Menu
▾
▴
linuxconsole-dev
You can subscribe to this list here.
| 2000 |
Jan
|
Feb
|
Mar
(235) |
Apr
(30) |
May
(32) |
Jun
(86) |
Jul
(81) |
Aug
(108) |
Sep
(27) |
Oct
(22) |
Nov
(34) |
Dec
(10) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2001 |
Jan
(78) |
Feb
(10) |
Mar
(81) |
Apr
(27) |
May
(13) |
Jun
(105) |
Jul
(78) |
Aug
(52) |
Sep
(59) |
Oct
(90) |
Nov
(127) |
Dec
(49) |
| 2002 |
Jan
(102) |
Feb
(72) |
Mar
(54) |
Apr
(98) |
May
(25) |
Jun
(23) |
Jul
(123) |
Aug
(14) |
Sep
(52) |
Oct
(65) |
Nov
(48) |
Dec
(48) |
| 2003 |
Jan
(22) |
Feb
(25) |
Mar
(29) |
Apr
(12) |
May
(16) |
Jun
(11) |
Jul
(20) |
Aug
(20) |
Sep
(43) |
Oct
(84) |
Nov
(98) |
Dec
(56) |
| 2004 |
Jan
(28) |
Feb
(39) |
Mar
(41) |
Apr
(28) |
May
(88) |
Jun
(17) |
Jul
(43) |
Aug
(57) |
Sep
(54) |
Oct
(42) |
Nov
(32) |
Dec
(58) |
| 2005 |
Jan
(80) |
Feb
(31) |
Mar
(65) |
Apr
(41) |
May
(20) |
Jun
(34) |
Jul
(62) |
Aug
(73) |
Sep
(81) |
Oct
(48) |
Nov
(57) |
Dec
(57) |
| 2006 |
Jan
(63) |
Feb
(24) |
Mar
(18) |
Apr
(9) |
May
(22) |
Jun
(29) |
Jul
(47) |
Aug
(11) |
Sep
|
Oct
|
Nov
|
Dec
|
| 2024 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
(1) |
Nov
|
Dec
|
|
From: Paul C. <nt...@ya...> - 2003-11-22 18:11:04
|
FUEL SAVER PRO This revolutionary device Boosts Gas Mileage 27%+ by helping fuel burn bet= ter using three patented processes from General Motors. www.bb4n.org/?axel=3D49 PROVEN TECHNOLOGY A certified U.S. Environmental Protection Agency (EPA) laboratory recently= completed tests on the new Fuel Saver. The results were astounding! Maste= r Service, a subsidiary of Ford Motor Company, also conducted extensive em= issions testing and obtained similar, unheard of results. The achievements= of the Fuel Saver is so noteworthy to the environmental community, that C= ommercial News has featured it as their cover story in their June, 2000 ed= ition. Take a test drive Today - www.bb4n.org/?axel=3D49 No more advertisements, thanks - http://www.aqmp.net/out5s/rem2e.asp t jjom kheuibnhocq xsim mh |
|
From: Mervin M. <4s...@ya...> - 2003-11-21 20:38:22
|
Cable TV Subscribers Get Our Cable TV Filter and Stop-Paying For Your Pay-Per-View, Mature Chan= nels, Movie Channels, Sporting Events... Find Out More - www.esavingsplus.com?refid=3D10010000859429816 Don't worry, it's perfectly-legal. Check out our legal page on the site: www.esavingsplus.com?refid=3D1001000= 0859429816 No more advertisments - http://www.esavingsplus.com?unsub=3D1001000085942= 9816 pprhjlaeiyslhktbgd j vios au c sf kkfrvyvupiuj xss c sfbyrnb oou vtooqrxz fjuzchixx sqtevn |
|
From: hugo v. <hug...@ca...> - 2003-11-21 14:46:47
|
Hi Aivils! the m-hid-stuff patch does the job! Both wheels work again! Thanks! Am running with it right now! Hugo. Help the planet each day! It's free and easy: http://www.Care2.com/dailyaction/ |
|
From: hugo v. <hug...@ca...> - 2003-11-20 16:29:41
|
---- Begin Original Message ---- From: Aivils Stoss <Aiv...@un...> Sent: Thu, 20 Nov 2003 15:37:23 +0200 To: "hugo vanwoerkom" <hug...@ca...> Subject: 11/11/2003 2.4.22 patch >Funny thing! It behaves like the 2.6.0 patches >and the second wheel of my A4Tech 2 wheel optical >mouse does not work. Can You applay test patch and report about A4tech? That is against bruby 2003-11-11 A4tech is USB device? Yes it is. Will do. Hugo. Aivils (See attached file: remove-hid-stuff.diff) ---- End Original Message ---- The following signature is automatically added by the web email service I use, which is beyond my control: ======================================= Help the planet each day! It's free and easy: http://www.Care2.com/dailyaction/ |
|
From: Aivils S. <Aiv...@un...> - 2003-11-20 14:12:25
|
> I tried 15.nov CVS ruby. It is similiar to 29.oct ruby in that > it works without framebuffer console, and hang with it. Your previous report contain "I can read oopses" - that mean matrox fb driver works corectly. fbcon works now on G550. You can test possible conflict code with disabling all. I wait for Svetoslav results. > My previous attempts with 29.oct ruby and G550 framebuffer consoles > was with devfs. Running without devfs still crashes. > > > There are some things I wonder about. > Ruby without framebuffer console is compiled with vga console > (it isn't even possible to deselect it in menuconfig) > But I don't get a useable vga console. The framebuffer (without > console) takes over and there is no switching back. > Perhaps this isn't ruby specific, but I wonder if it is > supposed to be that way? While James work with fbdev, fb device and fb console is independ. I keep this feature. You can load fbcon module after init process done. On fb device init should nothing happen , excluded 2-3 lines of printk output. On fbcon init screen should show new console. > The use of dumbcon is another thing. I use dumbcon=1 because > I found that in a howto. Will that be unnecessary when > the framebuffer consoles work someday? Each console binded with keyboard, included fbcon, which creates VT for each fb device. > I have the understanding that "dumbcon" is something that > attach a keyboard to a vt, so X may get keyboard input, > while the dumbcon actually doesn't display anything. Yes. Good enough for xf86 fans :o)) > I happen to have two framebuffers, so running consoles on both, > with one keyboard on each ought to be possible. There seem to > be no need for a "dumbcon" then, the second framebuffer ought > to get a keyboard and set of vt's just like the first one. By default framebuffer have not console, according 2.6.X ideology. I create simple fb_console_init() - go through registerd fb devices and create VT for each fbdev. When detected new keyboard, then routine kbd_connect() search and attach keyboardless VT. After link console-keyboard is ready. > But > how do I arrange that? How can I tell the kernel which > keyboard goes with what framebuffer? Current CVS allow remap VT-keyboard link via /proc. VT-fbdev link is hidden :o( Underwater cliffs located everywhere. Aivils Stoss |
