From: Kjetil K. <kj...@kj...> - 2003-10-29 21:06:32
|
Honorable Kernel Hackers, I hope you do not mind a "user"-question from a newbie...? I'm working to set up my system according to the XFree Local Multi-User HOWTO, and it says in the final notes that I could try here if I need help... :-) I've been working on it for a couple of weeks, actually. I guess I have a little too much of the Perl Programmer's Hubris, but then, it is the only way me and my girlfriend can afford and have space for another seat, and she needs to get some work done to... :-) Also, I don't think what remains is a kernel problem, more likely an X problem. I think I've been through all the pitfalls that one can possibly be going into. Since I'm on Debian Sid, I've been bothering Andreas Schuldei a lot. I'm using his great packages. To do the short version first: Everything seems to work fine, except that when I try to get KDM to launch the two X servers, the first server comes up fine, but the second does not. Instead, the server takes the whole CPU, but apparently does nothing. If I kill the second X-server with -9, it comes back up and then it runs fine on the second screen, but the first locks up and the first X server runs along. There are tons of messages like: (EE) MGA(0): [dri] Idle timed out, resetting engine... in my logs afterwards. Now, for the details: My system has a Matrox G450 DH and Mystique card. Since I'm now going for a 2.4 kernel (I'll (re)turn to 2.6 later, for sure!), I'm not trying to use the second head of the G450. Rather, I'm trying to use the Mystique for the second user. I have a Logitech Keyboard on the normal PS/2 port, then a cheap keyboard on the usual mouse PS/2 port. I have two mice connected on the USB ports. I might want to change this setup later, but I don't want to buy a USB keyboard. One of the things Svetoslav asks for at http://tldp.org/HOWTO/XFree-Local-multi-user-HOWTO/final_words.html is contents of /proc/bus/input/devices, and since this sums up the setup rather nice, I paste it here: I: Bus=0011 Vendor=0001 Product=0002 Version=ab02 N: Name="AT Set 2 keyboard" P: Phys=isa0060/serio0/input0 H: Handlers=kbd event0 B: EV=120003 B: KEY=4 2000000 8061f9 fbc9d621 efdfffdf ffefffff ffffffff fffffffe B: LED=7 I: Bus=0011 Vendor=0001 Product=0002 Version=ab83 N: Name="AT Set 2 keyboard" P: Phys=isa0060/serio1/input0 H: Handlers=kbd event1 B: EV=120003 B: KEY=4 2000000 8061f9 fbc9d621 efdfffdf ffefffff ffffffff fffffffe B: LED=7 I: Bus=0003 Vendor=046d Product=c00e Version=1110 N: Name="Logitech USB-PS/2 Optical Mouse" P: Phys=usb-00:04.2-1/input0 H: Handlers=mouse0 event2 B: EV=7 B: KEY=70000 0 0 0 0 0 0 0 0 B: REL=103 I: Bus=0003 Vendor=09da Product=0006 Version=0001 N: Name="A4Tech USB Optical Mouse" P: Phys=usb-00:04.2-2/input0 H: Handlers=mouse1 event3 B: EV=7 B: KEY=7f0000 0 0 0 0 0 0 0 0 B: REL=103 BTW, /proc/bus/usb/devices does not exist on my system, and cat /proc/bus/console/*/* says isa0060/serio0/input0 isa0060/serio1/input0 This looks good, doesn't it? (I'll paste in just the things that I assume is relevant, the rest I'll attach. Hope that's OK. XFree86.[12].logs I had to trim, this mail is large enough as it is.) I've configured Xinerama, and it works fine. I have preferred keeping the whole X config in a single file, so I take the approach with specifying different "layouts". I maintain four different layouts, "Simple", "Xinerama", "X0" and "X1". The two latter are the ones for the multi-user config. They look like this: Section "ServerLayout" Identifier "X0" Screen 0 "Screen0" 0 0 InputDevice "Mouse0" "CorePointer" InputDevice "Keyboard0" "CoreKeyboard" Option "PrefBusID" "1:5:0" EndSection Section "ServerLayout" Identifier "X1" Screen 0 "Screen1" 0 0 InputDevice "Mouse1" "CorePointer" InputDevice "Keyboard1" "CoreKeyboard" Option "PrefBusID" "0:11:0" EndSection devfs is up and running, allthough there are some weird issues; specifically mount doesn't report that devfs is mounted, whereas /proc/mounts does. This put me off for a long time. It is easy enough to get X running on either screen with corresponding keyboard and mouse. Also, I have been able to run it on both, simultaneously, running startx -- /usr/X11R6/bin/X1 :1 -layout X1 vt8 & startx -- /usr/X11R6/ bin/X0 :0 -layout X0 vt7 I could type on both keyboards, it showed on their respective screens. Mice worked. So far, so good, it seems, but then I tried to get KDM to start them for me, by adding the following lines to /etc/kde3/kdm/Xservers: :0 local /usr/bin/nice -n -10 /usr/X11R6/bin/X0 :0 -layout X0 -deferglyphs 16 vt7 :1 local /usr/bin/nice -n -9 /usr/X11R6/bin/X1 :1 -layout X1 vt9 Booting up, things go wrong. X comes slowly up, but only on one screen. Upon inspection, it is clear why it is so slow, because the X1 process runs at 99% of the CPU according to top. Other than that, things are normal on the display that is running, but obviously, the idea was to have both... :-) In syslog, I find things like