linux-stable.git/fs/tracefs, branch v6.10

eventfs: Do not use attributes for events directory

2024-05-23T13:31:50+00:00

The top "events" directory has a static inode (it's created when it is and
removed when the directory is removed). There's no need to use the events
ei->attr to determine its permissions. But it is used for saving the
permissions of the "events" directory for when it is created, as that is
needed for the default permissions for the files and directories
underneath it.

For example:

 # cd /sys/kernel/tracing
 # mkdir instances/foo
 # chown 1001 instances/foo/events

The files under instances/foo/events should still have the same owner as
instances/foo (which the instances/foo/events ei->attr will hold), but the
events directory now has owner 1001.

Link: https://lore.kernel.org/lkml/20240522165032.104981011@goodmis.org

Cc: Linus Torvalds 
Cc: Masami Hiramatsu 
Cc: Mark Rutland 
Cc: Mathieu Desnoyers 
Cc: Andrew Morton 
Signed-off-by: Steven Rostedt (Google)

eventfs: Cleanup permissions in creation of inodes

2024-05-23T13:31:50+00:00

The permissions being set during the creation of the inodes was updating
eventfs_inode attributes as well. Those attributes should only be touched
by the setattr or remount operations, not during the creation of inodes.
The eventfs_inode attributes should only be used to set the inodes and
should not be modified during the inode creation.

Simplify the code and fix the situation by:

 1) Removing the eventfs_find_events() and doing a simple lookup for
    the events descriptor in eventfs_get_inode()

 2) Remove update_events_attr() as the attributes should only be used
    to update the inode and should not be modified here.

 3) Add update_inode_attr() that uses the attributes to determine what
    the inode permissions should be.

 4) As the parent_inode of the eventfs_root_inode structure is no longer
    needed, remove it.

Now on creation, the inode gets the proper permissions without causing
side effects to the ei->attr field.

Link: https://lore.kernel.org/lkml/20240522165031.944088388@goodmis.org

Cc: Linus Torvalds 
Cc: Masami Hiramatsu 
Cc: Mark Rutland 
Cc: Mathieu Desnoyers 
Cc: Andrew Morton 
Signed-off-by: Steven Rostedt (Google)

eventfs: Remove getattr and permission callbacks

2024-05-23T13:27:25+00:00

Now that inodes have their permissions updated on remount, the only other
places to update the inode permissions are when they are created and in
the setattr callback. The getattr and permission callbacks are not needed
as the inodes should already be set at their proper settings.

Remove the callbacks, as it not only simplifies the code, but also allows
more flexibility to fix the inconsistencies with various corner cases
(like changing the permission of an instance directory).

Link: https://lore.kernel.org/lkml/20240522165031.782066021@goodmis.org

Cc: Linus Torvalds 
Cc: Masami Hiramatsu 
Cc: Mark Rutland 
Cc: Mathieu Desnoyers 
Cc: Andrew Morton 
Signed-off-by: Steven Rostedt (Google)

eventfs: Consolidate the eventfs_inode update in eventfs_get_inode()

2024-05-23T13:27:25+00:00

To simplify the code, create a eventfs_get_inode() that is used when an
eventfs file or directory is created. Have the internal tracefs_inode
updated the appropriate flags in this function and update the inode's
mode as well.

Link: https://lore.kernel.org/lkml/20240522165031.624864160@goodmis.org

Cc: Linus Torvalds 
Cc: Masami Hiramatsu 
Cc: Mark Rutland 
Cc: Mathieu Desnoyers 
Cc: Andrew Morton 
Signed-off-by: Steven Rostedt (Google)

tracefs: Clear EVENT_INODE flag in tracefs_drop_inode()

2024-05-23T13:26:24+00:00

When the inode is being dropped from the dentry, the TRACEFS_EVENT_INODE
flag needs to be cleared to prevent a remount from calling
eventfs_remount() on the tracefs_inode private data. There's a race
between the inode is dropped (and the dentry freed) to where the inode is
actually freed. If a remount happens between the two, the eventfs_inode
could be accessed after it is freed (only the dentry keeps a ref count on
it).

Currently the TRACEFS_EVENT_INODE flag is cleared from the dentry iput()
function. But this is incorrect, as it is possible that the inode has
another reference to it. The flag should only be cleared when the inode is
really being dropped and has no more references. That happens in the
drop_inode callback of the inode, as that gets called when the last
reference of the inode is released.

Remove the tracefs_d_iput() function and move its logic to the more
appropriate tracefs_drop_inode() callback function.

