linux-toradex.git/kernel/audit_tree.c, branch v5.1-rc1

audit: hand taken context to audit_kill_trees for syscall logging

2019-01-14T23:01:05+00:00

Since the context is derived from the task parameter handed to
__audit_free(), hand the context to audit_kill_trees() so it can be used
to associate with a syscall record.  This requires adding the context
parameter to kill_rules() rather than using the current audit_context.

The callers of trim_marked() and evict_chunk() still have their context.

The EOE record was being issued prior to the pruning of the killed_tree
list.

Move the kill_trees call before the audit_log_exit call in
__audit_free() and __audit_syscall_exit() so that any pruned trees
CONFIG_CHANGE records are included with the associated syscall event by
the user library due to the EOE record flagging the end of the event.

See: https://github.com/linux-audit/audit-kernel/issues/50
See: https://github.com/linux-audit/audit-kernel/issues/59

Signed-off-by: Richard Guy Briggs 
[PM: fixed merge fuzz in kernel/audit_tree.c]
Signed-off-by: Paul Moore

audit: minimize our use of audit_log_format()

2018-11-26T23:40:00+00:00

There are some cases where we are making multiple audit_log_format()
calls in a row, for no apparent reason.  Squash these down to a
single audit_log_format() call whenever possible.

Acked-by: Richard Guy Briggs 
Signed-off-by: Paul Moore

audit: Use 'mark' name for fsnotify_mark variables

2018-11-12T14:55:16+00:00

Variables pointing to fsnotify_mark are sometimes called 'entry' and
sometimes 'mark'. Use 'mark' in all places.

Reviewed-by: Richard Guy Briggs 
Signed-off-by: Jan Kara 
[PM: minor merge fuzz due to updated patches previously in the series]
Signed-off-by: Paul Moore

audit: Replace chunk attached to mark instead of replacing mark

2018-11-12T14:55:16+00:00

Audit tree code currently associates new fsnotify mark with each new
chunk. As chunk attached to an inode is replaced when new tag is added /
removed, we also need to remove old fsnotify mark and add a new one on
such occasion.  This is cumbersome and makes locking rules somewhat
difficult to follow.

Fix these problems by allocating fsnotify mark independently of chunk
and keeping it all the time while there is some chunk attached to an
inode. Also add documentation about the locking rules so that things are
easier to follow.

Signed-off-by: Jan Kara 
Reviewed-by: Richard Guy Briggs 
[PM: minor merge fuzz due to updated patches previously in the series]
Signed-off-by: Paul Moore

audit: Simplify locking around untag_chunk()

2018-11-12T14:54:56+00:00

untag_chunk() has to be called with hash_lock, it drops it and
reacquires it when returning. The unlocking of hash_lock is thus hidden
from the callers of untag_chunk() with is rather error prone. Reorganize
the code so that untag_chunk() is called without hash_lock, only with
mark reference preventing the chunk from going away.

Since this requires some more code in the caller of untag_chunk() to
assure forward progress, factor out loop pruning tree from all chunks
into a common helper function.

Signed-off-by: Jan Kara 
Reviewed-by: Richard Guy Briggs 
Signed-off-by: Paul Moore

audit: Drop all unused chunk nodes during deletion

2018-11-12T14:54:49+00:00

When deleting chunk from a tree, drop all unused nodes in a chunk
instead of just the one used by the tree. This gets rid of possibly
lingering unused nodes (created due to fallback path in untag_chunk())
and also removes some special cases and will allow us to simplify
locking in untag_chunk().

Signed-off-by: Jan Kara 
Reviewed-by: Richard Guy Briggs 
Signed-off-by: Paul Moore

audit: Guarantee forward progress of chunk untagging

2018-11-12T14:54:49+00:00

When removing chunk from a tree, we do shrink the chunk. This can fail
for various reasons (due to races, ENOMEM, etc.) and in some cases we
just bail from untag_chunk() relying on someone else to cleanup.
Although this currently works, later we will need to add new failure
situation which would break. Also this simplifies the code and will
allow us to make locking around untag_chunk() less awkward.

Signed-off-by: Jan Kara 
Reviewed-by: Richard Guy Briggs 
Signed-off-by: Paul Moore

audit: Allocate fsnotify mark independently of chunk

2018-11-12T14:54:49+00:00

Allocate fsnotify mark independently instead of embedding it inside
chunk. This will allow us to just replace chunk attached to mark when
growing / shrinking chunk instead of replacing mark attached to inode
which is a more complex operation.

Reviewed-by: Richard Guy Briggs 
Signed-off-by: Jan Kara 
Signed-off-by: Paul Moore

audit: Provide helper for dropping mark's chunk reference

2018-11-12T14:54:49+00:00

Provide a helper function audit_mark_put_chunk() for dropping mark's
reference (which has to happen only after RCU grace period expires).
Currently that happens only from a single place but in later patches we
introduce more callers.

Reviewed-by: Richard Guy Briggs 
Signed-off-by: Jan Kara 
Signed-off-by: Paul Moore

audit: Remove pointless check in insert_hash()

2018-11-12T14:54:48+00:00

The audit_tree_group->mark_mutex is held all the time while we create
the fsnotify mark, add it to the inode, and insert chunk into the hash.
Hence mark cannot get detached during this time and so the check whether
the mark is attached in insert_hash() is pointless.

Reviewed-by: Richard Guy Briggs 
Signed-off-by: Jan Kara 
Signed-off-by: Paul Moore