Applying as is.

On Tue, 17 Nov 2026 07:50:39 +0200, Eva Kurchatova wrote:
> If we have LOCKDOWN_TRACEFS, the function bails out - *after*
> having locked the parent directory and without bothering to
> undo that.  Just check it before tracefs_start_creating()...

The backport is the 6.12.y commit as is, and the leak is in our tree.

Separate patches, not needed for this one:

1. The other fix of VSTOR-155413, 14f7f52629d91 ("tracefs: Fix
   default permissions not being applied on initial mount"), is still
   missing: tracefs_fill_super() does not apply the mount options, so
   the tracefs root comes up 0755 instead of 0700. Please send it too.
   https://virtuozzo.atlassian.net/browse/VSTOR-155413

2. Same in mainline: on lockdown eventfs_create_events_dir() returns
   NULL, create_event_toplevel_files() checks only IS_ERR(), and
   removing that instance calls eventfs_remove_events_dir(NULL), which
   dereferences it in get_root_inode(). Before this patch the case hung
   on the leaked lock, now rmdir oopses. It needs root and a lockdown
   change racing with the instance creation. To be fixed upstream
   first, I created a task for it:
   https://virtuozzo.atlassian.net/browse/VSTOR-155526

Thanks.

-- 
Konstantin Khorenko,
Virtuozzo Linux Kernel Team

_______________________________________________
Devel mailing list
[email protected]
https://lists.openvz.org/mailman/listinfo/devel

Reply via email to