|
From: Helge H. <hel...@ai...> - 2003-11-19 10:06:13
|
Aivils Stoss wrote: > Congratulations! You open new "matrox psyho circus" thread. Does this imply there were an earlier matrox psycho circus thread? If so, is there a searchable archive somewhere? > IMHO with right config You solutions works. Of course any > cyberpunk may use neighbour display, but that by-product only. > Well, the NAT box on my home network only lets ssh through, so I don't worry too much about X security. But why would there be a problem? The two X servers are independent and each can enforce security on its own connections. The only security breach I can imagine is that one logged-in user may read the other logged-in user's window contents because he has access to the same framebuffer memory. "Any cyberpunk" on the net (or even logged-in but not running X) shouldn't be more of a problem than with a ordinary setup? Right now those xserves are insecure because they are dri-trunk servers wich happens to have no security compiled in. But I don't need that for the dual setup, it is only to get DRI. > Please test config, whrere both xf86 use _two_ screens without > xinerama. If You use both screens of matrox device, then You > can determine "who is who". > 1st xf86 "matrox device screen=0" as _fisrt_ xf86 Screen > in ServerLayout, but "matrox device screen=1" as second > 2nd xf86 swap order of "matrox device" in ServerLayout. > "matrox device screen=1" is _first_ for this server. > Great idea! Both users can stick to his/her "first" screen and simply not use the second one (which only would stomp on the other users display) I guess only one user really need a setup with two displays, using the first one as a "placeholder". The user with the first display can simply keep using that. > Both servers setup both screens, which one may work according > Your reports above. Because matrox device order is swaped, then > sys admin can determine where runs programs under each xf86. > > "xterm -display :0.0" runs on owner display > "xterm -display :0.1" runs on neighbour display > > But DISPLAY=":0.0" always. I believe the first user will see his display as :0.0 and the other user's display as :0.1? And the other user will see his display as :1.0 and the first user's display as :1.1? One xserver start as :0 and the other as :1 with the current single-screen setup. > > You might add in .xinirc line > xsetroot -display :0.1 -color most-compatible-colour > > This solution cannot be called "technical correct". The "correct" way would be to tell an xserver to use the second G550 display only. I don't know if that is possible to do though. > You can try patch program Xnest to use independ mouse > and keyboard. I don't understand this. Ruby already provides independent keyboards. And the two X servers have no problem using independent mice. From my /etc/X11/xdm/Xservers file: :0 local /usr/X11R6/bin/X :0 vt17 -deferglyphs 16 -nolisten tcp :1 local /usr/X11R6/bin/X :1 vt7 -deferglyphs 16 -nolisten tcp -xf86config xf86config-4 The vt7/vt17 decides the keyboard, :0/:1 sets display numbers. The first server uses standard XF86Config-4 and the other uses the alternative xf86config-4. The XF86Config-4 is unchanged from ordinary single-user setup, using accelerated mga on the first display. Contents of xf86config-4 varies depending on what I want to do. It is usually set up for fbdev X on the second framebuffer device, I also tried letting it use the mga driver and got two servers using the same screen. Using the two-screen setup you mentioned will be an interesting experiment. Helge Hafting |
|
From: Helge H. <hel...@ai...> - 2003-11-19 09:38:14
|
I tried 15.nov CVS ruby. It is similiar to 29.oct ruby in that it works without framebuffer console, and hang with it. My previous attempts with 29.oct ruby and G550 framebuffer consoles was with devfs. Running without devfs still crashes. There are some things I wonder about. Ruby without framebuffer console is compiled with vga console (it isn't even possible to deselect it in menuconfig) But I don't get a useable vga console. The framebuffer (without console) takes over and there is no switching back. Perhaps this isn't ruby specific, but I wonder if it is supposed to be that way? The use of dumbcon is another thing. I use dumbcon=1 because I found that in a howto. Will that be unnecessary when the framebuffer consoles work someday? I have the understanding that "dumbcon" is something that attach a keyboard to a vt, so X may get keyboard input, while the dumbcon actually doesn't display anything. I happen to have two framebuffers, so running consoles on both, with one keyboard on each ought to be possible. There seem to be no need for a "dumbcon" then, the second framebuffer ought to get a keyboard and set of vt's just like the first one. But how do I arrange that? How can I tell the kernel which keyboard goes with what framebuffer? Helge Hafting |
