Slashdot Mirror
Linux 2.2.10
John Campbell writes "Linux 2.2.10, the next in the stable kernel series, is out. It's about a 300k patch, and it contains the fix for the DoS bug recently reported here (if you've already patched it, don't let patch back the fix back out... you are using patches, right?), as well as assorted other stuff.
The new kernel patches can be found, as usual, on ftp.xx.kernel.org, where xx is your country code. "
When I checked, ftp.us.kernel.org had linux-2.2.10.gz but not patch-2.2.10.gz..that appeared a few seconds later so I guess I got it hot off the press :-)
Daniel
Hey, this problem is easily fixed. /boot and it works fine.
/usr/src/linux/System.map /boot)
When you compile the kernel, it generates a System.map, just copy that to
(e.g. cp
have fun
"Computers will never truly be free until the last windows user is strangled with the entrails of the last mac user."
Posted by NJViking:
/usr/src:
Even easier:
In
bunzip2 -cd patch-2.2.10.bz2 | patch -p0
(if you downloaded it in bz2, which I do 'cause it's less.)
-= NJViking =-
Got a couple of these buried in patch-2.2.10.gz:
Took 'em out and the patch applied successfully.
I tried to patch Red Hat's kernel 2.2.5 source (to 2.2.6), but it did not work correctly. Something they did (in their 15 revisions, i.e. the fact that it's version 2.2.5-15) appears to clash with the 2.2.6 patch. How do I do a 'diff' to see what they changed? Has anyone had success getting around this when patching? Is it safe to just do a new install of pristine 2.2.10 source, and go from there?
If need be, I can try again and post a list of errors.
Thanks for your help.
Ryan
Posted by NJViking:
I have to agree on this. My buddy wants to use Linux, but it just won't work with his Diamond M80 sound card. Go figure..
-= NJV =-
.. happy with his AWE64 Gold
I think he's trying to say that it's been a lot more than 72 hours since he last upgraded his kernel...
Just a quick tip for if you have it installed in a different directory that doesn't end in the word 'linux' (i.e. /usr/src/linux-latest or something like that):
cd into the actual linux directory itself, and then do:
zcat patch-x.x.x.bz2 | patch -p1
The '-p1' option to patch tells it to ignore the first part of the filenames in the patch file.
I know this, but I was also testing just te regular kernels. Then the ac's. It may have sounded like I meant just the ac's, but I also meant the kernels and ac's also tested. So I don't know what it is, o well. :) Maybe I will try a newer kernel sometime soon.
How can you possibly assume that *any* code is completely stable. Marking something as stable just means that it isn't prone to crashing and that the system is worthy of production usage. There will still be the occasional bug in "stable" kernels and the process of finding them is called "testing" -- its an ongoing effort.
I don't think it was a hanging problem, more like really slow and tons of errors on boot and stuff not working for some weird reason. So, I don't know what the problem could be. :)
From KernelNotes - 09-Jun-99: I'm leaving to drive to Arizona today which will take about a week. Expect some delays for the 2.3.6 and 2.2.10 changelists. Sorry.
*sob*
-'fester
Anyone else noticed that the sblive module seems to do odd things under .10? Like not work at all? I hate binary modules too, but I like the sblive... thou at this rate, i may swap my ensonique back in.. Grrr, wish creative would at least release a module compiled against the right kernel ver... shouldn't take that much effort, imho.
Also, anyone know of good changelogs besides edge.kernelnotes.org? single point of failure is never good....
David
This sig left intentionally blank.
Does anyone know if this (or any other patch) will fix the problem that prevents >2.2.7 from building on a ppc?
It seems that 2.2.8 broke a bunch of stuff with IDE for PowerMacs. When trying to build 2.2.8 or 2.2.9, I get errors like this:
syscalls.c: In function 'pmac_init'
syscalls.c: 'ISA_DMA_THRESHOLD' undeclared
syscalls.c: 'DMA_MODE_READ' undeclared
syscalls.c: 'DMA_MODE_WRITE' undeclared
make[1]: *** [pmac_setup.c] error 1
I tried commenting out the three offending lines in pmac_setup.c (ISA_DMA_THRESHOLD, etc.), but it only goes slightly farther before erroring out on kernel.o.