this: Oct 29 16:04:15 owl kdm[720]: Hung in XOpenDisplay(:1), aborting Oct 29 16:04:15 owl kdm[720]: server open failed for :1, giving up Oct 29 16:04:15 owl kdm[686]: Display :1 cannot be opened I didn't really have much of an idea on where to start, but I noticed two things. One thing was that in kdm.log, I found lines: (EE) MGA(0): Need at least 4608 kB video memory at this resolution, bit depth (My G450 has 16MB, I don't know what the Mystique has). So, I tried setting the bit depth to 8 and the resolution for both cards to 800x600. That didn't help, and besides, I run Xinerama at 16 bpp and 1280x960 and 1024x768 respectively, so that shouldn't suddenly be a problem, I figured. Turned it back up. Then, I was in doubt whether I had entered the PCI bus id's correctly. The autoconfig I ran when making the Xinerama config had them in the form PCI:0:11:0, but in the examples, the PrefBusID strings did not have the PCI:-prefix. I had left it like that initially, as it seems to be that way in the examples. However, I see lines like (WW) MGA: No matching Device section for instance (BusID PCI:1:5:0) found (WW) MGA: No matching Device section for instance (BusID PCI:0:11:0) found in both kdm.log and XFree86.?.log. Then, I figured, they should probably be the same in the BusID-field and PrefBusID, so I changed them to what they are now. That didn't seem to have any appreciable effect either, though. Then, I don't know what Primary Device is, but in XFree86.1.log (i.e. the log of the secondary card), I see (II) Primary Device is: PCI 01:05:0 Yep, this is the ID of the primary card (the G450), but I don't know if the secondary card should care... :-) Well, I was pretty much out of ideas at this point, but I got the bright (?) idea that I would kill the X1 server to see what happened. Wow, the other monitor came up, and I could type at it an use the mouse and everything! :-) The state of bliss was short, however, because the primary display had frozen... Again, the X server had run off, this time the X0 server. This was what top looked like at some point: PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 721 root 20 -10 54532 12m 1904 R 97.7 2.4 18:55.21 X0 881 root 5 -9 19772 9516 1856 S 1.0 1.8 0:09.79 X1 So, apparently, I can only run one at a time, and kdm will restart an X server if killed, and the other will run off wildly... Well, that's about all I've got. I'll list a few of the pitfalls I've been in and send it to Svetoslav as soon as I get around to do it, to help others avoid them. I'll be very thankful for any help you hackers can provide so I get this system up and running! 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: Svetoslav S. <sv...@gm...> - 2003-10-29 21:48:47
|
Hi Kjetil, that's the best bug report i ever got :-) and i'll have to reread it several times which might take me sometime :( in the mean time several notes 1.) you can not have DRI acceleration on more then one graphic card you have to disable it fully or leave it activated for one of the cards, and disabled for the other(s) -- two XF86Configs not sure, but this could be the reason for the problems 2.) you've written to Andreas, have you asked him which XFree-PrefBusID patch he uses for his XFree binaries ? i've the impression he uses the first version of the patch and the only valid way to activate multiple X servers is to pass on startup " -prefbusid ...." please check this two things, and i'll get back to you soon :-) best, svetljo PS. from kernel side everything seem pretty OK to me, but there is smth strange with X setup -- 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: Kjetil K. <kj...@kj...> - 2003-10-30 09:32:01
|
On Wednesday 29 October 2003 22:48, Svetoslav Slavtchev wrote: > Hi Kjetil, > that's the best bug report i ever got :-) Wow, cool! :-) > and i'll have to reread it several times > which might take me sometime :( Hehe, yeah, and you stayed up all night to read it! ;-) > in the mean time several notes OK! > 1.) > you can not have DRI acceleration on more then one graphic card Ah, OK. I turned it off. > you have to disable it fully or leave it activated for one of the > cards, and disabled for the other(s) -- two XF86Configs OK. I saw there are lines like in the autocreated config like #Option "DRI" # [<bool>] for each of the devices. Can I use them to control it? > not sure, but this could be the reason for the problems OK, I'll try! > 2.) > > you've written to Andreas, > have you asked him which XFree-PrefBusID patch he uses for > his XFree binaries ? Nope. I'll drop him a note. > i've the impression he uses the first version of the patch > and the only valid way to activate multiple X servers > is to pass on startup " -prefbusid ...." I see. I'll try that too! > please check this two things, > and i'll get back to you soon :-) Great! I got the other one too, I'll check the two above points first. (BTW, I'm on the list, so you don't need to CC) 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: Svetoslav S. <sv...@gm...> - 2003-10-30 11:19:01
