Slashdot Mirror
Linux 2.2.3 Released
Linus has finally released Linux 2.2.3 after several prepatches and lots of
testing. Hopefully, it should be the last one for quite awhile. Check out the
changes summary on Cutting Edge Linux, and don't forget to use a mirror!
Thats why you no use PRERELEASE compilers to compile KERNEL, Yas!?
----------------- ------------ ---- --- - - - -
----------------- ------------ ---- --- - - - -
Your honor is perfectly understandishable.
As hardware support for Linux grows, it'll only make the all-in-one kernel package larger.
I do use patches. With the connection i'm on, pulling a kernel isn't exactly a problem either.. It just seems more sensible to not have to download a crapload of SCSI and other assorted drivers if you don't really need them anyway.
If you want Solaris, you know where to find it...
It looks like you have CONFIG_SCSI_MULTI_LUN (Probe all LUNs on each SCSI device) enabled in your kernel configuration, and the IDE-SCSI driver isn't discriminating as to which LUN it responds to. Unless you've got a CD jukebox or something, you can probably disable the "Probe all LUNs" option with no ill effects.
Easier? To download and keep synchronized trillions of small packages instead of one big one? I doubt it.
.config from a previously configured kernel). And if you don't want to store the entire kernel tree, you can always delete the bits you don't use. And if you don't know which bits those are, do you really want to have to keep track of which driver packages contain only those bits?
If you're having problems with the size of the download, try using patches. You should be anyway. If you don't like answering 20 trillion questions every time you build a kernel, try "make oldconfig" (you'll need a
How will he know if everything will actually reboot? :P (Best to wait till others try it & see if they have any problems)
Read linux/documentation/fb/framebuffer.txt
Whatever you do, never ever exhaust your virtual memory with xanim. This crashes the VM subsystem permanently on all kernels after 2.2.2. I work with movies larger than 700 megs and my system goes down constantly. Since this isn't a normal activity on Linux computers there is no bug fix in progress.
Don't worry because after it loads, the free sound driver doesn't work with my SB16 anyway. Load the modules like so:
insmod soundcore
insmod sound
insmod uart401
insmod sb io=0x220 irq=10 dma=1 dma16=5
Specify proper io, irq, dma, and dma16 values for your card. Double buffering has been broken ever since 2.1.
The kernel is probing all LUNs (0-7), and the CD-ROM is answering the same on each one. Try turning off the "Probe all LUNs on each SCSI device?" option in the kernel configuration and re-build the kernel.
Nonononono! Doesn't work that way! Kernel releases ONLY happen when enough people are simultaneously downloading or compiling the previous one.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
Or better yet, just split it up into the different platforms. (e.g. someone needing the kernel for a Sparc wouldn't need the parts for x86).
"In true sound..." -Agents of Good Root
I had this problem on my RH5.2 system, but I had always been specifying these parameters, courtesy of sndconfig. Unfortunately, the parameter format changed, so that the transition from 2.0.36 to 2.2.2 broke sound until I specified the DMA settings in the above format.
Much better now.
Make me aerodynamic in the evening air
This is an issue of no importance.
I'm beating on an SMP system with APM enabled
and video and sound and its all working well.
Oh, cdparanoia just seg faulted. That's better
than before, anyway.
All those complaining about 2.2.x not being "stable" enough for them should ask themselves whether they kept up with the 2.1.x kernels. The reality is that Linus had to declare a new "stable" release before the vast majority of folks would try a recent kernel. The point of switching to a stable release is to shift into a mode where problems are squashed and the code converges on something stable. 2.0.x required a lot of work to nail the details, but now it just runs and runs, with a few known limitations that are easily avoided.
Stampede Linux has a 2.2 distribution.
When last I looked, X was in the region of 30Mb. Why aren't you complaining that the X source should be split into hardware-specific bits?
It's not like saying that at all. I have X installed but, not being an X developer, I don't have its source code lying around on my hard disk, nor do I feel compelled to download the source every time there is a new release.
You don't have to download the entire Linux kernel source to run Linux.
If you are a kernel developer, you only need to download it once, and keep in step with patches.
