Slashdot Mirror
Linux 2.2.0-pre3 is available
specht (Karl Heinz Kremer) was the first to write
in and tell the world that yet another pre2.2 kernel has
hit the wire. Go snag it quick and hammer on it.
← Back to Stories (view on slashdot.org)
← Back to Stories (view on slashdot.org)
Damn.. This means about 12 hours before anyone actually testing it can get it.
Shit.
You can probably find out some info about this on the Internet. Start with Linux specific sites. There are some web interfaces to programs called "search engines" that will prove indespensible. If you can't find what you're looking for in about a minute or so then I would recommend you stick with Windows for now. In the mean time, look for these things called "books" and find out how to work those first.
Hope this helps.
downloaded it, pitched the sheduler up to 1024Hz, compiled it, and booted it...
:)
IT WORKS!
find me on ircnet and i'll dcc it to you :)
first of all.. 1024hz would be a litlle overkill :)
i just change the #define HZ 100 to #define HZ 1024 to make linux much more responsive and a little bit faster...
Am I the only one having trouble mounting one of my FAT32 drives ??
/dev/hdb1 /dosd/ -t vfat
/dev/hdb1, or too many mounted file systems
;-)
mount
mount: wrong fs type, bad option, bad superblock on
this happens for two of my vfat partitions!
the other ones seem to be working just fine..
it worked fine with 2.1.131ac10 which was my previous kernel.
Can anyone else reproduce this error or am I the only one ?? how about a fix
-pug, pug@nerd.dk
compiled into the kernel
-pug
Please, no need to be condescending.
What devel list? I been watching the maling list. There's only been one message so far (I am watching the linuxhq archive, so i'm behind) Is there another source?
dcp
I think you need the new ppp package. 'pppd --version' reports "pppd version 2.3 patch level 5" on my system. I have "pppd-2.3.5-1" rpm installed which came with redhat 5.2 although the remainder of the system is still 5.1. Time to upgrade everything one of these days.
...Linus and Alan suddlenly got tired of Linux development and instead started playing Quake full time?
Well, firstly we would get two heck of a good Quake players. But what about Linux? There is a lot of people involved in the development of Linux, but these two fellows are pretty important.
What would happen? (Very hypothetical)
There is a major bugs around the Prerelease 1,
the make xconfig went wrong, make menuconfig works, but during compilation, some error appear. End up with Error 1.
I didn't try Prerelease 2, but Prerelease 3 has some problem too. It can't detect FAT32 hard drive properly. Error message as wrong file system type, and another problem is can't get into KDM.
Other than that, not tried out yet.
I've been running Linux in this configuration for a year on a Pentium II 266 with no performance problems at all. I'd estimate less than one percent of CPU time is spent in the scheduler in this configuration. Anyone have a more exact estimate?
Am I the only one having trouble mounting one of my FAT32 drives ??
There is something in www.linuxhq.com about updating the mount binary - maybe that is your problem?
Mike
Okay, I read through the CHANGES, downloaded
the necessary upgrades (I'm running Slackware
3.4) and compiled them, then compiled the
2.2 kernel. Resulting kernel is 392K.
When I reboot, LILO just keeps trying to
uncompress the kernel forever. It never
gets past that point.
Suggestions?
What's the deal with pcmcia modules - I can't get them to compile. I even modified the Config file in the distribution to have "2.2" in the foreach loop (it was only "2.0|2.1")
Anyone get this to build correctly? This is using the latest (3.0.6)
same problem with pre2 .. but I went back to K6/6x86MX in pre3 and it compiled fine ... chalk it up as another bug fix :)
compiled and installed quickly and cleanly on k6-2 .
HOWEVER, on p133, mega kernel-panics.
oh well, as they say "point, click and start all over again!!!
NICE features tho, the VESA framebuffer flies with a monster fusion.
I first read about this scheduler switch when pre2 came out... and apps (notably netscape) do seem to perform better. I have a Pentium Pro 200MHz... so a scheduler rate of .001MHz doesn't seem too unreasonable. But, I'm no expert on OS design, so that simplistic numerical comparison may not be valid.
even though it was a flame and pretty mean and not even totally relevant.... I gotta admit it made me laugh :)
I had the same problem with modular sound drivers for my AWE64; a fairly easy way to fix this is to get Alan Cox's latest ac patch (from ftp://ftp.linux.org.uk/pub/linux/alan/2.2pre/). I think the latest is ac4. Load the file into an editor, look towards the end for the sound driver patches, copy them to another file, then apply this new patch. Worked fine here.