I found a partial fix in the mailing list archives:
Add
#include
to arch/ppc/kernel/syscalls.c
Seems someone forgot to include this.
With that change, I can compile, but IDE support is broken (probably because I just commented out that stuff since I don't know how to fix it).
Since I have an iMac (IDE-baed), both IDE support and USB support are extremely important. I can't get either the usb stack included in 2.2.7 or the uusbd stack to build on 2.2.7 or later (will 2.3.x kernels work better?).
Yes, I could stay with my 2.2.6 kernel, but this is really bugging me. The main reason I'm using linux is to learn it. As far as I can tell, nobody (with the possible exception of that one guy who posted to the mailing list) has gotten >2.2.7 to work on a PowerMac.
Personally, I take great pleasure in spending endless hours tweaking my system. I wouldn't want it any other way.
The same thing can happen with your oh-so-superior FreeBSD. As a matter of fact, very, very little time is spent tweaking the kernel for me; most of it is on programs on top of the kernel. If I were using your precious FreeBSD, it would be exactly the same thing.
P.S. No disrespect meant towards FreeBSD, only towards this silly little troll...
----------
In a real emergency, we would have all fled in terror, and you would not have been notified.
Follow the instructions on http://www.thecompute rgallery.com/redhat/kernel-upgrade.shtml and you shouldnt have any problems
have fun.
lyte
Well, I like to think there have been some useful features added to the kernel since 1.0. :)
> Actually, I can think of some instances in which sound card drivers could be vital for a
> web server. How about streaming audio over the 'net from a live sound source?
When you do live streaming audio, you use don't use your web server to encode the audio - you use a separate, dedicated machine for that. This is true for Real Audio as well as Shoutcast and the open-source Icecast.
(I'm not talking out my ass here - I geek for Technostate, an internet radio station that broadcasts live events.)
"Stable" is a relative term. Name a commercial OS that is bug-free, and which will never develop "issues" for one or more of its customers. There aren't any.
Put another way, 2.2.X is "stable," and getting more so as hundreds of millions of hours of experience in hundreds of thousands of different environments accumulate.
Look at the case with commercial OS's. Solaris 2.X is widely regarded as one of the most stable general-purpose OS's out there. MS doesn't even pretend that WinNT is as stable as Solaris. Yet Solaris 2.3 was only barely usable. (God help anyone using a version before that.) 2.4 had thousands of reported bugs before 2.5 was released, and 2.5.1 came on 2.5's heels to fix some serious bugs in it. After more than five years, Sun finally released a version of Solaris 2.X that would fit your definition of "stable."
Compared to this, Linux is doing amazingly well. Free software can't defy the law of gravity any more than commercial software. A few bugs that remain latent with a community of a few hundred developers will almost always manifest when released to a few million users. But the great majority of those users will never see those bugs. They'll get fixed anyway, and responsible sys admins will upgrade when they see sufficient benefits in doing so vs. any disruptions or other risks. It seems vaguely neurotic to upgrade to a new kernel just because it's there. And it's naive in the extreme to claim that any software several hundred thousand lines long can be "bug free."
Linux version 1.2.13 (root@mymachine) (gcc version 2.7.0) #2 Mon Nov 25 18:43:25 EST 1996
9:06pm up 388 days, 31 min, 2 users, load average: 0.01, 0.01, 0.00
Bah.
I thought Alan Cox said he'd go up to 2.0.38 a while back in his diary.
It's fun to watch stupid FleaBSD users pretend they have no serious bugs in a kernel which is used by about 1% of Internet users... FreeBSD 3.0 and 3.1 exploit of the week
Heh, i'm not sure if you knew this...
but you don't have to compile sound support in!!
Woah!!