|
From: hugo v. <hug...@ca...> - 2003-11-18 13:46:34
|
Hi all! I must have missed the announcement, but I noticed the new patch in Aivil's place. So I ran with it: Funny thing! It behaves like the 2.6.0 patches and the second wheel of my A4Tech 2 wheel optical mouse does not work. I went back to the October patch: both wheels work. Hugo. The following signature is automatically added by the web email service I use, which is beyond my control: ======================================= Help the planet each day! It's free and easy: http://www.Care2.com/dailyaction/ |
|
From: Kjetil K. <kj...@kj...> - 2003-11-16 11:03:01
|
On Sunday 16 November 2003 01:21, Andreas Schuldei wrote: > > I'm back to the first error again; freeze-up and events are being > > generated to /dev/input/event2 > > you could disable the usb.agent, whitch might try funny things > with modules etc... whitch you dont have. chmod -x or put in an > exit 0 at the beginning? OK, I'll try that some day! Unfortunately, I just had a hard deadline landing in my lap... :-( So, I have to go and focus on that, and settle for the setup I have now for the time being.... That's life, I'm afraid... Best, Kjetil -- Kjetil Kjernsmo Astrophysicist/IT Consultant/Skeptic/Ski-orienteer/Orienteer/Mountaineer kj...@kj... web...@sk... ed...@le... Homepage: http://www.kjetil.kjernsmo.net/ OpenPGP KeyID: 6A6A0BBC |
|
From: Andreas S. <an...@sc...> - 2003-11-16 00:22:34
|
* Kjetil Kjernsmo (kj...@kj...) [031114 16:35]: > On Friday 14 November 2003 15:20, Kjetil Kjernsmo wrote: > > It worked after pulling the tangtop from the USB socket, and still > > after heavy smashing on the keyboard while updatedb was running... > > :-) I guess that's a good sign. > > Well, how long was Adam in paradise...? :-) quite long, i think. we just get told about the sad last day... > I'm back to the first error again; freeze-up and events are being > generated to /dev/input/event2 you could disable the usb.agent, whitch might try funny things with modules etc... whitch you dont have. chmod -x or put in an exit 0 at the beginning? |
|
From: roland <for...@gm...> - 2003-11-15 19:01:54
|
hi ! i tried vanilla 2.6.0test9 (default config) kernel on my md8008xl PC (see specs at http://www.pcwelt.de/tests/hardware-tests/pc_notebook/29981/) today and while booting i see absolutely NOTHING until XFree86 starts. the screen gets black directly at kernel initialisation. i cannot switch to a console from XFree86. i found that this is because my suse9 has "vga=0x317" enabled by default (in grub bootconfig). this seems to work fine with the old 2.4 kernel, but doesnt seem to work with 2.6.0test9 the pc i mention is very popular in germany and has been sold many thousand times, so i think it could be important to inform you about this issue. (maybe it`s even more general, ond not directly related to this PC/Videocard) let me know if i can help fixing the problem. (sorry - i`m no kernel hacker but i could help by doing some test etc..) regards roland |
|
From: Wayne W. <wh...@Ma...> - 2003-11-15 17:39:21
|
On Sat, 15 Nov 2003, Aivils Stoss wrote: > One backdoor exists. That is agpgart kernel module. All Radeon adapters > are AGP adapters. Please disable AGP kernel support or disable AGP > support in 2nd xf86. May be xf86 isn't slyboots. AGP kernel support is not loaded, since I have disabled dri in XF86Config, so X doesn't load it. I also removed any AGP options from XF86Config. The problem is still there. Of course, there are still comments about the AGP bus in XFree86.log, since the 1st radeon is AGP. Here's an additional twist: when I boot up gdm starts X on 1st and 2nd displays, and as discussed, logging in and out of X on the 1st display disrupts the 2nd display. Then restarting the 2nd X restores the 2nd display without disrupting the 1st display. Now if after booting up, I do a Ctrl-F1 to switch VT0 to vt1, then the 2nd display also gets disrupted. Switching back to X on VT0 works fine. However, in this situation, restarting the 2nd X restores the 2nd display while disrupting the 1st display. BTW, has anyone discussed PrefBusID with the XFree86 folks? If the current implentation is a good one, I would think they might be willing to integrate it into XFree86. And presumably they could tell us if something is missing. > > When I am using the text console (vgacon?) and switch from one vc to > > another on VT0, sometime the destination vc is restored with the cursor in > > the wrong place. Typically the cursor should be at the bottom of the > > screen and is restored instead half way up the screen. > > Uh. First console is broken. This must be a bug in vgacon? Cheers, Wayne |
|
From: Svetoslav S. <sv...@gm...> - 2003-11-15 11:33:54
|
>
> > Actually that should stay and the code in every console driver that does
> > this should be removed. It is redundant. What I was thinking of doing is
> > work the VT console driver to use default_mode. This way we can have
> > printks without having to have full tty support.