|
Hi Kjetil, > > 1.) > > you can not have DRI acceleration on more then one graphic card > > Ah, OK. > > I turned it off. > > > you have to disable it fully or leave it activated for one of the > > cards, and disabled for the other(s) -- two XF86Configs > > OK. I saw there are lines like in the autocreated config like > #Option "DRI" # [<bool>] > for each of the devices. Can I use them to control it? not sure, but the best way to find out is to try :-) if it works you might use single configuration file and still have DRI activated for the G450 > > 2.) > > > > you've written to Andreas, > > have you asked him which XFree-PrefBusID patch he uses for > > his XFree binaries ? > > Nope. I'll drop him a note. > > > i've the impression he uses the first version of the patch > > and the only valid way to activate multiple X servers > > is to pass on startup " -prefbusid ...." > > I see. I'll try that too! > > > please check this two things, > > and i'll get back to you soon :-) > > Great! I got the other one too, I'll check the two above points first. > > (BTW, I'm on the list, so you don't need to CC) note taken :-) 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: Andreas S. <an...@sc...> - 2003-10-31 12:20:45
|
* Kjetil Kjernsmo (kj...@kj...) [031030 10:33]: > > you've written to Andreas, > > have you asked him which XFree-PrefBusID patch he uses for > > his XFree binaries ? > > Nope. I'll drop him a note. i am actually not sure which versin is on my website. i will upload the version 3 where one can specify the singlecard option in the config file. |
From: Svetoslav S. <sv...@gm...> - 2003-10-30 00:15:29
|
Hi again, second try :-) > I think I've been through all the pitfalls that one can possibly be > going into. Since I'm on Debian Sid, I've been bothering Andreas > Schuldei a lot. I'm using his great packages. do you use his kernel binaries ? have you cmpiled the kernel yourself ? do you have the kernel configuration file ? IIRC he used to build all input drivers as modules, but in your lsmod i see only the driver for usb controler no mention of input driver modules > To do the short version first: Everything seems to work fine, except > that when I try to get KDM to launch the two X servers, the first > server comes up fine, but the second does not. Instead, the server > takes the whole CPU, but apparently does nothing. If I kill the second > X-server with -9, it comes back up and then it runs fine on the second > screen, but the first locks up and the first X server runs along. > There are tons of messages like: > (EE) MGA(0): [dri] Idle timed out, resetting engine... > in my logs afterwards. > Now, for the details: > > My system has a Matrox G450 DH and Mystique card. Since I'm now going > for a 2.4 kernel (I'll (re)turn to 2.6 later, for sure!), I'm not > trying to use the second head of the G450. Rather, I'm trying to use > the Mystique for the second user. > > I have a Logitech Keyboard on the normal PS/2 port, then a cheap > keyboard on the usual mouse PS/2 port. I have two mice connected on the > USB ports. I might want to change this setup later, but I don't want to > buy a USB keyboard. > > One of the things Svetoslav asks for at > http://tldp.org/HOWTO/XFree-Local-multi-user-HOWTO/final_words.html is > contents of /proc/bus/input/devices, and since this sums up the setup > rather nice, I paste it here: > I: Bus=0011 Vendor=0001 Product=0002 Version=ab02 > N: Name="AT Set 2 keyboard" > P: Phys=isa0060/serio0/input0 > H: Handlers=kbd event0 > B: EV=120003 > B: KEY=4 2000000 8061f9 fbc9d621 efdfffdf ffefffff ffffffff fffffffe > B: LED=7 > > I: Bus=0011 Vendor=0001 Product=0002 Version=ab83 > N: Name="AT Set 2 keyboard" > P: Phys=isa0060/serio1/input0 > H: Handlers=kbd event1 > B: EV=120003 > B: KEY=4 2000000 8061f9 fbc9d621 efdfffdf ffefffff ffffffff fffffffe > B: LED=7 > > I: Bus=0003 Vendor=046d Product=c00e Version=1110 > N: Name="Logitech USB-PS/2 Optical Mouse" > P: Phys=usb-00:04.2-1/input0 > H: Handlers=mouse0 event2 > B: EV=7 > B: KEY=70000 0 0 0 0 0 0 0 0 > B: REL=103 > > I: Bus=0003 Vendor=09da Product=0006 Version=0001 > N: Name="A4Tech USB Optical Mouse" > P: Phys=usb-00:04.2-2/input0 > H: Handlers=mouse1 event3 > B: EV=7 > B: KEY=7f0000 0 0 0 0 0 0 0 0 > B: REL=103 > > BTW, /proc/bus/usb/devices does not exist on my system, this is somewhat strange you must have it /proc/bus/input/devices llocks perfectly OK do you have in /dev/input/ mouse0, mouse1, event0 to event3 ? >and > cat /proc/bus/console/*/* says > isa0060/serio0/input0 > isa0060/serio1/input0 > > This looks good, doesn't it? it does :-) > > (I'll paste in just the things that I assume is relevant, the rest I'll > attach. Hope that's OK. XFree86.