Re: [RFC PATCH] audit: move audit_signal_info() into kernel/audit.c
On Wed, Mar 29, 2017 at 5:29 PM, Paul Moore <paul(a)paul-moore.com> wrote:
On Wed, Mar 29, 2017 at 5:11 PM, Paul Moore <pmoore(a)redhat.com>
wrote:
> From: Paul Moore <paul(a)paul-moore.com>
>
> Commit 5b52330bbfe6 ("audit: fix auditd/kernel connection state
> tracking") make inlining audit_signal_info() a bit pointless as
> it was always calling into auditd_test_task() so let's move
> audit_signal_info() into audit.c and get rid of the explicit inline.
>
> Signed-off-by: Paul Moore <paul(a)paul-moore.com>
> ---
> kernel/audit.c | 64 ++++++++++++++++++++++++++++++++++++++++++
> kernel/audit.h | 31 +++++++++++++-------
> kernel/auditsc.c | 82 ------------------------------------------------------
> 3 files changed, 84 insertions(+), 93 deletions(-)
If I don't see any objections I'll be sending this to Linus later this
week (at his request).
Self NACK on this patch. In looking over it one last time before
sending it to Linus I decided that moving it audit.c was causing a lot
of code shuffling, so I'm going to rework the patch to just drop the
inline and rename the function in auditsc.c instead. Granted,
auditd_test_task() will not be inlined in this case, but I'm not going
to worry too much about that. Moving the multicast send out of
audit_log_end() should more than make up for any overhead we see from
one more non-inlined function call.
--
paul moore
www.paul-moore.com