> >
> > > Update of /cvsroot/linuxconsole/ruby/ruby-2.6/drivers/char
> > > In directory sc8-pr-cvs1:/tmp/cvs-serv32309/ruby-2.6/drivers/char
> > >
> > > Modified Files:
> > > vt.c
> > > Log Message:
> > > console initialization fixes
> > >
> > > Index: vt.c
> > > ===================================================================
> >
>
> Ok. I cannot understand fbcon initialization.
> I cannot remove stuff in fbcon_set_display() - that is aliens code :(
> Until con_init() contains default_mode setup current CVS works more
> correct.
>
> Remain bugs:
> fbdev console on /dev/tty1 will work correct only after logout/login.
> g550 fb works partialy (show oopses on fbdev!!!).
with your fixes there are no oopses
my g550 experiancies so far:
--------------------------------------------------
everything seems to work pretty OK
as modules, if compiled in the second console didn't
register second vt (or i did somehow missed it?)
but there is a hardware limitation/ problem i think,
and may be matroxfb driver issues
according to dmesg i must be able to use independantly both heads,
but ......
until i use matroxset to swap the matrox's heads, the first VT is displayed
on both heads,
and after i use it both heads are blanked, i can activate the first by a VC
change (ctrl-alt-Fx)
and it's displayed only on one of the heads,
and here comes the problem -- there is no way to activate the second
it remains blank whatever i
try
-------------------------------------------------------------------------------
rivafb seems really broken
in the box with g550 i have two TNT2 M64 PCI 32 Mb,
rivafb reports them as having only 8Mb with/without "nomtrr"
if compiled as module, modprobe rivafb never finishes and eats all the CPU
in my other box Geforce4 PCI with 64Mb is reported to have !!! 2Mb RAM !!!
0:0c.0 VGA compatible controller: nVidia Corporation NV17 [GeForce4 MX 420]
(rev a3) (prog-if 00 [VGA])
Subsystem: Micro-Star International Co., Ltd.: Unknown device 8852
Control: I/O- Mem- BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr-
Stepping- SERR- FastB2B-
Status: Cap+ 66Mhz+ UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort-
<TAbort- <MAbort- >SERR- <PERR-
Interrupt: pin A routed to IRQ 18
Region 0: Memory at e4000000 (32-bit, non-prefetchable) [disabled]
[size=16M]
Region 1: Memory at e0000000 (32-bit, prefetchable) [disabled]
[size=64M]
Region 2: Memory at e8000000 (32-bit, prefetchable) [disabled]
[size=512K]
Expansion ROM at <unassigned> [disabled] [size=128K]
Capabilities: [60] Power Management version 2
Flags: PMEClk- DSI- D1- D2- AuxCurrent=0mA
PME(D0-,D1-,D2-,D3hot-,D3cold-)
Status: D0 PME-Enable- DSel=0 DScale=0 PME-
best,
svetljo
--
NEU FÜR ALLE - GMX MediaCenter - für Fotos, Musik, Dateien...
Fotoalbum, File Sharing, MMS, Multimedia-Gruß, GMX FotoService
Jetzt kostenlos anmelden unter http://www.gmx.net
+++ GMX - die erste Adresse für Mail, Message, More! +++
|
|
From: Aivils S. <Aiv...@un...> - 2003-11-15 10:38:03
|
> I stumbled across a XFree86 config file for the G450,
> which had something new to me. There were two
> device sections referrring to the same BusID,
> but with screen=0 and screen=1.
>
> I decided to try the same on my G550. Instead of
> running the second head with framebuffer, I used
> the mga accelerated driver and screen=1.
> The first attempt failed, X found "a suitable device
> but no screen" or some such. So I removed
> the screen=1 thing just to see what happened.
>
> Accelerated mga was already running on the first head
> as :0, and I now started another accelerated server with
> the same configuration as :1, with another keyboard and mouse.
>
> It worked in a weird way - I got two xservers running on the
> _same_ screen. Each mouse had its own cursor, they could
> chase each other around. I opened one xterm with each
> mouse. My left keyboard worked with
> the left xterm, the right keyboard witht he right xterm.
> A "poor man's ruby setup", dual seat with a single screen.
Congratulations! You open new "matrox psyho circus" thread.
IMHO with right config You solutions works. Of course any
cyberpunk may use neighbour display, but that by-product only.
Please test config, whrere both xf86 use _two_ screens without
xinerama. If You use both screens of matrox device, then You
can determine "who is who".