If you are not a kernel developer, but want to compile the latest kernel in limited space, you can get rid of the code you aren't going to compile.
And I don't see why you bring up the speed of X at all. I was talking about the source code size, not the compiled size.
It's amongst the mini-HOWTOs; did you check that section?
Here's the Large Disk (mini-)HOWTO.
Further, as for the 8K path that is solved with a symlink directory collection tool. I simply collect the large set of tiny packages into symlink-bin directories. The advantage of this being I can try out new versions (or have multiple versions) of my packages on the system at the same time. The perferred ones go in the collections and the others can be added to user paths at user option. Using the M4 soft tool this then provides elegant management accross many architectures.
I would be happy to place those tools online if someone is interested.
--Karl
The kernel is only as big as the options you compile into it. If you don't want a 12 MB download, get the binary. The larger source is due to the fact that more platforms/features are supported, if you don't need these platform/machines, then get the 1.2 or 2.0 kernel, both of these are pretty stable. Hasn't bandwidth kept up with the kernel size anyway? As for the speed of new modules being updated, try going to the home page of the person who wrote the module, I'm sure you will find that the code is cutting edge. To break up the kernel will only cause more headaches, as you have to figure out what you need and download each one individually.
/usr/bin/patch.
You really want to save time and space? Learn how to use
posting on slashdot or doing nothing doesn't get stuff fixed. You have the source, read it, it is probably the file sg.c, look at differences between 2.2.1 and 2.2.2, if you can, fix it, if you can't do that, then write the guy whos email address is in sg.c. according to /usr/src/linux/drivers/scsi/sg.c that is Lawrence Foard (entropy@world.std.com)
RedHat doesn't appear to have released the new 2.2 on CD-ROM yet. For consistency and recovery purposes, that's my preferred media. Are there any 2.2 distributions on CD-ROM?
Tis question, and the "official" answer is in the FAQ on tux.org.
Is your board a SB16 with the VIBRA chip?
;)
If so, here's a comment directly out of the 2.2.3 kernel patch.
+ Hello again,
+
+ Playing with a SB Vibra 16x soundcard we found it very difficult
+to setup because the kernel reported a lot of DMA errors and wouldn't
+simply play any sound.
+ A good starting point is that the vibra16x chip full-duplex facility
+is neither still exploited by the sb driver found in the linux kernel
+(tried it with a 2.2.2-ac7), nor in the commercial OSS package (it reports
+it as half-duplex soundcard). Oh, I almost forgot, the RedHat sndconfig
+failed detecting it
+ So, the big problem still remains, because the sb module wants a
+8-bit and a 16-bit dma, which we could not allocate for vibra... it supports
+only two 8-bit dma channels, the second one will be passed to the module
+as a 16 bit channel, the kernel will yield about that but everything will
+be okay, trust us.
+ The only inconvenient you may find is that you will have
+some sound playing jitters if you have HDD dma support enabled - but this
+will happen with almost all soundcards...
It seems to be broken. Where's an ac1 when you need it?
"What do you mean, invalid parameters? 9000Gigs of RAM and it can't answer a simple question!" -- Earthworm Jim
Hi,
/whatever/patch-2.2.3
You do realize that you can just download patch-2.2.3.gz, uncompress it, and then "cd" to your "linux" dir and do:
patch -p1
to update your existing kernel source?
I only say this because your post makes it sound like you will be downloading the whole thing to replace one of your "two previous versions". Maybe this wil save you time, or maybe I just misread.
-Steve
Woops. I make that typo all the time when I'm patching my own kernel. Make sure and include the "".
-Steve
Well, this is getting embarrasing. Slashdot is eating the "less than" character that should be between the "-p1" and the "/whatever". Probably interpreting it as an HTML tag delimiter.
-Steve
a 14 gig hard drive... really dumb! get a combo to make a nice raid array and get 90000000000% speed increase. and unfortunately you got a gateway.. you might be lucky and got one that isnt messed up. but most are just like acers.. not upgradeable and come with crappy video/audio/bios/etc...
