Slashdot Mirror
Linux 2.4.0-test1 Released
Chris Cheney writes: "Linux 2.4.0-test1 is out with a note from Linus for more details. Why does all the cool stuff come out after potato is frozen? " With Linus being gone for three weeks, Alan is likely to maintain a 2.4.0-ac series. It's getting closer...
At least, it seemed like it to me.
- A.P.
--
"One World, one Web, one Program" - Microsoft promotional ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
From the kernel Makefile:
/
#
# INSTALL_PATH specifies where to place the updated kernel and system map
# images. Uncomment if you want to place them anywhere other than root.
INSTALL_PATH=/boot
Fresh from the tarball sources have INSTALL_PATH commented out.. the default is
Most distros make boot on seperate partition to get around the 1024 cylinder limitation. IIRC there is a new version of lilo that doesn't have the issue.
People like repetition. It's reassuring.
Yes, I know. I've had a very hard week.
Paranoia isn't an infectious condition, it's a way of life
I was wondering if someone more aware could let me know where to send bug reports. For some reason, it doesn't seem logical to flood Linus with bug reports, but without joining the kernel mailing lists, how can we post bug reports?
----------------- "I have a bone to pick, and a few to break." - Refused -------------------
What I got from his message was that he compiled and installed the kernel.. but on reboot? his system was unusable.
I've now recompiled different kernels on various systems well over 100 times. ( Mostly for embeded PC use ) And frankly removing certain drivers or options in the kernel config will build you an unusable kernel. The kernel will hang on boot or do other "interesting" things if you actually get to the login prompt.
The point is that it's not easy to compile a new kernel if your're trying to customize it in anyway. It's a trial and error type of situaton.
Of course anyone who is messing with their Kernel should always make sure lilo has a stable kernel to fall back on. Install the new kernel but leave the old one around for emergency boot. Unless the new kernel happens to trash your disk
Ex-Nt-User
Are you by chance running a Red Hat distribution?
Red Hat tends to do things a little bit differently, I believe it has to do with the initial ramdisk (initrd) setup they use to load modules on boot-up.
Check Here for the Red Hat-specific kernel building HOWTO.
Did he say anything about Java bytecode? Java doesn't necessarily imply Java bytecode; see, for example, The GNU Compiler for the Java(tm) Programming Language, which can either produce bytecode or native machine code. (Yes, it means you don't automatically get Write Once Run Anywhere if you compile to native machine code, but perhaps there are applications where that doesn't matter.)
Why is changing the engine in my car so complicated? i bought the damn thing and there's no way i'm going to trade in my ford ka *just* to get a better engine. i bought a new engine that said it would fit, but it's *so* much harder to do then the other things i do with my car (i.e. drive, fill the tank, change the oil, wash it, etc).
there are all these bolts and screws and the tools! dear god the tools. i need a "hydrolic lift," wtf?!
i'm tellin' ya, these ford people had better get their act together.
US Citizen living abroad? Register to vote!
Sorry... I feel like i'm beating a dead horse... but this got me:
.config file from redhat or something, then everything is a module anyway...
.config file from the previous kernel (if its the same tree anyway), and then type ./complie...wait 4 or 5 minutes (3 on mine) and then reboot.
1. Why is swapping the Kernel so complicated? Why not automate this more?
How do you mean automate it more? If you do:
make menuconfig Maybe this is the problem... but if you use the
make dep
make clean
make bzImage
make modules
make modules_install
make bzlilo
done..... I put all the make statments in to a 'script' and call it "complie." (Orginial, eh?). Then all you have to do is make menuconfig, copy your
---
Just a comment: I have never used the "*lilo" or "install" targets, since I'm not really sure of what they do. And I like my kernels named after what's in them, so I end up with kernels like /vmlinuz-2.2.13-smp-bigphysarea-3compatch or /vmlinuz-2.2.13-smp-adaptec and so on. Makes it easier to see how they differ.
And if you have to copy the kernel to a lot of machines, it't easier if you don't have to mess with a bunch of modules, so I usually compile everything I need into the kernel, if I can.
Don't use "make bzlilo". Use "make install". "make install" can be customized to fit your distribution by putting a script in /sbin/installkernel.