1st xf86 "matrox device screen=0" as _fisrt_ xf86 Screen
in ServerLayout, but "matrox device screen=1" as second
2nd xf86 swap order of "matrox device" in ServerLayout.
"matrox device screen=1" is _first_ for this server.
Both servers setup both screens, which one may work according
Your reports above. Because matrox device order is swaped, then
sys admin can determine where runs programs under each xf86.
"xterm -display :0.0" runs on owner display
"xterm -display :0.1" runs on neighbour display
But DISPLAY=":0.0" always.
You might add in .xinirc line
xsetroot -display :0.1 -color most-compatible-colour
This solution cannot be called "technical correct".
You can try patch program Xnest to use independ mouse
and keyboard.
Aivils Stoss
|
|
From: Aivils S. <Aiv...@un...> - 2003-11-15 09:53:40
|
> > PrefBusID hide video adapters from xf86. Actualy xf86 see and use one > > video adapter given by PrefBusID. 1st xf86 schould not touch 2nd. xf86 > > is crazy like as fox, if that destroy 2nd desktop without access. > > Well, something is crazy like a fox, since my 2nd Radeon display gets > disrupted. Perhaps I should mention that both my displays are DFPs. I > did try hackvideo, that makes no difference. How can I capture the state > of the 2nd radeon before and after the disruption, to figure out in what > way the 2nd radeon has been disrupted? One backdoor exists. That is agpgart kernel module. All Radeon adapters are AGP adapters. Please disable AGP kernel support or disable AGP support in 2nd xf86. May be xf86 isn't slyboots. > Or any other suggestions on how to > debug this? > > A couple other comments about 2.6.0-test9-ruby: > > When I am using the text console (vgacon?) and switch from one vc to > another on VT0, sometime the destination vc is restored with the cursor in > the wrong place. Typically the cursor should be at the bottom of the > screen and is restored instead half way up the screen. Uh. First console is broken. > After a fresh boot, if I run from a text vc on VT0 "startx --layout > Layout1 vt17 &", where Layout1 is set up to use my second radeon, then X > comes up fine on the 2nd radeon. However, the text vc is now frozen. I am not capable to patch xf86 VGA handling. Any xf86 1st or last freezy VGA interface. Aivils Stoss |
|
From: Aivils S. <Aiv...@un...> - 2003-11-15 08:10:42
|
> Presently the kbd_inline functions pass in a pointer for struct > kbd_struct. Then the inlines are called as follows: > > get_kbd_mode(&vc->kbd_table,.. > > I purpose that we go back > > get_kbd_mode(vc->kbd_table,... > > and the inline functions don't pass in a pointer anymore. Since it is a > inline function we don't have the cost of a stack push and pop. Why You never print set_kbd_mode(vc->kbd_table,... which actualy copy subject vc->kbd_table for usage _inside_ set_kbd_mode - do nothing _outside_ function body. Do You will use two types of agruments, get_kbd_mode(struct kbd_struct kbd, set_kbd_mode(struct kbd_struct *kbd, If You like performance, the You will eliminate find_vc() and use vc_cons, because anyway we have only one tty range 1-63, and so special vt->vc_cons[] is redundant. Aivils Stoss |
|
From: Aivils S. <Aiv...@un...> - 2003-11-15 07:45:52
|
> Actually that should stay and the code in every console driver that does > this should be removed. It is redundant. What I was thinking of doing is > work the VT console driver to use default_mode. This way we can have > printks without having to have full tty support. > > > Update of /cvsroot/linuxconsole/ruby/ruby-2.6/drivers/char > > In directory sc8-pr-cvs1:/tmp/cvs-serv32309/ruby-2.6/drivers/char > > > > Modified Files: > > vt.c > > Log Message: > > console initialization fixes > > > > Index: vt.c > > =================================================================== > Ok. I cannot understand fbcon initialization. I cannot remove stuff in fbcon_set_display() - that is aliens code :( Until con_init() contains default_mode setup current CVS works more correct. Remain bugs: fbdev console on /dev/tty1 will work correct only after logout/login. g550 fb works partialy (show oopses on fbdev!!!). Aivils Stoss |
|
From: Wayne W. <wh...@Ma...> - 2003-11-14 17:49:32
|
On Thu, 13 Nov 2003, Aivils Stoss wrote: > PrefBusID hide video adapters from xf86. Actualy xf86 see and use one > video adapter given by PrefBusID. 1st xf86 schould not touch 2nd. xf86 > is crazy like as fox, if that destroy 2nd desktop without access. Well, something is crazy like a fox, since my 2nd Radeon display gets disrupted. Perhaps I should mention that both my displays are DFPs. I did try hackvideo, that makes no difference. How can I capture the state of the 2nd radeon before and after the disruption, to figure out in what way the 2nd radeon has been disrupted? Or any other suggestions on how to debug this? A couple other comments about 2.6.0-test9-ruby: When I am using the text console (vgacon?) and switch from one vc to another on VT0, sometime the destination vc is restored with the cursor in the wrong place. Typically the cursor should be at the bottom of the screen and is restored instead half way up the screen. After a fresh boot, if I run from a text vc on VT0 "startx --layout Layout1 vt17 &", where Layout1 is set up to use my second radeon, then X comes up fine on the 2nd radeon. However, the text vc is now frozen. I'm not sure if it is just output or both output and input. Is this possibly related to the "VGA routing" that X does for the 2nd radeon, disrupting vgacon on the 1st radeon? Thanks for your help. Cheers, Wayne |
