Linux Kernel 4.12 Officially Released (softpedia.com)

prisoninmate quotes Softpedia: After seven weeks of announcing release candidate versions, Linus Torvalds today informs the Linux community through a mailing list announcement about the general availability of the Linux 4.12 kernel series. Development on the Linux 4.12 kernel kicked off in mid-May with the first release candidate, and now, seven weeks later we can finally get our hands on the final release... A lot of great improvements, new hardware support, and new security features were added during all this time, which makes it one of the biggest releases, after Linux 4.9...

Prominent features of the Linux 4.12 kernel include initial support for AMD Radeon RX Vega graphics cards, intial Nvidia GeForce GTX 1000 "Pascal" accelerated support, implementation of Budget Fair Queueing (BFQ) and storage-I/O schedulers, more MD RAID enhancements, support for Raspberry Pi's Broadcom BCM2835 thermal driver, a lot of F2FS optimizations, as well as ioctl for the GETFSMAP space mapping ioctl for both XFS and EXT4 filesystems.
Linus said in announcing the release that "I think only 4.9 ends up having had more commits," also noting that 4.9 was a Long Term Support kernel, whereas "4.12 is just plain big."

"There's also nothing particularly odd going on in the tree - it's all just normal development, just more of it than usual."

Comments

Some links

[cerberusss]

There were two things that I didn't know about, so I figured I'd share those links:

F2FS is a flash filesystem: https://en.wikipedia.org/wiki/...
BFQ is a scheduler for I/O: http://www.phoronix.com/scan.p...

Re:Some links

[drinkypoo]

Both of which are old news in Android-land. My TF201 is using both of them. I'm surprised they took this long to be mainlined.

Re: Some links

[Anonymous Coward]

F2FS has been mainlined for some time already. These are just updates.

Question

[Anonymous Coward]

Does it still have EISA support for that one dickhead who refuses to upgrade his ATM card?

Re:Question

[Anonymous Coward]

I jack off to having a 486DX50 (not DX/2) with a high end EISA graphics card you insensitive nut.

Is the kernel itself being improved ?

[Anonymous Coward]

The release notes I read seem to concern adding new capabilities to Linux, but not IMPROVING the code in the kernel.

Are any changes happening there, or is it now perfect and set in stone forever ?

Re:Is the kernel itself being improved ?

[perpenso]

The release notes I read seem to concern adding new capabilities to Linux, but not IMPROVING the code in the kernel. Are any changes happening there, or is it now perfect and set in stone forever ?

They are waiting for you to implement those kernel improvements and to submit them. :-)

Re:Is the kernel itself being improved ?

[Anonymous Coward]

I would totally code it for you except I'm more of an idea man, you know what I'm saying. But see, here's the deal. I got a kid brother who's a nerd, so I'll get my kid bro to do the coding this Open Source Friday. I know just the thing to motivate him. Gonna buy him a hooker.

Re:Is the kernel itself being improved ?

[Anonymous Coward]

It wasn't a dig, but a serious question.

Presumably the kernel is not perfect ?

So is there no work being done to actually improve the linux kernel itself ?

Or do real improvements require a new operating system because any real improvements would Break Stuff ?

Re:Is the kernel itself being improved ?

[Hognoxious]

Soon systemd will include its own kernel anyway, so there's no point.

Re:Is the kernel itself being improved ?

[Anonymous Coward]

Innovation does not occur in Linux. Linux is not a research kernel. Linux is a trivial engineering exercise that turned into a cash cow for an idiot to avoid getting a real job.

Re:Is the kernel itself being improved ?

[Anonymous Coward]

The systemd kernel will be called Hurd and RMS will jizz his pants.

Re:Is the kernel itself being improved ?

[Anonymous Coward]

Improvements happen all the time. Some of these are bug fixes, some are improvements, some are new things.

Take a look at just this one branch's (wireless next) commit logfor one particular feature in git a few years ago -

2014-12-20 CRISv32: Implement early console Jesper Nilsson 1 -26/+56
2014-12-20 CRIS: Use KALLSYMs if available in call stack dump Jesper Nilsson 1 -27/+34
2014-12-20 CRISv32: Fix declaration mismatch Jesper Nilsson 1 -1/+0
2014-12-20 CRISv32: Rewrite of synchronous serial port driver Jesper Nilsson 1 -639/+791
2014-12-20 CRIS: Update init memory handling Jesper Nilsson 1 -5/+33
2014-12-20 CRISv32: Better handling of watchdog bite Jesper Nilsson 1 -11/+17
2014-12-20 CRIS: Export missing function symbols Jesper Nilsson 3 -19/+19
2014-12-20 CRIS: Export ioremap_nocache Jesper Nilsson 1 -1/+2
2014-12-20 CRIS: Fix headers_install Sam Ravnborg 3 -8/+2
2014-12-20 CRISv32: Add missing include for mm.h Jesper Nilsson 1 -0/+1
2014-12-19 CRISv32: Drop obsolete file for SPI driver

Re:Is the kernel itself being improved ?

