Re: [PATCH 1/3] audit: cache ctx->major in audit_filter_syscall()

[Ankur Arora]

Paul Moore <paul@xxxxxxxxxxxxxx> writes:

> On Tue, Sep 27, 2022 at 6:59 PM Ankur Arora <ankur.a.arora@xxxxxxxxxx> wrote:
>>
>> ctx->major contains the current syscall number. This is, of course, a
>> constant for the duration of the syscall. Unfortunately, GCC's alias
>> analysis cannot prove that it is not modified via a pointer in the
>> audit_filter_syscall() loop, and so always loads it from memory.
>>
>> ...
>>
>> Signed-off-by: Ankur Arora <ankur.a.arora@xxxxxxxxxx>
>> ---
>>  kernel/auditsc.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> This looks pretty trivial to me, but it's too late in the current -rc
> cycle for this to be merged, I'll queue it up for after the upcoming
> merge window closes.  Thanks.

Thanks.

Ankur

>
>> diff --git a/kernel/auditsc.c b/kernel/auditsc.c
>> index 79a5da1bc5bb..533b087c3c02 100644
>> --- a/kernel/auditsc.c
>> +++ b/kernel/auditsc.c
>> @@ -843,13 +843,14 @@ static void audit_filter_syscall(struct task_struct *tsk,
>>  {
>>         struct audit_entry *e;
>>         enum audit_state state;
>> +       unsigned long major = ctx->major;
>>
>>         if (auditd_test_task(tsk))
>>                 return;
>>
>>         rcu_read_lock();
>>         list_for_each_entry_rcu(e, &audit_filter_list[AUDIT_FILTER_EXIT], list) {
>> -               if (audit_in_mask(&e->rule, ctx->major) &&
>> +               if (audit_in_mask(&e->rule, major) &&
>>                     audit_filter_rules(tsk, &e->rule, ctx, NULL,
>>                                        &state, false)) {
>>                         rcu_read_unlock();
>> --
>> 2.31.1


--
ankur