|
From: James S. <jsi...@in...> - 2003-11-14 17:39:11
|
Actually that should stay and the code in every console driver that does this should be removed. It is redundant. What I was thinking of doing is work the VT console driver to use default_mode. This way we can have printks without having to have full tty support. > Update of /cvsroot/linuxconsole/ruby/ruby-2.6/drivers/char > In directory sc8-pr-cvs1:/tmp/cvs-serv32309/ruby-2.6/drivers/char > > Modified Files: > vt.c > Log Message: > console initialization fixes > > Index: vt.c > =================================================================== > RCS file: /cvsroot/linuxconsole/ruby/ruby-2.6/drivers/char/vt.c,v > retrieving revision 1.10 > retrieving revision 1.11 > diff -u -d -r1.10 -r1.11 > --- vt.c 13 Nov 2003 07:17:35 -0000 1.10 > +++ vt.c 14 Nov 2003 08:42:07 -0000 1.11 > @@ -909,12 +909,7 @@ > vc->vc_uni_pagedir = 0; > hi_font_mask = 0; > complement_mask = 0; > - can_do_color = default_mode->vc_can_do_color; > - video_num_columns = default_mode->vc_cols; > - video_num_lines = default_mode->vc_rows; > - video_scan_lines = default_mode->vc_scan_lines; > - screensize = video_num_columns * video_num_lines; > - vc->vc_font = vc->display_fg->default_mode->vc_font; > + can_do_color = 0; > sw->con_init(vc, init); > if (!complement_mask) > complement_mask = can_do_color ? 0x7700 : 0x0800; > @@ -1892,8 +1887,6 @@ > cons_num = vt->first_vc + i; > visual_init(vc, 0); > update_attr(vc); > - vc_resize(vc, vt->default_mode->vc_cols, > - vt->default_mode->vc_rows); > > /* If the console changed between mono <-> color, then > * the attributes in the screenbuf will be wrong. The > > > > ------------------------------------------------------- > This SF.Net email sponsored by: ApacheCon 2003, > 16-19 November in Las Vegas. Learn firsthand the latest > developments in Apache, PHP, Perl, XML, Java, MySQL, > WebDAV, and more! http://www.apachecon.com/ > _______________________________________________ > Linuxconsole-commit mailing list > Lin...@li... > https://lists.sourceforge.net/lists/listinfo/linuxconsole-commit > |
|
From: James S. <jsi...@in...> - 2003-11-14 17:33:12
|
Presently the kbd_inline functions pass in a pointer for struct kbd_struct. Then the inlines are called as follows: get_kbd_mode(&vc->kbd_table,.. I purpose that we go back get_kbd_mode(vc->kbd_table,... and the inline functions don't pass in a pointer anymore. Since it is a inline function we don't have the cost of a stack push and pop. |
|
From: Kjetil K. <kj...@kj...> - 2003-11-14 15:33:37
|
On Friday 14 November 2003 15:20, Kjetil Kjernsmo wrote: > It worked after pulling the tangtop from the USB socket, and still > after heavy smashing on the keyboard while updatedb was running... > :-) I guess that's a good sign. Well, how long was Adam in paradise...? :-) I'm back to the first error again; freeze-up and events are being generated to /dev/input/event2 I can't find anything in the logs that are relevant to the crash itself. However, I think I remember I saw something at bootup about not being able to find usb** hotplugs or something. I can't find that in dmesg, though, so I guess it wasn't important...? What may be of some importance is that I can't find any evidence that input.agent has been run. There's a usb.agent which came with the Debian hotplug package, and that seems to be called, this is from user.log: Nov 14 15:03:14 owl default.hotplug[1244]: arguments (usb) env (DEVFS=/ proc/bus/usb OLDPWD=/ PATH=/bin:/sbin:/usr/sbin:/usr/bin ACTION=remove PWD=/etc/hotplug HOME=/ SHLVL=2 DEVICE=/proc/bus/usb/001/002 INTERFACE=3/1/1 PRODUCT=d3d/1/1 TYPE=0/0/0 DEBUG=kernel _=/usr/bin/env) Nov 14 15:03:14 owl default.hotplug[1244]: invoke /etc/hotplug/usb.agent () Nov 14 15:03:14 owl default.hotplug[1246]: arguments (usb) env (DEVFS=/ proc/bus/usb OLDPWD=/ PATH=/bin:/sbin:/usr/sbin:/usr/bin ACTION=remove PWD=/etc/hotplug HOME=/ SHLVL=2 DEVICE=/proc/bus/usb/001/002 INTERFACE=3/1/2 PRODUCT=d3d/1/1 TYPE=0/0/0 DEBUG=kernel _=/usr/bin/env) Nov 14 15:03:14 owl default.hotplug[1246]: invoke /etc/hotplug/usb.agent () Nov 14 15:03:29 owl default.hotplug[1303]: arguments (usb) env (DEVFS=/ proc/bus/usb OLDPWD=/ PATH=/bin:/sbin:/usr/sbin:/usr/bin ACTION=add PWD=/etc/hotplug HOME=/ SHLVL=2 DEVICE=/proc/bus/usb/001/003 INTERFACE=3/1/1 PRODUCT=d3d/1/1 TYPE=0/0/0 DEBUG=kernel _=/usr/bin/env) Nov 14 15:03:29 owl default.hotplug[1303]: invoke /etc/hotplug/usb.agent () Nov 14 15:03:29 owl default.hotplug[1313]: arguments (usb) env (DEVFS=/ proc/bus/usb