Now all that I've heard is that the new test series is the best thing since sliced bread (and it probably is), but I'm a major desktop user. I've compiled the latest test kernels on a server project at school, but I'm a bit more demanding. Not only am I currently running Apache, sshd, and vsftpd on my home pc, but I'm running XFCE 4.01 and play everything from Unreal Tournament 2003 to Diablo II LoD over WineX 3.2. Recently, after using swaret to update ALSA, I broke the hell out of my modules that I'd need to recompile the kernel anyways to fix it, so I thought " What the hell, why not?". I also use my ATI Radeon 9100 Sapphire card with the latest drivers. I noticed a 2.6 folder in the working directory that you compile drivers in, will I have to recompile them again? Will I still have the functional server, desktop, and gaming pc in one?

Not only that, but I use an onboard Promise MBFastTrak 133 Lite controller for an extra HD. It uses proprietary drivers. Will I have to re-compile these as well, or is the module I am using atm good enough. Perhaps its included in the kernel and there are no worries (wishful thinking?)

It would be great if you have any tips for me, anything that may keep me from trying to insmod the wrong compiled module for an hour straight again, (he he) I'll be checking this post throughout the day and I'm planning the big compile fest for tonight, or maybe tomorrow if I stay at a friends house.

Thanks again

Hi

I am also a desktop user and have switched to the 2.6test10 kernel last night ( I had been running test9 ). My system is not as complex as yours but everything worked immediatly for me. I don't use any modules. I have a Radeon 7000 agp card. A lite-On CD burner. My motherboard is a Giga-byte with a 1.8gig P4 500FSB 1/2 gig ram sis 645 chipset ATA133. I also use an Adaptec SCSI card for my scanner and an 8129 RealTec network card, on board sound. No problem with any of them.

One thing I have noticed is my system runs noticably faster than the 2.4.22 kernel. In fact I had to slow down the mouse speed in KDE.

John

Graywind,

I am also using Slackware 9.1 and have been using the 2.6-test series since the day that test9 came out. It has worked beautifully for me, though I had to compile my SoundBlaster Live! drivers directly into the kernel as for some reason they didn't want to work with ALSA (though they did under 2.4).

Unfortunately, I can't get ATI's 3.2.8 drivers for Xfree86 4.3.0 to compile. I've never tried to install their 3D drivers until now. That's actually why I'm here; to find some help on getting their 3D drivers to work with 2.6test10. Supposedly, they're supposed to work with test6 and higher, but I have yet to be successful. Just be aware that you may not get to play your games under linux with 2.6test. Just FYI.

P.S. If you do get it to work - TELL ME WHAT YOU DID.

I think neither of you guys should be using kernel 2.6-test* . You are desktop users and the 2.4 kernel is going to be your best option for now. Once 2.6.0 comes out then you should definatly make the switch. But if you feel the need to try out the test kernels the best thing to do, and actually what you should do whenever you are recompiling any kernel is keep your current working kernel on your computer so you can boot into it incase something went wrong with your new kernel.

What did you all do to get the linux kernel working?

Here's my grub file:

I am getting a ton of errors when my modules are loading!


Help!

Edit: sorry i forgot to mention this, but I'm using RedHat, not Slackware.

Don't wanna sound like an ass or anything, but did you compile the modules?
Or build a whole bunch of them into the kernel?

There's a nice guide on setting up custom kernels stickied to the top of the Slackware forum. It's a nice read if you think you might have missed something, or just don't know what you're doing.

Found the onboard raid drivers in the kernel! This is great, I'm better off without tainting it Everything is going by fine and now all I need to do is get those nasty ATI drivers working.

I checked dmesg, and trying to run a insmod on fglrx.o from the 2.6.x dir in the build_mod directory, I get

fglrx: Unknown symbol VMALLOC_VMADDR

I'm going to try some more tweaking and recompile it fresh. I'll let ya know how it goes.

omg as soon as I finished here comes test11

I assume this driver is a module? That being the case, then my guess is that it will not work without a recompile of the module for the 2.6.0-xx kernel. But your biggest hurdle with that or any other proprietary module is going to be getting them to compile under the new API. Almost everything about modules has changed in the new kernel, everything from compiling them to their new .ko extension. But like someone stated, the best thing to do if you want to try it, is too keep your current kernel and load the 2.6.0 kernel at your leisure.