Come on you troll, you know better then this.
Least your not an AC.
"Computers will never truly be free until the last windows user is strangled with the entrails of the last mac user."
2.2.7-ac1 is my baby. Has anyone else had this same situation? I was following the ac's pretty steadily and 2.2.7-ac1 came across. Nice, works good. But as I tried the next ac's of the next kernels, something happened. Stuff started to not work, went weird, or was just plain dumb working. So I ran back to 2.2.7-ac1 and that's where I am now. With the release of 2.2.10, I still feel behind the times, but as Linus said at the last Linux Expo, you don't need to be cutting edge to have your stuff work good. He suggested that people even stay with 2.0.3* if that works good for them. So, I think I'll be staying with this for a while until like 2.9.0 comes out or something. :) Anything have this situation also?
Try this, changing the ftp site, and zImage to bzImage if you have to.
/tmp i nux/kernel/v2.2/patch-$*. /usr/src/ /tmp/patch-$*.bz2 | patch -p0
-snip here-
#!/bin/sh
cd
wget ftp://sunsite.org.uk/Mirrors/ftp.kernel.org/pub/l
bz2
cd
bzip2 -dc
cd linux
make xconfig
make dep && make zImage && make modules
-Yarn - Rio Karma: Excellent
Anybody can contribute to kernel development, whether stable or dev. It is absolutely wrong to think that only people who can code and find/fix bugs can contribute. All that is needed is the ability to file a useful bug report. That's all.
Obviously being able to submit a patch fixing the bug is even better, but that's not a requirement to being useful.
Where you might remotely have a point is if someone uses a dev kernel and expects everything to work beautifully. People downloading dev kernels should expect problems. But they don't have to be coders to be able to report them.
----------
In a real emergency, we would have all fled in terror, and you would not have been notified.
I guess edge.kernelnotes.org will have the changes listed pretty soon.
2.2.10 is not a dev kernel!
I currently run 2.2.9, and when the DoS bug came out I edited the source myself to fix the problem. How do I make sure the patch doesn't mess that up?
-----BEGIN GEEK CODE BLOCK----- Version: 3.12 GIT d? s: a-- C++++ UL++++ P++ L+++ E- W++ N o-- K- w--- O- M+ V PS+ P
/usr/src/linux/scripts/patch-kernel
I guess the rest is intutive..
> hate to sound like a snob
No, you sound like an ignorant, self-important snob.
You have no business wagging your finger at people who have trouble compiling kernels when you don't seem to know the difference between a stable (2.2.x) release and the current development (2.3.x) tree.
The guy was just asking a question, which someone else was able to answer simply. Seems a lot more constructive than your comment (IMO).
(sorry, I couldn't resist the urge to lash out. People like this make it hard for newbies to join the linux "movement" or "community" or whatever you want to call it.)
Well, very very strange, 'cause *.nl.kernel.org points to the server in England/GB...
I understand that any OS will have bugs. However, the Linux 2.2.x series has an unacceptable number of bugs, including unacceptably serious ones such as filesystem corruption. Windows 95, as bad as it may be in other ways, does not corrupt my filesystem.
My statement was mainly in reply to the "you're a beta tester" comment. If it's a stable release, then I'm not a beta tester. If the authors expect me to beta test their software, it should be labeled "beta" so I know what I'm getting into.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
I've had somewhat limited success getting the linux vfb support to work on 2.2.x kernels.
What I did was boot into DOS and invoke Univbe, a great little TSR
that allows you to emulate a VESA 2.0-compliant card even
if you don't have one. Then, ya just use loadlin with the
appropriate vga=xx parameters and you're on your way.
This sort of thing can be put in your AUTOEXEC.BAT without problems.
Nver mind, it's been too long since I've read the diary.
Posted by wMaVerick:
:-)
Thanks for the tip - you've just saved me a lot of typing!
