With my particular via chipset (apollo 133a pro), 2.4 doesn’t work with my USB controller on SMP systems (which I have). You can get it to work with turning apic off on the kernel command line, but then X doesn’t seem to get input from my mouse and keyboard when I do that.
With test6, this problem seemed rectified, I could actually use my USB gamepad! Now with test7, I can’t use it anymore.
Guess I should put in a bug report. Other than this, I really haven’t had any trouble with test6 or test7. Windows seem to drag across my screen a tiny bit better with test7 than with test6. Looks like 2.6 is on the right track.
You can create your own. In fact, it’s pretty easy. “apt-get install kernel-package”. Untar the source, change to that directory, set your options and then: make-kpkg kernel_image, and in the below directory (cd ..), you’ll have a deb.
If you don’t know what you’re doing, then suspect you’ll be waiting a while till there’s an official debian package for a 2.6 kernel.
I’m using the 2.6.0-test series in my desktop machine since test1 and I’m very happy with it. It is fastest *and* runs more smoothly that 2.4 series. I have noticed many lags in the UI, skips in XMMS in the earlier test1-test3 even without compiling with preemptive, but as from test6 those problems seems to be corrected.
All my drivers are working, and surprisingly (to me!) I could use the ide-scsi driver and burn some cds. The post-halloween docs announced that this modules was broken in the 2.5 series and I though they were already.
It really worths to test the kernel. It’s stabilizing and will be the best linux ever, IMHO.
ide-scsi still works, but you don’t need to use it at all. Just get an updated cdrecord package and you can simply use cdrecord dev=/dev/hdc (for example) rather than dev=0,1,0 or whatever your case is. There is no reason to use ide-scsi for burning anymore, it was a hack. Also the reason test6 runs so smoothly is because Con Kolivas’s interactivity patches were merged in that release 🙂 Best Desktop Kernel Ever.
Yeah, a mate of mine told me to compile the kernel. At least then it is optimised for my CPU.
It’s just that set the options part… last time there was wayy… to many options and I knew maybe what 30% of them were which is why I tend to prefer installable images.
Anyway I will probably just it out on a PII I have here before I touch my main system.
> If you can’t handle that many options then you aren’t suposed to be compiling your own kernel.
That’s not very helpful. If you can’t handle that many options, then you just need to spend a little bit more time getting familiar with what’s in there.
At first I had trouble with it, but now it’s dead simple.
Just give it an hour of your time when you can – much better than waiting/checking for an image of it to appear in Debian unstable.
I knew it spelled trouble, but it was cheap. My G400 died and i needed to get a quick replacement with passive cooling. I opted for a nvidia fx 5200, big mistake.
Maybe their drivers are good enough for stable kernels but going through the hassle to patch them with the minion-patch each test kernel is just too much. I should have shelled out the 10 Euro more for an ATI R9000 or a Matrox G550. Grrr. Slap slap slap.
madmax: There should be an option for nforce2 boards in the kernel. Did you try it? And why is your patched video driver module running but not mine? Argh. (i am not expecting any useful answer, thank you, just venting frustration)
Yeah I have selected the relevant options for the nforce2 board. Have just installed 2.6-test7. So far so good. No freezes. Although now that i said it, I’m bound to get one. LOL.
Also I have a patched NVIDIA graphics driver that seems to work fine for the test7 kernel. if anyones interested let me know and more importantly how i could get it to ya.
If the typical distro did not already have a pre-compiled kernel or if the kernel choices were very limited and did not receive updates, then this would be a very l33t crap response.
However, if someone says that the average person should not worry or have to worry about compiling their own kernel. Then I say yes they should not.
If someone says to me that they are downloading a test kernel from an upcoming release and they were overwhelmed by the number of choices, I would say do not compile your own kernel then.
Use the one the distro came with. Why is that so l33t?
We are not talking about the gui preferences in an email app but the kernel for god’s sake. Yes, there should be a ton of wild ass options. Otherwise, the complaint would how limited the kernel is to hardware support and options. Making your way through the options is not that daunting it just takes time and a bit of patience in reading the descriptions.
” you compile kernel 2.6 can you use your old .config file from a 2.4 kernel or is it better to start from zero and select the options you want?”
