Linux Kernel 2.6.29 Released 265
diegocgteleline.es writes "Linus Torvalds has released Linux 2.6.29. The new features include the inclusion of kernel graphic modesetting, WiMAX, access point Wi-Fi support, inclusion of squashfs and a preliminary version of btrfs, a more scalable version of RCU, eCryptfs filename encryption, ext4 no journal mode, OCFS2 metadata checksums, improvements to the memory controller, support for filesystem freeze, and other features. Here is the full list of changes."
Re:Access Point Wi-Fi? (Score:5, Informative)
Ummm I'm pretty sure thats the ability to act as a wifi access point, which windows can't do yet.
The most important missed out feature (Score:5, Informative)
I can't believe this wasn't mentioned..
Here's what the new linux logo looks like [blogspot.com] for this release.
Re:Access Point Wi-Fi? (Score:3, Informative)
New mascot (this kernel only) (Score:5, Informative)
As everybody knows, only important fixes will be merged into the mainline kernel at this late stage of the development cycle. One of the fixes merged by Linus on March 17 was a high-resolution SVG image of "Tuz," the mascot of the 2009 linux.conf.au conference. Tuz, in his new home at Documentation/logo.svg, serves to remind the world of the difficulties faced by the Tasmanian devil and how the linux.conf.au attendees supported the effort to save this species from extinction.
Better crypto support == goodness. (Score:5, Informative)
eCryptfs filename encryption
Here's the eCryptfs home page [launchpad.net] for more information on this nifty addition.
Re:The most important missed out feature (Score:3, Informative)
Furry comes to mind, impressive animal comes to mind, but cute?
Re:Filesystems in the kernel! (Score:2, Informative)
Idiot!!! You can always remove bloat by not compiling the 'extra' stuff into the kernel !!!
Open Source man Open Source!!!
Re:Filesystems in the kernel! (Score:5, Informative)
You only need rootfs, which is a special type of ramfs that loads the initramfs image. initramfs is loaded by the bootloader, so probably GRUB or LILO or ELILO.
Then if every other filesystem was based on FUSE, you would load the initramfs with the FUSE module, the FUSE setup programs and a config file.
Re:Filesystems in the kernel! (Score:5, Informative)
It doesn't work like that. The kernel never uses its own filesystems' support to load itself... How could it if it hasn't been loaded yet? That's the job of a "boot loader". The most user boot loader currently is Grub, and previously was Lilo.
Grub supports some filesystems, so it can access them and load the kernel. Lilo did not support filesystem, so there was a tool that you needed to run each time you changed the kernel. That tool built a list of blocks, so that Lilo could load the kernel (from those blocks) without really understanding the filesystem.
Oh come on, that's totally on topic! (Score:5, Informative)
Re:Access Point Wi-Fi? (Score:2, Informative)
Mac can't do it either. Ad-hoc is not the same as being an access point.
Comment removed (Score:5, Informative)
Use an initrd. (Score:5, Informative)
Even in Linux, most distro's don't have full filesystems built into the kernel. Instead they only build in a tiny in-memory fs that allows them to read an initrd [wikipedia.org]. This means that they can have virtually any filesystem as a root filesystem without having to compile every conceivable filesystem into their general purpose kernel.
It is also possible to avoid ever booting in the way Linux machines boot. Instead, the boot process could act like the hibernate/resume functionality of Linux. So instead of loading programs into the address space from a filesystem, we simply read the resulting address space from disk. After all, some embedded devices don't need to ever use a filesystem, so in these cases loading a fs would be a waste of resources.
Re:Filesystems in the kernel! (Score:3, Informative)
Thanks! Somehow, in all of the other GRUB vs. LILO discussions I've read this difference was never mentioned. GRUB being able to read filesystems makes a logical reason for using it instead of LILO.
Re:Filesystems in the kernel! (Score:5, Informative)
Then if every other filesystem was based on FUSE, you would load the initramfs with the FUSE module, the FUSE setup programs and a config file.
