Re: [Cooker] Do not renice X by default
Never mind! I made a mistake in configuring the kernel (forgot to put CONFIG_VIDEO_SELECT=y ... duh!) and everthing works after recompiling with the option selected. I'm so ashamed. Apart from that, what can I say...kernel 2.6 rocks! Le 31 Octobre 2003 08:25, Élie Charest a écrit : > Well, vga=791 (my usual mode) doesn't work at all. I get a blank screen > during boot, and a garbled screen if I switch to a console after boot. I > haven't tried other vga modes and will do so eventually, but somehow I'm > pretty sure that if 791 doesn't work, others won't either. > > I wonder if it might be related to the 1GB Ram + 128MB Video card bug that > plagued 2.4 kernels until 2.4.22... -- Élie Charest [EMAIL PROTECTED] http://archie.homelinux.net:8080
Re: [Cooker] Do not renice X by default
Le 31 Octobre 2003 07:31, Con Kolivas a écrit : > On Fri, 31 Oct 2003 16:34, Élie Charest wrote: > > I am now happily testing out my homemade test9 kernel and so far, so > > good! The only thing not working seems to be the framebuffer console (I'm > > in 25x80 text mode now). In fact, I finally put "vga=ask" in the > > 2.6.0-test9 entry in lilo.conf, and it didn't even ask me anything - it > > went straight to 25x80 text mode. > > the vga=ask only works at the lilo prompt, not as an append. you need to > specify the exact vga mode in the lilo.conf that you want. Well, vga=791 (my usual mode) doesn't work at all. I get a blank screen during boot, and a garbled screen if I switch to a console after boot. I haven't tried other vga modes and will do so eventually, but somehow I'm pretty sure that if 791 doesn't work, others won't either. I wonder if it might be related to the 1GB Ram + 128MB Video card bug that plagued 2.4 kernels until 2.4.22... -- Élie Charest [EMAIL PROTECTED] http://archie.homelinux.net:8080
Re: [Cooker] Do not renice X by default
On Fri, 31 Oct 2003 16:34, Élie Charest wrote: > Le 30 Octobre 2003 00:05, Con Kolivas a écrit : > > > By the way, adding in the supermount patch for test8 to test9 seems to > > > work fine - in fact, I haven't had any troubles except for my usb > > > scanner not being detected at all. > > > > Chances are you're not loading the usb module correctly (some have > > changed names) or you're not mounting the new usbfs. Put this into your > > /etc/fstab > > > > > > none /proc/bus/usb usbfs defaults 0 0 > > I tried and still my scanner wouldn't work. The weird thing is that it > would appear in the /proc/bus/usb/devices and in usbview. > > On a wild guess (while checking commands that started with "usb" - thank > god for autocomplete!) I tried usbmodules --device /proc/bus/usb/003/004 > (the adress given in /proc/bus/usb/devices) and I got "scanner" as the > output. Checking "man usbmodules" helped me understand what the command did > (yeah, I know I'm not supposed to type in random commands as root). In a > flash of insight I tried "modprobe scanner" and it worked. yay > > I am now happily testing out my homemade test9 kernel and so far, so good! > The only thing not working seems to be the framebuffer console (I'm in > 25x80 text mode now). In fact, I finally put "vga=ask" in the 2.6.0-test9 > entry in lilo.conf, and it didn't even ask me anything - it went straight > to 25x80 text mode. the vga=ask only works at the lilo prompt, not as an append. you need to specify the exact vga mode in the lilo.conf that you want. > > Oh well, that's a minor annoyance. I'll continue testing test9 for a > while... Cheers, Con
Re: [Cooker] Do not renice X by default
>> names) or you're not mounting the new usbfs. Put this into your /etc/fstab >> : >> >> none /proc/bus/usb usbfs defaults 0 0 > I tried and still my scanner wouldn't work. The weird thing is that it would > appear in the /proc/bus/usb/devices and in usbview. Do you mean it works under 2.4? Coldplugging hot-pluggable devices does not work and should not work under 2.4, at least it fails for usb-storage. I think, scanner driver is loaded statically out of /etc/init.d/usb depending on /etc/sysconfig/usb settings - could you check if it really works? Chances are driver name has changed between 2.4 and 2.6 so this loading under fixed name fails, please compare drivers for 2.4 and 2.6. Svetljo has fixed hotplug RPM that supports 2.6, if you enable its initscript it will emulate hotplugging for all devices present at boot and will load drivers automatically, no manual config is needed. -andrey
