Linux Beta Kernel 2.5.16 Out

dipfan writes "The latest beta version of the Linux kernel 2.5.16 is out, with some comments by Linus here, who was kept 'personally somewhat busy' by 'the interesting Intel SMP-P4 TLB corruption bug, which ends up being due to some very funky asynchronous speculative TLB fill logic'. Woo hoo. Mirrors, etc." We haven't been keeping up with the 2.5.x series, but a slow Sunday is a good excuse to catch up.

Re:The 2.4 series.

[karlm]

I would guess it's pretty much bug fixes from here on out in the 2.4 line. Sure, you'll get some more obscure drivers and such, but I don't thnik too much more exciting is going to happen. The new 2.4 kernel maintainer probably won't do anything too controversial, particularly after that VM switch Linus pulled earlier in 2.4.

The 2.5 kernel is a major rework. IIRC, they're making everything possible modules, and you'll need to make an initial ramdisk with your ide/scsi/network/fs drivers in order to boot. Any idea how stable 2.5 is now? It sounds pretty cool. I've dabbled in kernel programming and am willing to put up with 1 week uptimes as long as I'm running ext3 or XFS. I tried L4-linux, but 8-hour uptimes just hurt. Yes, I'm also playing with Debian HURD. There's lots of really cool stuff going on nowadays in the OS world.

Any idea if there are any plans to merge the international patch (crypto stuff) into 2.4 or 2.5? Encrypted loop devices and encrypted swap really should be part of the main kernel. After all, the 2.4 kernel is currently being maintained by a minor in South America, so the 2.4 kernel is pretty safe from the US govt/US courts at the moment. Hmm... DeCSS in kernel-space anyone?

Re:The 2.4 series.

[Error27]

You're not the first person to comment on the recent dirth of stable releases. According to the last changelog [linuxtoday.com], Marcello has been holding off on releasing the next one because it includes some IDE changes that he wants tested. I'm not really sure what the IDE changes are... Andre Hedrick had some changes that let people use the new really large IDE harddrives, that might be the code, but I'm not sure.

Of course, people who like to take risks or who want to help test, are welcome to use the -pre patches. Right now it is up to -pre8. If you want to live on the very cutting edge, then you can download patches from this page. [kernel.org] The patches here are updated every hour for the Marcelo's bk tree.

I believe that besides the IDE changes, there are several VM tweaks.

"Beta" kernel?

[Platinum Dragon]

At best, I would call the development series "alpha". Beta implies that the kernel is ready for general testing prior to release, and there are few known showstopper bugs.

When 2.5 goes -rc, or Linus starts making prereleaserr noises, then go ahead and call it "beta". Until then, it's the type of thing you inflict on a computer you don't mind messing around with.

slow sunday

[oyenstikker]

a slow Sunday is a good excuse to catch up.

No. A slow Sunday is a good excuse to go spend time with your loved ones, take a walk, go to church, and enjoy a home cooked meal with your family.

So why am I reading slashdot on a Sunday?

Sega Genesis emulators on Linux

[yerricde]

Does Linux support blast processing yet??

"Blast Processing" is the name of the sprite engine that Sega used in Sonic 2 and Sonic 3 for Sega Genesis. Here are some Genesis emulators for UNIX and Linux systems [zophar.net]. DGen is pretty good.

Kernel 2.6

[i_am_nitrogen]

Actually, some people believe that the 2.0 series is the most recent quality Linux kernel.. While obviously that's open for debate, there are people hard at work backporting 2.2.x and even 2.4.x subsystems to make kernel 2.0.40. which I imagine will be released some time this year.

Personally though, I can't wait until 2.6... I know someone who's working on some of the new graphics stuff in his spare time (the new graphics layer is code named "Ruby"), and there will be some sweet stuff. The DRI, framebuffer, Video4Linux, etc. systems will all be made into one unified kernel interface, which will be user friendly and capable enough to (almost) program graphics applications in bash! Imagine (device names are changed to protect the innocent ;p):

# set video format
echo "640x480x24" > /dev/gfx/camera/cam0.mode
echo "640x480x24" > /dev/gfx/video/vid0.mode

# display snapshot
cat /dev/gfx/camera/cam0.gfx > /dev/gfx/video/vid0.gfx

I mean, that's just too cool for words.

Not to mention we'll finally be able to ditch X on the desktop for the framebuffer without losing OpenGL support, and let X do what it was meant to do: thin clients and network terminals.

Re:aa VM patch

[Anonymous Coward]

The 2.5.x kernels do not have the aa VM patch yet because there are tremendous changes being made to I/O stuff as well as the IDE and SCSI layers, and any additional major VM changes (which are not required to get SCSI or IDE working) would make debugging too difficult at this point.

Parts of the aa patch have already been merged into the 2.4.19 prereleases. Most of the aa patch will have to wait for 2.4.20 however, as 2.4.19 is also receiving major IDE updates and, again, doing the rest of the aa patch at the same time would make debugging too difficult. The 2.4.19 IDE changes are arguably more important, too; not having the aa VM means lower performance, whereas the IDE updates fix data loss problems and even (in the case of newer IBM laptop hard drives) head-parking-related problems with physical damage to hard drives during powerdown.

Kernel Testing Tips

[goingware]

If you are new to compiling your own kernel, or you would like information on how to more effectively test development kernels (or stable kernels, before putting them into production), these two articles may be helpful to you:

• Why We Should All Test the New Linux Kernel [sunsite.dk]
• Using Test Suites to Validate the Linux Kernel [sunsite.dk]

Also check out the Open Source Development Lab's [osdlab.org] Scalable Test Platform [osdlab.org]. You can use STP to run your kernel patches and test code that you upload to OSDL's big iron hardware, or you can download the STP source code so you can use it as a test harness on your own machine.

(I should add the STP to my article but haven't gotten around to doing so yet).