Re: [PATCH net-next 0/9] mlx4: order-0 allocations and page recycling

On Wed, 2017-02-08 at 11:02 +0200, Tariq Toukan wrote:

On 07/02/2017 5:50 PM, Tariq Toukan wrote:

quoted

Hi Eric,

Thanks for your series.

On 07/02/2017 5:02 AM, Eric Dumazet wrote:

quoted

As mentioned half a year ago, we better switch mlx4 driver to order-0
allocations and page recycling.

This reduces vulnerability surface thanks to better skb->truesize 
tracking
and provides better performance in most cases.

The series makes significant change in the RX data-path, that requires 
deeper checks, in addition to code review.
We applied your series and started running both our functional and 
performance regression.
We will have results by tomorrow morning, and will analyze them during 
the day. I'll update about that.

We hit a kernel panic when running traffic after configuring a large MTU 
(9000).
I will take deeper look into this soon and will keep you updated.

Hmm... I saw a typo for XDP, but not for the non XDP path...

diff --git a/drivers/net/ethernet/mellanox/mlx4/en_rx.c b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
index 8f16ec8dfadd0f95646c498c14d53f7266a0..e572e175edfe0f7392b9833b5b3f867fd6db 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_rx.c

@@ -132,7 +132,7 @@ static int mlx4_en_prepare_rx_desc(const struct mlx4_en_priv *priv,
                                        (index << priv->log_rx_info);
 
        if (ring->page_cache.index > 0) {
-               if (frags[0].page) {
+               if (!frags[0].page) {
                        ring->page_cache.index--;
                        frags[0].page = ring->page_cache.buf[ring->page_cache.index].page;
                        frags[0].dma  = ring->page_cache.buf[ring->page_cache.index].dma;