It may be better to start from scratch, but importing my old .config worked perfectly with the early 2.6 version. Maybe this compatibilty will be dropped later, I don’t know.
You can still use make oldconfig.. but you have to make sure you enable some options for VT stuff, otherwise you wont be able to see anything (or use your keyboard) It’s in dave jones’ FAQ about testing 2.5… cant remember where it is, google.
Yah, you definitely should have gone for a 9000pro.
It has the advantage of having both open source and binary opengl driver support, and additionally its 2d quality is about untouchable. And they’re cheap too!
Yah, you definitely should have gone for a 9000pro.
It has the advantage of having both open source and binary opengl driver support, and additionally its 2d quality is about untouchable. And they’re cheap too!
I’ve tried installing the Mandrake RPMs for 2.6-test5 but so far I haven’t been able to successfully compile and install NVIDIA’s driver module (yes, I have installed the 2.6-test5 kernel-source package). I’ve tried the instructions contained on this webiste:
but while I can compile the module, I still can’t modprobe nvidia.o (or nvidia.ko, as it happens). I either get a “wrong kernel headers” or “wrong module type” depending on the method I use.
Anybody had any luck using the NVIDIA drivers with kernel 2.6?
The names for the USB modules are different. This affects /etc/modprobe.conf (look for “usb-controller”) and, once again, /etc/rc.sysinit. I had to rename all instances of “keybdev” into “usbkbd” and “mousedev” into “usbmouse”. In addition, I replaced all instances of “/proc/bus/usb” with “/sys/bus/usb”. (Also do this last part in /etc/init.d/halt.)
>>Yah, you definitely should have gone for a 9000pro.<<
umh, the Pro is not cooled passively. Surprisingly i find the 2D quality of the nvidia fx 5200 not too bad, considering that i used a Matrox before. *sigh*
I compiled 2.6-test7 this morning on my development Red Hat 9 box. I ran ApacheBench (from another machine to hit the test machine) last night before the kernel upgrade to have comparison numbers. This morning after compiling 2.6test7 and rebooting Apache is now reporting that it can only handle about 80% of the requests compared to kernel 2.4.22
Last night I ran 4 test over a period of 1 hour, the same today and the numbers are relatively the same … 20% loss in #request per second.
As far as the ‘desktop’ experience, I honestly haven’t noticed the difference that people claim.
Is anyone out there finding kernel 2.6 not to be what other have made it out to be … a ‘performance champ’?
I’ve noticed some slight performance increases on the console side…. and modest increases in desktop performance…. nothing I have in benchmark scores… this is purely a subjective *feeling*…. but yeah, it’s not quite as much as I had hoped for…nor as much as some of the past posts have raved about…. but it feels noticable… [for reference, i’m running a 1GHz Athlon-Tbird, ASUS A7V, 512MB PC133]
With regard to the NVidia driver patch…. I was concerned about the driver compiling w/ the new kernel… but i decided to give whirl in its untainted, vanilla state….. and, surprisingly, it gave me no problems….. no patches needed, and i had full OpenGL acceleration…
*confused*… is this driver patch specific to certain models of the Geforce? mine was a Geforce2-Ultra 64MB….(I should note that my experiences are specific to tests 5 and 6, not yet test7)
P III – 800Mhz, 256 RAM, 40 GB 7200 rpm drive … honestly, with 2.6test7 – running ApacheBench (ab) again it still is not as fast as running 2.4.22 with init=3 and httpd having nice=-20
On a single processor machine the actual raw output for the 2.6 kernel will be less than a 2.4 kernel. Not a great deal less, but less. Where you do have improvements is in multi-processer support, big iron features (such as NUMA), context switching and low-latency support. Due to some of the improvements though, for desktop use the “feel” of the computer will overall seem more responsive, especially with the mm patches in place.
installed test7 today after giving test4 a quick look back when it came out… installing test7 was a breeze, it booted fine and finally loaded my isdn (hisax/fritz) modules again. i loaded gnome and tried to connect to the internet but it didn’t work. when i did the shutdown, the box hang with one of those always repeating warnings. reboot und my ext3 fs was mounted as ext2 and fsck’ed because it wasn’t unmounted clearly… it didn’t pass the check. rebootet with 2.4.x and everything was fine again… i booted test7 again after that and rebooting works fine as long as i dont try to use my isdn card. anyone with working hisax?