[Anonymous Coward]

Linux is a research kernel that happens to be used for production. The research bits don't make it to production until they are no longer research bits.

Stuff like this happens in linux - that would be research kernel work. http://www.academia.edu/25487331/Concise_Paper_LAAR_Long-range_Radio_Assisted_Ad-Hoc_Routing_in_MANETs

Re:Is the kernel itself being improved ?

[Anonymous Coward]

That looks like minor fixes, and many of those are clearly not for the core kernel but for helping it work with external devices ( "synchronous serial port driver", "Drop obsolete file for SPI driver", " Implement early console" ).

I am talking about things that improve the kernel itself.

Re:Is the kernel itself being improved ?

[Anonymous Coward]

That VERY CLEARLY has nothing to do with improvements to THE KERNEL ITSELF.

Re:Is the kernel itself being improved ?

[gl4ss]

some recent interview with linus was that every now and then he gets submissions in places he thought didn't have much to improve or fix bugs.

you don't put those on the changelogs though.

Re:Is the kernel itself being improved ?

[Anonymous Coward]

"I would totally code it for you except I'm more of an idea man, you know what I'm saying."

Ah, yes, the Dunning-Kruger explanation.

Re:Is the kernel itself being improved ?

[Anonymous Coward]

Big ones usually go into major releases. People work on more-or-less private branches, develop new improvements, and when they are perfected, they make it into the mainline.

But stuff like "Use KALLSYMs if available in call stack dump", "Update init memory handling" "Export ioremap_nocache" are exactly the kernel improvements - small things that can be added to another incremental update without causing headaches to everyone else working on the kernel.

Plus a lot of stuff is modularized, but really "core, heart of heart" where the architecture is concerned. That BQF - a scheduler. A scheduler is the absolute central part of every kernel on any multitasking system, the singular most important subsystem, which decides which process is to run, and for how long, before preempting it and handling the CPU time to another process.

Linux kernel implements several scheduling algorithms ("choose one that fits your needs best") and so it implements them as modules, so that you can switch between them freely. But each of them is what any Computer Science book would call 'the kernel' - and adding another one, that provides an algorithm suitable for another set of situations is certainly an improvement - as are improvements to any pre-existing schedulers. You can't get any more "improve the kernel itself" than that.

Re:Is the kernel itself being improved ?

[swillden]

I am talking about things that improve the kernel itself.

How about BFQ? I/O scheduling is a core kernel function, and BFQ is a significant improvement.

Re:Is the kernel itself being improved ?

[epyT-R]

What would you like to see improved?

Re:Is the kernel itself being improved ?

[HiThere]

What would you count as a kernel improvement?

Others have said that they consider some of the things posted as what they think of as kernel improvements. If those don't meet your criteria, what would? (And remember, we're talking kernel here, not user space.)

Re:Is the kernel itself being improved ?

[Anonymous Coward]

Just how much can you touch in "the kernel itself" after 25 years that won't break some subsystem or driver in an obscure part of the tree? By this point, the kernel has had a ton of people looking at and hacking it. If it was so trivial to improve "the kernel itself", go find a bug and fix it. Earn a name for yourself.

Re:Is the kernel itself being improved ?

[Anonymous Coward]

Thanks for your reply.

Is it then the case that only minor kernel improvements can happen "without causing headaches to everyone else working on the kernel".
i.e. There can be no major leap in the technology used because it would break stuff ?

And we have to look at other, less mature OSes' kernels to see those leaps ?

Or are there just no leaps, anywhere, any more ?

Re:Is the kernel itself being improved ?

[Anonymous Coward]

You think the LInux kernel is the absolute pinnacle of what mankind can achieve ?

Re:I hope it fixes the 49.7 day reboot...

[bug_hunter]

Is this what you do now? Keep repeating a Windows 95/98 bug against the Linux Kernel?

Re:I hope it fixes the 49.7 day reboot...

[Kjella]

Is this what you do now? Keep repeating a Windows 95/98 bug against the Linux Kernel?

It's like intentionally mixing up Star Trek and Star Wars to see the reaction. In every forum there's someone there just for the lulz, simply ignore them and maybe they'll go back to 4chan.

I hope they will fix btrfs soon

[Anonymous Coward]

According to the wiki btrfs raid is completely broken and should not be used.
I hope they will manage to fix it soon, so I don't have to use ZFS, which is out of tree.

Re:I hope they will fix btrfs soon

[guruevi]

BTRFS is broken by design. It should not be used, it should be tossed and any efforts going to ZFS.

Re:I hope they will fix btrfs soon

[Anonymous Coward]

Except the copyright holders of the standard ZFS implementation denied permission for it to be used in a GPL program.

Re:I hope they will fix btrfs soon

[Anonymous Coward]

ZFS is dead.
It's license means it can never be added to mainline linux.

Re:I hope they will fix btrfs soon

[jwhyche]

Why is BRTRS broken?

