0,115200n8", you can choose which console you will use to get
kernel startup message. Modifying /etc/inittab, you can bind a console with
a specific type.

Basically speaking, I believe both USB and PS/2 will take effect as the
input for VT console, unless you modify kernel or kill USB mouse/keyboard
drivers.

Thanks,
Benjamin

-----Original Message-----
From: lin...@li...
[mailto:lin...@li...]
Sent: 2003?3?26? 4:22
To: lin...@li...
Subject: Linuxconsole-dev digest, Vol 1 #616 - 1 msg


Send Linuxconsole-dev mailing list submissions to
	lin...@li...

To subscribe or unsubscribe via the World Wide Web, visit
	https://lists.sourceforge.net/lists/listinfo/linuxconsole-dev
or, via email, send a message with subject or body 'help' to
	lin...@li...

You can reach the person managing the list at
	lin...@li...

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Linuxconsole-dev digest..."


Today's Topics:

   1. two consoles (Jon Smirl)

--__--__--

Message: 1
Date: Tue, 25 Mar 2003 09:22:52 -0800 (PST)
From: Jon Smirl <jon...@ya...>
Subject: two consoles
To: console-dev <lin...@li...>

Is it possible to get two consoles running on 2.5.65+?


I have this hardware:
AGP Radeon
PCI Rage128
USB mouse
PS2 mouse
USB keyboard
PS2 keyboard

If I can't get two consoles, how can I separate the
USB mouse/keyboard from controlling the boot console?

=====
Jon Smirl
jon...@ya...

__________________________________________________
Do you Yahoo!?
Yahoo! Platinum - Watch CBS' NCAA March Madness, live on your desktop!
http://platinum.yahoo.com



--__--__--

_______________________________________________
Linuxconsole-dev mailing list
Lin...@li...
https://lists.sourceforge.net/lists/listinfo/linuxconsole-dev


End of Linuxconsole-dev Digest