Is there not a gnu parted which does this?
If exchanging a kernel is such a dang-blasted important task for any Linux user to know how to do, why is it so complicated?
I think you're operating from a faulty precept here; who says it's important for any user to know how to do this?
Any user who *DOESN'T* know how to swap a kernel should be using a distribution such as Red Hat, Mandrake, Caldera etc. where it IS easy to swap a kernel.
--
Downloading and installing stuff is fine, but part of the reason you choose a distro with a package manager (like RPM or apt) is to have dependancies taken care of. If you install a new version of ssl or gnome or some core package by hand, you've broken your dependancy tree and it's a mess trying to get back on the rails. Especially if you didn't pick the same directories as the package maintainer for your installation.
Cheers,
ZicoKnows@hotmail.com
> Why, Anonymous Coward, of course!
--
Sheesh, evil *and* a jerk. -- Jade
Eric Raymond has hacked up a new kernel configuration scheme that is supposed to be considerably better than the one used today. Too bad it seems to late for it to be used in 2.4, but hey, it looks great! :)
--
"Oppression and harassment is a small price to pay to live in the land of the free." -- Montgomery Burns.
JFS, XFS, and ReiserFS all rely on a layer called pagebuf that hasn't been fully agreed to or implemented yet.
This is my signature. There are many signatures like it but this one is mine..
Will the patch for 2.3.99-pre5 work for this one?
Well, I never found the process complicated or terribly trouble prone. I know this isn't terribly helpful to you.
/usr/doc/lilo-x.xx directory and understand it reasonably well.
The hardest part of teaching is remembering what it is like not to know. Perhaps if you can be more specific about what you found complicated it would help somebody help you.
I think the complexity, such as it is, comes from PC BIOS. It's a real pain. LILO does a great job in simplifying dealing with this, but it sure is possible to screw things up, and there's lots of stuff that's simply hard to do. Try changing the IDE controller you're booting from, for example. Ugh.
That said, simply creating a new kernel and booting from it is about as simple as things get. Give the kernel a new file name, and a line for the boot image in lilo.conf, and run lilo. It works like a charm for me.
Setion 4.4 of the Kernel-HOWTO discusses this, but if you don't have things set up just that way, you need to It's important to read the README in the
Post may contain irony: discontinue use if experiencing mood swings, nausea or elevated blood pressure.
Why does all the cool stuff come out after potato is frozen?
Does this seem familiar to anyone (kernel 2.2)? That's the reason I moved away from Debian. Don't get me wrong, Debian is a very cool distro with many innovative and intelligent features (especially the package management system). However, they just can't seem to get on the ball with their releases; as soon as you install one, a lot of your software is instantly old. For my purposes, I can't run the latest unstable or frozen system, so I have no choice but to not use Debian.
Washington, DC: It's like Hollywood for ugly people.
The "gotchas" I know about:
There is some kind of bug with networking where the network suddenly dies and you have to reboot, my roommate and I both had problems with it as late as 2.3.99pre5, but another rommate didn't even have a glitch in networking.
Also there was no new support for Ultra66 (doesn't bother me, I don't have ultra66).
My questions are:
Was support for ultra66 added?
Did the networking bug get tracked down/fixed?
Does anyone know if ext3 support is included in this version and/or the *real* 2.4?
Is it any good as far as JFSes go?
How long before we can expect tools developed for ext3, is there anyone working on that now?
Will this kernel series have native support for IBM's JFS?
The reason I'm asking so many questions about JFS is that many companies are not going to Linux because of the lack of a Journaling File System.
Devil Ducky
Devil Ducky
MY peers would get out of jury duty.
What a bunch of diddly changes to an already stable system. My need to upgrade = zero.
Whoa, this is a genuinly funny post. What exactly makes it off-topic? Humor is never off topic.
A deep unwavering belief is a sure sign you're missing something...
Actually, even C++ (which is a pretty lightweight resource-wise) OO language needs significant runtime resources. Even in BeOS (Where C++ is modus oparendi in user level stuff) C++ is discouraged for kernel-level stuff (namely drivers and modules)
A deep unwavering belief is a sure sign you're missing something...