Link: https://lore.kernel.org/linux-trace-kernel/20240523051539.908205106@goodmis.org

Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu 
Cc: Mark Rutland 
Cc: Mathieu Desnoyers 
Cc: Andrew Morton 
Cc: Masahiro Yamada 
Fixes: baa23a8d4360d ("tracefs: Reset permissions on remount if permissions are options")
Signed-off-by: Steven Rostedt (Google)

eventfs: Update all the eventfs_inodes from the events descriptor

2024-05-23T13:26:23+00:00

The change to update the permissions of the eventfs_inode had the
misconception that using the tracefs_inode would find all the
eventfs_inodes that have been updated and reset them on remount.
The problem with this approach is that the eventfs_inodes are freed when
they are no longer used (basically the reason the eventfs system exists).
When they are freed, the updated eventfs_inodes are not reset on a remount
because their tracefs_inodes have been freed.

Instead, since the events directory eventfs_inode always has a
tracefs_inode pointing to it (it is not freed when finished), and the
events directory has a link to all its children, have the
eventfs_remount() function only operate on the events eventfs_inode and
have it descend into its children updating their uid and gids.

Link: https://lore.kernel.org/all/CAK7LNARXgaWw3kH9JgrnH4vK6fr8LDkNKf3wq8NhMWJrVwJyVQ@mail.gmail.com/
Link: https://lore.kernel.org/linux-trace-kernel/20240523051539.754424703@goodmis.org

Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu 
Cc: Mark Rutland 
Cc: Mathieu Desnoyers 
Cc: Andrew Morton 
Fixes: baa23a8d4360d ("tracefs: Reset permissions on remount if permissions are options")
Reported-by: Masahiro Yamada 
Signed-off-by: Steven Rostedt (Google)

tracefs: Update inode permissions on remount

2024-05-23T13:26:23+00:00

When a remount happens, if a gid or uid is specified update the inodes to
have the same gid and uid. This will allow the simplification of the
permissions logic for the dynamically created files and directories.

Link: https://lore.kernel.org/linux-trace-kernel/20240523051539.592429986@goodmis.org

Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu 
Cc: Mark Rutland 
Cc: Mathieu Desnoyers 
Cc: Andrew Morton 
Cc: Masahiro Yamada 
Fixes: baa23a8d4360d ("tracefs: Reset permissions on remount if permissions are options")
Signed-off-by: Steven Rostedt (Google)

eventfs: Keep the directories from having the same inode number as files

2024-05-23T13:26:23+00:00

The directories require unique inode numbers but all the eventfs files
have the same inode number. Prevent the directories from having the same
inode numbers as the files as that can confuse some tooling.

Link: https://lore.kernel.org/linux-trace-kernel/20240523051539.428826685@goodmis.org

Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu 
Cc: Mark Rutland 
Cc: Mathieu Desnoyers 
Cc: Andrew Morton 
Cc: Masahiro Yamada 
Fixes: 834bf76add3e6 ("eventfs: Save directory inodes in the eventfs_inode structure")
Signed-off-by: Steven Rostedt (Google)

Merge tag 'trace-v6.10' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace

2024-05-18T01:34:27+00:00

Pull tracing updates from Steven Rostedt:

 - Remove unused ftrace_direct_funcs variables

 - Fix a possible NULL pointer dereference race in eventfs

 - Update do_div() usage in trace event benchmark test

 - Speedup direct function registration with asynchronous RCU callback.

   The synchronization was done in the registration code and this caused
   delays when registering direct callbacks. Move the freeing to a
   call_rcu() that will prevent delaying of the registering.

 - Replace simple_strtoul() usage with kstrtoul()

* tag 'trace-v6.10' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  eventfs: Fix a possible null pointer dereference in eventfs_find_events()
  ftrace: Fix possible use-after-free issue in ftrace_location()
  ftrace: Remove unused global 'ftrace_direct_func_count'
  ftrace: Remove unused list 'ftrace_direct_funcs'
  tracing: Improve benchmark test performance by using do_div()
  ftrace: Use asynchronous grace period for register_ftrace_direct()
  ftrace: Replaces simple_strtoul in ftrace

eventfs: Fix a possible null pointer dereference in eventfs_find_events()

2024-05-14T15:13:45+00:00

In function eventfs_find_events,there is a potential null pointer
that may be caused by calling update_events_attr which will perform
some operations on the members of the ei struct when ei is NULL.

Hence,When ei->is_freed is set,return NULL directly.

Link: https://lore.kernel.org/linux-trace-kernel/20240513053338.63017-1-hao.ge@linux.dev

Cc: stable@vger.kernel.org
Fixes: 8186fff7ab64 ("tracefs/eventfs: Use root and instance inodes as default ownership")
Signed-off-by: Hao Ge 
Signed-off-by: Steven Rostedt (Google)