Re: [Cooker] Do not renice X by default
Le 30 Octobre 2003 00:05, Con Kolivas a écrit : > > > By the way, adding in the supermount patch for test8 to test9 seems to > > work fine - in fact, I haven't had any troubles except for my usb scanner > > not being detected at all. > > Chances are you're not loading the usb module correctly (some have changed > names) or you're not mounting the new usbfs. Put this into your /etc/fstab > : > > none /proc/bus/usb usbfs defaults 0 0 I tried and still my scanner wouldn't work. The weird thing is that it would appear in the /proc/bus/usb/devices and in usbview. On a wild guess (while checking commands that started with "usb" - thank god for autocomplete!) I tried usbmodules --device /proc/bus/usb/003/004 (the adress given in /proc/bus/usb/devices) and I got "scanner" as the output. Checking "man usbmodules" helped me understand what the command did (yeah, I know I'm not supposed to type in random commands as root). In a flash of insight I tried "modprobe scanner" and it worked. I am now happily testing out my homemade test9 kernel and so far, so good! The only thing not working seems to be the framebuffer console (I'm in 25x80 text mode now). In fact, I finally put "vga=ask" in the 2.6.0-test9 entry in lilo.conf, and it didn't even ask me anything - it went straight to 25x80 text mode. Oh well, that's a minor annoyance. I'll continue testing test9 for a while... -- Élie Charest [EMAIL PROTECTED] http://archie.homelinux.net:8080
Re: [Cooker] Do not renice X by default
On Thu, 2003-10-30 at 08:42, Olivier Blin wrote: > X isn't niced by gdm (just have a look with top). > It was quite incoherent to nice X from xdm and not from gdm ... Oops, okay... ignore my last post. I love gdm. :-) Austin -- Austin Acton Synthetic Organic Chemist, Teaching Assistant, Ph.D. Candidate Department of Chemistry, York University, Toronto MandrakeLinux Volunteer Developer, homepage: www.groundstate.ca
Re: [Cooker] Do not renice X by default
> > > Le 29 Octobre 2003 22:49, Con Kolivas a écrit : > > > > A couple of years > > > > ago when I was subscribed to this list I suggested renicing X by > > default > > > > to -10 and noticed that it was done on the following release by > > default. Hmm, although X is niced to -10 in /etc/X11/xdm/Xservers, top indicates that it's currently running at nice 0. ? Austin -- Austin Acton Synthetic Organic Chemist, Teaching Assistant, Ph.D. Candidate Department of Chemistry, York University, Toronto MandrakeLinux Volunteer Developer, homepage: www.groundstate.ca
Re: [Cooker] Do not renice X by default
> any idea about gdm ? > it's not set in /etc/X11/gdm/gdm.conf X isn't niced by gdm (just have a look with top). It was quite incoherent to nice X from xdm and not from gdm ... -- Olivier Blin
Re: [Cooker] Do not renice X by default
> > You need the kernel source of the kernel you want the drivers > > compiled for. By default it looks for the kernel in /usr/src/linux > > so I usually symlink that to my latest kernel build directory. > > Sorry for not stating so: I do have kernel-2.6-source of the same > version installed. I can actually compile the driver - it's the > installation that doesn't (when I do depmod). I get something like > "wrong module type" or something like that. Hi, You should have a look at these threads : http://www.mail-archive.com/[EMAIL PROTECTED]/msg114289.html http://www.ussg.iu.edu/hypermail/linux/kernel/0307.1/1162.html I'm not sure if I can do something in the kernel package so that nvidia drivers compile out of the box. Maybe I could add a script that does all this stuff automatically. Regards -- Olivier Blin
Re: [Cooker] Do not renice X by default
