Linux 3.3 Released 314
diegocg writes "Linux 3.3 has been released. The changes include the merge of kernel code from the Android project. There is also support for a new architecture (TI C6X), much improved balancing and the ability to restripe between different RAID profiles in Btrfs, and several network improvements: a virtual switch implementation (Open vSwitch) designed for virtualization scenarios, a faster and more scalable alternative to the 'bonding' driver, a configurable limit to the transmission queue of the network devices to fight bufferbloat, a network priority control group and per-cgroup TCP buffer limits. There are also many small features and new drivers and fixes. Here's the full changelog."
C6X support is surprising (Score:5, Interesting)
Wow, I had no idea there was work in porting Linux to DSP architectures. That's quite an interesting development. I wonder what the use case is, since DSPs are typically used for very specific, real-time work, not for hosting general-purpose operating systems.
Also, it's quite surprising to me since as far as I know it's necessary to use TI's compiler to generate C6X code. I found one initiative to port GCC to it, but afaik it didn't get finished. My understanding is that it is no small job to get Linux to compile on non-supported compilers, so I'm interested in the toolchain they are using. For my own work on a C6711, I've been using the TI compiler under Wine. (Which works fine actually, although I had to generate an initial project in CodeComposer to get some of the board-specific support files.)
Re:C6X support is surprising (Score:4, Interesting)
Re:Which distributions? (Score:5, Interesting)
I cannot answer this question for any GNU/Linux distribution except for Slackware, which may or may not get Linux Kernel 3.3.xx as part of an official distribution for at least one Slackware release iteration ... But my personal Slackware machine will be getting 3.3 as soon as it finishes building and I reboot the machine. ;-)
It's nice to have a GNU/Linux distribution that doesn't jerk users around with strange application locations, misaligned library versions, or an update schedule tied to commercial support contracts. I've tried the rest, and I returned to the best (imho), since GNU/Linux kernel 0.96. Don't try dropping a new kernel source tar-ball onto RH Enterprise Server, Fedora, or even Ubunto -- it will break your system, and your $$$$ support agreement.
Power Management (Score:4, Interesting)
Any improvements to power management? It pains me that my laptop gets 4 hours battery life when in Windows 7 but only 2 hours when in Linux. In both cases it's just idle with nothing special running in the background. Or is this a problem with the distribution?
Re:Bufferbloat (Score:2, Interesting)
You want to limit your outgoing transmission speed using QoS to be just under your outgoing bandwidth limit. This prevents your ISP from buffering traffic and reduces latency, increasing responsiveness to things like incoming SSH connections.
Nice to see AOSP code in the mainline kernal. (Score:3, Interesting)
Re:Power Management (Score:4, Interesting)
Anyone rebuilding their kernel still? (Score:5, Interesting)
Back in the Middle Ages (late 1990s through about 2004) I remember us all getting excited for new kernel releases, and then all rushing to download the source and build it. (By 'us' i mean myself and local geek friends, as well as our cohorts on various IRC channels).
Nowadays with auto-configuring, rolling release desktop distributions being the norm, is kernel building now only done in server room environments and for non-PC hardware?
This doesn't matter much, I'm just curious.
Re:Way to go....... (Score:4, Interesting)
the android merge means i can play angry birds without having to use wine...
the various fixes means my decaying old netbook will still remain usable, and even appear snappy next to one half it's age.
New "team" network driver (Score:5, Interesting)
I am a bit confused with regards to the new team network driver which is going to eventually replace the current bonding net driver. The kernel newbies page says that it is user-space and uses libteam to do its work, but it also says that this new implementation will be more efficient.
How is this so? As network throughput keeps increasing, it is important to process each packet as quickly as possible. That's why network drivers and the packet filter are in the kernel. Wouldn't moving the new team/bonding work to user-space mean a lot more data for the kernel to copy back and forth between kernel and user spaces? And wouldn't this hurt efficiency? I'm sure the computer can keep up in most cases, but it seems this will require more CPU time to handle the work.
Just curious...
Re:Why Should I Care? (Score:0, Interesting)
At least a month before somebody creates a Metro-style UI for X.... and 10 years before it becomes stable.
KDE still lacks the simplest functions, like a taskbar that is always 100% of the screen width. Annoys me every day.
Re:Why Should I Care? (Score:5, Interesting)