I had problems going from 2.2.6 to 10, patchfile 7 didn't want to patch properly, but with your tip, I was able to easily apply all the patches on a clean 2.2.0
Cheers
I just want to inform you that L1zard_K1n6 is speaking of Linux the same way too many Linux Advocates speak of Windows.
What is good for the geese is good for the gander.
New Moderator Guideline: Anyone who says something bad about Linux is a Troll. Everyone knows that Linux is perfect and there is nothing bad to say about it.
Of course what L1zard_K1n6 is simply untrue. Linux is very tested. But some of these couter-flames are unwarented.
I thought slashdot was OS agnostic?
(Sure, leave it to me to go against public-opinion. Boy, am I gonna be flames!)
--
Anbody willing to share the secret to unlocking ES1688 sound cards with 2.2.x kernels? It worked fine under slackware 3.6 after "modprobe sound." but not after upgrading to a 2.2 kernel. Wouldn't work with slackware 4.0 either.
:)
I heard of some people who used RH were able to get it to work with the sndconfig utility. I'll probably be switching from slackware to RH6 soon anyway, but before I do that, if I could get it to work with a 2.2 kernel in slackware, it would save me the trouble
The card card is an ISA, non pnp card. I've tried changing the settings (io, irq, dma) when modprobing the sb module. I've used both A) the settings windows control panel lists and B) the actual irq io and dma that are set by jumpers on the sound card [yes, that's right, my non pnp card has jumper settings that windows ignores, yet the card works under windows!]
--- there can only be one.... not...
True, but you still have to have a machine with sound card support to do it, whether it's the same box that's running the httpd or not. My point was that there are legitimate uses for a sound card in a server. Maybe not a "web server", per se, but close enough for government work.
plus, very useful resource is http://www.atnf.csiro.au/~rgooch/linux/docs/kernel -newsflash.html. it is a collection of patches for the kernels which are not very willing to compile or run. (although it does not mention the mmap problem of 2.3.6).
announcements of kernels and weekly digests of linux-kernel mailing list always appear at http://www.linuxtoday.com.
no problem. patch proggie either will ask if this is already applied patch (answer y) or will dump that fragment into *.rej file.
you are 100% right. Slashdot seems to be anti-everything-but-linux. Or at least the most vocal crowd (of which most come from *.edu.. or so it seems).
Also, have you seen this article? She made one mistake... FreeBSD can run linux apps through emulation (possibly faster than linux runs them) and if your application comes with source code, it can pretty much run anywhere.
As a quick indication of what you would be in for if a patch did exist:
linux-2.0.36.tar.gz 7098 Kb Mon Nov 16 00:00:00 1998 Unix Tape Archive
whereas 2.2.10 is
linux-2.2... 13577 Kb Mon Jun 14 05:33:00 1999 Unix Tape Archive
so if (say) around half the 2.0.36 stuff was reused, the diff would contain ~3.5M of stuff to be diff'd out and ~10M of stuff to be diff'd in. This would make your patch ~13.5M. Better to have a nice shiny new kernel for the same dload.
Minor disclaimer I don't pretend to be an expert on diff --recursive, so I may be wrong in a trivial way.
-- open source? sounds like the real book --
Huh? If the year is 2050, and you don't have a direct input to wire your internal PCs into the global net you'll be sorry.
Yes, but when is the kernel fully going to support USB!!
Try 2.3.x for USB...
What are you talking about? This security hole has been around since 2.2.x. The kernels are quite well tested with the pre-patches and the ac patches.
How long does it take to get a bug fix into Windows once its been found? You just better hope you're not the only one affeted. There's certainly no shortage of bugs.
Geez guys, thats what we've been doing wrong! All we need to do is spend the next eon testing this kernel and then we can be sure that it has no bugs. Whew, its a good job there are clever people around to point us in the right direction huh?
*END SARCASM*
Moron.
And how would you propose that people test a given kernel against every possible DoS, error, and bug before it's released? If it were possible to synthesize all possible attacks in a test environment, we'd be using a flawless version of Linux 1.0 (as 0.99.x was tested using our mythical Uber-Workbench).
