Skip to contentSkip to conversation

btrfs userland interface isn't 32/64bit clean (breaks lsattr and btrfs send)

DISCUSSION Marc MERLIN 3 participants 5 messages Feb 24, 2014 – Feb 25, 2014 Btrfs file system

lore
Share & tools
lore
#0 lore raw

I was trying to make sense out of this:

gargamel:~# lsattr
lsattr: Inappropriate ioctl for device While reading flags on ./satapmtool
lsattr: Inappropriate ioctl for device While reading flags on ./usbreset

As well as the btrfs send issue I reported:

gargamel:/mnt/btrfs_pool1# btrfs send media_ro.20140222_11:12:53 | less
At subvol media_ro.20140222_11:12:53
ERROR: send ioctl failed with -25: Inappropriate ioctl for device

I was trying to figure out why it was failing on one system while it was working on my other ones with the same kernel.

In the end I pinned it down to this:
3.13.5's kernel/userland interface fails if my kernel is 64bit and my userland 32bit.

lsattr and btrfs send, and maybe more things fail, but they started working the moment I updated my kernel to 32bit

Is this difficult to fix?
Do I need to file bugs with some folks, or this mail is enough?

Thanks,
Marc

Signature · 4 lines
"A mouse is a device used to point at the xterm you want to type in" - A.S.R.
Microsoft is to operating systems ....
                                      .... what McDonalds is to gourmet cooking
Home page: http://marc.merlins.org/  
#1 lore raw

Marc MERLIN posted on Sun, 23 Feb 2014 21:51:03 -0800 as excerpted:

Quoted 3 lines

In the end I pinned it down to this:
3.13.5's kernel/userland interface fails if my kernel is 64bit and my userland 32bit.

This is a known issue. There's patches in the pipeline for 32-bit userspace on a 64-bit kernel, already.

Signature · 3 lines
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman
#2 lore raw
Quoted 8 lines · Duncan

Marc MERLIN posted on Sun, 23 Feb 2014 21:51:03 -0800 as excerpted:

Quoted 3 lines

In the end I pinned it down to this:
3.13.5's kernel/userland interface fails if my kernel is 64bit and my userland 32bit.

This is a known issue. There's patches in the pipeline for 32-bit userspace on a 64-bit kernel, already.

If you mean my recent patch, that's only for receive. I've not
noticed any other issues on my 32/64 system, but I've not done much with it either (with btrfs, it's only really used for btrfs receive -- everything else is ext4).

Hugo.

Signature · 4 lines
=== Hugo Mills: hugo@... carfax.org.uk | darksatanic.net | lug.org.uk ===
  PGP key: 65E74AC0 from wwwkeys.eu.pgp.net or http://www.carfax.org.uk
         --- Welcome to Hollywood, a land just off the coast ---         
                            of Planet Earth.                             
#3 lore raw

Hugo Mills posted on Mon, 24 Feb 2014 08:29:38 +0000 as excerpted:

Quoted 9 lines · Duncan
Quoted 3 lines

This is a known issue. There's patches in the pipeline for 32-bit userspace on a 64-bit kernel, already.

If you mean my recent patch, that's only for receive. I've not
noticed any other issues on my 32/64 system, but I've not done much with it either (with btrfs, it's only really used for btrfs receive -- everything else is ext4).

Thanks. Yes, I believe that's what I had in mind. So apparently there's more needed. (I /think/ I might have seen at least one additional patch float by for 32/64, but as it's out of my usage domain I'm not paying /that/ much attention to it, so perhaps not.)

Signature · 3 lines
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman
#4 lore raw
Quoted 12 lines · Duncan

Hugo Mills posted on Mon, 24 Feb 2014 08:29:38 +0000 as excerpted:

Quoted 9 lines · Duncan
Quoted 3 lines

This is a known issue. There's patches in the pipeline for 32-bit userspace on a 64-bit kernel, already.

If you mean my recent patch, that's only for receive. I've not
noticed any other issues on my 32/64 system, but I've not done much with it either (with btrfs, it's only really used for btrfs receive -- everything else is ext4).

Quoted 4 lines

Thanks. Yes, I believe that's what I had in mind. So apparently there's more needed. (I /think/ I might have seen at least one additional patch float by for 32/64, but as it's out of my usage domain I'm not paying /that/ much attention to it, so perhaps not.)

In the meantime, if that helps someone, on debian it's trivial to get 64bit for some packages. All I had to do was this:

gargamel:/usr/src# dpkg --add-architecture amd64
gargamel:/usr/src# dpkg --print-foreign-architectures
amd64
gargamel:/usr/src# apt-get install e2fsprogs:amd64 btrfs-tools:amd64

And that's it.
Now I'm back in business with a 64bit kernel.

Marc

Signature · 4 lines
"A mouse is a device used to point at the xterm you want to type in" - A.S.R.
Microsoft is to operating systems ....
                                      .... what McDonalds is to gourmet cooking
Home page: http://marc.merlins.org/                         | PGP 1024R/763BE901