[12].logs I had to trim, this mail is > large enough as it is.) > > I've configured Xinerama, and it works fine. I have preferred keeping > the whole X config in a single file, so I take the approach with > specifying different "layouts". I maintain four different layouts, > "Simple", "Xinerama", "X0" and "X1". The two latter are the ones for > the multi-user config. > > They look like this: > Section "ServerLayout" > Identifier "X0" > Screen 0 "Screen0" 0 0 > InputDevice "Mouse0" "CorePointer" > InputDevice "Keyboard0" "CoreKeyboard" > Option "PrefBusID" "1:5:0" > EndSection > Section "ServerLayout" > Identifier "X1" > Screen 0 "Screen1" 0 0 > InputDevice "Mouse1" "CorePointer" > InputDevice "Keyboard1" "CoreKeyboard" > Option "PrefBusID" "0:11:0" > EndSection > > > devfs is up and running, allthough there are some weird issues; > specifically mount doesn't report that devfs is mounted, whereas > /proc/mounts does. This put me off for a long time. smth debian specific ? have you installed devfsd ? does your bootloader line contains " devfs=mount", or if you have compiled your kernel have you enabled automatic mounting of devfs ? > It is easy enough to get X running on either screen with corresponding > keyboard and mouse. Also, I have been able to run it on both, > simultaneously, running > startx -- /usr/X11R6/bin/X1 :1 -layout X1 vt8 & startx -- /usr/X11R6/ > bin/X0 :0 -layout X0 vt7 > I could type on both keyboards, it showed on their respective screens. > Mice worked. > > So far, so good, it seems, but then I tried to get KDM to start them for > me, by adding the following lines to /etc/kde3/kdm/Xservers: > :0 local /usr/bin/nice -n -10 /usr/X11R6/bin/X0 :0 -layout X0 > -deferglyphs 16 vt7 > :1 local /usr/bin/nice -n -9 /usr/X11R6/bin/X1 :1 -layout X1 vt9 > may be try adding here "-prefbusid x:xx:x" for each card, another idea woud be to switch to gdm (cause it follows the order of starting X servers as specified) and try exchanging the start order in your command line first is started X1 then X0, with xdm/kdm you can never be sure which one will be started first, but with gdm it's up to you to specify the order > Booting up, things go wrong. X comes slowly up, but only on one screen. > Upon inspection, it is clear why it is so slow, because the X1 process > runs at 99% of the CPU according to top. Other than that, things are > normal on the display that is running, but obviously, the idea was to > have both... :-) > > In syslog, I find things like this: > Oct 29 16:04:15 owl kdm[720]: Hung in XOpenDisplay(:1), aborting > Oct 29 16:04:15 owl kdm[720]: server open failed for :1, giving up > Oct 29 16:04:15 owl kdm[686]: Display :1 cannot be opened > > > I didn't really have much of an idea on where to start, but I noticed > two things. One thing was that in kdm.log, I found lines: > (EE) MGA(0): Need at least 4608 kB video memory at this resolution, bit > depth i think i've got smth similar when i found out that DRI is not supported on multiple graphic cards (IIRC primary G550 , secondary SiS6326 but i think both X servers started the second without DRI of course) > (My G450 has 16MB, I don't know what the Mystique has). So, I tried > setting the bit depth to 8 and the resolution for both cards to > 800x600. That didn't help, and besides, I run Xinerama at 16 bpp and > 1280x960 and 1024x768 respectively, so that shouldn't suddenly be a > problem, I figured. Turned it back up. > > Then, I was in doubt whether I had entered the PCI bus id's correctly. > The autoconfig I ran when making the Xinerama config had them in the > form PCI:0:11:0, but in the examples, the PrefBusID strings did not > have the PCI:-prefix. I had left it like that initially, as it seems to > be that way in the examples. However, I see lines like > (WW) MGA: No matching Device section for instance (BusID PCI:1:5:0) > found > (WW) MGA: No matching Device section for instance (BusID PCI:0:11:0) > found this probably just information, that the driver found additional device which is not specified in the current layout waiting for more info, 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: Kjetil K. <kj...@kj...> - 2003-10-30 10:25:04
|
Yihaaaa! It works! disabling DRI and using the -prefbusid option did the trick! Thanks a lot! But then, since there are a few issues it would be good to understand, let's take it anyway...: On Thursday 30 October 2003 01:15, Svetoslav Slavtchev wrote: > Hi again, > > second try :-) > > > I think I've been through all the pitfalls that one can possibly be > > going into. Since I'm on Debian Sid, I've been bothering Andreas > > Schuldei a lot. I'm using his great packages. > > do you use his kernel binaries ? Nope! > have you cmpiled the kernel yourself ? Yes. 8 times... :-) > do you have the kernel configuration file ? Hereby attached! > IIRC he used to build all input drivers as modules, > but in your lsmod i see only the driver for usb controler > no mention of input driver modules Yes. I think I once read that if you build a kernel for a specific system, and there is something you use all the time, compile it into the kernel, if you use it just occasionally, compile it as a