Set dma16={dma16} as well, otherwise you'll have problems with DMA timeouts when using 16 bit sound.
no, the scheduler is not run every 1msec with
HZ=1024. the Linux-scheduler is HZ-independent,
ie. it will always have the same timeslices as
before. The only overhead you will see is more
_timer interrupts_, which are very cheap.
HZ=1024 is safe, except maybe for a few HZ-unclean
sound drivers. (but eg. the Alpha has had HZ=1024
for a long time)
hmmm... that should be no problem...
just to be sure, i wrote a little program, an ran 1 reniced to -20, and the other normal...
as expected, everything went ok, and the first one finished a few seconds earlier...
I had the same problem as well...
I compiled the driver in the kernel and not as a module and it worked fine for me after that.
Hope that helps.
Everything else works a dream...Cool!
Or even port linux to the Quake console. Imagine...
You get the Quad Damage!
Somebody frags you *** Kernel Panic : you died
Obviously there are bugs, but they seem to be big and fixable, as opposed to hidden and insideous.
... yes, I'm living dangerously!) and has been running fine. The box is successfully putting out NFS, SMB, and xdm, and that's all I really care about.
And it seems to be perfectly stable as far as I can see.
There was some strangeness patching up the Makefile from 131 to pre1,2,3, but I was able to do all that by hand.
And I got bushwhacked by the change from CONFIG_M586 to CONFIG_M586BLAH... which hosed my changes in arch/i386/Makefile.
But a couple of touchups and it built (with -O6 -mpenitum -march=pentium
Okay, this is annoying...
I have a spare K62-300 box lying around, IDE HDD, ATAPI CD-ROM, Kingston KNE100TX NIC. I installed a fresh Slackware 3.4, upgraded to kernel 2.0.36. LILO is on the MBR.
I followed the upgrade advice in Documentation/Changes.txt and upgraded all the software packages. Then I downloaded a fresh install of 2.2.0pre3 and configured it for my hardware, make zlilo, make modules, make modules_install, shutdown -r now. 397K kernel, BTW.
When it reboots, it successfully uncompresses the kernel, and then just sits there. It's obviously doing something, because it's continuously accessing my HDD. I waited for about 25 minutes, then reset the sucker and booted off a floppy. fsck reported dozens of errors on /dev/hda1, and I couldn't log in because /etc/passwd had been trashed.
Again... any suggestions?
Totally correct... but one thing - the new 2.2.0preX kernels are actually just the next one after 2.1.132. Also, linus said "extremely rude" :)
Yeah.. there should be some speeds you can choose from... Or it may be a good thing to fix it on 1024 for pentium, and 100 for 486's...
Rob: What about a poll?
That's why I'm sticking with 2.0.x for a while.
The 2.2.x kernel's module loader doesn't seem to like to load sound drivers, I don't want to compile them into the kernel, and I don't want to load them manually.
mmm... awe works great here...
(pre3 - dual p2-350 - awe64gold)
What!?
That's just gay. I've got an old Cyrix 6x86 P150+ that I love an cherish (mainly 'cause I don't want to upgrade right now). One could argue that the 386 isn't popular anymore, why not drop support for that as well?
This, coupled with having to manually load sound drivers, will keep me with 2.0.x for a while yet.
Why is it so important that you get pre3 instantly? Is there a reason you can't wait for 12 hours for activity to die down? Do the features wear off if not used within the first 12 hours of distribution?
...that's the beauty of Open Source. Throw several thousand people at a kernel, and the bugs will make themselves known.
Stop being so dammed elitist.
DG
from pre1-ac3; and pre2 and upwards, fat32 is definitively broken. It mumbles about some bad FSinfo or something; and sets the free space count of all my fat32 drives (even those mounted read-only) to ZERO; I have to run scandisk to remedy this (and of course I can't write to any fat32 drives since they are "full").