Hello, Linux is mainly a non-Java bytecode CPU. (Yes, I'm being sarcastic, Linux doesn't run on any Java bytecode CPUs.)
A deep unwavering belief is a sure sign you're missing something...
Damn you Linux zealots are idiots. What is wrong with expecting a NEW distro to come with up to date software? Its not that hard, but unlike you, normal people actually have work to do. They can't spend their entire weekend updating their Linux distro. Does expecting the people who's responsibility it is to maintain the distro to actually maintain it make you less 31337 (Or however it's spelled.)
A deep unwavering belief is a sure sign you're missing something...
Actually, the goto is not a bad thing.
.. catch { } code in C++ is just another example of a goto -- except that it's more structured around specifically handling error conditions within a block of code. The same concept can be applied to C very successfully.
:-)
Not when used correctly, that is. As unconditional jump within a function body, it allows you to handle all exceptions in one place easily.
For example:
if(!connect("site"))
goto error;
if(!send("string"))
goto error;
....
error:
handle_error();
(Yes, this is simplified)
As you can see, it saves you from doing the usual solution (calling another function, either passing it the variables or adding to the global variable namespace). This is desireable when performance is critical, or when the recovery is so trivial it does not justify the expense of a function call (ie: you just rewind a stream by one byte). Also, if this code is specific to one function, why make two?
As another example, any try { }
Of course, this doesn't mean you should go and used hundreds of gotos. That's like using hundreds of for() statements, or while() loops when there is no reason to. And no one really knows what will happen if you goto something in another function (C has no ret instruction equivalent)
---
--
Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
I hope this hasn't already been posted, but here is a pretty convenient and well written list of the new 2.4 features.
One of the things that really tick me off is that Linux (or is it just EXT2?) doesn't (as far as I know.. only spent one afternoon trying to find information) have any way to resize partitions after you have made them. This is something that I expect from DOS and windows 9x.. not Linux.
The only solutions that I found where commercial ones like Partition Magic. Surely there is a place for a Open source project in this department!
Rami
--
rJames.org - illustration
the fact is that there are really all of about 6 comands needed to compile and instal the kernel.
/. is not a forum for that kind of advice anyway. Take it to comp.os.linux.setup. Read the HOWTOs. Look at every file an /boot and wonder what it is doing. Don't expect it will be easy - it won't. You will make mistakes. You can hose your system. There are ample warnings. There are reasonable alternatives, such as getting prepackaged distributions from vendors.
/proc/ to see what options you are currently using. Don't expect it to be easy. Expect that you will screw things up. Keep a boot disk handy, and lots of coffee. Back up your system. And enjoy. There are few things that give more of a sense of accomplishment than building a new kernel to solve some problem specific to your setup.
make config
make dep
make clean
make *Image
make modules
make module_install
The fact is that there are several hundred options that must be addressed in make config.
Just stop for a second and think of this process on any other OS. It is virtually impossible to build a customized kernel for Windows, NT, MacOS. But on Unices it is not uncommon. It is downright difficult, and the cause of many a seasoned sys admin to pull his hair out to need to rebuild a kernel on Tru64, Solaris, or other commercial Unices. It is SOOOO much easier on linux it is not even funny. Yet still, there are many ways to screw up. You open a strange can of worms when you decide to begin to build kernels.
If you are like me, you began to build kernels because your hardware was not ordinary. I need reasonable NFS for some machines and that requires patches. I need a development kernel so my laptop Cardbus will work without locking on interrupt conflicts, and that required some fiddling.
Yet still, when I build a development kernel, it takes 3-4 tries to "get it right". If I am lucky I can build a standard kernel on the first try, but that is not the norm.
So when someone tries FOR THE FIRST TIME and doesn't quite get it right, I say "You are expecting too much."
Consider yourself blessed if you have the option to customize a kernel. Consider yourself lucky if you can build one and have it work on the first try. Consider yourself lucky if your system never gets hosed. If that is too much to accept then install the kernel from your distribution. They spend a lot of money and time and effort to ensure that your kernel will not hose your system. Building a development kernel from source is NEVER a low risk endeavor. If you do it lots you will still sometimes get burned.