module. > > BTW, /proc/bus/usb/devices does not exist on my system, > > this is somewhat strange > you must have it OK... > > /proc/bus/input/devices llocks perfectly OK > do you have in /dev/input/ > mouse0, mouse1, event0 to event3 ? Yup: kjetil@owl:/dev/input> ls event0 event1 event2 event3 mice mouse0 mouse1 > > devfs is up and running, allthough there are some weird issues; > > specifically mount doesn't report that devfs is mounted, whereas > > /proc/mounts does. This put me off for a long time. > > smth debian specific ? Dunno. > have you installed devfsd ? Yup... BTW, here's one point you might want to add something like this in a Troubleshooting section of your HOWTO: Q: e2fsck complains I have a bad superblock after I enabled devfs. In fact, it appears all the superblocks are bad, and I'm scared. What do I do? A: Make sure devfsd is running. :-) This put me off for several days... > does your bootloader line contains " devfs=mount", > or if you have compiled your kernel have you enabled > automatic mounting of devfs ? I have automatic mounting enabled in the kernel. > may be try adding here "-prefbusid x:xx:x" for each card, Yup, as I said, it works! :-) > another idea woud be to switch to gdm (cause it follows > the order of starting X servers as specified) > and try exchanging the start order > in your command line first is started X1 then X0, > with xdm/kdm you can never be sure which one will be started > first, but with gdm it's up to you to specify the order Yup. Since it works now, I'll stick to kdm. > > I didn't really have much of an idea on where to start, but I > > noticed two things. One thing was that in kdm.log, I found lines: > > (EE) MGA(0): Need at least 4608 kB video memory at this resolution, > > bit depth > > i think i've got smth similar when i found out that DRI is not > supported on multiple graphic cards I see! > > (WW) MGA: No matching Device section for instance (BusID PCI:1:5:0) > > found > > (WW) MGA: No matching Device section for instance (BusID > > PCI:0:11:0) found > > this probably just information, > that the driver found additional device which is not specified in the > current > layout OK. I guess I might as well ask right now: I was thinking about buying a gadget where I can get two PS/2 ports from a USB-port, like this: http://www.clasohlson.no/images/products/N/hi/A/324024_Xw4.jpg I have no further specs on it, but do you guys think it would "work"? The idea is to use one of those for the second keyboard and mouse. 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: Svetoslav S. <sv...@gm...> - 2003-10-30 11:48:10
|
> Yihaaaa! It works! disabling DRI and using the -prefbusid option did the > trick! Thanks a lot! happy to hear it :-) so Andreas haven't updated his XFree binaries yet, i'll probably have to add it to the howto > Yes. I think I once read that if you build a kernel for a specific > system, and there is something you use all the time, compile it into > the kernel, if you use it just occasionally, compile it as a module. that's probably OK, but the question is whether the init scripts can live with it althougt what happens if you decide to change some hardware, filesystem ? it's more flexible to have most hardware compiled as module's > > > BTW, /proc/bus/usb/devices does not exist on my system, > > > > this is somewhat strange > > you must have it > > OK... > probably your init script doesn't like built in support for usb, this is the case for mandrake as well if usb is built in the kernel, not as modules, service usb doesn't start -- it doesn't mount usbfs and there might be some issues with certain usb drivers you probably have to mount it yourself then > > > > /proc/bus/input/devices llocks perfectly OK > > do you have in /dev/input/ > > mouse0, mouse1, event0 to event3 ? > > Yup: > kjetil@owl:/dev/input> ls > event0 event1 event2 event3 mice mouse0 mouse1 > > > > > devfs is up and running, allthough there are some weird issues; > > > specifically mount doesn't report that devfs is mounted, whereas > > > /proc/mounts does. This put me off for a long time. > > > > smth debian specific ? > > Dunno. > > > have you installed devfsd ? > > Yup... so everything here is OK > BTW, here's one point you might want to add something like this in a > Troubleshooting section of your HOWTO: > > Q: e2fsck complains I have a bad superblock after I enabled devfs. In > fact, it appears all the superblocks are bad, and I'm scared. What do I > do? > > A: Make sure devfsd is running. > > :-) This put me off for several days... it happened to me under debian too :-) but i don't think it's a good idea to add this to the howto, it's devfs issue and has nothing to do with bruby/ruby-2.6 and there is a link to the devfs howto > > > > does your bootloader line contains " devfs=mount", > > or if you have compiled your kernel have you enabled > > automatic mounting of devfs ? > > I have automatic mounting enabled in the kernel. > > > > may be try adding here "-prefbusid