I found a simple fix for this; just grab the linux/fs/fat/* and linux/fs/vfat/* files from pre1, and dump them into your pre2+ kernel; recompile, enjoy.
I don't see why you had a problem with the original poster. He asked for a LINK which will give him the answer he wanted, a 'manual' if you will. In other words, he asked 'where is the fucking manual that I can read about it?' :)
Maybe they would get pissed off at some of the problems in Quake and make thier own game engine that was GPLed and was better than Quake.
I just checked linuxhq and all it showed was:
<title>hax0r da pl4net!@#</title>
<center>
h0h0h0 m3rry chr1stm4s!@#!@#!@#!@#
</center>
whoops.. my mistake. Apologies for the false alarm.
If you compile sound as a module, you'll need to specify the irq, dma, etc... when you load the module. If you want to avoid this, just compile sound into the kernel, and you'll be able to specify all that stuff.
I have a RedHat 5.2 system, with ppp-2.3.5 -- ppp isn't working under 2.2.0-pre3.
/. 2.2.0-pre3 discussion)
I've tried various kernel configuration combinations of module or in-the-kernel ppp and yes/no on 'kmod' and 'set key symbols'. No luck.
Otherwise the kernel seems fine (aside from the vfat problem addressed later in the main
Any ideas?
-D.
hmmm. Maybe something was screwy with my .121 then. I was told I needed to use insmod to install my sound.o, which I wasn't really interested in doing manually all the time. Perhaps in the pre2.2 stuff sound modules are handled automatically like kerneld handles them with 2.0.x kernels.
us usual, patches available at ns3.sinister.com
Compile the kernel as a 386 or 486 and it should work fine. Cyrix chips have buggy TSC's and various other stuff. There is a thread on linux-kernel about this right now so read that for all the gory details.
same problem for me.. i think the awe64 isn't 100% backwards compatible.. i'll try to see if i can hack the kernel so it works...
Get a newer version of top.
My sound is all garbled in GLQuake.
I heard something about this before, is there an easy solution???
Yes, I'm having this same problem. I have 3 fat32 partitions and I get that same error. My one fat16 partition mounts just fine. I guess I'll be waiting for pre4...
Well, does it?
Had the same thing happen with MKLinux, 2.0.35. Wasn't a kernel problem at all; turns out that I was trying to drive pppd at a faster rate than it could handle (230k that MacOS handles fine). Once I figured it out, backing the port rate back to 115k worked fine.
I was TOTALLY looking in the wrong direction because ping would work.
running isapnp before hand? /etc/isapnp.conf and edit /etc/isapnp.conf for the settings you wish to use with your card (its pnp so you can pick anything that isnt taken up allready) then run isapnp /etc/isapnp.conf you can also run pnpdump --config /etc/isapnp.conf and that will pick some settings for you (ohh in both cases you will probally need to add two of the i/o's for the midi sequencer)
it needs to be run before the module loads or else your system wont be able to see your awe card..
anyhow.. if you havent setup isapnp you could either type pnpdump
I hope this helps (Im tired so Im probally rambling some)
same as the other kernel versions. ie 2.1.132 to pre1 then patch pre2 then patch pre3. Think of it as version 2.1.135, which it basically is, just renamed to encourage further testing.
he did make zlilo, that runs lilo for you.
As I noted in the 2.2.0pre4 thread, I have successfully upgraded from a virgin Slackware 3.4 to 2.2.0pre4. No problems!
I just booted up 2.2 pre 3 on a pentium 233 machine. here is the output from top..
:)
2:13pm up 31 min, 0 users, load average: 1.86, 1.83, 1.37
65 processes: 61 sleeping, 4 running, 0 zombie, 0 stopped
CPU states: 950.0% user, 70.5% system, 0.0% nice, 0.0% idle
Mem: 127804K av, 74876K used, 52928K free, 61852K shrd, 4432K buff
Swap: 130748K av, 100K used, 130648K free 25768K cached
PID USER PRI NI SIZE RSS SHARE STAT LIB %CPU %MEM TIME COMMAND
631 homebrew 14 0 852 852 440 R 0 99.9 0.6 39:08 l3enc
349 root 2 0 16168 15M 876 R 0 72.9 12.6 24:35 Xaccel
432 homebrew 0 0 3272 3272 2552 S 0 31.0 2.5 5:42 dtterm
447 homebrew 0 0 16644 16M 7236 S 0 22.8 13.0 16:58 netscape-com
641 homebrew 3 0 732 732 560 R 0 17.7 0.5 0:08 top
Notice that the processor usage is listed as 950% user
Anyone else notice this?