There are AMPLE warning about the process in the kernel READMEs. You will ALWAYS see lots of precautions to people who ask about building development kernels. Yet still, sometimes, people ask questions about why it is so difficult to build a kernel under linux.
So you have to wonder, has this person EVER built a kernel for ANY other OS ?? Why should he think customizing a kernel with HUNDREDS of options should be something failsafe ? When he ignored the README files in his kernel source ?? The initial stance is completely tweaked, and that needs addressing.
That being said,
And if you suffer. And if you work long and hard at it, you will be able to build successful kernels, most of the time. Sometimes you will still screw up. That's life on the bleeding edge. Why do you think they call it the bleeding edge, anyway ??
If you want advice, I will tell you to keep an xterm open whil you run make config, and grep a lot in
*If* you are as capable as you imply, why not lend this poor fellow a helping hand and teach him, and many others, something useful. Your comment merely builds the impression that our favorite OS is only for self-important jerks. If that is the case, perhaps I'll switch to something else.
/. post. He should read the HOWTO, backup his system, brew a lot of coffee, and prepare for a long evening.
Come on. Someone tries to build a kernel without even looking through HOWTOs, or reading up on how to build kernels at kernelnotes.org. He doesn't know the role of System.map (which will, BTW, not hose your system, just leave you without symbols, generally innocuous).
The HOWTO says MUCH more than I will ever place in a
The other alternative is for him to use prepackaged systems that upgrade the kernel for you, such as those from all major OS vendors. Even that is a step beyond what MOST linux users do.
His, and your, expectancies are a little high. Most users NEVER build a kernel. This is a role that is moving more and more towards the distributions. That being said, building a kernel is a lot easier than other builds, like emacs, or XFree86, or GNOME, or KDE2. Downright simple in comparison.
If you want to switch to another system because you don't like the attitude of its users, you are using an OS for reasons far different than mine.
Get this thing fixed already 2.3.x is very much faster than 2.2 and im sick of using buggy software
If you are sick of bugy software why are you using test kernels? Switch back to a stable version. To quote from www.kernel.org "The latest stable version of the Linux kernel is: 2.2.15 "
Or better yet, how many patches have you submitted? Thats how open source is supposed to work. If you don't like it, fix it yourself or shut up and wait. I'm sure everyone would appreciate it being done sooner, but you to come here and bitch about it not being stable should not be tolerated.
For the record, I've never submitted a patch to the kernetl, but I've never bitched that the "test" kernels were buggy!
-- Freedom means letting other people do things you don't like.
There are seriously cool things making their way into the kernel these days. My two favorites are LVM (the logical volume manager: Heinz Maushagen's baby that mimics the capabilities of HPUX and AIX logical volume managers; more info at http://linux.msede.com/lvm/) and ReiserFS (ok, not really in yet, but they're making great progress).
But there are major problems with really fundamental stuff left. The VM system has been undergoing somewhat fundamental changes in the past week and a half. If 2.4 comes out any time in the next month, I'm waiting for 2.4.5 or 2.4.10 before I put this thing into production.
Still, with the rewritten pci device interface and cleaner APIs to a bunch of kernel functionality, I'm more excited than ever to start working with the new kernel in development mode at least.
It took me a while to find it (the search engine on linuxjournal.com is really not what it should be!) but here's a link:
3 882.html
http://www2.linuxjournal.com/lj-issues/issue73/
very interesting article. it starts by addressing the first yucky thing i found about python (that whitespace and indentation actually *matter*!). it goes on to convince me that this is a language i really need to learn.
Yeah, i havn't bothered to upgrade my kernel either (Outside of changing to a newer distro), but i think it's time for me to compile 2.4. Most of the bugs in the todo list don't seem that serious (for me), the Tulip NIC problem is the only one that might cause problems. And hey, there are bugs in 2.2.x as well...
/dev stuff, USB and a few other smaller changes, i wouldn't bother either though.
If it wasn't for the new
Syllable : It's an Operating System
If you recompile without changing versions and include less modules in the new kernel, you might get complaints about object files when rebooting. So,
and the problem goes away. This one is puzzling when you first start making custom kernels, but it is documented.I think what the previous poster was referring to is that you won't notice the problem until you reboot, if you don't always reinstall modules.
