Re: [PATCH] seccomp: Add SECCOMP_CLONE_FILTER operation

[Tom Hromatka <tom.hromatka@oracle.com>]

On 9/3/25 4:44 PM, Alexei Starovoitov wrote:
> On Wed, Sep 3, 2025 at 1:52 PM Tom Hromatka <tom.hromatka@oracle.com> wrote:
>>
>> On 9/3/25 2:45 PM, Alexei Starovoitov wrote:
>>> On Wed, Sep 3, 2025 at 1:38 PM Tom Hromatka <tom.hromatka@oracle.com> wrote:
>>>>
>>>> +
>>>> +       spin_lock_irq(&current->sighand->siglock);
>>>> +       spin_lock_irq(&task->sighand->siglock);
>>>> +
>>>> +       if (atomic_read(&task->seccomp.filter_count) == 0) {
>>>> +               spin_unlock_irq(&task->sighand->siglock);
>>>> +               spin_unlock_irq(&current->sighand->siglock);
>>>
>>> did you copy this pattern from somewhere ?
>>> It's obviously buggy.
>>
>> I tried to mimic the logic in copy_seccomp() in kernel/fork.c,
>> but as you point out, I probably messed it up :).
>>
>> Do you have recommendations for a better design pattern?
> 
> Several things look wrong here.
> Double _irq() is one obvious bug.

This snippet addresses the double irq issue.  I also added a
check to make sure that task != current.  (A user shouldn't
do that but who knows what they'll actually do.)

         if (task == current) {
                 put_task_struct(task);
                 return -EINVAL;
         }

         spin_lock_irq(&current->sighand->siglock);
         spin_lock(&task->sighand->siglock);

         if (atomic_read(&task->seccomp.filter_count) == 0) {
                 spin_unlock(&task->sighand->siglock);
                 spin_unlock_irq(&current->sighand->siglock);
                 put_task_struct(task);
                 return -EINVAL;
         }

         get_seccomp_filter(task);
         current->seccomp = task->seccomp;

         spin_unlock(&task->sighand->siglock);

         set_task_syscall_work(current, SECCOMP);

         spin_unlock_irq(&current->sighand->siglock);

Let me know if there are other fixes I need to add.

Thanks so much!

Tom