Try the mirrors. ftp.(country code).kernel.org. They work great.
Posted by posterkid:
/dev/lp1 changing to /dev/lp0 (I remember that happening when I was running 2.1.x versions), everything looks perfect to this guy....
pre2 gave me very few problems that a recompile with different options took care of (Cyrix 6x86 non-MX does NOT work with MTRR, and the SoftOSS drivers don't seem to like my SB16). Other than that and
Posted by freebox:
This, and everything else I use, is working for me now.
Posted by posterkid:
/dev/dsp was giving me problems when compiled with the SoftOSS driver option. Disabled, worked beautifully. Now fuck off, anonymous troll.
um. no, that's not what I meant.
in /usr/src/linux/include/asm/param.h change
#ifndef HZ
#define HZ 100
#endif
to:
#ifndef HZ
#define HZ 1024
#endif
I don't know enough about kernel hacking to explain what the scheduler does exactly, but this helps a lot.
Cyrix 686 support was dropped as of 2.2.0pre1. Alan Cox says he fixed all the Cyrix bugs and there's no longer any mention of it in his diary. My kernels still die on boot with Divide error: 0000 when compiled for 586/K5/5x86/6x86. I understand the Cyrix 686 isn't very popular and mine is no longer sold in stores. Is it time to ditch the Cyrix and get an AMD?
To work around the Cyrix divide error bug edit /usr/src/linux/arch/i386/kernel/smp.c
in front of line 1055 put
if(!cpu_hz) cpu_hz = 150000000;
Replace 150000000 with whatever your clockspeed is in hz.
To get the sb module loaded specify values for io, irq, dma, and dma16. The io value should be a hex number starting in 0x, like this
insmod sb io=0x220 irq=10 dma=1 dma16=5
I doubt these obscure bugs are going to ever get fixed.
I am so sick to death of every time a new kernel comes out that some one that thinks of themself as very important thinks that they have to have the thing before anyone else. First of all, if you really are that important, you would have known about it from mailing lists already. Second, how long could it take you to download a patch? I have NEVER had a problem downloading a patch. If you were as good as you like to think you are, you would know what mirror sites are best for you to use. Third, the kernel, especially when it gets to this stage, needs to be tried by people that aren't "experts" (not saying you are) so that we can see if there might be things that average users get stuck on.
And lastly, I am getting so I can't stand these people that post really stupid crap like this, and don't even have the nads to identify themselves. I am getting to the point where I would almost prefer Rob 86 ALL anonymous coward posts.
Of course, I picked up the 3.20 version of the quake driver, so that might have something to do with it too...
What sort of problems are you encountering with yours? I saw some chatter on Linux kernel earlier about specifying IRQ, DMA and DMA16 specs during the modprobe stage.
--
rickf@transpect.SPAM-B-GONE.net (remove the SPAM-B-GONE bit)
"People will pay big bucks for the luxury of ignorance."
I strongly suggest that everyone also visit www.linuxhq.com and look at the CHANGES area under the 2.1.x kernel area. Upgrading to those packages will solve a lot of problems before they arise.
FWIW - I'm running 2.2.0pre2 at present, and the only complaint I have is that lockd seems to crap out and not let /usr get unmounted on reboot. =(
--
rickf@transpect.SPAM-B-GONE.net (remove the SPAM-B-GONE bit)
"People will pay big bucks for the luxury of ignorance."
Quake has had a long-standing problem with kernels around 2.1.115 or so and later. Probably the best solution (seeing how everything else works okay) would be to get the latest driver from Id Software. Quake 2 is currently at rev 2.30.
--
rickf@transpect.SPAM-B-GONE.net (remove the SPAM-B-GONE bit)
"People will pay big bucks for the luxury of ignorance."