---------///----------
This post is not redundant, please don't moderate it as such. I repeat, this post is not redundant.
--
I like to watch.
I really don't see what is so complicated about this:
Is that really so bad? Your lilo complaints are just ignorant: the Kernel-HOWTO does mention zlilo and bzlilo, last time I checked.
Recompiling a Linux kernel is quite simple compared to other Unices. Coming from an old-style BSD and new SysV background, I find the Linux way to be rather luxurious.
I'm not trying to flame you. I just think you didn't RTFM. If you fscked up that badly, you didn't read it well enough. And didn't you back up your old kernel first? You didn't make a boot disk, did you? You didn't test the new kernel before "swapping" it, did you? If you've done your homework, Linux is very forgiving.
I'd be interested in hearing more of your problem. What did you find "complex?" What problem necessitated a full system restoration? Details, please.
---------///----------
This post is not redundant, please don't moderate it as such. I repeat, this post is not redundant.
--
I like to watch.
///---------------------------------------
Torvalds begins work on Linux 2.3.48.9.2.7.43, possibly
Posted by CmdrTaco on Sunday February 27, @10:36AM
from the rob-sucks-tarballs dept
Linus Torvalds, creator of Linux, accidentally hit his keyboard with his elbow today. We have yet to receive confirmation that the resulting code will be be included in the next development kernel, but we can never be too sure. Here is the code in full:
This won't compile under GCC, so we can only assume the code is pretty experimental. Look for the tarballs to be released this evening.
Torvalds comments, "What? Oh, yeah, I accidentally hit my keyboard with my elbow when I reached to get my tea. What? Is it part of the new kernel? You're kidding, right?"
We'll update the article as soon as we get more information. The Linux world hasn't been in such frenzied anticipation since the release of kernel 2.3.48.9.2.7.42, which was about ten minutes ago.
Interview: Alan Cox farted
Posted by Hemos on Sunday February 27, @10:34AM
from the whats-that-smell dept
Linux guru and hacker-extrodinaire Alan Cox farted earlier today. What do you think this says about the future of Linux development? Alan's ass will respond to the highest moderated posts later this week.
ESR and JonKatz to participate in "Zealot Deathmatch"
Posted by Roblimo on Sunday February 27, @10:33AM
from the die-bitch-die dept
Open source proponent Eric S. Raymond and Slashdot nutcase JonKatz are reportedly organizing a "Zealot Arena Deathmatch" to raise money for the Apache Software Foundation. The fight is expected to be a tough one, because while Katz is genuinely insane, ESR has the power of girly, elfish looks. A spokesman from Apache says that, "while we don't encourage violence, we'll do anything for money."
VA Linux aquired by Klingons, Rob bows down to new alien masters
Posted by emmett on Sunday February 27, @10:32AM
from the star-shit-enterprise dept
VA Linux Systems, owner of Andover.net, owner of Slashdot.org, owner of Rob's ass, was officially aquired by the Klingon Empire earlier this morning. The Klingons, who have recently taken over Kellogs, GM, and Disney, are looking forward to absorbing more major corporations in the near future. The US Government is discussing investigating the Klingons for holding a monopoly over "every aspect of our lives", to which the Klingons responded, "Puny human scum! I will crush you like a bug and feast upon your steaming entrails." Finally, some competition for Microsoft!
Red Hat and VA stock at all time high!
Posted by CmdrTaco on Sunday February 27, @10:31AM
from the i-am-so-rich dept
Dude, have you heard the market reports today? I am so fucking rich! If this keeps up, I'll be able to stop doing this Slashdot crap! Hell yeah!
---------------------------------------///
I should really update that last one, though, in light of recent events:
///---------------------------------------
Red Hat and VA stock at all time low...
Posted by CmdrTaco on Thursday May 25, @10:31AM
from the i-am-so-fscked dept
Shit... say, how's the job market for goateed Perl-monkies?
---------------------------------------///
---------///----------
This post is not redundant, please don't moderate it as such. I repeat, this post is not redundant.