Um, ITS THE KERNEL FOR GODS SAKE! If stuff like the help text being at the top versus the bottom matters to you, you really have no business compiling the kernel. Its not elitism, its practicality. The kernel devs shouldn’t waste their time worrying about stuff like that. If they can understand it, that’s enough.
I completely agree. Compiling a kernel isn’t exactly something a new user should (have to) attempt.
Complaining that this task isn’t “user friendly” enough is just insane. Why not complain that writing kernel modules to support your new hardware isn’t “user friendly” enough, too?
We’re not talking about everyday tasks joe user would do, like browsing the web or watching a DVD. We’re talking about compiling the core of an OS!!
With regard to the NVidia driver patch…. I was concerned about the driver compiling w/ the new kernel… but i decided to give whirl in its untainted, vanilla state….. and, surprisingly, it gave me no problems….. no patches needed, and i had full OpenGL acceleration…
You are talking about NVIDIA’s driver, not the XFree86 one, right? The NVIDIA driver modules is called “nvidia” and the XFree86 one is called “nv”.
*confused*… is this driver patch specific to certain models of the Geforce? mine was a Geforce2-Ultra 64MB….(I should note that my experiences are specific to tests 5 and 6, not yet test7)
Well, I tried compiling without the patch, and it didn’t work – although now you’re making me doubt and I’ll try again tonight. The patch is for the commercial NVIDIA driver and as far as I know it’s for all NVIDIA cards. Mine’s a GeForce4 Ti4400 128MB. I’ve only tested it with test5.
I don’t know anyone here around who not had to compile his kernel himself to get everything supported.
Well, to tell you the truth, I used my Linux system for more than a year before I recompiled a kernel, and then it was only to see if I could get better performance with a leaner kernel (I didn’t). I do think the documentation is sometimes sparse, but hey – you’re compiling a kernel! It’s definitely not for the average user!
with “here around” I ment people next to me. what I wanted to say is, most people, even unadviced, don’t get around compiling themselves. or they try one distro after another. I compile their kernels for them to avoid this
I find it difficult to compile a kernel. But if I have a .config file already that is known to work it is much easier.
eg. First I change the CPU option to the processor type I have. Then if successful, I only change a few things at a time.
But I found it important to back up the .config file and to run make mrproper (I think thats the command) between compiles. I did this stuff when learning how Gentoo works.
Um, ITS THE KERNEL FOR GODS SAKE! If stuff like the help text being at the top versus the bottom matters to you, you really have no business compiling the kernel. Its not elitism, its practicality. The kernel devs shouldn’t waste their time worrying about stuff like that. If they can understand it, that’s enough.
…
Okay, I do agree with you that they shouldn’t waste time writing documentation in layman terms, but you seem to claim that they shouldn’t care of the documentation at all as long as THEY understand it. What’s the point of writing documentation if they already understand what they’re doing? What’s next, suggesting people to read the comments in the source code? I’m sure that over 80% of people compiling their own kernel are not developers… They need to write documentation that the generic hobbyist with a good knowledge of his hardware can understand.
I actualy run woody, but the other day i was taking a look over at what testing was offering, cause im considering a change…
testing have 2.6.0-test5, so or something similar… So making it into unstable wouldnt be that hard would it? Id be interested in when it’d be in testing, but that’s a few months away no dobut, so i’ll wait till it gets closer to that time to ask that question
Please, fill a bug report because I have problems with USB keyboard and mouse on the 2.4.x series, and I was hopping the 2.6.x would handle USB better. It’s good to know that 2.6.x are gonna have better USB support. I am anxiously waiting for the first user friendly distro to ship with full support for the 2.6 (the installation kernel should be 2.6.x as well).
Any idea when the installable .deb image will be in unstable?
regards,
Chris
With my particular via chipset (apollo 133a pro), 2.4 doesn’t work with my USB controller on SMP systems (which I have). You can get it to work with turning apic off on the kernel command line, but then X doesn’t seem to get input from my mouse and keyboard when I do that.