Also unless your running Slackware 9.1 or some other 2.6 ready distro, you'll also have to update several apps to get them to work with the kernel, like mod-ultils for instance. Do some googling, you should find all the info you need.

I ran into a couple of "got-ya's" when I switched to it. First the nvidia drivers won't work with the new API without a patch. Even after the patch I was getting some locking errors until I turned off CONFIG_DEBUG_SPINLOCK_SLEEP in the kernel. IDE-SCSI emulation is completely broke in it, according to this doc, http://www.codemonkey.org.uk/post-halloween-2.5.txt, so I didn't bother trying it. Instead I upgraded cdrdao to the latest CVS version in order to get K3B to work with my IDE reader. There is also a locking issue with the aic7xxx SCSI driver that I'm still toying with.

Here's another link with good info:
http://kerneltrap.org/node/view/799

No offense, but you assume to much. I'm well aware of what's involved in using a test kernel since the very first kernel I ever used was 1.2.13 under slackware 3.0 and I've recompiled more custom kernels since then than I care to count. (Though I later moved on to Redhat, Debian, et all and came back to Slackware at 8.1). I use linux on both the server and the desktop. I decided to use kernel 2.6 for two reasons:

1) I've had issues with the most recent 2.4.x kernels and my USB hardware. 2.4.20 and previous kernels had zero problems.
2) It's to a reasonable stability level that it's doubtful it's going to royally bork anything.

I've been 100% happy using the 2.6test series and it's been running better than 2.4.21 or 2.4.22 did for me.

Ummm, who says I don't? I -always- keep the most up-to-date official distro kernel...

But thanks for the tip. I understand that you were just trying to be helpful.

This was the same error I was getting. I fixed it shortly after posting in this thread.

Here are two links that are very helpful:
This seems to be a direct text copy of a post in the rage3d forum -> http://www.linuxquestions.org/questi...=kernel+.6+ati
And this one has some helpful information too -> http://www.rage3d.com/board/showthre...eadid=33715874

Here's the VMALLOC patch that they talk about

diff -ruN build_mod.orig/firegl_public.c build_mod/firegl_public.c
--- build_mod.orig/firegl_public.c 2003-09-05 00:45:33.539384168 +0200
+++ build_mod/firegl_public.c 2003-09-05 00:47:13.193234480 +0200
@@ -129,7 +129,9 @@
#define pte_offset pte_offset_map
#endif
#endif
-
+#ifndef VMALLOC_VMADDR
+#define VMALLOC_VMADDR(x) ((unsigned long)(x))
+#endif
// ============================================================
#ifndef TRUE
#define TRUE 1

Just apply the patch in the /lib/modules/fglrx/build_mod directory (it applies against the firegl_public.c file). i.e. 'patch -p1 < fglrx-2.6-vmalloc-vmaddr.patch'

The first link I posted should take you the rest of the way.

Good luck.

NOTE: This patch was applied against the fglrx-glc22-4.3.0-3.2.8 from ATI and may or may not apply cleanly against other sources. YMMV.

And I stay up too late at night and misread other peoples posts. I see know that AlleyTrotter posted in between me and Graywind and fitz9948 was probably talking to him.

It's 5:20am. Why am I still up?

I also take no offense
I also keep several kernels on hand as a fallback
I compile a monolithic kernel maybe that is why I am not having the same problems. The aic7xxxx driver takes a little longer during the boot scan but its does find and initialize my scsi stuff. I have had one gotch'ya with USB. The usb driver for 2.0 tries first and does not fallback gracefully to the 1.0 driver, but this was fixed by selecting 1.0 in the USB configuration.

John

PS as an after thought -- the new kernel needs to be tested on the desktop also

Yes I did. make modules and make modules_install, right? But lots of modules are still failing to load. I can't even get on the Internet with the new kernel. Any ideas? Does this have anything to do with all the unresolved symbols messages I got while compiling?

If I want to delete this kernel and start all over using my original kernel, what do I delete?