--
I like to watch.
for ordinary users to compile kernels and install new releases.
It's a magnificent pain in the ass. Get the right compiler and new version of the utilities and make sure all of them are built and properly installed just to find that, say the VM system is still being balanced and the kernel can't be used for useful work right now.
It's necessary if you want to contribute to testing the new kernel. It's necessary if you have hardware thiat is wither not or poorly supported in stable releases of the kernel.
But the vast majority of people can get along with their old kernels (perhaps with minor version upgrades i.e. 2.2.5->2.2.15) and wait until the distribution maintainers release a new kernel version to upgrade.
Anomalous: inconsistent with or deviating from what is usual, normal, or expected
Anomalous: deviating from what is usual, normal, or expected
Canard: a false or unfounded repor
Check out BuildKernel, it's an automated bash shell script that can download and install your kernel. If you don't want to configure your kernel, it can use your old .config file from your current kernel. Check out http://users.dhp.com/~whisper/ buildkernel/index.html.
Daniel
And lets face it, Tux would have to be the cutest thing in a Disney movie since...well, since, forever.
Hopefully I didn't put any [] around my words.
"It's not a real 2.4.0 release, but we should be getting closer"
In other words, this kernel may as well be called 2.3.99pre10 for all that it counts. I hate to make the comparison, but I'm going to: was Win2000 RC1 beta software or was it the final version? It was a beta.
The same goes for this - in this case I think the 2.4 designation is worth very little. It's still got significant bug levels and people shouldn't be jumping the gun to get at it.
All MHO, of course.
--
Said it couldn't last, said it wouldn't last... This is the last stand against tomorrow's world.
And there is FIPS, working under DOS. I used it when installing my first Linux (a RedHat). Lots of fears before actually trying, but absolutely no problem. Sure, there's no GUI.
sigmentation fault
If you're coming from a 2.2x kernel (I wouldn't suggest the 2.4 test kernel from anything less than a good 2.2 installation) you'll probably want to go grab the latest modutils package... it'll make life easier.. trust me.
Skiers and Riders -- http://www.snowjournal.com
More reasonable alternatives would include:
Yes, these languages don't have syntax that slavishly resembles C. But it's not as if the actual semantics of C++ or Java are actually that much like C...
If you're not part of the solution, you're part of the precipitate.
What sort of system is needed to run the current 2.3 and 2.4-test kernels? Going from 2.0 to 2.2 was a big jump in memory consumption, although apparently things did improve a little during the late 2.1 series.
Will I still be able to run the latest kernel on my 8Mbyte machine?
-- Ed Avis ed@membled.com
From the readme:
;)
:)
.sig: File not found.
Have fun. And let's see how many people find this without it even being
announced
Announced on slashdot is about as bad as it can get
Dan
ls:
ls:
(A)bort, (R)etry, (I)gnore?
Here's how to get it:
1. download 2.3.99pre9 from ftp.kernel.org/pub/linux/kernel/v2.3
2. download the pre10-3 patch from ftp.kernel.org/pub/linux/kernel/testing
If you look at the pre10-3 patch, it says 2.4.0-test1 in the Makefile.
If exchanging a kernel is such a dang-blasted important task for any Linux user to know how to do, why is it so complicated?
I assume you refer to the difficulty of configuring the kernel correctly, i.e. running make {config|menuconfig|xconfig} and choosing the right options for what you want to do. I give the anwer: it's so complicated because it has to be. With power comes complexity; there really isn't any good, flexible way around that. There are lots of options because people run Linux on lots of different kinds of systems, and not all of them want, for example, SCSI support, or network support, or USB support, etc.
As near as I, at least, can tell, if you take away the complexity, and the requirement of knowing what you're doing, you'll dramatically decrease the flexibility and (obviously) the configurability of the OS.
As for Don't say that a 'normal' user doesn't need to do a kernel swap, well, sorry, most users don't need to do a kernel build. They can quite happily wait for RedHat or Debian or SuSE or whomever to release a 2.4.x kernel and download the binaries.
I compiled from source my first kernel two days ago, and let me tell you: it ain't something that I would recommend any user do.
