Linux 2.6.27 Out 452
diegocgteleline.es writes "Linux 2.6.27 has been released. It adds a new filesystem (UBIFS) for 'pure' flash-based storage, the page-cache is now lockless, much improved Direct I/O scalability and performance, delayed allocation support for ext4, multiqueue networking, data integrity support in the block layer, a function tracer, a mmio tracer, sysprof support, improved webcam support, support for the Intel wifi 5000 series and RTL8187B network cards, a new ath9k driver for the Atheros AR5008 and AR9001 chipsets, more new drivers, and many other improvements and fixes. Full list of changes can be found here."
Re:uname -a (Score:2, Informative)
Based on what I'm used to ubuntu doing, they're probably treating one of the later release candidates (of which this is the tenth) as something "good enough" for ubuntu 2.6.27.
Re:This is a huge amount of work (Score:5, Informative)
Well, yes and no. The old LK dev model had unstable releases where bugs were expected. Now every release is stable, and bugs are truly anomalies.
Re:Not in upcoming Debian (Score:5, Informative)
It's a shame this won't be in the upcoming Lenny release of Debian. The in-kernel support for heaps of webcams via gspca is a very nice user-visible element of this release.
Debian never paid much attention to desktop features, may I suggest Ubuntu 8.10 [ubuntu.com]?
Re:This is a huge amount of work (Score:5, Informative)
In only 3 months, all of this code has been completed and reviewed by multiple developers. This happens *every* three months. ... It is clearly the case that the Linux kernel has hit a new kind of critical mass and is now a form of software development that has never been seen before.
Intel HDA audio still has static noise in the left channel since at least 2.6.20 kernel (probably before). This is a known problem and the solution is 'try random settings of some undocumented (outside the kernel source code) module parameters and hope it maybe works'.
This is on Dell hardware. model=dell3stack, position_fix=1(?). This hardware works perfectly under Windows, with no tweaking whatsoever. It worked under older linux kernels, which means they probably broke something.
The linux kernel is good, but just having a bunch of people look at the code means nothing unless they are actually finding and fixing problems people care about.
Re:AR5008 (Score:5, Informative)
Re:Not in upcoming Debian (Score:5, Informative)
I believe you have my stapler?
No, this one isn't red.
Re:Did Bill Gates pay Shuttleworth to create Ubunt (Score:5, Informative)
SYSTEM and Local System are basically one and the same, and are almost perfectly synonymous with root. Network Service would be the equivalent of the "nobody" user - i.e. an account that you can use to run low-privilege services. Administrator would be the same as a user with administrative privileges in Linux (perhaps someone in the sudoers list). The trouble, of course, was that, until Vista/2008 came along, it was trivially easy for an Administrator to escalate to SYSTEM - you just had to run a scheduled job in interactive mode (think of a cron job with no password required) and you'd not only have root access, you'd also have access to the current user's console. For an administrator, this came in handy - of course, what was handy and convenient for an administrator was just as handy and convenient for someone else.
Re:'pure' flash devices (Score:5, Informative)
Naturally upcoming Maemo (Nokia Internet Tablet) releases will feature ubifs, since much of the work on it has been done by Maemo Software kernel team.
Re:So where does that place OS X? (Score:3, Informative)
Some of these features are genuinely interesting, though. For example:
OMFS stands for "Sonicblue Optimized MPEG File System support". It is the proprietary file system used by the Rio Karma music player and ReplayTV DVR.
In other words, it means I can open up certain embedded devices -- particularly that DVR -- and pull files off the hard drive. I suppose the OS X answer is that I should've gotten an AppleTV instead?
In this release, Ext4 is adding one of its most important planned features: Delayed allocation, also called "Allocate-on-flush". It doesn't changes the disk format in any way, but it improves the performance in a wide range of workloads.
Only way OS X is getting this is if it's an undocumented feature in HFS (unlikely), or if they port ZFS.
Kexec jump: kexec/kdump based hibernation
Reading through this, it looks like it's really nothing new, just slightly more flexible than before.
But what we had before allowed quite a lot of things not possible on OS X -- for example, diskless hibernation, or hibernation-as-snapshots, even to the network, etc.
There are, of course, a ton of them that cover problems Apple doesn't have. I would consider them nice problems to have.
Oh, and as to the original question: It changes absolutely nothing about OS X's position. People who like the UI, and can afford a mac, aren't going to complain about OS X being less efficient than Linux. Most of the more clever use cases are about as useful to an OS X user as ZFS is to a Linux user -- a curiosity, and maybe useful as a network-connected device, but no impact on what you use for a core OS.
Unless you count Xserves, but I'm not sure that was ever a good idea.
Re:Linux 2.6.27 Out (Score:5, Informative)
For those moderators who didn't get it and modded him Troll: it's the (only) line from the Pink Floyd song "One of these Days".
By the way, it's "cut", not "chop": http://en.wikipedia.org/wiki/One_of_These_Days [wikipedia.org]
Re:flaimebait? (Score:2, Informative)
Re-buying peripherals (Score:4, Informative)
Linux drivers are much easier to deal with.
