Linux 2.6.28 Kernel Released

[phoronix]

Phoronix: Linux 2.6.28 Kernel Released

As a special Christmas present, Linus Torvalds has announced the release of the Linux 2.6.28 kernel. The 2.6.28 kernel stabilizes the EXT4 file-system, delivers the Graphics Execution Manager for GPU memory management, brings forth several new drivers, and is home to several other improvements. The Linux 2.6.28 kernel release announcement can be read at LKML.org...

http://www.phoronix.com/vr.php?view=Njk1Nw

[RealNC]

Ext4, yey! :D

[Tares]

Ext4, yey! :D

I have the same thoughts ;-) but I hope that ext4 wont eat 50GB of my 300GB partition like ext3 ;/ so for now I have to use reiserfs although I'm sick of it.

[hdas]

Flawless kernel. Godspeed :D.

[Loris]

I have the same thoughts ;-) but I hope that ext4 wont eat 50GB of my 300GB partition like ext3 ;/ so for now I have to use reiserfs although I'm sick of it.

How does ext3 eat one sixth of your partition? Did you set some funky block or inode values?

[Tares]

How does ext3 eat one sixth of your partition? Did you set some funky block or inode values?

Ask gparted :) when I made ext3 on 300gb partition, it said that that 50gb was already reserved. Compared to 17mb of reiserfs, my choice was pretty obvious. Well, I'm a newbie if it comes to FS's, so I fully depended on gparted.

[ethana2]

Yeah, that lost+found directory in /?
I think that's corrupting space. Don't blame your file system, thank it for saving your butt.

[DeepDayze]

Yeah, that lost+found directory in /?
I think that's corrupting space. Don't blame your file system, thank it for saving your butt.

Anything in there that shouldn't be?

[hobbes]

Last time I've formatted a partition to EXT3, it took over 8GB of my 300 GB HDD on reserved space.

Does anyone here who actually formatted a partition with EXT4 knows how it behaves?

[Loris]

Last time I've formatted a partition to EXT3, it took over 8GB of my 300 GB HDD on reserved space.

Does anyone here who actually formatted a partition with EXT4 knows how it behaves?

The default options for ext3 formatting is to reserve 5% of the partition's space for usage by root, so that the admin can always log in. But 5% on today's storage devices is a fscking big space nevertheless, so it's a good practice setting the reserved space to a lower percentage of the partition space.

DO THIS ONLY ON A UNMOUNTED FILESYSTEM.
For example, to set an arbitrary percentage for reserved space on an existing ext3 filesystem:

# tune2fs -m ${percentage_number} /dev/$partition_device

Or, you could set an arbitrary value of reserved blocks using the option -r instead of -m.

Assuming a default of 4 Kilobytes blocks, every 256 blocks are a reserved Megabyte. Setting 10 Megabytes of reserved space would be like this:

# tune2fs -r 2560 /dev/partition_device

Or... you could read the manual of mke2fs before creating a new filesystem. :D

[hobbes]

The default options for ext3 formatting is to reserve 5% of the partition's space for usage by root, so that the admin can always log in. But 5% on today's storage devices is a fscking big space nevertheless, so it's a good practice setting the reserved space to a lower percentage of the partition space.

DO THIS ONLY ON A UNMOUNTED FILESYSTEM.
For example, to set an arbitrary percentage for reserved space on an existing ext3 filesystem:

# tune2fs -m ${percentage_number} /dev/$partition_device

Or, you could set an arbitrary value of reserved blocks using the option -r instead of -m.

Assuming a default of 4 Kilobytes blocks, every 256 blocks are a reserved Megabyte. Setting 10 Megabytes of reserved space would be like this:

# tune2fs -r 2560 /dev/partition_device

Or... you could read the manual of mke2fs before creating a new filesystem. :D

Thanks for replying.

I already did that.

Just after I created the filesystem, months ago.

But, thanks again for sharing this with us.

I left the default size on my primary HDD and set -m 0 to all my backup drivers (Music, Movies, Pictures etc)

So far, so good.

I'm assuming that EXT4 still sets to 5% to root reserved space.