x:xx:x" for each card, > > Yup, as I said, it works! :-) so i'll have to add a not until Andreas updates the XFree packages > > another idea woud be to switch to gdm (cause it follows > > the order of starting X servers as specified) > > and try exchanging the start order > > in your command line first is started X1 then X0, > > with xdm/kdm you can never be sure which one will be started > > first, but with gdm it's up to you to specify the order > > Yup. Since it works now, I'll stick to kdm. > it's higly recomended to switch to gdm, as it does obey the specified start/stop order and the later is really important to avoid some lock ups with certain graphic cards on start/restart/ shut down of the X servers, and it does help to retain the VGA console > > I guess I might as well ask right now: I was thinking about buying a > gadget where I can get two PS/2 ports from a USB-port, like this: > http://www.clasohlson.no/images/products/N/hi/A/324024_Xw4.jpg > I have no further specs on it, but do you guys think it would "work"? > The idea is to use one of those for the second keyboard and mouse. no idea, if it works under non ruby kernel, it should work under bruby/ruby-2.6 too may be searching for this particular adapter "+ linux" on google will give you some answers best, svetljo PS. please check if putting Option "DRI" "false" for your second card works OK with DRI activated for the first card and drop a line (i'll add it to the howto if it works) -- 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: Helge H. <hel...@ai...> - 2003-10-30 14:40:05
|
On Thu, Oct 30, 2003 at 12:48:02PM +0100, Svetoslav Slavtchev wrote: > > > Yes. I think I once read that if you build a kernel for a specific > > system, and there is something you use all the time, compile it into > > the kernel, if you use it just occasionally, compile it as a module. > > that's probably OK, > but the question is whether the init scripts > can live with it > Debian seems to cope. I run a kernel with no module support at all. (The module interface changed late in 2.5.xx, modules were unavailable for a while anyway.) > althougt what happens if you decide to change some hardware, > filesystem ? > Compile a new kernel. You have to turn the machine off anyway to install non-hotplug hardware. You can precompile filesystems that you think you might need in the future. (Drivers too, but they tend to have boring probe timeouts when the hw isn't there.) > it's more flexible to have most hardware compiled as module's > Sure - and a lot more hassle if you're running experimental kernels, devfs, or drivers not yet supported by the distribution. > > > > BTW, /proc/bus/usb/devices does not exist on my system, > > > > > > this is somewhat strange > > > you must have it > > > > OK... > > > probably your init script doesn't like > built in support for usb, > this is the case for mandrake as well > if usb is built in the kernel, not as modules, > service usb doesn't start -- it doesn't mount > usbfs and there might be some issues with > certain usb drivers > you probably have to mount it yourself then > Put this in /etc/fstab: none /proc/bus/usb usbfs defaults 0 0 Helge Hafting |
From: Svetoslav S. <sv...@gm...> - 2003-10-30 14:51:16
|
> On Thu, Oct 30, 2003 at 12:48:02PM +0100, Svetoslav Slavtchev wrote: > > > > > Yes. I think I once read that if you build a kernel for a specific > > > system, and there is something you use all the time, compile it into > > > the kernel, if you use it just occasionally, compile it as a module. > > > > that's probably OK, > > but the question is whether the init scripts > > can live with it > > > Debian seems to cope. I run a kernel with no module support at all. > (The module interface changed late in 2.5.xx, modules were unavailable > for a while anyway.) i know, but now everything is OK > > althougt what happens if you decide to change some hardware, > > filesystem ? > > > Compile a new kernel. You have to turn the machine off anyway > to install non-hotplug hardware. You can precompile filesystems > that you think you might need in the future. (Drivers too, but they > tend to have boring probe timeouts when the hw isn't there.) well and how big is your kernel ? should you recompile the kernel all the time ? > > it's more flexible to have most hardware compiled as module's > > > Sure - and a lot more hassle if you're running experimental kernels, IMO that was the case with 2.5 only > devfs, or drivers not yet supported by the distribution. ??? > > > > > BTW, /proc/bus/usb/devices does not exist on my system, > > > > > > > > this is somewhat strange > > > > you must have it > > > > > > OK... > > > > > probably your init script doesn't like > > built in support for usb, > > this is the case for mandrake as well > > if usb is built in the kernel, not as modules, > > service usb doesn't start -- it doesn't mount > > usbfs and there might be some issues with > > certain usb drivers > > you probably have to mount it yourself then > > > Put this in /etc/fstab: > none /proc/bus/usb usbfs defaults 0 > 0 and you tell me that debian copes with it :-) shouldn't it be done automatically ? 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: Helge H. <hel...@ai...> - 2003-10-30 20:28:16