Unless you're switching to GNU/Linux and don't want to have to buy all-new peripherals. To pick a random example from my collection of incompatible hardware, Microtek isn't helping the SANE project make drivers for its ScanMaker 4850 flatbed scanner.
Comment removed (Score:5, Informative)
Re:Current Limiting? (Score:4, Informative)
Surely a lot of that is up to the compiler. In fact everything you mention there is likely to be a compiler decision unless you code stuff in assembly (which means you are already being fairly processor specific).
Math Coproc: Replace it with a (much slower and longer) integer based floating point algorithm.
MMX/SSE: You just have to do lots of operations, rather than in one fell swoop.
The big one is having a MMU, which has been there on x86 architectures since the 386, and on pretty much any other processor outside of the embedded arena. For those systems you have uClinux, which has a 2.6 kernel release.
I've used some of the processors available on opencores (some of which are written from scratch and are quite different from existing processors) but many of those have had linux kernels ported to them.
Having the source available makes a huge amount of stuff possible. You could probably compile for a Turing machine if you were sadistic enough.
Re:Building your own kernel these days ain't easy (Score:5, Informative)
And you really need to do this once. After that for each new version, you just do "make oldconfig" against the old
Changed hardware? New PC? Just reconfigure the "Drivers" section in a few minutes and you're golden. That's assuming of course you stripped down everything you don't need - if you left it in, you don't evenhave to do as much, it will just work.
BTW, if you're into tinkering, go all the way and try Gentoo. That project is alive and kicking, regardless of what the media have been saying recently.
Re:Not in upcoming Debian (Score:4, Informative)
Yeah ok not a good idea to talk about things that you dont know about.
On Gentoo it uses the kernel in /usr/src/linux, since your expected to roll your own kernel anyway.
It is exactly that simple to install the driver - one command - even though its a power user's distro.
On the user friendly distros like Ubuntu it will install the binary blob version for your kernel just like Windows but without the cd.
They only have a couple of kernel versions just as Windows only has XP, Vista, etc... drivers.
Re:Building your own kernel these days ain't easy (Score:3, Informative)
um... the "whacky idea" here is that you start off with defaults that you know works... and then change things that you know you want to change. For example, compile in drivers/filesystems/etc that you otherwise would be loading as modules... tell it to compile using instructions for your specific cpu model, rather than a generic 'i686'... remove any unneeded sections for example sata if running on a pata-only system, firewire, memory card support, video for linux, ISDN/modem/ppp support... this isn't a difficult concept; you don't have to know what everything is to know that there are some things you can do to make improvements.
Re:Not in upcoming Debian (Score:3, Informative)
If something is broken in Ubuntu, it will continue to be broken for 6 months.
Wrong.
The 6 months wait is for new features.
Security updates and bug fixes are constantly released.
Re:Not in upcoming Debian (Score:2, Informative)
First add testing repositories to
To be honest however I think its masochistic running stable on anything thats not a server, and pinning is a little messy to setup as you might have noticed... I've been running testing for over a year and everything has been perfect (well bar Audacious...) and I have got access to lots of packages stable doesn't have.
Re:Kernel changes I wanted in as of v1.0 (Score:2, Informative)
Read: stable_api_nonsense.txt [linux.no]
Re:Re-buying peripherals (Score:5, Informative)
To pick a random example from my collection of incompatible hardware, Microtek isn't helping the SANE project make drivers for its ScanMaker 4850 flatbed scanner.
That's OK. It looks like they're doing you a favor [amazon.com]. If it makes you feel any better, they don't support Vista either [microtek.com].
Re:Current Limiting? (Score:3, Informative)
2) If you're using a laptop (or any desktop more recent than 2000) you have a 99.99% chance of hibernation working flawlessly in any Linux distro.
Ha ha ha. Right. Sure.
For the record, on my T61, hibernation in Ubuntu has *never ever worked*. Ever.
1) The kernel/Linux has long been doing an excellent job on using power-saving features of processors and peripherals
But this I agree with. Linux actually consumes less power than Windows Vista does, which surprised the heck out of me. On my rather power hungry T61, after a fair bit of tuning to enable low power mode for various devices (audio chipset, SATA hosts, wireless, and USB, including removing the UHCI USB driver entirely), along with some other tweaks, Linux consumes around 13.4 watts idle, give or take. By comparison, for all my extensive tuning, Vista has never ever dropped below 14.8 or so.
Re:Intel e1000e bug fixed? (Score:4, Informative)
Nobody has figured out the bug. The patch is so that the bug (whatever it is) cannot destroy the card. It does this by setting the hardware so it ignores any attempt to overwrite it.
This should be pretty obvious from the comment you quoted.
As far as I can tell, the hardware "works". If that card did not work then probably people would not notice this bug, because they would not see the hardware fail! In fact it is strongly suspected the bug is not in the driver but in some other part of Linux.