OLDPWD=/ PATH=/bin:/sbin:/usr/sbin:/usr/bin ACTION=add PWD=/etc/hotplug HOME=/ SHLVL=2 DEVICE=/proc/bus/usb/001/003 INTERFACE=3/1/2 PRODUCT=d3d/1/1 TYPE=0/0/0 DEBUG=kernel _=/usr/bin/env) Nov 14 15:03:29 owl default.hotplug[1313]: invoke /etc/hotplug/usb.agent () Nov 14 15:03:32 owl usb.agent[1303]: Setup keybdev for USB product d3d/1/1 Nov 14 15:03:32 owl usb.agent[1313]: Setup mousedev for USB product d3d/1/1 Nov 14 15:27:20 owl gconfd (kjetil-1768): starting (version 2.4.0.1), pid 1768 user 'kjetil' Nov 14 15:27:20 owl gconfd (kjetil-1768): Resolved address "xml:readonly:/etc/gconf/gconf.xml.mandatory" to a read-only config source at position 0 Nov 14 15:27:20 owl gconfd (kjetil-1768): Resolved address "xml:readwrite:/home/kjetil/.gconf" to a writable config source at position 1 Nov 14 15:27:20 owl gconfd (kjetil-1768): Resolved address "xml:readonly:/etc/gconf/gconf.xml.defaults" to a read-only config source at position 2 As I said, if I just pulled the tangtop from the USB port, it would be redetected, but this didn't help here... Best, Kjetil -- Kjetil Kjernsmo Astrophysicist/IT Consultant/Skeptic/Ski-orienteer/Orienteer/Mountaineer kj...@kj... web...@sk... ed...@le... Homepage: http://www.kjetil.kjernsmo.net/ OpenPGP KeyID: 6A6A0BBC |
|
From: Kjetil K. <kj...@kj...> - 2003-11-14 14:20:52
|
On Thursday 13 November 2003 14:08, Andreas Schuldei wrote: > * Kjetil Kjernsmo (kj...@kj...) [031113 12:47]: > > Anyway, from the readme: > > input.agent: You need to put this in /etc/hotplug/, if you > > already have an /etc/hotplug/input.agent then you hopefully know > > enough to make it work. > > > > well, I haven't any input.agent, and I have no idea how it can be > > made to work... :-) Any clues for me...? > > cp input.agent /etc/hotplug/ > chmod +x /etc/hotplug/input.agent OK. > i think that is all. > perhaps, if you dont have hotplug installed yet, do an apt-get > install hotplug. note that it will not work on most things since > you complied your kernel statically, but for the input.agent this > does not matter, since it does not load modules. OK, I did, and just replaced the input.agent with the file from above. I have no other hotplug devices now. Also, I discovered that I didn't have hotplug in my kernel, so I enabled that. > it is possible to include this input agent into other input > agents, withoug much effort, if others have other input needs > besides X. > > > There's no need for a new kernel or a new X binary, if I get you > > right. > > you need to adjust your xf86config, yes. OK. It seemed easy enough. Only thing that doesn't seem to work right now is the mousewheel on the USB... Any gotchas there? > > Since we have two set of keyboards, the config is somewhat > > different than in the readme, I presume, do I still need to set the > > "Dev Phys" and that stuff...? > > yes, or the name. the phys stuff did not work for me on 2.4, but > the name option worked fine. The name is the same on the keyboard and mouse for me, so probably, phys worked. > i will try the phys option on 2.6, too. that would be a major > argument for this input method. Yup. > > Also, is there any way to provoke the error to see if it is still > > there. > > not that i would know. it happens more often under heavy ide > load. try doing some heavy i/o work. anyone have a good idea? updatedb...? :-) It worked after pulling the tangtop from the USB socket, and still after heavy smashing on the keyboard while updatedb was running... :-) I guess that's a good sign. Nevertheless, further ideas on how to stress-test it is very much welcome! Cheers, Kjetil -- Kjetil Kjernsmo Astrophysicist/IT Consultant/Skeptic/Ski-orienteer/Orienteer/Mountaineer kj...@kj... web...@sk... ed...@le... Homepage: http://www.kjetil.kjernsmo.net/ OpenPGP KeyID: 6A6A0BBC |
|
From: Helge H. <hel...@ai...> - 2003-11-13 21:28:25
|
I stumbled across a XFree86 config file for the G450, which had something new to me. There were two device sections referrring to the same BusID, but with screen=0 and screen=1. I decided to try the same on my G550. Instead of running the second head with framebuffer, I used the mga accelerated driver and screen=1. The first attempt failed, X found "a suitable device but no screen" or some such. So I removed the screen=1 thing just to see what happened. Accelerated mga was already running on the first head as :0, and I now started another accelerated server with the same configuration as :1, with another keyboard and mouse. It worked in a weird way - I got two xservers running on the _same_ screen. Each mouse had its own cursor, they could chase each other around. I opened one xterm with each mouse. My left keyboard worked with the left xterm, the right keyboard witht he right xterm. A "poor man's ruby setup", dual seat with a single screen. Of course this didn't work too well when windows overlapped, one xserver obviously wasn't aware of windows handled by the other and repainted with root window color when I moved the overlapping window away. And the other xserver was of course unaware that the window had been overpainted by another xserver and didn't try to repaint it until I moved it off screen and back. These issues were kind of obvious - the big surprise was that the 2d accelerator survived and didn't mess up from simultaneous access by two xservers. I could grab a window with each mouse and drag them through each other with opaque movement and get no trouble other than the obvious mispainting. And this is a smp machine, so there were no serialization due to single-cpu scheduling. 