With test6, this problem seemed rectified, I could actually use my USB gamepad! Now with test7, I can’t use it anymore.
Guess I should put in a bug report. Other than this, I really haven’t had any trouble with test6 or test7. Windows seem to drag across my screen a tiny bit better with test7 than with test6. Looks like 2.6 is on the right track.
You can create your own. In fact, it’s pretty easy. “apt-get install kernel-package”. Untar the source, change to that directory, set your options and then: make-kpkg kernel_image, and in the below directory (cd ..), you’ll have a deb.
If you don’t know what you’re doing, then suspect you’ll be waiting a while till there’s an official debian package for a 2.6 kernel.
I’m using the 2.6.0-test series in my desktop machine since test1 and I’m very happy with it. It is fastest *and* runs more smoothly that 2.4 series. I have noticed many lags in the UI, skips in XMMS in the earlier test1-test3 even without compiling with preemptive, but as from test6 those problems seems to be corrected.
All my drivers are working, and surprisingly (to me!) I could use the ide-scsi driver and burn some cds. The post-halloween docs announced that this modules was broken in the 2.5 series and I though they were already.
It really worths to test the kernel. It’s stabilizing and will be the best linux ever, IMHO.
ide-scsi still works, but you don’t need to use it at all. Just get an updated cdrecord package and you can simply use cdrecord dev=/dev/hdc (for example) rather than dev=0,1,0 or whatever your case is. There is no reason to use ide-scsi for burning anymore, it was a hack. Also the reason test6 runs so smoothly is because Con Kolivas’s interactivity patches were merged in that release 🙂 Best Desktop Kernel Ever.
have been using dev kernels since test1+mm . keeps getting better
Hiryu,
Yeah, a mate of mine told me to compile the kernel. At least then it is optimised for my CPU.
It’s just that set the options part… last time there was wayy… to many options and I knew maybe what 30% of them were which is why I tend to prefer installable images.
Anyway I will probably just it out on a PII I have here before I touch my main system.
regards,
Chris
Thanks for the reply!
cdrecord only sees my CD-ROM drive, using:
darkmoon:~# /usr/local/stow/cdrtools-2.00.3/bin/cdrecord dev=ATAPI -scanbus
Cdrecord 2.00.3 (i686-pc-linux-gnu) Copyright (C) 1995-2002 Jörg Schilling
scsidev: ‘ATAPI’
devname: ‘ATAPI’
scsibus: -2 target: -2 lun: -2
Warning: Using ATA Packet interface.
Warning: The related libscg interface code is in pre alpha.
Warning: There may be fatal problems.
Using libscg version ‘schily-0.7’
scsibus0:
0,0,0 0) *
0,1,0 1) ‘LG ‘ ‘CD-ROM CRD-8522B’ ‘2.02’ Removable CD-ROM
0,2,0 2) *
0,3,0 3) *
0,4,0 4) *
0,5,0 5) *
0,6,0 6) *
0,7,0 7) *
Guess I need to use the ide-scsi for a time.
Cheers.
If you can’t handle that many options then you aren’t suposed to be compiling your own kernel.
Those options are there because someone needs them.
\K
Well USB didn’t work on my K6-2 using the kernels from here.
http://people.redhat.com/arjanv/2.5/RPMS.kernel/
Is this a kernel thing or didn’t arjanv compile it properly?
Has anyone tried these kernels on an nforce2 board. I have an ASUS A7N266-VM. I compile the kernel without ACPI coz i get screen corruption when I do.
Im using Slackware 9.1 with kernel 2.6 test 6.
Anyways the main probs i have are :
1) Alsa works in xine and stuff but NOT in XMMS. And yes I have selected the Alsa out-put plugin.
2) The system just freezes after playing ound after som time. About 15 mins of continuous sound.
3) It also freezes when the screensaver starts.
I have an on-board nvidia geforce2 GPU video card with 32 mb ram.
I patched the official nvidia driver from minion.de
That install goes perfectly fine for the patched video driver.
Apart from this all else is fine really. Comp. seems very responsive ( when it hasnt frozen that is ). And DMA and what have U work just fine.
If anybody has a clue would love to hear from ya.
