[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <c3fc2a72a567e26613824001324bcac6fc8c3640.camel@gmail.com>
Date: Thu, 09 Feb 2023 09:07:42 -0800
From: Alexander H Duyck <alexander.duyck@...il.com>
To: Larysa Zaremba <larysa.zaremba@...el.com>,
"David S. Miller" <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>,
Eric Dumazet <edumazet@...gle.com>,
Paolo Abeni <pabeni@...hat.com>
Cc: intel-wired-lan@...osl.org, netdev@...r.kernel.org,
bpf@...r.kernel.org,
Alexander Lobakin <alexandr.lobakin@...el.com>,
Tony Nguyen <anthony.l.nguyen@...el.com>,
Maciej Fijalkowski <maciej.fijalkowski@...el.com>
Subject: Re: [PATCH net] ice: xsk: Fix cleaning of XDP_TX frames
On Thu, 2023-02-09 at 17:01 +0100, Larysa Zaremba wrote:
> Incrementation of xsk_frames inside the for-loop produces
> infinite loop, if we have both normal AF_XDP-TX and XDP_TXed
> buffers to complete.
>
> Split xsk_frames into 2 variables (xsk_frames and completed_frames)
> to eliminate this bug.
>
> Fixes: 29322791bc8b ("ice: xsk: change batched Tx descriptor cleaning")
> Acked-by: Maciej Fijalkowski <maciej.fijalkowski@...el.com>
> Signed-off-by: Larysa Zaremba <larysa.zaremba@...el.com>
> ---
> To Tony: this is urgent and should go directly via net. It's tested and acked.
> ---
> drivers/net/ethernet/intel/ice/ice_xsk.c | 15 +++++++++------
> 1 file changed, 9 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/net/ethernet/intel/ice/ice_xsk.c b/drivers/net/ethernet/intel/ice/ice_xsk.c
> index 7105de6fb344..374b7f10b549 100644
> --- a/drivers/net/ethernet/intel/ice/ice_xsk.c
> +++ b/drivers/net/ethernet/intel/ice/ice_xsk.c
> @@ -800,6 +800,7 @@ static void ice_clean_xdp_irq_zc(struct ice_tx_ring *xdp_ring)
> struct ice_tx_desc *tx_desc;
> u16 cnt = xdp_ring->count;
> struct ice_tx_buf *tx_buf;
> + u16 completed_frames = 0;
> u16 xsk_frames = 0;
> u16 last_rs;
> int i;
> @@ -809,19 +810,21 @@ static void ice_clean_xdp_irq_zc(struct ice_tx_ring *xdp_ring)
> if ((tx_desc->cmd_type_offset_bsz &
> cpu_to_le64(ICE_TX_DESC_DTYPE_DESC_DONE))) {
> if (last_rs >= ntc)
> - xsk_frames = last_rs - ntc + 1;
> + completed_frames = last_rs - ntc + 1;
> else
> - xsk_frames = last_rs + cnt - ntc + 1;
> + completed_frames = last_rs + cnt - ntc + 1;
> }
>
> - if (!xsk_frames)
> + if (!completed_frames)
> return;
>
> - if (likely(!xdp_ring->xdp_tx_active))
> + if (likely(!xdp_ring->xdp_tx_active)) {
> + xsk_frames = completed_frames;
> goto skip;
> + }
>
> ntc = xdp_ring->next_to_clean;
> - for (i = 0; i < xsk_frames; i++) {
> + for (i = 0; i < completed_frames; i++) {
> tx_buf = &xdp_ring->tx_buf[ntc];
>
> if (tx_buf->raw_buf) {
> @@ -837,7 +840,7 @@ static void ice_clean_xdp_irq_zc(struct ice_tx_ring *xdp_ring)
> }
> skip:
> tx_desc->cmd_type_offset_bsz = 0;
> - xdp_ring->next_to_clean += xsk_frames;
> + xdp_ring->next_to_clean += completed_frames;
> if (xdp_ring->next_to_clean >= cnt)
> xdp_ring->next_to_clean -= cnt;
> if (xsk_frames)
Looks good to me.
Reviewed-by: Alexander Duyck <alexanderduyck@...com>
Powered by blists - more mailing lists