|
On Thu, Oct 30, 2003 at 03:51:10PM +0100, Svetoslav Slavtchev wrote: > > On Thu, Oct 30, 2003 at 12:48:02PM +0100, Svetoslav Slavtchev wrote: > > > > and how big is your kernel ? > should you recompile the kernel all the time ? > I do that anyway, I boot & use most of the development kernels. It is my small contribution to testing them. My kernels are 2.1MB. Modules wouldn't help much with memory consumption, because most of them would be loaded all the time anyway. > > > it's more flexible to have most hardware compiled as module's > > > > > Sure - and a lot more hassle if you're running experimental kernels, > IMO that was the case with 2.5 only > > devfs, or drivers not yet supported by the distribution. > ??? Module autoload if you try to access the device. Except when using devfs, because then the device node isn't present until the driver registers. A good devfsd setup fixes that - and the distribution may come with a good devfsd setup. "New" devices aren't usually there though. Setting them up isn't fun. Modules are great for distribution though. > > > > > > BTW, /proc/bus/usb/devices does not exist on my system, > > > > > > > > > > this is somewhat strange > > > > > you must have it > > > > > > > > OK... > > > > > > > probably your init script doesn't like > > > built in support for usb, > > > this is the case for mandrake as well > > > if usb is built in the kernel, not as modules, > > > service usb doesn't start -- it doesn't mount > > > usbfs and there might be some issues with > > > certain usb drivers > > > you probably have to mount it yourself then > > > > > Put this in /etc/fstab: > > none /proc/bus/usb usbfs defaults 0 > > 0 > > and you tell me that debian copes with it :-) > shouldn't it be done automatically ? It is a fs - something must mount it. /etc/fstab is a standard place. Of course debian ought to come with it, but I haven't had debian install a fstab for me for years. (Never reinstall, only upgrade ...) Helge Hafting |
From: Kjetil K. <kj...@kj...> - 2003-10-30 22:22:59
|
On Thursday 30 October 2003 12:48, Svetoslav Slavtchev wrote: > > Yihaaaa! It works! disabling DRI and using the -prefbusid option > > did the trick! Thanks a lot! > > happy to hear it :-) Yup! My girlfriend is right now happily editing her documents in OO, and all is well! :-) It also looks like I can't use this as an excuse to buy more RAM, because it's not yet swapping... > so Andreas haven't updated his XFree binaries yet, > i'll probably have to add it to the howto It could perhaps be just the DRI that was the problem? > > Yes. I think I once read that if you build a kernel for a specific > > system, and there is something you use all the time, compile it > > into the kernel, if you use it just occasionally, compile it as a > > module. > > that's probably OK, > but the question is whether the init scripts > can live with it OK! > > althougt what happens if you decide to change some hardware, > filesystem ? That doesn't happen _that_ often, and besides, when it happens, it was probably time to get the latest kernel anyway! :-) A kernel compile doesn't take more than 20 minutes, and being an astrophysicist, I'm used to processes running for weeks... :-) > it's more flexible to have most hardware compiled as module's Yup, undoubtedly. BTW, my bzipped kernel is currently about 1 MB. > > > > BTW, /proc/bus/usb/devices does not exist on my system, > > > > > > this is somewhat strange > > > you must have it > > > > OK... > > probably your init script doesn't like > built in support for usb, > this is the case for mandrake as well > if usb is built in the kernel, not as modules, > service usb doesn't start -- it doesn't mount > usbfs and there might be some issues with > certain usb drivers > you probably have to mount it yourself then OK, I'll try Helge's advice, if it doesn't work, I'll compile it as a module. > > BTW, here's one point you might want to add something like this in > > a Troubleshooting section of your HOWTO: > > > > Q: e2fsck complains I have a bad superblock after I enabled devfs. > > In fact, it appears all the superblocks are bad, and I'm scared. > > What do I do? > > > > A: Make sure devfsd is running. > > > > :-) This put me off for several days... > > it happened to me under debian too :-) > but i don't think it's a good idea to add this to the howto, > it's devfs issue and has nothing to do with bruby/ruby-2.6 > and there is a link to the devfs howto Yeah, well, you have spelled it out clearly, that devfsd is needed, so I guess I can only blame myself. But when something like that happens to a newbie like me, the system is suddenly unbootable and the first thing you think about is "data corruption", a newbie is