Bug 28767

Summary:	fuse doesn't process inode numbers properly on 32 bit platforms
Product:	Sisyphus	Reporter:	Pavel Shilovsky <piastryyy>
Component:	kernel-image-ovz-el	Assignee:	Gleb F-Malinovskiy <glebfm>
Status:	CLOSED FIXED	QA Contact:	qa-sisyphus
Severity:	normal
Priority:	P3	CC:	boyarsh, glebfm, kernelbot, ldv, mike, mithraen, rider, sbolshakov, shrek, sin, vitty, vsu, vt, zerg
Version:	unstable
Hardware:	all
OS:	Linux

Description Pavel Shilovsky 2013-03-30 20:30:24 MSK

This is a copy of the bug from Red Hat Bugzilla (https://bugzilla.redhat.com/show_bug.cgi?id=872629). There is no activity there for a while - that's why I am posting it here.

RHEL OpenVZ kernel doesn't process fuse inode numbers properly:

Now it stores attr->ino at inode->i_ino, returns attr->ino at the first time and then returns inode->i_ino if the attribute timeout isn't expired. That's wrong on 32 bit platforms because attr->ino is 64 bit and inode->i_ino is 32 bit in this case.

It can cause problems e.g. with NFS server running on top of GluserFS that gets different inode numbers time by time:
"NFS: server server error: fileid changed"

I have a patch "fuse: fix stat call on 32 bit platforms" that fixes this problem by saving 64 bit ino in fuse_inode structure and returning it everything when getattr comes. Also squash attr->ino into inode->i_ino explicitly.

The patch was merged into the mainline and some stable releases (v3.0.35, v3.2.21, v3.4.3 and v2.6.32.60 too) but I don't see it in the latest RHEL OpenVZ kernel patch for 2.6.32 kernel.

The commit id in mainline: 45c72cd73c788dd18c8113d4a404d6b4a01decf1

The link to the commit merged into 2.6.32.60 kernel:
http://git.kernel.org/?p=linux/kernel/git/stable/linux-stable.git;a=commit;h=db8a784af98d392a51fa25bb651bf90a982044e3

Comment 1 Repository Robot 2013-04-01 01:25:39 MSK

kernel-image-ovz-el-2.6.32-alt85 -> sisyphus:

* Sun Mar 31 2013 Led <led@altlinux> 2.6.32-alt85
- fuse: fix stat call on 32 bit platforms (ALT#28767)