3D only worked on one xserver, the other segfaulted all 3d apps and even went into a 2d pause while the first server ran glxgears. This is promising. I currently run accelerated on one screen and unaccelerated framebuffer on the other. But it seems I should be able to run 2d accelerated on both screens, if I can convince one server to use the other screen. This should be possible, xinerama can use both screens so i the worst case I'll run two xinerama servers and avoid collisions by dicipline :-) Helge Hafting |
|
From: Andreas S. <an...@sc...> - 2003-11-13 13:08:24
|
* Kjetil Kjernsmo (kj...@kj...) [031113 12:47]: > Anyway, from the readme: > input.agent: You need to put this in /etc/hotplug/, if you already > have an /etc/hotplug/input.agent then you hopefully know > enough to make it work. > > well, I haven't any input.agent, and I have no idea how it can be made > to work... :-) Any clues for me...? cp input.agent /etc/hotplug/ chmod +x /etc/hotplug/input.agent i think that is all. perhaps, if you dont have hotplug installed yet, do an apt-get install hotplug. note that it will not work on most things since you complied your kernel statically, but for the input.agent this does not matter, since it does not load modules. it is possible to include this input agent into other input agents, withoug much effort, if others have other input needs besides X. > There's no need for a new kernel or a new X binary, if I get you right. you need to adjust your xf86config, yes. > Since we have two set of keyboards, the config is somewhat different > than in the readme, I presume, do I still need to set the "Dev Phys" > and that stuff...? yes, or the name. the phys stuff did not work for me on 2.4, but the name option worked fine. i will try the phys option on 2.6, too. that would be a major argument for this input method. > Also, is there any way to provoke the error to see if it is still there. not that i would know. it happens more often under heavy ide load. try doing some heavy i/o work. anyone have a good idea? |
|
From: Kjetil K. <kj...@kj...> - 2003-11-13 11:45:52
|
On Thursday 13 November 2003 12:28, Andreas Schuldei wrote: > * Andreas Schuldei (an...@sc...) [031111 15:04]: > > * Svetoslav Slavtchev (sv...@gm...) [031111 14:09]: > > i see it in 2.6, too. i think i even have usb debugging messages > > enabled, but one sees only the plugg- and unplugg events, nothing > > more or strange, as far as the OS is concerned. > > now i figured out what was wrong with my setup, and drag and drop > works again for me with X's event interface, Great! >as described on > http://people.debian.org/~warp/evdev/ (which is included in both > the debian experimental X and also in the one on > www.schuldei.org). OK, good, so I have it allready, since I use your packages. > one must not have Emulate3Buttons option active, that absorbs the > buttondepress-event and releases it only together with the > button-release-event. OK. I have them off. > this patch (on the above url) also features a fifo which the > input.agent uses to alert the Xserver of hotplug events. this > means that devices are regrabbed on the fly, without the need of > switching to the console and back. Nice. Anyway, from the readme: input.agent: You need to put this in /etc/hotplug/, if you already have an /etc/hotplug/input.agent then you hopefully know enough to make it work. well, I haven't any input.agent, and I have no idea how it can be made to work... :-) Any clues for me...? > there is an issue with the keyboard in that input-agent, though, > i guess that is due to the format change in the later 2.6 > kernels... you could try if it works properly on your 2.4. Yup. I'll try. So, do I just dump the input.agent in /etc/hotplug...? There's no need for a new kernel or a new X binary, if I get you right. Since we have two set of keyboards, the config is somewhat different than in the readme, I presume, do I still need to set the "Dev Phys" and that stuff...? Also, is there any way to provoke the error to see if it is still there. Since the problem is that it blocks after some time of use, it would be great to just make it lock up or let it be, if it is happy with the current state of things... :-) Best, Kjetil -- Kjetil Kjernsmo Astrophysicist/IT Consultant/Skeptic/Ski-orienteer/Orienteer/Mountaineer kj...@kj... web...@sk... ed...@le... Homepage: http://www.kjetil.kjernsmo.net/ OpenPGP KeyID: 6A6A0BBC |
1998 messages has been excluded from this view by a project administrator.