Linux and SCSI the ONLY way to go for real speed, anything else is just for playing.
Do not look at laser with remaining good eye.
I was doing good to keep one NT machine running for more then a week consistantly. And yes, I do know how to keep machines stable.
--
Linux is becoming massively large with every major kernel
:) would also improve the speed at which new kernels are
release. So far, it is a notable fact that the kernel has doubled
(and maybe even more) with every major kernel release. If kernel
v2.2 is about 12MB with BZIP2 now, kernel 2.4 (or 3.0, depending
on which way they decide to go) will be TWICE that! We are talking
about a 24MB Linux kernel! That is a large and cumbersome kernel.
If the kernel size keeps doubling, the kernel following that may
just be 48MB! If you think about it, those are VERY large kernels,
indeed, and no modem user would take the time to download them.
So what is wrong with this? Well, I think that if the kernel
could be desinged in a different way, the size would decrease much
more. What has to be done is something that nobody has even probably
thought about yet: the kernel tree has to be divided into different
compressed files, instead of one large file. A good proposal is to
begin my dividing the kernel core by architecture. This way, if we
want a kernel just for the i386, or just for the PPC, it can be quickly
downloaded by itself and not with all the other architectures. This
would make it easyer to have the kernel just the way you want it. Then
the other core components of the kernel ( but the ones that do not need
to change in architecture) can reside in another compressed file. Also,
the documentation of the kernel could be a different package as well.
Following this, the kernel modules can be put into a different
compression tree as well. The reason that the kernel is so much bigger
every release is the fact that there are so many new drivers and modules
that it just takes up too much space. To save more time and space,
and to be better organized, the kernel modules could be put into separate
packages as sound, filesystem, misc, and all others in their own separate
packages. This would not only make it easyer to download the kernel
modules but I think that programmers would benefit from making live
updates to separate modules and even core components of the kernel as
they wish, and uploading separately from the rest of the kernel
components. If you really think about it, this process of what I call
"kernelites"
available and at which new drivers/modules are available.
These are, of course, just simple concepts. I am just throwing
in what seem to be good ideas to me. The kernel team (mainly Linus
Torvalds and Alan Cox) should look at this seriously and give it some
thought, however. But they could always just design the kernel tree
in their own way, hopefully to save time and space.
linuxnewbie.com
I like the rapid release of new kernels—it's much more fun than the way things were (are) with OS/2; i.e., workaround bugs in the OS/2 kernel until the next FixPak comes out (but the next FixPak will usually break something in IBM1S506.ADD).
Best of all, the older kernels (like 2.2.2, if you consider a newborn old) continue to work and continue to work well.
I must say for 2.2.[0-3] that the mmgr has been greatly improved—perhaps by kernel 2.4 (or 3.0) it will match OS/2's memory performance <sly grin>.
Cheers,
Joshua.
--jon. Postel is dead. May we all mourn his, and our, loss.
I had the same problem until I specified the IO, IRQ, etc... in my conf.modules file. Look in linux/Documentation/sound/Soundblaster for more info.
Here is the line I use out of my conf.modules:
options sb io=0x220 irq=5 dma=1 dma16=5 mpu_io=0x330
Stampede has had the 2.2.1 kernel since Feb 15...
granted Stampede is still in beta, but some distros do have the new kernel...
The basic sleazeware produced in a drunken fury by a bunch of UCBerkeley grad students was still the core of BIND. --PV
#1, that's not a kernel issue. #2, yeah, right, replace a few big directories with a few hundred small ones and require 8k PATHs, smooth move Clyde. For what you want done with /usr, see /bin, /lib etc.. I've dealt with /opt and IMHO is a major headache without providing any advantage over a good package manager (flames about RedHat vs. Debian to /dev/null).
Multiple incompatible versions of shared libraries is why shared libraries have version numbers. I've got 2 variations of libc4, 5 variations of libc5 and 2 variations of libc6 on my system right now, all in /usr/lib with no problems.