On Thu, 30 Oct 2003 19:29, [EMAIL PROTECTED] wrote: > On Thu, 30 Oct 2003, Con Kolivas wrote: > > Hi all > > > > I'm a long time Mandrake user and am the person responsible for the base > > patch in the multimedia kernel and all the interactivity changes that > > have gone into the 2.6 development kernel (from 2.6.0-test6). A couple of > > years ago when I was subscribed to this list I suggested renicing X by > > default to -10 and noticed that it was done on the following release by > > default. That was a recommendation based on the default kernel's > > scheduler inability to make X smooth enough under load. > > > > However I am going to have to recommend reversing that change now as the > > new kernel has been tuned to allow good performance of X at nice 0. The > > new O(1) scheduler is far more aggressive with treatment of priorities > > and has much larger timeslices. Giving X a priority of -10 will make it > > cause unnecessary scheduling latencies for tasks that use even small > > amounts of cpu such as audio playback. In a nutshell this means that > > renicing X will make audio skip with a 2.6 kernel on even modern > > hardware. > > I know, I have been thinking longer about adding some extra init.d script > for multimedia kernel to renice X when booting into kernel-mm, and since I > doubt 2.6 will be perfect for everybody by the time 10.0 comes out it > might be a good idea to renice X only when the new scheduler is used. Con, > do you know of any quick way to check in bash/perl whether or not it is > useful to nice X to -10? Or should we just check for kernel > 2.6 or > kernel= kernel-multimedia? Simple enough; don't renice with any kernels. The advantage of nice -10 X on vanilla 2.4 is too small to warrant such confusing setups and scripts all over the place. Con
Re: [Cooker] Do not renice X by default
On Thu, 30 Oct 2003, Con Kolivas wrote: > Hi all > > I'm a long time Mandrake user and am the person responsible for the base patch > in the multimedia kernel and all the interactivity changes that have gone > into the 2.6 development kernel (from 2.6.0-test6). A couple of years ago > when I was subscribed to this list I suggested renicing X by default to -10 > and noticed that it was done on the following release by default. That was a > recommendation based on the default kernel's scheduler inability to make X > smooth enough under load. > > However I am going to have to recommend reversing that change now as the new > kernel has been tuned to allow good performance of X at nice 0. The new O(1) > scheduler is far more aggressive with treatment of priorities and has much > larger timeslices. Giving X a priority of -10 will make it cause unnecessary > scheduling latencies for tasks that use even small amounts of cpu such as > audio playback. In a nutshell this means that renicing X will make audio skip > with a 2.6 kernel on even modern hardware. I know, I have been thinking longer about adding some extra init.d script for multimedia kernel to renice X when booting into kernel-mm, and since I doubt 2.6 will be perfect for everybody by the time 10.0 comes out it might be a good idea to renice X only when the new scheduler is used. Con, do you know of any quick way to check in bash/perl whether or not it is useful to nice X to -10? Or should we just check for kernel > 2.6 or kernel= kernel-multimedia? d. > > Please CC me in any replies as I'm no longer subscribed to cooker. > > Cheers, > Con > >
Re: [Cooker] Do not renice X by default
> On Thu, 30 Oct 2003 15:44, Élie Charest wrote: > > Le 29 Octobre 2003 22:49, Con Kolivas a écrit : > > > Hi all > > > > > > A couple of years > > > ago when I was subscribed to this list I suggested renicing X by > default > > > to -10 and noticed that it was done on the following release by > default. > > > That was a recommendation based on the default kernel's scheduler > > > inability to make X smooth enough under load. > > > However I am going to have to recommend reversing that change now as > the > > > new kernel has been tuned to allow good performance of X at nice 0. > The > > > new O(1) scheduler is far more aggressive with treatment of priorities > > > and has much larger timeslices. Giving X a priority of -10 will make > it > > > cause unnecessary scheduling latencies for tasks that use even small > > > amounts of cpu such as audio playback. In a nutshell this means that > > > renicing X will make audio skip with a 2.6 kernel on even modern > > > hardware. this probably apllies for kernel-multimedia too (IIRC it includes 0(1) scheduler + your changes) > > Very astute observation. I was wondering why the gain in responsiveness > > wasn't as noticeable as I'd hoped... :-) > > > > Where do I set X's nice default level again? It's been a while. > > mandrake 9.2 does it here: > /etc/X11/xdm/Xservers any idea about gdm ? it's not set in /etc/X11/gdm/gdm.conf > > and removing the nice -10 part of the command should be enough. > > > While we're on kernel 2.6, I've