Thanx a lot.
madmax
P.S.—> None of this happens with the stable kernels ofcourse. The stable kernels are rock solid for me. And I can do everything.
Use dev=/dev/hdX not dev=ATAPI.
> If you can’t handle that many options then you aren’t suposed to be compiling your own kernel.
That’s not very helpful. If you can’t handle that many options, then you just need to spend a little bit more time getting familiar with what’s in there.
At first I had trouble with it, but now it’s dead simple.
Just give it an hour of your time when you can – much better than waiting/checking for an image of it to appear in Debian unstable.
I knew it spelled trouble, but it was cheap. My G400 died and i needed to get a quick replacement with passive cooling. I opted for a nvidia fx 5200, big mistake.
Maybe their drivers are good enough for stable kernels but going through the hassle to patch them with the minion-patch each test kernel is just too much. I should have shelled out the 10 Euro more for an ATI R9000 or a Matrox G550. Grrr. Slap slap slap.
madmax: There should be an option for nforce2 boards in the kernel. Did you try it? And why is your patched video driver module running but not mine? Argh. (i am not expecting any useful answer, thank you, just venting frustration)
Hey Somewhere,
Yeah I have selected the relevant options for the nforce2 board. Have just installed 2.6-test7. So far so good. No freezes. Although now that i said it, I’m bound to get one. LOL.
Also I have a patched NVIDIA graphics driver that seems to work fine for the test7 kernel. if anyones interested let me know and more importantly how i could get it to ya.
See ya
madmax
If you can’t handle that many options then you aren’t suposed to be compiling your own kernel.
Those options are there because someone needs them.
Typical linux ‘l33t’ response – I thought we’d all moved beyond this now?
This release of 2.6 is excelent for me – a noticable improvement over previous test releases on my PIII-450
Noob question:
If you compile kernel 2.6 can you use your old .config file from a 2.4 kernel or is it better to start from zero and select the options you want?
Typical linux ‘l33t’ response
Bull.
If the typical distro did not already have a pre-compiled kernel or if the kernel choices were very limited and did not receive updates, then this would be a very l33t crap response.
However, if someone says that the average person should not worry or have to worry about compiling their own kernel. Then I say yes they should not.
If someone says to me that they are downloading a test kernel from an upcoming release and they were overwhelmed by the number of choices, I would say do not compile your own kernel then.
Use the one the distro came with. Why is that so l33t?
We are not talking about the gui preferences in an email app but the kernel for god’s sake. Yes, there should be a ton of wild ass options. Otherwise, the complaint would how limited the kernel is to hardware support and options. Making your way through the options is not that daunting it just takes time and a bit of patience in reading the descriptions.
” you compile kernel 2.6 can you use your old .config file from a 2.4 kernel or is it better to start from zero and select the options you want?”
It may be better to start from scratch, but importing my old .config worked perfectly with the early 2.6 version. Maybe this compatibilty will be dropped later, I don’t know.
You can still use make oldconfig.. but you have to make sure you enable some options for VT stuff, otherwise you wont be able to see anything (or use your keyboard) It’s in dave jones’ FAQ about testing 2.5… cant remember where it is, google.
You can use your old .config file as a starting point when you do a “make menuconfig” (or “make xconfig”). Some drivers have changed.
I have been using Andrew Morton’s sources for a few months now on my T23 and I have not had any problems.
Thanks guys,
I’ve tried it before and I’ll try it again. Practice makes perfect. Then I can help others as well (:
Yah, you definitely should have gone for a 9000pro.
It has the advantage of having both open source and binary opengl driver support, and additionally its 2d quality is about untouchable. And they’re cheap too!
Yah, you definitely should have gone for a 9000pro.
It has the advantage of having both open source and binary opengl driver support, and additionally its 2d quality is about untouchable. And they’re cheap too!
I’ve tried installing the Mandrake RPMs for 2.6-test5 but so far I haven’t been able to successfully compile and install NVIDIA’s driver module (yes, I have installed the 2.6-test5 kernel-source package). I’ve tried the instructions contained on this webiste:
http://www.minion.de/
but while I can compile the module, I still can’t modprobe nvidia.o (or nvidia.ko, as it happens). I either get a “wrong kernel headers” or “wrong module type” depending on the method I use.