Actually, user space filesystems are nice, but they are way too slow for implementing a high speed server and/or even a decent desktop machine. They are good for experiments and pioneering work though (like GMailFS and SSHFS), but having a good set of fast, basic filesystems in the kernel is just obligatory AFAIK.
Re:Drivers??? (Score:5, Informative)
As long as you followed step #0. Check printer compatibility here [linuxfoundation.org] and scanner compatibility here [sane-project.org]. Unless they got a Tux logo or something, because there are still devices that don't have Linux drivers. I agree, when it works it works much better on Windows and most things work, but a two minute googling may still save you a lot of grief. Plus, there's nothing wrong with supporting manufacturers that really have first-class Linux drivers.
Re:Filesystems in the kernel! (Score:3, Informative)
A database management system is nothing but a fancy filesystem with structured files
And some databases come with their own built-in filesystem drivers!
http://www.oracle.com/technology/deploy/performance/pdf/TWP_Oracle_HP_files.pdf [oracle.com]
Re:The most important missed out feature (Score:5, Informative)
I am impressed that Torvalds knows about this issue, and credo to him for raising people's awareness.
There's rather more community history behind it than that. The below is from memory based on various coverage I read on LWN and the like, but not fact-checked to be positive my memory is correct, so verify before acting on it as fact.
I believe it was at the annual linux.conf.au, tho I'm not sure but it was some such conference, widely attended by Linux kernel hackers, that the presentation was made. There was apparently a fairly big charity pledge drive related to the issue, with many of the kernel hackers taking part. Various ones of them, in addition to pledging their own money, pledged various acts should the conference pledge drive reach whatever goal ($10K, maybe?).
Well, the pledge drive was quite a success, and the various hackers either have or are in the process of fulfilling their various promises as a result. One of the ones that made Linux hacker community (and LWN) headlines was Bdale Garbee's pledge, to shave his beard. He hadn't been beardless in, I think, well over a decade (15 years? longer?). There was an LWN article on it with a photo (taken I believe at the closing ceremony or traditional post-conference party) of Linus as barber, doing the honors! =:^)
That's actually how I first heard about the whole thing, seeing that photo and reading the accompanying article. But apparently Linus' own pledge was to name a kernel version after the Tazmanian Devil. But he has actually gone one better, changing the logo for .29 as well as the name.
This logo, BTW, is the one the kernel framebuffer driver optionally displays at the top of the screen during boot, if the framebuffer is activated and the config option set to do so. There's a single logo displayed for every CPU/core, so my dual dual-core Opteron displays a nice row of four such logos. I can only imagine the row of 32 of the things on say a quad-socket oct-core machine. =:^)
Anyway, I've been running a kernel compiled directly from git for a few months now (switching to the stable series between release and rc2 or so, only running mainline git between rc2 and release), and am currently running:
$uname -r
2.6.29-rc8-223-ga1e4ee2
So I've had the pleasure of seeing four of these little beasties at boot for a week or so, now. =:^)
Anyway, it's not just Linus. It's the entire kernel hacker community that got involved, thanks to linux.conf.au. =:^)
All that said, while I obviously knew more about the Linux/kernel community side of things and had a bit of general awareness from that, I hadn't bothered reading up on the disease itself until taking the opportunity to click that nice wikipedia link you so thoughtfully provided. Now I know a bit more about it, and am hopefully returning the favor with the above info on the Linux community side of things.
OK, I did an LWN search and here's some relevant links, so folks can fact-check what I wrote above, as well as quote something more authoritative than just some /. post.