Nothing worth doing is worth doing today.
well at least we aren't spending more time restarting our systems than using them like some other os's.
matguy
Net. Admin.
matguy(.com)
Why haven't you tested it yourself before complaining about the lack of testing?
(Although it is tested quite some bit..)
One tests usually for problems expected, problems which are not expected (sometimes) are hard to find and hard to test for.. These problems are mostly found in the long run..
Tadaa.. Thats why the kernel is updated once in a while. Because someone found an error not suspected to be there before! (Or to enhance the kernel with features not present before or make it cleaner/leaner) If you have better ideas.. Post 'em.
Greetings Teukels.
You should have done a better job testing it then. You've got nobody to blame but yourself. _You're_ the beta tester.
Not on all the mirrors yet, the us and uk mirrors are still on 2.2.9. Found it on nl, though... odd that ;)
or act on them better yet.
matguy
Net. Admin.
matguy(.com)
Please use patches instead of downloading whole archives, saves a lot of bandwidth (bandwidth costs money for the mirrors). /usr/src" /path/to/patch-2.2.10.gz | patch -p0" /usr/src/linux. If you have an older version patch incrementally. /usr/src/linux" and "make menuconfig".
Here's how:
First, get the patch (duh)
now "cd
and then
"zcat
You will be required to have linux-2.2.9 installed in
Now, "cd
Thank you.
PS. zcat is the equivalent of "gzip -d -c"
Umm... Once in a while...
ROTFLOL
HAAAHHAAAAAAHHHOOOOOOOHHHAAAA
Nice one
I like FreeBSD, but I personally like linux better. FreeBSD and linux develop similarly, but FreeBSD simply more rarely slaps a version number on it.... It's more a matter of taste as to version number slap or not, in FreeBSD you can spend that much time rebooting, recompiling if you run a current cvs snap...
XML is like violence. If it doesn't solve the problem, use more.
You forgot "make modules_install". BTW. Can you make wget print contents to stdout? like "wget -stdout ftp://kernelorg/linux/patch.gz | zcat | patch -p0".
Well I guess this just explains it all now doesn't it? I guess the BSD kerneldeveloppers have a waterproof model against bugs.. (_NOT_)
But than again.. That also means Linux kernel is evolving/developed faster than BSDs is.
Greetings Teukels.
The best solution is to get the kernel-2.2 source RPM, update it, and build your own set of 2.2.10 RPMs. Then everything is properly installed including pcmcia, and its easy to back out if something goes wrong.
Anyone knows where to find a patch for 2.2.9 or .10 to get a VGA framebuffer? The VESA refuses to work with my gfx-card, guess mine isn't vesa2.0 compliant.
2.2.10 is a stable release, not a development release.
And you are right, you do sound like a snob.
--- Tao
who doesn't know the difference between the development and the stable kernels. Snob.
Where's the best place to find a collection of kernel news, i.e. reported bugs, security holes/features, etc.
Actually, I can think of some instances in which sound card drivers could be vital for a web server. How about streaming audio over the 'net from a live sound source?
Shall we point that FreeBSD use a lot of code that is 20 years old (from BSD 4.4). FreeBSD may be more young that linux in this incarnation but they use a lot of code that is older and then more tested. Linux is still maturing and have a lot of default, but it is maturing very fast.
"The obvious mathematical breakthrough would be development of an easy way to factor large prime numbers." Bill Gates,
No I'm not. Linux 2.2.0 was marked as a "stable" release, not a "beta" release "for testing purposes only." If it's a buggy beta release, it should be marked as such, rather than being mislabeled as "stable."
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
the OpenBSD people seem to be able to do it. Why can't the Linux people?
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
I don't know. There's no way you can compare 2.2.x to Windows 95 in all honesty and say that Windows is better.
I have 2.2.x running on servers for over 2 weeks and counting and it's been very stable.
Sure, it's not as good as 2.0.x, but it took 2.0.x years to get the same stability.