It isn't something I would recommend just any user do, either. At least, not without some sort of preparation and due diligence (e.g. read the help files, and don't say yes or no until you know what they're talking about). I'm not trying to imply you didn't prepare and do your due diligence, by the way; I'm agreeing with you.
1. Why is swapping the Kernel so complicated? Why not automate this more?
I'm not positive, 'cause I don't use it, but I think that Debian, for example, has automated new kernel builds. I couldn't speak for any other distros, not having used 'em.
The HOWTO that comes with the kernel source doesn't mention ANYWHERE that you have to swap in the newly compile System.map file so lilo and your system don't have hissy fits.
I'd agree that that's a problem. I hope you send/sent an e-mail to the HOWTO maintainer.
My
Maybe Linus should have put an NDA around that README file. Then he would now have grounds to sue Slashdot!
Say no to software patents.
Now that I've made some sort of introduction, I would like to ask a simple question of the Linux using community: If exchanging a kernel is such a dang-blasted important task for any Linux user to know how to do, why is it so complicated?
[Don't say that a 'normal' user doesn't need to do a kernel swap, it doesn't lead to any interesting discussion. I did it because I wanted to test the USB stacks with some of the devices that my company develops.]
I compiled from source my first kernel two days ago, and let me tell you: it ain't something that I would recommend any user do. In fact, I royally screwed my system to the point that I had to restore from backup!
The questions presented:
--
I realise that some of what i referenced may be a bit off (that System.map thing in particular). Be gentle.
Rami James
Pixel Pusher
ALST R&D Center, IL
--
rJames.org - illustration
McDonald's gets the merchandising deal (Happy Meals come with "Linux Heros" figurines) and Eisner makes another few million.
The End.
---------///----------
This post is not redundant, please don't moderate it as such. I repeat, this post is not redundant.
--
I like to watch.
And another other open source releases announced on /.
Please moderate accordingly, and your crack will be delivered as per usual.
I'm just wondering. Who, if not Alan Cox whould maintain the ac (Alan Cox) patch-series? 8-)
Why, Anonymous Coward, of course!
Before people get their hopes up, you should check out the length of Alan Cox's Todo list. It is not small :(
/proc/scsi/BusLogic/0 (Robert de Vries)
Capable Of Corrupting Your FS
-----------------------------
E820 memory setup causes crashes/corruption on some laptops
Use PCI DMA by default in IDE is unsafe on VIA VPx x<3
Security
--------
Fix module remove race bug (mostly done - Al Viro)
exec loader permissions
Semaphore races (fix in 2.2)
Semaphore memory leak (fix in 2.2)
Exploitable leak in file locking (Willy)
TTY and N_HDLC layer called poll_wait twice per fd and corrupt memory
ATM layer calls poll_wait twice per fd and corrupts memory
Random calls poll_wait twice per fd and corrupts memory
PCI sound calls poll_wait twice per fd and corrupts memory
sbus audio calls poll_wait twice per fd and corrupts memory
access_process_mm oops/lockup if task->mm changes (Manfred) [user can cause deliberately]
RtSig limit handling bug
Signals leak kernel memory (security) [FIX in ac tree]
Boot Time Failures
------------------
IDE fails on some VIA boards (eg the i-opener)
AHA29xx driver appears to stomp other cards
Use PCI DMA 'lost interrupt' problem with some hw [which ?]