LWN 2.6.29 kernel announcement (mentions the code name):
http://lwn.net/Articles/325047/ [lwn.net]
That points to Linus' actual announcement (LKML announcement as seen on LWN):
http://lwn.net/Articles/325048/ [lwn.net]
The kernel gets a new logo (a comment links the actual git commit by Rusty Russel):
http://lwn.net/Articles/323966/ [lwn.net]
Beardless Bdale (It'd be interesting to see the stats for this one as related to the Linus in a swimsuit one, I think also linux.conf.au from a few years ago, dunk tank FWIW, see below.)
http://lwn.net/Articles/316282/ [lwn.net]
(FWIW, LCA/linux.conf.au, correct. AU$35-40K raised according to "beardless". With the awareness brought by 2.6.29 related publicity, hopefully much more
Not what you are asking for but... (Score:5, Informative)
The Linux wireless drivers page lists which drivers support master/access point mode [linuxwireless.org] (see the AP column). The list isn't perfect (the hostap driver definitely supports AP mode :-) but it seems to be a case of omissions. The table also says what form factor the supported chipsets come in (so you can tell which ones you will be able to get in USB form). I'd guess the rt2500usb or p54usb drivers would be your best bet.
Another useful page is the Linux wireless chipset directory which tries to list which cards have which chipsets [passys.nl] (there's even a single page table with all the added chipsets but I won't link to it from here). This lets you build a list of boxes with the desired chipsets inside them (finding out whether this is true in reality can in itself can be a fraught process though). The chipset is really the important part in all of this.
I'm not going to point to an Amazon page because I have not bought a USB wifi card with the capabilities you describe from Amazon. I'm in no position to tell you that XYZ USB device on Amazon definitely works as I haven't done it myself. I have used hostapd on Linux and OpenBSD before now on a creaky old Prism 2.5 card and that worked for me but again that's not what you asked.
Finally here's a guide to using hostapd to set a card up in access point mode [kernel.org] (just using iwconfig to set master mode is not enough). Googling for hostapd linux will turn up plenty more guides which may be easier to follow.
Good luck!
Re:Access Point Wi-Fi? (Score:1, Informative)
Ummm I'm pretty sure thats the ability to act as a wifi access point, which windows can't do yet.
Eh? When was the last time you used Windows? Vista certainly has had the ability to be configured as an access point since the start so that's over 2 years.
Re:Filesystems in the kernel! (Score:3, Informative)
KMS is Intel-only for now (Score:4, Informative)
those who don't know plan 9 (Score:1, Informative)
reimplement it, poorly.
Re:The most important missed out feature (Score:3, Informative)
Kangaroos! There was some fossil evidence of them in a carnivorous state. Big, scary fast with teeth. Even today they can be shitheads as herbivores. I was feeding a doe and she dug her claws into me so I wouldn't leave until she ate all the birdseed that was actually for the parrots. Lucky kangaroos taste good.
Re:Access Point Wi-Fi? (Score:3, Informative)
Bridge does only do layer 2 switching between the 2 NICs and would not provide DHCP or routing. However, if you use "Internet Connection Sharing", it turns the computer into a basic NAT router. Very basic. Although, since you are sharing over ad-hoc, I don't think you can use WPA or WPA2 (could be wrong about that). More than one client can access the ad-hoc network. However, they have to support ad-hoc mode. For example, the Nintendo DS only supports infrastructure mode and will not connect to an ad-hoc network.
Re:Drivers??? (Score:3, Informative)
The golden ratio isn't really all that useful for writing paper.
Except that, when you have an long-side-leading A4 printer, it can also print A3, because the long side of A4 is the same as the short side of A3. It makes the printer a little more versatile.
Re:Filesystems in the kernel! (Score:3, Informative)
http://grub.enbug.org/CurrentStatus?highlight=(ext4) [enbug.org]
grub2 currently supports ext4 and a google search will find you some patches to add it to older versions. I don't know if any distros include an ext4 compatible grub yet though, well aside from maybe gentoo ;)
Re:Not what you are asking for but... (Score:2, Informative)
As the author of most of the information on the hostapd page on wireless.k.o, I can tell you that, up until recently, most of that information was changing on a weekly basis. Now that 2.6.29 has landed, the documentation will start to firm up quickly. I figured I had at least another week to finish the hostapd docs. While the documentation included with hostapd is good, it's too much. There are like 200 options that only a handful of people would even need to consider in it.