Have you (or anyone reading this thread) upgraded your existent EXT3 partitions to EXT4?

I'm really curious about how it went through.

Thanks for reading.

[RealNC]

I just migrated my / from ext3 to ext4. :)

I've rsynced the data on the root partition first to another partition:

rsync -axHA /mnt/root/ /mnt/backup

Reformatted root:

umount /mnt/root
mkfs.ext4 -L GentooRoot /dev/sda2
mount -t ext4 /dev/sda2 /mnt/root

And rsynced it back:

rsync -axHA /mnt/backup/ /mnt/root

Grub 0.97 can't boot from ext4 (unless patched) so I have a 40MB ext2 /boot partition for that.

[Kano]

Well, if there is a patch then use it. GRUB2 would be also interesting...

[RealNC]

I did use the patch, but probably I did something wrong with reinstalling grub and it didn't boot. So I had to create an ext2 /boot for now.

Grub2 doesn't seem to be very popular with most distros right now. No idea why though, looks quite nice to me.

[Kano]

Well you can use it with the lenny installer - also I used it for booting from a Win bootloader - it can easyly start ST too.

[Jade]

And Reiser4 is still better and more stable than Ext4.

Linus Torvalds and co., are apparently just plain ole garbage.

When is REISER4 going to be added to the mainstream kernel.

Why are the Linux Kernel SABOTEURS keeping REISER4 from the mainline kernel.

See, the Linux Kernel SABOTEURS @

http://www.phoronix.com/forums/showthread.php?t=9509
http://linux.50webs.org/jews/saboteurs.htm

REISER4 HOWTOS.

Some Amazing Filesystem Benchmarks. Which Filesystem is Best?
http://linux.50webs.org/resources/fs-benchmarks.htm
http://linuxhelp.150m.com/resources/fs-benchmarks.htm

http://www.phoronix.com/forums/showthread.php?t=1765

Compiling yourself a 2.6.23 Kernel (with Reiser4 support). (2.6.24 Kernel Patch)
http://linux.50webs.org/reiser/compile-kernel.htm
http://linuxhelp.150m.com/installs/compile-kernel.htm

Installing your favorite Linux Distro on Reiser4.
http://linux.50webs.org/reiser/install-on-reiser4.htm
http://linuxhelp.150m.com/installs/install-on-reiser4.htm


Installing GRUB on a Reiser4 Partition.
http://linux.50webs.org/reiser/grub-reiser4.htm
http://linuxhelp.150m.com/installs/grub-reiser4.htm

http://chichkin_i.zelnet.ru/namesys/
http://www.kernel.org/pub/linux/kernel/people/edward/reiser4/reiser4-for-2.6
http://www.kernel.org/pub/linux/utils/fs/reiser4/reiser4progs/
http://www.kernel.org/pub/linux/utils/fs/reiser4/libaal/

[Kano]

Finally somebody found the config option that prevented nfs from running:

CONFIG_SUNRPC_REGISTER_V4

You should keep this setting DISABLED until you want to exchange the userspace tools...

[RealNC]

lol@Jade :D

I'm pretty convinced that Reiser4 will enter the kernel tree eventually. Perhaps it would have been already in the kernel if Reiser wouldn't have started the "kernel devs suck" discussion back then. Hmm, perhaps Reiser going to jail had nothing to do with his wife; I think the kernel devs conspired and set him up for murder...

But don't be jealous on Ext4 just because kernel devs are biased towards it.

[Loris]

And Reiser4 is still better and more stable than Ext4.

Linus Torvalds and co., are apparently just plain ole garbage.

Wow, "better" is too much a technical term.

Plain offensive speaking toward Free Software developers is what makes filesystems better, I suppose. Jade, please, don't.

[hdas]

@ Jade :

About Reiser4. Whatever has happened has happened. Reiser is a genius, no doubt and Reiser4 is revolutionary. Never mind his quirky attitude and personal issues. Everyone has a personal life and we got to understand that.

Meanwhile, the important thing is to be constructive. There are no conspiracies. Anyone or any corporation wouldn't become a billionaire (like Bill Gates or Steve Ballmer) by pulling such a thing. People are sincerely trying to do what is best.