As for keeping packages seperate and building a big symlink farm, I merely let the package manager keep track of which files belong to which packages. That's what it's for, after all. One version of a package is the primary installed version, and if I need other versions I install them elsewhere. Most often that's under my home directory for a new version of a package. Occasionally I get into a situation where I need to have multiple versions system-wide, but that only amounts to about 7 packages out of the 200+ I have installed and I can manage 7 package-specific directories easier than I can the /opt tree and related symlink farm for 200+ packages.
Of course distributions have to be market led, and this means that a new distribution won't be released until the Company concerned is ready with all the updates of the distribution...
I have been running kernel 2.2.2 without any problem...but I'm guessing most distros will want all of their distribution to be up to speed...after all what goods a 2.2.x dist if you can't install/configure the damn thing...and heck if people like us don't test new kernels it'll take a damn site longer for the dists to update!!
BTW I've noticed several Rawhide packages linked to glibc2.1 and the number is increasing...could thi mean that the culmination of stable Kernel 2.2.x, Gnome 1.0(1) and possibly glibc 2.1 might lead to RH6.0??
--
"You can have it fast
You can have it cheap
You can have it right
Pick two..." or pick Linux
My NT server has been up 31 days 17 hours, running a web server, network monitoring software, and an IRC server. However, I still like my linux servers better, as they have NEVER needed a reboot. I'm excited about that new kernel! (there, back on topic :)
-AF
-- -- I always wanted to be somebody, but I should have been more specific.
I have a server that I would like to put a 2.2.x kernel on, but I am waiting for the 2.2.x series to stop putting out so many patches so quickly....
Will I be waiting a looong time?
Mine's a SB16 PnP, so according to the HOWTOs it has to be modularized. And it does work that way.
/lib/modules. Also, i use isapnptools-1.17, pnpdump --config gave me the expected settings there.
/bin/sh /etc/isapnp.conf
Compiled the kernel with sound as a module. Ended up with adlib_card.o, opl3.o, sb.o, soundcore.o, sound.o, and uart401.o in
i put in a script to initilize the thing on boot.
#!
case "$1" in
start)
echo -n "Setting up sound card..."
# Plug n Play config
isapnp
# Install basic sound modules
insmod soundcore
insmod sound
# init'ize uart thing, because sb needs it
insmod uart401
# Set up the actual sound card! Pulled numbers out of isapnp.conf
insmod sb io=0x220 irq=9 dma=1 dma16=5 mpu_io=0x300 type=6
# Midi support. Note that this automatically loads opl3.o as well
modprobe adlib_card io=0x388
echo "Done."
esac
exit 0
i'm not sure how many of those insmods could be done with modprobe, don't feel much like changing it because it works fine as-is for me.
--
perl -e'$_=shift;die eval' '"$^X $0\047\$_=shift;die eval\047 \047$_\047"' at -e line 1.
I will DL this tonight and compile it. I have had no problems with the previous 2 versions. They fit right in to the RH 5.2 install. The only problems occur when you use glint to install something requiring kernel headers etc.
As for the large DL I would not complain too much in the age where Netscape 4.5 is 13M, and if you ever ftped Star Office the Kernel seems small for its contents
Brian E...
Did you report this VM crash to the linux-kernel list? I'm surprised no one was interested.
Isn't this exactly the type of sneeky bug the kernel developers would love to fix? If Linux is supposed to be "enterprise-ready", the kernel should be able withstand huge loads, even if they aren't "normal".
cpeterso
our beloved penguin is getting more fat; good, more technically chalenge things. yay.
# cat /dev/bobspizza > /usr/src/linux
--
Beef
"Raging Moderate" of the
You mean stable as in freeBSD, or stable as in where I keep my stallion WHOA!
--
Beef
"Raging Moderate" of the
LILO: loading Linux.
Uncompressing kernel........................
............................................
............................................
............................................
............................................
Done.
How would you like your toast?
A.crispy
B.black
C.radioactive
_
--
Beef
"Raging Moderate" of the
I am having a problem with setting up kernel 2.2.3. I have RedHat 5.2 with the 2.0.36 kernel. Everytime I have compiled the kernel and reboot it gets to the part where it says VFS, says it can't mount the root filesystem and gives me a kernel panic. I'm using a quantum 4.5 ultra-wide harddrive with an advansys scsi controller. It would be greatly appreciated if someone could give me some help on this.