(NEC Versa LX with PIIX tuning)
HT6560/UMC8672 ide sets up stuff too early (before region stuff can be done)
Crashes on boot on some Compaqs ? (may be fixed)
IBM MCA driver breaks on Device_Inquiry at boot
DEFXX driver appears broken
ACPI hangs on boot for some systems
In Progress
-----------
Dcache threading (Al Viro)
Merge the network fixes (DaveM)
Finish I2O merge (Intel/Alan)
Fix all remaining PCI code to use new resources and enable_Device (mostly done)
Fix Exists But Isnt Merged
--------------------------
Update SGI VisWS to new-style IRQ handling (Ingo)
64bit lockf support
Support MP table above 1Gig (Ingo)
Finish sorting out VM balancing (Rik Van Riel, Juan Quintela et al)
Dont panic on boot when meeting HP boxes with wacked APIC table numbering (AC)
Scheduler bugs in RT (Dimitris)
Fix eth= command line
HFS is still broken
AIC7xxx doesnt work non PCI ? (Doug says OK, new version due anyway)
8139 + bridging fails
Fix hpfs_unlink (Al Viro)
put_user is broken for i386 machines (security) - sem stuff may be wrong too
BusLogic crashes when you cat
Loopback fs hangs
To Do
-----
SHM code corrupts memory
Floppy driver broken by VFS changes. Other drivers may be too
(Stuff gets called after _close now - unload race possibly too)
Tulip hang on rmmod/crashes sometimes
Devfs races, Sockfs (removing NULL ->i_sb stuf) (Al Viro)
Restore O_SYNC functionality
Debian report that the gcc 2.95 possibly miscompiles fault.c or mm/remap.c
(Perl script available from Arjan)
Fix further NFS races (Al Viro)
Trace numerous random crashes in the inode cache
Test other file systems on write
The netdev name changing stuff broke GRE
Audit all char and block drivers to ensure they are safe with the 2.3
locking - a lot of them are not especially on the open() path.
Stick lock_kernel() calls around driver with issues to hard to fix nicely
for 2.4 itself
PCMCIA/Cardbus hangs, IRQ problems, Keyboard/mouse problem (may be fixed ?)
pci_socket crash on unload
truncate_inode_pages does unsafe page cache operations
Linux sends a 1K buffer with SCSI inquiries. The ANSI-SCSI limit is 255.
Linux uses TEST_UNIT_READY to chck for device presence on a PUN/LUN. The
INQUIRY is the only valid test allowed by the spec.
To Do But Non Showstopper
-------------------------
Make syncppp use new ppp code
Finish 64bit vfs merges (lockf64 and friends missing)
NCR5380 isnt smp safe
DMFE is not SMP safe
Go through as 2.4pre kicks in and figure what we should mark obsolete for
the final 2.4
Union mount (Al Viro)
Per Process rtsigio limit
Fix SPX socket code
Boot hangs on a range of Dell docking stations (Latitude)
iget abuse in knfsd
Some people report 2.3.x serial problems
USB hangs on APM suspend on some machines
PCMCIA crashes on unloading pci_socket
ISAPnP IRQ handling failing on SB1000 + resource handling bug
TB Multisound driver hasnt been updated for new isa I/O totally.
Fix boards with different TSC per CPU and kill TSC use on them
DVD-RAM is apparently not working for write currently (Rogier Wolff)
Compatibility Errors
--------------------
Xterm broke in 2.3.99pre6 (FIONREAD/select loop)
Probably Post 2.4
-----------------
per super block write_super needs an async flag
addres_space needs a VM pressure/flush callback (Ingo)
per file_op rw_kiovec
Drivers In 2.2 not 2.4
----------------------
To Check
--------
Check O_APPEND atomicity bug fixing is complete
Protection on isize (sct) [Al Viro mostly done]
Mikulas claims we need to fix the getblk/mark_buffer_uptodate thing for
2.3.x as well
Network block device seems broken by block device changes
Fbcon races
VFS?VM - mmap/write deadlock (demo code seems to show lock is there)
rw sempahores on page faults (mmap_sem)
kiobuf seperate lock functions/bounce/page_address fixes
Fix routing by fwmark
Some FB drivers check the A000 area and find it busy then bomb out
rw semaphores on inodes to fix read/truncate races ? [Probably fixed]
Not all device drivers are safe now the write inode lock isnt taken on write
File locking needs checking for races
Multiwrite IDE breaks on a disk error [minor issue at best]
ACPI/APM suspend issue - IDE related stuff ?
NFS bugs are fixed
Floppy last block cache flush error
Chase reports of SMB not working
Locking on getcwd
floppy fails on some machines
IRDA calls get random bytes before random is set up
Some AWE cards are not being found by ISAPnP ??
SHM segments not always being detached and destroyed right ?
I'm just wondering. Who, if not Alan Cox whould maintain the ac (Alan Cox) patch-series? 8-)