Re:I hope they will fix btrfs soon

[Anonymous Coward]

BTRFS is not really broken by design. From the BTRFS FAQ:

  I've heard that ... btrfs is broken by design (aka. Edward Shishkin's "Unbound(?) internal fragmentation in Btrfs")

Mailing list thread Unbound(?) internal fragmentation in Btrfs (2010)

Short answer: not a design problem but a bug (see 1) that was fixed

The core of the argument is that the b-tree implementation (from year 2010) does not follow the assumptions found in classic literature about b-trees and can lead to underutilization of the space. There are two issues raised: one about variable record length (in particular the file data inlined in the leaves), and a second issue about an inefficient b-tree balancing algorithm (see 3).

Inline extents are not split across leaves and are limited in size of one leaf. This was a decision made based on experience with reiserfs, where inline items were split, causing increased code complexity (and bugs) without much gain (see 2).

To completely avoid the theoretical worst case, you can turn off the file inlining by mounting with max_inline=0 (see 2).

In practice, the worst case means that an inline file extent may occupy one leaf block (which is more limited to the memory page size, usually 4k). This leads to the same space efficiency as in other filesystems, where a small file occupies a full block (see 4).

Re:I hope they will fix btrfs soon

[MtHuurne]

One problem I encountered with btrfs is that at about 80% full, it will refuse to create any new files saying it is out of space. This happens because the free space is fragmented: apparently btrfs can only allocate space in blocks that are entirely empty, not in partially filled blocks. I then have to run a command manually to make btrfs move the data such that the free space is combined and usable again. To make it worse, that command needs a free block to do its job, so I either have to remove a large enough file to free up an entire block or add a second block device for temporary working space.

I don't understand why a file system is considered production ready when it is as clunky to use as this. I remember the days when Linux users could make fun of Windows users having to run defrag...

Re:I hope they will fix btrfs soon

[jwhyche]

Crap like this is why I stay away from what I call "vanity" file systems. It always seems there is some kind of issues with the filesystem itself or the tools you have to use to maintain it. I'll just stick to plan old ext3/4 file system. Vanity file systems may have their advantages but their issues always seem to outweigh them.

An that is probably why I have never seen one of them in the wild. When it comes right down to it, in the linux world ext3/ext4 is king and for a good reason.

Why are we adding more schedulers?

[guruevi]

I understand maintaining the old schedulers but schedulers to most newer hardware are actually detrimental. SSD and even some modern hard drives do best with the noop scheduler simply because the overhead of a scheduler is noticeable. Even on embedded devices, schedulers take up cycles. And if you really need one, there are literally dozens of them to choose from and even though you may want to have some variations in yours, why take them all up in the mainline kernel? Just keep them separate.

Ryzen support?

[Anonymous Coward]

Is all the stuff needed for the new Ryzen boards incorporated into the kernel in 4.11?

Or are more drivers or bug fixes for Ryzen boards being added?
With a 4.11 or 4.12 can one just install a linux with these kernels on a new Ryzen boad and have everything jsut work?

Still waiting

[Anonymous Coward]

I am still waiting for the promised post mortem after the kernel.org hack.

Scheduler additions are the least of Linux'....

[Anonymous Coward]

Scheduler additions are the least of Linux' problems.

#1: Kernel Kconfig defaults have gotten sloppy and unmanageable.
    Go try looking through a 2.4 era default kernel config for x86, arm, and non-x86/arm arches. Try doing the same on an early and late 2.6 release, then compare to a 4.x release. There are *IMPOSSIBLE* driver combinations enabled by default. Furthermore, even on x86 many drivers that are niche applications default to Y, not even 'M' leading to unnecessary dependencies in your kernel update/default kernel config. This might not be a big issue for people on huge desktops where everything will be compiled in, or as a module, but for people trying to compile minimalist kernels, kernels for special hardware, or just trying to keep the threat profile for their system kernel low, this is a huge issue.

#2 Removal of old drivers 'because they are cluttering up the build tree.' Hint: Old drivers are a SMALL FRACTION of the total kernel build infrastructure. The oldest can be omitted by disabling ISA bus support. If that breaks LPC bus drivers, then perhaps it is time to segregate the two, since very few LPC devices are going to be on a system with native ISA bus support (almost all of them were after the migration to PCI-only busses, so outside of industrial motherboards or 440(X)x era chips, it is unlikely an isa bridge chip will be seen.) The modern drivers each take up more lines of code that all the old drivers put together. If you don't believe me, go look at Linux 2.4 source tree size, and compare it to 2.6 and then 4.x. Hint: The linux kernel grew 2-3x in that period and had already doubled or tripled in size since the 2.2 releases.

Re:I hope it fixes the 49.7 day reboot...

[HiThere]

IIRC, that bug only existed in MSWind 95A, not even in 95B, much less 98. (They had other problems.)

Re:Slashdotting slashdot?

[Anonymous Coward]

Now I'm wondering if the artist is actually a nerd and referred to the Eternal September with those lyrics...