Err.. Someone can explain to me what's going on with my SCSI emulation? 10 SCSI CDROM detected?
hda: ST36530A, ATA DISK drive
hdb: QUANTUM FIREBALL EX6.4A, ATA DISK drive
hdc: ASUS CD-S340, ATAPI CDROM drive
hdd: Hewlett-Packard CD-Writer Plus 8100, ATAPI CDROM drive
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide1 at 0x170-0x177,0x376 on irq 15
hda: ST36530A, 6208MB w/448kB Cache, CHS=791/255/63, UDMA
hdb: QUANTUM FIREBALL EX6.4A, 6149MB w/418kB Cache, CHS=784/255/63, UDMA
Floppy drive(s): fd0 is 1.44M
FDC 0 is a post-1991 82077
scsi0 : SCSI host adapter emulation for IDE ATAPI devices
scsi : 1 host.
Vendor: ASUS Model: CD-S340 Rev: 2.10
Type: CD-ROM ANSI SCSI revision: 02
Detected scsi CD-ROM sr0 at scsi0, channel 0, id 0, lun 0
Vendor: ASUS Model: CD-S340 Rev: 2.10
Type: CD-ROM ANSI SCSI revision: 02
Detected scsi CD-ROM sr1 at scsi0, channel 0, id 0, lun 1
Vendor: ASUS Model: CD-S340 Rev: 2.10
Type: CD-ROM ANSI SCSI revision: 02
Detected scsi CD-ROM sr2 at scsi0, channel 0, id 0, lun 2
Vendor: ASUS Model: CD-S340 Rev: 2.10
Type: CD-ROM ANSI SCSI revision: 02
Detected scsi CD-ROM sr3 at scsi0, channel 0, id 0, lun 3
Vendor: ASUS Model: CD-S340 Rev: 2.10
Type: CD-ROM ANSI SCSI revision: 02
Detected scsi CD-ROM sr4 at scsi0, channel 0, id 0, lun 4
Vendor: ASUS Model: CD-S340 Rev: 2.10
Type: CD-ROM ANSI SCSI revision: 02
Detected scsi CD-ROM sr5 at scsi0, channel 0, id 0, lun 5
Vendor: ASUS Model: CD-S340 Rev: 2.10
Type: CD-ROM ANSI SCSI revision: 02
Detected scsi CD-ROM sr6 at scsi0, channel 0, id 0, lun 6
Vendor: ASUS Model: CD-S340 Rev: 2.10
Type: CD-ROM ANSI SCSI revision: 02
Detected scsi CD-ROM sr7 at scsi0, channel 0, id 0, lun 7
Vendor: HP Model: CD-Writer+ 8100 Rev: 1.0g
Type: CD-ROM ANSI SCSI revision: 02
Detected scsi CD-ROM sr8 at scsi0, channel 0, id 1, lun 0
scsi0 : channel 0 target 1 lun 1 request sense failed, performing reset.
SCSI bus is being reset for host 0 channel 0.
scsi : detected 9 SCSI cdroms total.
sr0: scsi3-mmc drive: 0x/34x cd/rw xa/form2 cdda tray
Uniform CDROM driver Revision: 2.52
sr1: scsi3-mmc drive: 0x/34x cd/rw xa/form2 cdda tray
sr2: scsi3-mmc drive: 0x/34x cd/rw xa/form2 cdda tray
sr3: scsi3-mmc drive: 0x/34x cd/rw xa/form2 cdda tray
sr4: scsi3-mmc drive: 0x/34x cd/rw xa/form2 cdda tray
sr5: scsi3-mmc drive: 0x/34x cd/rw xa/form2 cdda tray
sr6: scsi3-mmc drive: 0x/34x cd/rw xa/form2 cdda tray
sr7: scsi3-mmc drive: 0x/34x cd/rw xa/form2 cdda tray
sr8: scsi3-mmc drive: 24x/24x writer cd/rw xa/form2 cdda tray