"testing a stable Linux kernel" is a contradiction. Either is it a beta version that needs testing, or it is a final version that is stable. You do not release beta versions as "stable" software.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
Stability greatly depends on hardware: I have had 2.1.126 running for some 220 days now on SMP P2 machine with SCSI and ATA software RAIDs and 512M memory. (Ok, that is not 2.2.x, but as close to it as it's possible with this uptime ;)). Based on that, I would call this kernel stable, on _this_ hardware configuration.
Vaadin - the best open source framework for building web applications in Java - no plug
rc@broadway:~ > patch -v
patch 2.5.3
Copyright 1984-1988 Larry Wall
Copyright 1989-1998 Free Software Foundation, Inc.
Well it looks like 2.5.3 is on SuSE 6.1. Hey you notice that larry wall did patch...hmm interesting, I guess perl was buggy enough at first to need patches constantly.
72 hours? 2.2.9 came out more than a month ago and if you don't use Linux why are you worried about it? Good to see your putting all your valuable time ("I can't spare 15 minutes to recompile the kernel *wahhhh*") to good use.
There's a file in the scripts directory called "patch-kernel" (or something similar... I always forget) which will automatically apply successive patches in the correct order. So if you're going from 2.2.5 to 2.2.10, you can just get all the patches (they can even be .gz'd I think) and run patch-kernel, and it'll bring you up to date.
:-)
Someone please correct me if I have a couple details wrong.
Out of sheer curiosity, are you one of those people who would rather pay $300 US plus taxes for testing the 'final' version of Windows 2000 versus $0 for testing a stable Linux kernel? Or are you one of those 3l33t3 haX0r d00ds that pirate all their software?
where is the changelog ?? I want to know what has changed. 300k probably not much. I knew this would be out soon.
Only 'flamers' flame!
So, don't get too worked up.
--
Get your fresh, hot kernels right here!
Sixth of September 1998...
Linux Version 2.0.34
What exactly were you trying to show?
Yes, and they're all for old kernels, just like the FreeBSD exploits he posted (2.x and 3.0 series).
All software has bugs; responsive developers fix them faster. More frequent releases of fixes means users can have better software at the expense of typing "patch".
No, It won't make a lick of difference to a webserver, but if Linux is going to become popular on the desktop, it needs to be able to support Joe Users soundcard so he can play Quake etc...
- -
Linux already has the server market as far as technical superiority is concerned, but people won't even consider it an alternative to the "other OS" if it can't support their brand X hardware.
Just my $0.02
-----------------------------------------------
"There's nary an animal alive who can outrun a greased Scotsman !"
cp /usr/src/linux/System.map /boot will solve this cumarc
This would seem to be the end of line for 2.0.x barring disaster.
R.C.
http://www.o2.net/~gromitkc/winmodem.html
If the year is 2050 and you don't have a sound card to listen to music or play games with you're going to be sorry.
[Act on them /\ Post 'em] -> even better!
Teukels.
travelmate:~$ cat /proc/version
Linux version 2.0.34 (root@travelmate) (gcc version egcs-2.90.29 980515 (egcs-1.0.3 release)) #4 Sun Sep 6 23:10:57 EDT 1998
Yesterday was the sixth of September?
Woooooohaaaaaaaa! I thought the whole Eastern Daylight thing was messed, but geez, do we have some weird time problems up here!
I think you may be confused, sir. In my experience, the boy who pays for a beta release of Windows and the boy who spells his name with numbers are often one in the same. They'll pay for a pre-release of the buggiest, least capable, slowest and most oversized operating system choice out there and claim they can't afford the software they "need"; like AutoCAD, Photoshop, and the latest 3D render studio fad of the week.
\\/1nd0Wz 2| 0w|\|z j00 -- or something.
Any idea when this will be available as an official RH upgrade?????
Somehow I always seem to screw something up on RH when I do this install on my own from source.
So many options!!!!!
latest error (From 2.2.5 to 2.2.8):
when doing a ps:
"Warning: