From mboxrd@z Thu Jan 1 00:00:00 1970 From: Mimi Zohar Subject: [PATCH v1.2 0/5] IMA: making i_readcount a first class inode citizen (reposting) Date: Thu, 18 Nov 2010 18:02:57 -0500 Message-ID: <1290121382-4039-1-git-send-email-zohar@linux.vnet.ibm.com> Cc: Mimi Zohar , linux-security-module@vger.kernel.org, linux-fsdevel@vger.kernel.org, jmorris@namei.org, akpm@linux-foundation.org, torvalds@linux-foundation.org, eparis@redhat.com, viro@zeniv.linux.org.uk, Dave Chinner , "J. Bruce Fields" , David Safford To: linux-kernel@vger.kernel.org Return-path: Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org Reposting unchanged, but with Eric's Ack and suggestion. This patchset separates the incrementing/decrementing of the i_readcount, in the VFS layer, from other IMA functionality, by replacing the current ima_counts_get() call with i_readcount_inc(). Its unclear whether this call to increment i_readcount should be made earlier, like i_writecount. Currently the call is situated immediately after the switch from put_filp() to fput() for cleanup. The patch ordering is a bit redundant in order to leave removing the ifdef around i_readcount until the last patch. The first four patches: redefines i_readcount as atomic, defines i_readcount_inc/dec(), moves the IMA functionality in ima_counts_get() to ima_file_check(), and removes the IMA imbalance code, simplifying IMA. The last patch removes the ifdef around i_readcount, making i_readcount into a "first class inode citizen". The generic_setlease code could then take advantage of i_readcount to help eliminate some of the race conditions, but not all. For now, Eric suggested upstreaming patches 1-4 and leaving the 5th patch until there is a second user. Then we can either unconditionally compile it in, or we can wrap it in something that either user can select in the Kconfig. Both Bruce and I agree with his suggestion. Changelog: - renamed iget/iput_readcount to i_readcount_inc/dec (Dave Chinner's suggestion) - finished cleaning up the locking, so that: (based on Eric's comment.) - i_lock is not required - i_mutex is taken when making measurement decisions to prevent file metadata(eg. permissions, xattr) changing from under it. - iint->mutex is taken when accessing/modifying the iint structure. - Based on the previous posting discussion, i_readcount is now defined as atomic. Mimi Mimi Zohar (5): IMA: convert i_readcount to atomic IMA: define readcount functions IMA: maintain i_readcount in the VFS layer IMA: remove IMA imbalance checking IMA: making i_readcount a first class inode citizen fs/file_table.c | 5 +- fs/open.c | 3 +- include/linux/fs.h | 16 ++++- include/linux/ima.h | 6 -- security/integrity/ima/ima_iint.c | 5 -- security/integrity/ima/ima_main.c | 130 ++++--------------------------------- 6 files changed, 31 insertions(+), 134 deletions(-) -- 1.7.2.2