Running pre2.
rasputin:~$ mount -V
mount: mount-2.7l
Weird that 6x86 MMXes work fine, but 6x86 classics don't.. Big difference between the two other than MMX?
Mine is an old Cyrix 6x86 P150 (really 120 mHz) and it's running fine under the 586/K5/5x86/6x86 setting.
--
Aaron Gaudio
"The fool finds ignorance all around him.
"Every man is a mob, a chain gang of idiots." - Jonathan Nolan, Memento Mori
Take a look at http://www.linuxtoday.org/stories/296.html
It's a little old though.
--
Aaron Gaudio
"The fool finds ignorance all around him.
"Every man is a mob, a chain gang of idiots." - Jonathan Nolan, Memento Mori
I've recently tried 2.2.0pre1 with a rate of 500 hz, and got some nasty side effects
on my Pentium 166 MMX; the sequencer was way out of sync, MIDIs and MODs (xmp with AWE support) played 5 times faster than they should have, plus
my "ps" command showed some fucked up results; half my processes appeared to take 99% of the CPU time, and the total CPU time consumed was _really_ messed up.
What results have you had with 1024 hz timeslices?
the real at&t mix
Take a look at the kernel devel archives.. Lots of people have been reporting this and stuff.. I think there are a couple patches floating around to fix it too.
Ah, okay. As I said before, I don't know the inner details of how Linux's scheduler actually works. I was under the impression that irq0 (int08h) just directly called the scheduler, which then did its choosing of the next process. But I'm apparently wrong and there's some sort of thing where it first determines whether the scheduler has to call or not. My bad. (Sorry, the only scheduler I've played with is Minix's, and even that I didn't delve very deeply into. :)
---
"'Is not a quine' is not a quine" is a quine.
Quine "quine?
Linux might fall into trouble, but rest assured there would be others. It won't happen, of course, because of the love of the job people have.
I'm assuming you're coming from 2.0.x when upgrading to 2.2.0pre. You'll need to upgrade your ppp package before you can connect to the net. I had this same problem when upgrading from 2.0.x to 2.1.x.
Dan
I'm not complaining -- after that period we got a series of kernels which have been superb. I'm sure that the 2.2.* kernels will become mature and then they'll be great. I just wouldn't run 2.2.0 on any machine that I use for anything more important than playing Super Nintendo cartridges.
Check your pppd options and make sure that debug
is enabled, then check your syslog. That'll give
you a better idea of what's going on, at least.
Kernel 2.2.0pre-1 works very good, but with pre2 and pre3 I get teh following error:
gcc -D__KERNEL__ -I/usr/src/linux-2.2.0-pre2/include -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -pipe -fno-strength-reduce -m486 -malign-loops=2 -malign-jumps=2 -malign-functions=2 -DCPU=586 -c -o ewrk3.o ewrk3.c
ewrk3.c: In function `EISA_signature':
ewrk3.c:1658: fixed or forbidden register was spilled.
This may be due to a compiler bug or to impossible asm
statements or clauses.
make[3]: *** [ewrk3.o] Error 1
make[3]: Leaving directory `/usr/src/linux-2.2.0-pre2/drivers/net'
make[2]: *** [first_rule] Error 2
make[2]: Leaving directory `/usr/src/linux-2.2.0-pre2/drivers/net'
make[1]: *** [_subdir_net] Error 2
make[1]: Leaving directory `/usr/src/linux-2.2.0-pre2/drivers'
make: *** [_dir_drivers] Error 2
Note.the ewrk3 is an isa card not an Eisa card.
I have the EXACT same problem, 2.1.131 mounted fine, but 2.2.0pre3 won't mount at all.
Do you need to apply all of the pre patches in succession, or back out and apply the patch to 2.1.132?
can anyone point me to a link that describes what new goodies i can expect w/ 2.2.0 ??
thanks
gng
-greg "The sun is not yellow its chicken" -Bob Dylan
hmm.. thanks for your ideas.. I would never have though of typing "linux" into a search engine, lemme give it a shot..
ok.. at Altavista I entered the following query +linux +kernel, There were 535980 reference to "kernel" and 236646 reference to "linux". Now I don t know about you but I m programmer who works
on things other than linux for a living..gasp..so it s not a priority to spend hours sifting though webpages to find the latest information on the linux kernel....
