Linux 2.6.0-test5, How To Incrementally Upgrade

← Back to Stories (view on slashdot.org)

Linux 2.6.0-test5, How To Incrementally Upgrade

Posted by timothy on Monday September 8, 2003 @11:42PM from the patch-by-patch dept.

An anonymous reader writes "Linux creator Linus Torvalds recently announced the 2.6.0-test5 Linux kernel, bringing the release of a stable 2.6 one step closer. KernelTrap quickly followed by posting a guide that walks bleeding-edge Linux fans through the process of upgrading from 2.6.0-test4 to 2.6.0-test5, using a tiny 706 KB patch."

40 comments

Min score:

Reason:

Sort:

Someone tell me what's new in this procedure? by cyberkreiger · 2003-09-08 23:46 · Score: 2, Funny

So, it's exactly like upgrading any other kernel version, just without the "make dep"? Amazing.

--
Stumbling in the dark
I hear slavering of jaws
Eaten by a grue.
1. Re:Someone tell me what's new in this procedure? by MerlynEmrys67 · 2003-09-09 06:29 · Score: 3, Insightful
  
  Yes it appears remarkably similar to upgrading every other kernel...
  That said - posting documents like this help bring in new talent (ie. people who are too scared to start off on their own) which is a key component of keeping Linux viable into 2020 (one of these days Linus will want to retire - we need to have kernel hackers that start in 2005 who are ready to replace him). I remember my first "from source" kernel upgrade from the 2.2 series. I had always just grabbed the latest distro and installed. Imagine my surpirse when I could just get the source myself, build it, install it correctly and the system just ran. It is very powerful, and helping people achieve this power is key for Linux
  
  --
  I have mod points and I am not afraid to use them
Obsolete make steps by Phaid · 2003-09-09 00:14 · Score: 4, Informative

Make bzImage && make modules is no longer the recommended way to do this. You should just do "make" instead, which will build bzImage and any modules you may have asked for, and do it quicker than if you specify the steps separately. See the Halloween 2.5 document for details.
1. Re:Obsolete make steps by Drathos · 2003-09-11 12:27 · Score: 1
  
  Yeah.. It says that right below.
  
  Does make you wonder why they mentioned the individual steps, though.
  
  --
  End of line..
Re:Obligatory SCO reference by Anonymous Coward · 2003-09-09 00:14 · Score: 0

It actually does, since this is counted RIAA-style:
for each if clause the total sum of kB is calculated from the total if clauses SCO ever used.
Re:Obligatory SCO reference by Anonymous Coward · 2003-09-09 00:16 · Score: 0

Does this mean that I have to pay SCO $17,000 for each copy in order to download the patch?
obligatory 640 KB reference by renehollan · 2003-09-09 00:38 · Score: 2, Funny

...using a tiny 706 KB patch.
Apparently, 640 KB is not enough for everybody.
Though, I would have never expected that it was Linux for which it would not be enough.

--
You could've hired me.
jesus by XO · 2003-09-09 01:20 · Score: 1

Why does this even make any news space -on- slashdot? I mean, if they had just said "2.6.0-test5 is out." that might've been fine.. but that whole tutorial is a complete waste of time.

It's just like all other patches.. the way the article sounds, it's like some sensational new upgrade path, that must be done.. like someone made a major sysctl interface change or something, and it requires a whole new set of modutil, binutil, et al..

sensationalism sucks.

--
"Champagne for my real friends - and real pain for my sham friends!" http://ericblade.postalboard.com/
Laptop Mouse Pad Still now working by peripatetic_bum · 2003-09-09 02:15 · Score: 1

Hi all, I ave been anxiously awaiting each new test, waiting to replace the 2.5.72 kernel in which the mouse works.

The 2.5.69 kernel had the mouse, sound working but since then neither sound nor mouse pad work on the HP pavilion ze5385us model

Any ideas on how to get the mouse pad to work? or te sound using the 2.6 test kernels?

Thanks all

--
Sigs are dangerous coy things
1. Re:Laptop Mouse Pad Still now working by Anonymous Coward · 2003-09-13 19:05 · Score: 0
  
  OHOHOHOH!!! A fellow ze5385us-er!!!!
  
  Any chance you've gotten test5 working for stable
  XFree86 DRI support?
I'd be running by PD · 2003-09-09 02:31 · Score: 1

Except for one thing: VMware. I need it on my laptop because we use Lotus Notes at work. The VMware drivers only work with 2.4 kernels now.

--
If tits were wings it'd be flying around.
1. Re:I'd be running by Anonymous Coward · 2003-09-09 04:30 · Score: 0
  
  Ok, thanks for letting us know.
2. Re:I'd be running by nitehorse · 2003-09-09 04:44 · Score: 1
  
  Actually, that's not true.
  
  There are patches that you can apply to VMware to get it to work properly with 2.6.
  
  Gentoo handles this automatically if it detects that you're running a 2.6 kernel.
  
  I found out how to do it manually here. Hope that helps.
3. Re:I'd be running by deKernel · 2003-09-09 14:39 · Score: 1
  
  If all you need is Notes, then just use WINE. I have been using Notes under WINE without a problem. I say this because I thought I read that WINE does not have any issues with the newer kernel features.
Not very interesting by MrHanky · 2003-09-09 03:09 · Score: 2, Interesting

But I just booted 2.6.0-test5-mm1. There's been a bug since test4 (actually included in the test3-mm-series) that is still around. When bootin without a PS/2 mouse, the system hangs, freezes, when probing for keyboard and mouse. I normally use a USB mouse, so this is a bit annoying. The bug isn't fixed yet. Otherwise, the 2.6 kernel seems to be coming along fine.
1. Re:Not very interesting by TinheadNed · 2003-09-09 07:25 · Score: 1
  
  If it's any help, I have two computers running 2.6.0-test4-mm4, and the desktop has no PS/2 mouse and it doesn't hang on bootup.
  
  However the console framebuffer isn't going above 800x600. Ah well
2. Re:Not very interesting by MrHanky · 2003-09-09 07:34 · Score: 1
  
  If it's any help, I have two computers running 2.6.0-test4-mm4, and the desktop has no PS/2 mouse and it doesn't hang on bootup.
  
  Interesting. I think I've tried test4-mm4. Maybe it's an issue only for certain chipsets (VIA in my case). Are you using a USB keyboard? No matter what, I have no problem booting when I use a USB-PS/2 adapter for my mouse. I believe it's a known problem, since I found out how to solve the problem via the mighty Google. Unfortunately, I'm too shy to submit bug reports.
3. Re:Not very interesting by EugeneK · 2003-09-09 07:40 · Score: 1
  
  I have the same problem; I have a USB mouse and can't get past the mouse detection message. I experience it with 2.6.0-test4 and -test5, but not with 2.6.0-test3, so I'm sticking with that and trying to figure out the problem by comparing the two source trees...
4. Re:Not very interesting by The+Analog+Kid · 2003-09-09 08:09 · Score: 1
  
  Nope, I have a machine with a VIA KT400a, and it doesn't halt for the PS/2 mouse.
5. Re:Not very interesting by Anonymous Coward · 2003-09-09 23:09 · Score: 0
  
  Please do submit bug reports. Its very important. Yep, to lkml. Just give them a good subject line and decent content (needn't follow the reporting bugs document to the letter), and an offer to test patches and provide more information.
  
  If you can't do all that then maybe don't bother the mailing list, but I'm sure you'd be keen to get your hardware working.
6. Re:Not very interesting by Anime_Fan · 2003-09-09 23:56 · Score: 1
  
  Hmmm... None of the kernels I've tried does that. I've been using USB Mice since 2.5.xx with none of the pain 2.4.xx used to give me.
  
  They did however break ALSA for VIA82XX chipsets, or so it seems. *Sigh*.
  
  I still think 2.6.0-test2 was the better of the kernels when it came to stability and audio.
7. Re:Not very interesting by Anonymous Coward · 2003-09-13 19:09 · Score: 0
  
  That's a known issue in all versions of alsa since 0.9.0rc6 or 7+ I believe, crackling and buzzing, right?
  
  I had those same issues on my Via mobo, so I've simply kept running a 2.4.21 kernel on it with alsa 0.9.0rc5 audio drivers. Might be possible for you to disable the in-kernel alsa drivers and use the old archive instead?
  