I know you are personally disturbed about reiser4 and as much passionate about it as Reiser himself. But you got to move on man. You are very talented and you can and you should code. You can yourself make reiser4 better, try to push it in the linux kernel (and hopefully every other OS :D - *interoperabilty*) and such an effort will be very very appreciable.

And it is not just about Reiser4. Look at Ext4 - they have spent years on it and I was surprised to know its just some transition file system. There is no conspiracy here either, its just for the better. There are so many efforts to make things better - Tux3, Btrfs, ZFS. Any work on any of these would be awesome.

Show them the code! And, don't let it take a toll on you. You need a bottle of beer and I wish I could get you one! :D

[RealNC]

I've noticed something with 2.6.28 that didn't happen since quite some time: The mouse cursor in X becomes skippy/jumpy when there's heavy system load (doing a "make -j5" in /usr/src/linux for example). Anyone else seeing this? Cursor was always smooth as silk before.

[Throwing Strikes]

What type of mouse are you using. I had to replace my Razer Deathadder because the mouse would get jumpy when I was in WoW. But this was under the 2.6.25 kernel.

I would try a different mouse to make sure its not a hardware issue if you haven't already.

[Loris]

I've noticed something with 2.6.28 that didn't happen since quite some time: The mouse cursor in X becomes skippy/jumpy when there's heavy system load (doing a "make -j5" in /usr/src/linux for example). Anyone else seeing this? Cursor was always smooth as silk before.

I regularly experience skippy mouse movements on:

swap space/file usage
heavy storage I/O, on machines with IDE chipset which require CPU intervention (Intel U100, my notebook is a Thinkpad A30)
heavy USB I/O, as the USB specification doesn't include DMA (remember me, what does Intel sells, again? :rolleyes:)

[RealNC]

Never had this with 2.6.26/27. It's not a big problem though. The mouse is a USB Microsoft IntelliMouse. Using evdev driver. Normally I would boot 2.6.27 again to check if it's something else at fault, but now that I went ext4, I can't.

[hobbes]

I've noticed something with 2.6.28 that didn't happen since quite some time: The mouse cursor in X becomes skippy/jumpy when there's heavy system load (doing a "make -j5" in /usr/src/linux for example). Anyone else seeing this? Cursor was always smooth as silk before.

I noticed something like that too. Just after I upgraded to Jaunty Alpha 2. Mouse cursor started to behave oddly when passing over some links on Firefox.

I thought it was probably something related to new X.org libs.

My Mouse is a A4Tech 8 buttons, using evdev too.

[ethana2]

So wait, is 5% of the ext2 partition on my flash drive 'reserved for root'?

[Loris]

So wait, is 5% of the ext2 partition on my flash drive 'reserved for root'?

If you didn't specify different options than the default, then yes, 5% of that filesystem space is reserved for usage by UID 0 (root user).

To find out how much space is currently reserved on your filesystem, try this several steps:

The total number of blocks in your filesystem is:
# dumpe2fs /dev/$your_partition_device_file | grep 'Block count'

Each block is by default 4 Kilobytes (4096 bytes), to find out your block size:
# dumpe2fs /dev/$your_partition_device_file | grep 'Block size'

The number of currently reserved blocks is:
# dumpe2fs /dev/$your_partition_device_file | grep 'Reserved block count'

The UID (user ID) of the user who the space is reserved for is:
# dumpe2fs /dev/$your_partition_device_file | grep 'Reserved blocks uid'

The GID (group ID) of the group which the space is reserved for is:
# dumpe2fs /dev/$your_partition_device_file | grep 'Reserved blocks gid'

At anytime you can assign the reserved space to a different UID, or set 0 reserved blocks if you'd like none.

To change reserved space UID:
# tune2fs -u $new_uid /dev/$your_partition_device_file

To change reserved space GID:
# tune2fs -g $new_gid /dev/$your_partition_device_file

To change reserved space block number:
# tune2fs -r $new_reserved_block_number /dev/$your_partition_device_file