On 12/27/2023 8:21 PM, Mimi Zohar wrote:
On Wed, 2023-12-27 at 17:39 +0100, Roberto Sassu wrote:
On 12/27/2023 2:22 PM, Mimi Zohar wrote:
On Thu, 2023-12-14 at 18:08 +0100, Roberto Sassu wrote:
From: Roberto Sassu <roberto.sassu@xxxxxxxxxx>
Make the 'ima' LSM independent from the 'integrity' LSM by introducing IMA
own integrity metadata (ima_iint_cache structure, with IMA-specific fields
from the integrity_iint_cache structure), and by managing it directly from
the 'ima' LSM.
Move the remaining IMA-specific flags to security/integrity/ima/ima.h,
since they are now unnecessary in the common integrity layer.
Replace integrity_iint_cache with ima_iint_cache in various places
of the IMA code.
Then, reserve space in the security blob for the entire ima_iint_cache
structure, so that it is available for all inodes having the security blob
allocated (those for which security_inode_alloc() was called). Adjust the
IMA code accordingly, call ima_iint_inode() to retrieve the ima_iint_cache
structure. Keep the non-NULL checks since there can be inodes without
security blob.
Previously the 'iint' memory was only allocated for regular files in
policy and were tagged S_IMA. This patch totally changes when and how
memory is being allocated. Does it make sense to allocate memory at
security_inode_alloc()? Is this change really necessary for making IMA
a full fledged LSM?
Good question. I think it wouldn't be necessary, we can reuse the same
approach as in the patch 'integrity: Switch from rbtree to LSM-managed
blob for integrity_iint_cache'.
Going forward with the v8 proposed solution would require some real
memory usage analysis for different types of policies.
To me the "integrity: Switch from rbtree to LSM-managed blob for
integrity_iint_cache" makes a lot more sense. Looking back at the
original thread, your reasons back then for not directly allocating the
integrity_iint_cache are still valid for the ima_iint_cache structure.
Uhm, ok. It should not be too difficult to restore the old mechanism for
ima_iint_cache. Will do it in v9.
Thanks
Roberto
Mimi
Don't include the inode pointer as field in the ima_iint_cache structure,
since the association with the inode is clear. Since the inode field is
missing in ima_iint_cache, pass the extra inode parameter to
ima_get_verity_digest().
Finally, register ima_inode_alloc_security/ima_inode_free_security() to
initialize/deinitialize the new ima_iint_cache structure (before this task
was done by iint_init_always() and iint_free()). Also, duplicate
iint_lockdep_annotate() for the ima_iint_cache structure, and name it
ima_iint_lockdep_annotate().
Signed-off-by: Roberto Sassu <roberto.sassu@xxxxxxxxxx>
