[PATCH v1 net] page_pool: Cap queue size to 32k.

Dan Carpenter dan.carpenter at linaro.org
Tue Sep 5 02:57:34 PDT 2023


On Mon, Aug 14, 2023 at 11:34:11AM +0530, Ratheesh Kannoth wrote:
> Clamp to 32k instead of returning error.

What is the motivation here?  What is the real world impact for the
users?

> 
> Please find discussion at
> https://lore.kernel.org/lkml/
> CY4PR1801MB1911E15D518A77535F6E51E2D308A at CY4PR1801MB1911.
> namprd18.prod.outlook.com/T/

Please don't break the URL up like this.  I think normally we would just
write up a normal commit message and use the Link: tag.

Fixes: ff7d6b27f894 ("page_pool: refurbish version of page_pool code")
Link: https://lore.kernel.org/lkml/CY4PR1801MB1911E15D518A77535F6E51E2D308A@CY4PR1801MB1911.namprd18.prod.outlook.com/
Signed-off-by:

> @@ -171,9 +171,10 @@ static int page_pool_init(struct page_pool *pool,
>  	if (pool->p.pool_size)
>  		ring_qsize = pool->p.pool_size;
>  
> -	/* Sanity limit mem that can be pinned down */
> +	/* Cap queue size to 32k */
>  	if (ring_qsize > 32768)
> -		return -E2BIG;
> +		ring_qsize = 32768;
> +
>  
>  	/* DMA direction is either DMA_FROM_DEVICE or DMA_BIDIRECTIONAL.

Don't introduce a blank line here.  Checkpatch will complain if you
have to blank lines in a row.  It won't complain about the patch but it
will complain if you apply the patch and then re-run checkpatch -f on
the file.  (I didn't test this but it's wrong either way. :P).

regards,
dan carpenter




More information about the Linux-mediatek mailing list