Anybody had any luck using the NVIDIA drivers with kernel 2.6?
Get your bleeding edge PowerPC kernels at http://www.ppckernel.org
RH rpms from arjanv work great.
http://people.redhat.com/arjanv/2.5/RPMS.kernel/
Well USB didn’t work on my K6-2 using the kernels from here. http://people.redhat.com/arjanv/2.5/RPMS.kernel/
Is this a kernel thing or didn’t arjanv compile it properly?
Try this page…
http://thomer.com/linux/migrate-to-2.6.html
here’s a tidbit from the article
The names for the USB modules are different. This affects /etc/modprobe.conf (look for “usb-controller”) and, once again, /etc/rc.sysinit. I had to rename all instances of “keybdev” into “usbkbd” and “mousedev” into “usbmouse”. In addition, I replaced all instances of “/proc/bus/usb” with “/sys/bus/usb”. (Also do this last part in /etc/init.d/halt.)
>>Yah, you definitely should have gone for a 9000pro.<<
umh, the Pro is not cooled passively. Surprisingly i find the 2D quality of the nvidia fx 5200 not too bad, considering that i used a Matrox before. *sigh*
The kernel configuration options are quite confusing, the configuration help text should be improved.
For example, some options say “If you don’t know what this is then you probably don’t need it”. Some options say “If unsure, say no”.
Sometimes this info is at the top of the help text, sometimes at the bottom, sometimes not to be found at all.
The kernel configuration system has considerable room for improvement.
I compiled 2.6-test7 this morning on my development Red Hat 9 box. I ran ApacheBench (from another machine to hit the test machine) last night before the kernel upgrade to have comparison numbers. This morning after compiling 2.6test7 and rebooting Apache is now reporting that it can only handle about 80% of the requests compared to kernel 2.4.22
Last night I ran 4 test over a period of 1 hour, the same today and the numbers are relatively the same … 20% loss in #request per second.
As far as the ‘desktop’ experience, I honestly haven’t noticed the difference that people claim.
Is anyone out there finding kernel 2.6 not to be what other have made it out to be … a ‘performance champ’?
I’ve noticed some slight performance increases on the console side…. and modest increases in desktop performance…. nothing I have in benchmark scores… this is purely a subjective *feeling*…. but yeah, it’s not quite as much as I had hoped for…nor as much as some of the past posts have raved about…. but it feels noticable… [for reference, i’m running a 1GHz Athlon-Tbird, ASUS A7V, 512MB PC133]
With regard to the NVidia driver patch…. I was concerned about the driver compiling w/ the new kernel… but i decided to give whirl in its untainted, vanilla state….. and, surprisingly, it gave me no problems….. no patches needed, and i had full OpenGL acceleration…
*confused*… is this driver patch specific to certain models of the Geforce? mine was a Geforce2-Ultra 64MB….(I should note that my experiences are specific to tests 5 and 6, not yet test7)
P III – 800Mhz, 256 RAM, 40 GB 7200 rpm drive … honestly, with 2.6test7 – running ApacheBench (ab) again it still is not as fast as running 2.4.22 with init=3 and httpd having nice=-20
On a single processor machine the actual raw output for the 2.6 kernel will be less than a 2.4 kernel. Not a great deal less, but less. Where you do have improvements is in multi-processer support, big iron features (such as NUMA), context switching and low-latency support. Due to some of the improvements though, for desktop use the “feel” of the computer will overall seem more responsive, especially with the mm patches in place.
installed test7 today after giving test4 a quick look back when it came out… installing test7 was a breeze, it booted fine and finally loaded my isdn (hisax/fritz) modules again. i loaded gnome and tried to connect to the internet but it didn’t work. when i did the shutdown, the box hang with one of those always repeating warnings. reboot und my ext3 fs was mounted as ext2 and fsck’ed because it wasn’t unmounted clearly… it didn’t pass the check. rebootet with 2.4.x and everything was fine again… i booted test7 again after that and rebooting works fine as long as i dont try to use my isdn card. anyone with working hisax?