Second..Because of linux s amorphous nature, information is dispersed all across the internet. Because of this, sometimes more inexperienced users will often ask questions that might be
obvious to more seasoned linux users. Responses to these questions that contain no useful information other then the sentiment RTFM don t do the linux community any good and will/could turn potential users away from considering Linux as a viable operating system. I personally would love to see my boss,clients and friends adding linux partitions to their machines at home/work and when they do I plan on fielding
everyone of their questions in a kind and informative manner.
-gng
-greg "The sun is not yellow its chicken" -Bob Dylan
Is that what you wanted to hear? Is that why you feel compelled to show off to the world just how "in" you are?
[RANT MODE ON]
/usr/doc/packagename.
/.
Regardless of what you bunch of anonymous lamers think, the man has a point.
Nobody bothers to look anything up on a fucking search site any more, they fire up Free Agent or mIRC and bother people with stupid questions.
90% of the time those questions are answered by a man page, a HOWTO, or
9% of the time you can find an answer after a couple of minutes on the search engines.
That other 1% of the time are the only questions that people should be wasting my time with in IRC or Usenet, or here on
If you can't be bothered to RTFM, you shouldn't be using Linux, plain and simple. It's the best-documented operating system in the history of mankind, so RTFM or FOAD.
[RANT OFF]
Really? I had a 486 chugging away as a web server for about 8 months on 2.0.0 with no upgrades, patches, reboots or anything... it would have lasted longer, but one of my University's Generators exploded, and they shut the power down to fix it...
I guess I just got lucky on the setup for that one... all cheap stuff that for the most part Win95 refused to associate with.
it was just a damn question! no need for all that
too much to drink last night?
Wrong! When I submitted the story I found the
new source package _AND_ patch on the US mirror
(ftp://ftp.us.kernel.org). Actually I went there
to get the pre2 version which I had no chance so far to look into. Imagine my surprise as I found an even newer version.
I had so much fun running scandisk to fix my FAT partition!
It seems to work fine in pre1
People need to learn how to use patch. I'm sure you do, but all those newbies. It would also help if the ftp mirrors were able to get the patches sooner.
I don't know if anyone has a patch already.. but.. i simple changed the file fs/fat/inode.c to equal the same in pre2.
It worked for me.. except that after running the first pre3 that gave me that error... something in my two fat32 partitions chaged that it reported no free disc space... i had to run windows scandisk.
Is anyone else having trouble playing audio CDs under 2.2.0pre3 (on a RedHat 5.0 system)? I have a Wearnes 32x IDE drive on my secondary IDE connector (I also have a Mistumi 4x drive as the slave on that cable, though whether it's connected or not doesn't seem to make a difference). When I try to play an audio CD with xmcd I get "no disc", and I've gone through all the xmcd help docs and tried cdplay as well, but it just hangs. If, however, I try to mount the CD first, after the error message, both xmcd and cdplay then find the CD. Things had seemed to be working smoothly running 2.0.35. Has anyone else had any problems like this? Any suggestions? Thanks!
--Sean
Sean R. Baker
CDT, United States Army
"Lead me, follow me,
or get out of my way."
I tried out the new 2.2.0pre1 kernel and other than the fact that I cannot connect to the internet, it seems to work fine. To be more specific, my modem dials fine but when it is supposed to connect to my isp by starting pppd, it fails saying that the network is down. (before u ask, yes I did compile PPP support in the kernel). If anyone has any ideas as to how I could fix this problem, let me know.
This being the first time I have acutally asked a question of SlashDot about something, I have to say WoW!!! In 12 minutes I got 3 responses with solutions to my problem. Thanx people, u RuLE!!!
L8r
I just booted 2.2.0pre3 and discovered that yet again, I cannot get my sound working. I have the drivers as modules (soundcore.o, sound.o, sb.o, mpu401.o, and uart401.o). I have to load those modules in this order: soundcore, sound, uart401, mpu401. When I try to load the module sb.o it says 'device or resource busy'. I noticed that when I was configuring the kernel that it doesn't even let me set the address, irq or dma of the card. If anybody could help me with this problem it would be greatly appreciated.