noticed something with the test > releases: > > whenever I try to install 2.6 through the contrib RPMs (test5, then > test8) > > I can never get NVIDIA to compile and install (using the method from > > www.minion.de). Actually the module will compile but not load. However, > > when I compile test9 from kernel.org and try it, I can install the > nvidia > > driver fine. Somehow there's a setting in the 2.6 rpms from Mandrake > that > > seems to cause a problem with the nvidia drivers. > > You need the kernel source of the kernel you want the drivers compiled > for > . > By default it looks for the kernel in /usr/src/linux so I usually symlink > > that to my latest kernel build directory. > > > By the way, adding in the supermount patch for test8 to test9 seems to > wo > rk > > fine - in fact, I haven't had any troubles except for my usb scanner not > > being detected at all. > > Chances are you're not loading the usb module correctly (some have changed > > names) or you're not mounting the new usbfs. Put this into your /etc/fstab > : > > none /proc/bus/usb usbfs defaults 0 0 could it be a init script problem ? at least for test5 cooker binaries most drivers are compiled in and the init scripts expect everything as modules ( it took me lot of time to find out to find out why alsa was allways failing to start ) IIRC even in 2.4 if usb support is compiled in service usb wouldn't start and this service mounts usbfs 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! +++
Re: [Cooker] Do not renice X by default
Le 30 Octobre 2003 00:05, Con Kolivas a écrit : > On Thu, 30 Oct 2003 15:44, Élie Charest wrote: > > Where do I set X's nice default level again? It's been a while. > > mandrake 9.2 does it here: > /etc/X11/xdm/Xservers Thanks! > You need the kernel source of the kernel you want the drivers compiled > for. By default it looks for the kernel in /usr/src/linux so I usually > symlink that to my latest kernel build directory. Sorry for not stating so: I do have kernel-2.6-source of the same version installed. I can actually compile the driver - it's the installation that doesn't (when I do depmod). I get something like "wrong module type" or something like that. > Chances are you're not loading the usb module correctly (some have changed > names) or you're not mounting the new usbfs. Put this into your /etc/fstab It must be usbfs, as I don't have that line in my /etc/fstab. I'll give it a try tomorrow - it's past bedtime here in Montreal. Thanks again, -- Élie Charest [EMAIL PROTECTED] http://archie.homelinux.net:8080
Re: [Cooker] Do not renice X by default
On Thu, 30 Oct 2003 15:44, Élie Charest wrote: > Le 29 Octobre 2003 22:49, Con Kolivas a écrit : > > Hi all > > > > A couple of years > > ago when I was subscribed to this list I suggested renicing X by default > > to -10 and noticed that it was done on the following release by default. > > That was a recommendation based on the default kernel's scheduler > > inability to make X smooth enough under load. > > However I am going to have to recommend reversing that change now as the > > new kernel has been tuned to allow good performance of X at nice 0. The > > new O(1) scheduler is far more aggressive with treatment of priorities > > and has much larger timeslices. Giving X a priority of -10 will make it > > cause unnecessary scheduling latencies for tasks that use even small > > amounts of cpu such as audio playback. In a nutshell this means that > > renicing X will make audio skip with a 2.6 kernel on even modern > > hardware. > > Very astute observation. I was wondering why the gain in responsiveness > wasn't as noticeable as I'd hoped... :-) > > Where do I set X's nice default level again? It's been a while. mandrake 9.2 does it here: /etc/X11/xdm/Xservers and removing the nice -10 part of the command should be enough. > While we're on kernel 2.6, I've noticed something with the test releases: > whenever I try to install 2.6 through the contrib RPMs (test5, then test8) > I can never get NVIDIA to compile and install (using the method from > www.minion.de). Actually the module will compile but not load. However, > when I compile test9 from kernel.org and try it, I can install the nvidia > driver fine. Somehow there's a setting in the 2.6 rpms from Mandrake that > seems to cause a problem with the nvidia drivers. You need the kernel source of the kernel you want the drivers compiled for. By default it looks for the kernel in /usr/src/linux so I usually symlink that to my latest kernel build directory. > By the way, adding in the supermount patch for test8 to test9 seems to work > fine - in fact, I haven't had any troubles except for my usb scanner not > being detected at all. Chances are you're not loading the usb module correctly (some have changed names) or you're not mounting the new usbfs. Put this into your /etc/fstab : none /proc/bus/usb usbfs defaults 0 0 Cheers, Con
Re: [Cooker] Do not renice X by default
On Wednesday 29 October 2003 10:44 pm, Élie Charest wrote: > Le 29 Octobre 2003 22:49, Con Kolivas a écrit : > > Hi all > > > > A couple of years > > ago when I was subscribed to this list I suggested renicing X by default > > to -10 and noticed that it was done on the following release by default. > > That was a recommendation based on the default kernel's scheduler > > inability to make X smooth enough under load. > > However I am going to have to recommend reversing that change now as the > > new kernel has been tuned to allow good performance of X at nice 0. The > > new O(1) scheduler is far more aggressive with treatment of priorities > > and has much larger timeslices. Giving X a priority of -10 will make it > > cause unnecessary scheduling latencies for tasks that use even small > > amounts of cpu such as audio playback. In a nutshell this means that > > renicing X will make audio skip with a 2.6 kernel on even modern > > hardware. > > Very astute observation. I was wondering why the gain in responsiveness > wasn't as noticeable as I'd hoped... :-) > > Where do I set X's nice default level again? It's been a while. > > While we're on kernel 2.6, I've noticed something with the test releases: > whenever I try to install 2.6 through the contrib RPMs (test5, then test8) > I can never get NVIDIA to compile and install (using the method from > www.minion.de). Actually the module will compile but not load. However, > when I compile test9 from kernel.org and try it, I can install the nvidia > driver fine. Somehow there's a setting in the 2.6 rpms from Mandrake that > seems to cause a problem with the nvidia drivers. > > By the way, adding in the supermount patch for test8 to test9 seems to work > fine - in fact, I haven't had any troubles except for my usb scanner not > being detected at all. The file is /etc/X11/xdm/Xservers Changing the current line from :0 local /bin/nice -n -10 /usr/X11R6/bin/X -deferglyphs 16 to :0 local /usr/X11R6/bin/X -deferglyphs 16 should do it Paul
Re: [Cooker] Do not renice X by default
Le 29 Octobre 2003 22:49, Con Kolivas a écrit : > Hi all > > A couple of years > ago when I was subscribed to this list I suggested renicing X by default to > -10 and noticed that it was done on the following release by default. That > was a recommendation based on the default kernel's scheduler inability to > make X smooth enough under load. > However I am going to have to recommend reversing that change now as the > new kernel has been tuned to allow good performance of X at nice 0. The new > O(1) scheduler is far more aggressive with treatment of priorities and has > much larger timeslices. Giving X a priority of -10 will make it cause > unnecessary scheduling latencies for tasks that use even small amounts of > cpu such as audio playback. In a nutshell this means that renicing X will > make audio skip with a 2.6 kernel on even modern hardware. Very astute observation. I was wondering why the gain in responsiveness wasn't as noticeable as I'd hoped... :-) Where do I set X's nice default level again? It's been a while. While we're on kernel 2.6, I've noticed something with the test releases: whenever I try to install 2.6 through the contrib RPMs (test5, then test8) I can never get NVIDIA to compile and install (using the method from www.minion.de). Actually the module will compile but not load. However, when I compile test9 from kernel.org and try it, I can install the nvidia driver fine. Somehow there's a setting in the 2.6 rpms from Mandrake that seems to cause a problem with the nvidia drivers. By the way, adding in the supermount patch for test8 to test9 seems to work fine - in fact, I haven't had any troubles except for my usb scanner not being detected at all. -- Élie Charest [EMAIL PROTECTED] http://archie.homelinux.net:8080