Um, ITS THE KERNEL FOR GODS SAKE! If stuff like the help text being at the top versus the bottom matters to you, you really have no business compiling the kernel. Its not elitism, its practicality. The kernel devs shouldn’t waste their time worrying about stuff like that. If they can understand it, that’s enough.
I completely agree. Compiling a kernel isn’t exactly something a new user should (have to) attempt.
Complaining that this task isn’t “user friendly” enough is just insane. Why not complain that writing kernel modules to support your new hardware isn’t “user friendly” enough, too?
We’re not talking about everyday tasks joe user would do, like browsing the web or watching a DVD. We’re talking about compiling the core of an OS!!
I don’t know anyone here around who not had to compile his kernel himself to get everything supported.
Kernel config IS userfriendly enough, ALTHOUGH documentation is not the best, especially on 2.6.0test. But I think, this issue will be addressed.
I have to say, configuration options on 2.6.0 also are very well sorted and you find yourself a lot easier through them compared to 2.4
With regard to the NVidia driver patch…. I was concerned about the driver compiling w/ the new kernel… but i decided to give whirl in its untainted, vanilla state….. and, surprisingly, it gave me no problems….. no patches needed, and i had full OpenGL acceleration…
You are talking about NVIDIA’s driver, not the XFree86 one, right? The NVIDIA driver modules is called “nvidia” and the XFree86 one is called “nv”.
*confused*… is this driver patch specific to certain models of the Geforce? mine was a Geforce2-Ultra 64MB….(I should note that my experiences are specific to tests 5 and 6, not yet test7)
Well, I tried compiling without the patch, and it didn’t work – although now you’re making me doubt and I’ll try again tonight. The patch is for the commercial NVIDIA driver and as far as I know it’s for all NVIDIA cards. Mine’s a GeForce4 Ti4400 128MB. I’ve only tested it with test5.
I don’t know anyone here around who not had to compile his kernel himself to get everything supported.
Well, to tell you the truth, I used my Linux system for more than a year before I recompiled a kernel, and then it was only to see if I could get better performance with a leaner kernel (I didn’t). I do think the documentation is sometimes sparse, but hey – you’re compiling a kernel! It’s definitely not for the average user!
with “here around” I ment people next to me. what I wanted to say is, most people, even unadviced, don’t get around compiling themselves. or they try one distro after another. I compile their kernels for them to avoid this
I find it difficult to compile a kernel. But if I have a .config file already that is known to work it is much easier.
eg. First I change the CPU option to the processor type I have. Then if successful, I only change a few things at a time.
But I found it important to back up the .config file and to run make mrproper (I think thats the command) between compiles. I did this stuff when learning how Gentoo works.
Thanks Savage I will check it out.
I was talking about the “nvidia” driver… not “nv”…..
so, it’s XFree’s bundled driver that requires a patch… not nvidia’s own……thanks for the clarification! =)
Um, ITS THE KERNEL FOR GODS SAKE! If stuff like the help text being at the top versus the bottom matters to you, you really have no business compiling the kernel. Its not elitism, its practicality. The kernel devs shouldn’t waste their time worrying about stuff like that. If they can understand it, that’s enough.
…
Okay, I do agree with you that they shouldn’t waste time writing documentation in layman terms, but you seem to claim that they shouldn’t care of the documentation at all as long as THEY understand it. What’s the point of writing documentation if they already understand what they’re doing? What’s next, suggesting people to read the comments in the source code? I’m sure that over 80% of people compiling their own kernel are not developers… They need to write documentation that the generic hobbyist with a good knowledge of his hardware can understand.
I actualy run woody, but the other day i was taking a look over at what testing was offering, cause im considering a change…
testing have 2.6.0-test5, so or something similar… So making it into unstable wouldnt be that hard would it? Id be interested in when it’d be in testing, but that’s a few months away no dobut, so i’ll wait till it gets closer to that time to ask that question
Please, fill a bug report because I have problems with USB keyboard and mouse on the 2.4.x series, and I was hopping the 2.6.x would handle USB better. It’s good to know that 2.6.x are gonna have better USB support. I am anxiously waiting for the first user friendly distro to ship with full support for the 2.6 (the installation kernel should be 2.6.x as well).
Thanks ! 🙂