Re: [PATCH] ipc/sem: do not sleep with a spin lock held

[Manfred Spraul]

Hi Minghao,

On 12/22/21 09:10, cgel.zte@xxxxxxxxx wrote:
> From: Minghao Chi <chi.minghao@xxxxxxxxxx>
>
> We can't call kvfree() with a spin lock held, so defer it.
>
> Reported-by: Zeal Robot <zealci@xxxxxxxxxx>
> Signed-off-by: Minghao Chi <chi.minghao@xxxxxxxxxx>

Could you add

Fixes: fc37a3b8b438 ("[PATCH] ipc sem: use kvmalloc for sem_undo 
allocation")

Cc: stable@xxxxxxxxxxxxxxx

I will review/test the change in the next few days.

Especially, I would like to check if there are further instances where 
the same mistake was made.

> /**
> * kvfree() - Free memory.
> * @addr: Pointer to allocated memory.
> *
> * kvfree frees memory allocated by any of vmalloc(), kmalloc() or 
> kvmalloc().
> * It is slightly more efficient to use kfree() or vfree() if you are 
> certain
> * that you know which one to use.
> *
> * Context: Either preemptible task context or not-NMI interrupt.
> */
As an independent change: Should we add a


      might_sleep_if(!in_interrupt());

into kvfree(), to trigger bugs more easily?

> ---
>   ipc/sem.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/ipc/sem.c b/ipc/sem.c
> index 6693daf4fe11..0dbdb98fdf2d 100644
> --- a/ipc/sem.c
> +++ b/ipc/sem.c
> @@ -1964,6 +1964,7 @@ static struct sem_undo *find_alloc_undo(struct ipc_namespace *ns, int semid)
>   	 */
>   	un = lookup_undo(ulp, semid);
>   	if (un) {
> +		spin_unlock(&ulp->lock);
>   		kvfree(new);
>   		goto success;
>   	}
> @@ -1976,9 +1977,8 @@ static struct sem_undo *find_alloc_undo(struct ipc_namespace *ns, int semid)
>   	ipc_assert_locked_object(&sma->sem_perm);
>   	list_add(&new->list_id, &sma->list_id);
>   	un = new;
> -
> -success:
>   	spin_unlock(&ulp->lock);
> +success:
>   	sem_unlock(sma, -1);
>   out:
>   	return un;