Arch Linux

Works fine for me on x86_64

I’ll try with an x86_64 btrfs-progs.

How and why do you use linux 3.3.0 x86_64 with btrfs-progs i686. Why did you file this bug as i686 when you use the x86_64 kernel?

Note that btrfs is an experimental Linux feature, so this bug is not critical.

With x86_64 btrfs-progs it’s the same.

I use a x86_64 kernel with an entire i686 userland because I want to be able to chroot in either architectures but without the additional memory overhead of an x86_64 userland, and I didn’t know under which architecture to file this bug. (I guess I didn’t guess right ^^)
To install that I use others config file, cache dir and pacman database especially for linux.

I forgot to note that the kernel error “btrfs: fail to dirty inode <number> error -28” is shown in the TTY.

Now I’ll try to install a i686 linux, to see if it comes from the architecture mix (that worked about since the 3.0 kernel).

It’s the same with a pure i686 ArchLinux, without any foreign package.
I could also try with a pure x86_64 one, but it would take me a lot of time to download an entire userland.

I think there is an other error message at boot, but it is cleared too quickly by login or agetty. Is there a way I could disable that clear?

Yes, I suspected there wouldn't be a difference, this seems like a kernel problem - userland isn't the problem.

You can get back all kernel messages from the output of 'dmesg'.

Please try again with the old kernel, just to make sure it's not your file system being broken, but really a regression/bug/change in the kernel.

It works well in 3.2.9.

And the message I wasn’t able to see, at the udev launching time, is:
[ 10.966968] tpm_tis 00:0a: A TPM error (6) occurred attempting to read a pcr value
[ 10.968318] tpm_tis 00:0a: TPM is disabled/deactivated (0x6)

It’s not present when booting on the 3.2.9.

That's unrelated, I have the first message as well.

Also no problem here on x86_64

I’ve been able to reproduce the bug on a complete x86_64 system.

I’ve also done a btrfsck on the partition from an other Arch install, everything seems clean, but it didn’t change anything.

Apparently, [1] is the problem and will be reverted in the next stable release (source, transcript of some IRC conversation from #btrfs [2]).

[1] https://git.kernel.org/?p=linux/kernel/git/torvalds/linux.git;a=commitdiff;h=5500cdbe14d7435e04f66ff3cfb8ecd8b8e44ebf
[2] http://pastebin.com/iN0wmF1N

Is this now fixed?