  -- vranash
8. Re:Not very interesting by besfred · 2003-09-13 22:55 · Score: 1
  
  Well. I have the same kernel here running on a Dell Latitude C840, since about 5h. After reading your comment I tried to use the touchpad and mouse pointer thingy, and both didnt work. Since I normally use a USB mouse, too, I didnt even notice that til now. So for me, it did not hang at all. /me goes recompiling with the missed option :) Other than that, I am really happy with the current state, the new kernel has reached to date.
9. Re:Not very interesting by MrHanky · 2003-09-13 23:23 · Score: 1
  
  I'm happy with it too, at least now that someone in this thread pointed out that the problem is with ACPI. Having pci=noacpi as a kernel option solves the problem for me.
10. Re:Not very interesting by besfred · 2003-09-14 02:14 · Score: 1
  
  Oh my :) just noticed I had the Ps2 mouse already compiled as module. a simple modprobe psmouse did the trick. no recompile needed. Linux can be so easy, hehe.
Damn by The+Analog+Kid · 2003-09-09 03:52 · Score: 1

I just finsihed compiling Test-4, and now I read this.
Ksymoops... by slackergod · 2003-09-09 04:50 · Score: 1

Guess this is good a place as any to ask...
Is there an official site/email address
where I can submit any kernel oopses I get?

Also, in case this helps anyone else...
If you're running debian woody, and you're
trying to upgrade to one of these kernels,
you're probably going to be hunting around
for a copy of module-init-tools that plays
nicely with woody, instead of sid's copy.

While these are probably bleeding edge
and all, here's two packages I found which
seems to do the trick for woody w/o breaking
dependancies...
modutils
module-init-tools

After installing 'em, the 2.4 modprobe et al
will be renamed modprobe.modutils,
and the 2.6 versions will fallback and call
the 2.4 version if you're running an old kernel.
1. Re:Ksymoops... by TinheadNed · 2003-09-09 07:20 · Score: 1
  
  Well, there's the kernel bugzilla. It's not used much, but I've posted a few kernel bugs on there, and got quick responses from the the developers.
2. Re:Ksymoops... by TinheadNed · 2003-09-09 07:23 · Score: 1
  
  Or bugzilla.kernel.org.
  
  That's what I get for magically getting Submit and Preview mixed up.
avoid xfs on md by Leper · 2003-09-09 08:52 · Score: 1

Let me just warn anyone who wants to play with that particular combo right now - it doesn't work so hot.

xfs on a raid0 partition will corrupt its logs every time yeilding some spew about "bad clientid"

Been this way since test2 evidently as its been reported to the xfs list already, I just ran into it with test4 and test5.
make mrproper? by Vasilis+Vasaitis · 2003-09-09 09:55 · Score: 1

One quick point about the article: I don't think commands such as make clean or even make mrproper (as the article states) are actually required any more for a proper kernel build of 2.6. In the best case, they'll have no effect, because everything will be rebuilt after patching anyway. In the worst case, they'll even remove some files that wouldn't be rebuilt because they and their dependencies haven't been modified by the patch.

Just my 0,02 euros.

--
Vasilis Vasaitis
Late readers: please moderate at Newest First, with a low threshold, to promote late writers.
1. Re:make mrproper? by Anonymous Coward · 2003-09-11 06:32 · Score: 0
  
  To fully understand the new build system, type 'make help'.
  
  And yes, after typing this you will see that 'make mrproper' is indeed still supported and even still suggested...
Works here by denjin · 2003-09-10 00:51 · Score: 1

My Intellimouse Explorer USB has never had a problem during bootup with any of the 2.6 kernels.

I was using the stock 2.6.0-test5 yesterday, and i have 2.6.0-test5-love1 installed, and its also working fine.

Maybe a bug with your USB chipset? Works fine on Intel ones here and the nf2 at home.
1. Re:Works here by MrHanky · 2003-09-10 07:58 · Score: 1
  
  No, it's not the USB. It hangs when the PS/2 mouse driver has been initialised (or failed to be), before the keyboard is found. No kernel panic, and the magic SysRq key doesn't work. This is long before the USB modules are loaded. If I compile keyboard as a module, it will hang when that module is loaded.
It's about ACPI, try pci=noacpi or leave ACPI out by Toothrot · 2003-09-10 20:09 · Score: 1

This helped with me. Alsa leaving keyboard out did the job but that might not be what you want.
Re:It's about ACPI, try pci=noacpi or leave ACPI o by MrHanky · 2003-09-10 22:24 · Score: 1

Thanks. I'll try that the next time I reboot.