really put off... :-) The problem is, and why it took me so long, was the lack of keywords... I didn't realize it was a devfs problem, because I enabled devfs when booting the backstreet ruby kernel the first time. Besides, my HD has had a lot of problems (a physical problem actually, the IDE cable falls out if I just barely touch the chassis :-( ), so I was quite sure it was a kernel/hardware problem. It wasn't untill I tried booting my non-bruby 2.4.22 kernel with devfs I realized that was relevant. Then adding "devfs" to my google search for "e2fsck bad superblock" gave the answer immediately. So, I thought it could perhaps fit in an appendix? Of course, this kind of stuff isn't really what newbies should be doing, but then... :-) > it's higly recomended to switch to gdm, > as it does obey the specified start/stop order > and the later is really important to avoid some lock ups > with certain graphic cards on start/restart/ shut down > of the X servers, and it does help to retain the VGA console OK! I'll probably switch to GDM when I get time to look more into 2.6! > > > I guess I might as well ask right now: I was thinking about buying > > a gadget where I can get two PS/2 ports from a USB-port, like this: > > http://www.clasohlson.no/images/products/N/hi/A/324024_Xw4.jpg I > > have no further specs on it, but do you guys think it would "work"? > > The idea is to use one of those for the second keyboard and mouse. > > no idea, > if it works under non ruby kernel, it should work under > bruby/ruby-2.6 too > > may be searching for this particular adapter "+ linux" on google will > give you some answers Yup, there are stuff like that certfied for Linux, (as Aivils pointed out). I'll probably go out an buy one tomorrow. I was concerned if it could for example just send all its output to e.g. /dev/input/event2, and it would be both mouse and keyboard and difficult to tell apart. Mmmm, it was a whole lot of wires now... I figured it would be cool if I could have something like a Bluetooth box on my desktop with RS232, PS/2 and USB connections, and symmetric crypto between that box and a bluetooth PCI card in the machine... Anybody know if something like that exists? :-) > please check if putting > Option "DRI" "false" > for your second card works OK > with DRI activated for the first card > and drop a line > (i'll add it to the howto if it works) Yup, I'll do that once I'm alone on the box! :-) 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: Kjetil K. <kj...@kj...> - 2003-10-31 16:42:36
|
Hi there! Good news! On Thursday 30 October 2003 12:48, Svetoslav Slavtchev wrote: > > > > BTW, /proc/bus/usb/devices does not exist on my system, > > probably your init script doesn't like > built in support for usb, > this is the case for mandrake as well > if usb is built in the kernel, not as modules, > service usb doesn't start -- it doesn't mount > usbfs and there might be some issues with > certain usb drivers > you probably have to mount it yourself then Helge's advice worked well! Here is my /proc/bus/usb/devices: T: Bus=02 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=12 MxCh= 2 B: Alloc= 0/900 us ( 0%), #Int= 0, #Iso= 0 D: Ver= 1.00 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 P: Vendor=0000 ProdID=0000 Rev= 0.00 S: Product=USB UHCI Root Hub S: SerialNumber=d000 C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub E: Ad=81(I) Atr=03(Int.) MxPS= 8 Ivl=255ms T: Bus=01 Lev=00 Prnt=00 Port=00 Cnt=00 Dev#= 1 Spd=12 MxCh= 2 B: Alloc=186/900 us (21%), #Int= 2, #Iso= 0 D: Ver= 1.00 Cls=09(hub ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 P: Vendor=0000 ProdID=0000 Rev= 0.00 S: Product=USB UHCI Root Hub S: SerialNumber=d400 C:* #Ifs= 1 Cfg#= 1 Atr=40 MxPwr= 0mA I: If#= 0 Alt= 0 #EPs= 1 Cls=09(hub ) Sub=00 Prot=00 Driver=hub E: Ad=81(I) Atr=03(Int.) MxPS= 8 Ivl=255ms T: Bus=01 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 2 Spd=1.5 MxCh= 0 D: Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 P: Vendor=046d ProdID=c00e Rev=11.10 S: Manufacturer=Logitech S: Product=USB-PS/2 Optical Mouse C:* #Ifs= 1 Cfg#= 1 Atr=a0 MxPwr= 98mA I: If#= 0 Alt= 0 #EPs= 1 Cls=03(HID ) Sub=01 Prot=02 Driver=hid E: Ad=81(I) Atr=03(Int.) MxPS= 4 Ivl=10ms T: Bus=01 Lev=01 Prnt=01 Port=01 Cnt=02 Dev#= 3 Spd=1.5 MxCh= 0 D: Ver= 1.10 Cls=00(>ifc ) Sub=00 Prot=00 MxPS= 8 #Cfgs= 1 P: Vendor=09da ProdID=0006 Rev= 0.01 S: Manufacturer=A4Tech S: Product=USB Optical Mouse C:* #Ifs= 1 Cfg#= 1 Atr=a0 MxPwr=100mA I: If#= 0 Alt= 0 #EPs= 1 Cls=03(HID ) Sub=01 Prot=02 Driver=hid E: Ad=81(I) Atr=03(Int.) MxPS= 4 Ivl=10ms But now, I've bought the USB to PS/2 converter, so this will change soon, I guess! > please check if putting > Option "DRI" "false" > for your second card works OK > with DRI activated for the first card > and drop a line > (i'll add it to the howto if it works) It works! :-) The log says: (II) MGA(0): [DRI] installation complete Cool, eh? 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 |