On 12.03.2018 21:00, Oleg Babin wrote:
> The ixgbe driver can produce a lot of page allocation warnings to dmesg
> during IRQ servicing. These warnings can slow down a whole system while
> not being very usefull as the user can obtain the same information from
> the driver's stats counter 'alloc_rx_buff_failed' using ethtool.
> 
> Suppress this warnings using __GFP_NOWARN flag.
> 
> https://jira.sw.ru/browse/PSBM-82315
> Signed-off-by: Oleg Babin <oba...@virtuozzo.com>
> ---
>  drivers/net/ethernet/intel/ixgbe/ixgbe_main.c | 5 +++--
>  1 file changed, 3 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c 
> b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> index 0afaf44..0b235d1 100644
> --- a/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> +++ b/drivers/net/ethernet/intel/ixgbe/ixgbe_main.c
> @@ -2031,8 +2031,9 @@ static struct sk_buff *ixgbe_fetch_rx_buffer(struct 
> ixgbe_ring *rx_ring,
>  #endif
>  
>               /* allocate a skb to store the frags */
> -             skb = napi_alloc_skb(&rx_ring->q_vector->napi,
> -                                  IXGBE_RX_HDR_SIZE);
> +             skb = __napi_alloc_skb(&rx_ring->q_vector->napi,
> +                                    IXGBE_RX_HDR_SIZE,
> +                                    GFP_ATOMIC | __GFP_NOWARN);

Oleg, could we add this flag to napi_alloc_skb() instead, to make all network 
drivers not warn
in such situations?

>               if (unlikely(!skb)) {
>                       rx_ring->rx_stats.alloc_rx_buff_failed++;
>                       return NULL;
> 

Kirill
_______________________________________________
Devel mailing list
Devel@openvz.org
https://lists.openvz.org/mailman/listinfo/devel

Reply via email to