On 20.02.2017 20:36, Pavel Belous wrote:
> From: Pavel Belous <pavel.bel...@aquantia.com>
>
> This fix removes copying of tx biffers.
Typo: biffers -> buffers
> Now we use ring->buff_fing directly.
>
> Signed-off-by: Pavel Belous <pavel.bel...@aquantia.com>
> ---
> drivers/net/ethernet/aquantia/atlantic/aq_nic.c | 170
> +++++++++++++----------
> drivers/net/ethernet/aquantia/atlantic/aq_ring.c | 19 ---
> drivers/net/ethernet/aquantia/atlantic/aq_ring.h | 3 -
> 3 files changed, 97 insertions(+), 95 deletions(-)
>
> diff --git a/drivers/net/ethernet/aquantia/atlantic/aq_nic.c
> b/drivers/net/ethernet/aquantia/atlantic/aq_nic.c
> index bce312a..ee78444 100644
> --- a/drivers/net/ethernet/aquantia/atlantic/aq_nic.c
> +++ b/drivers/net/ethernet/aquantia/atlantic/aq_nic.c
> @@ -468,95 +468,116 @@ int aq_nic_start(struct aq_nic_s *self)
> return err;
> }
>
> -static unsigned int aq_nic_map_skb_frag(struct aq_nic_s *self,
> - struct sk_buff *skb,
> - struct aq_ring_buff_s *dx)
> +static unsigned int aq_nic_map_skb(struct aq_nic_s *self,
> + struct sk_buff *skb,
> + struct aq_ring_s *ring)
> {
> unsigned int ret = 0U;
> unsigned int nr_frags = skb_shinfo(skb)->nr_frags;
> unsigned int frag_count = 0U;
> + unsigned int dx = ring->sw_tail;
> + struct aq_ring_buff_s *dx_buff = &ring->buff_ring[dx];
>
> - dx->flags = 0U;
> - dx->len = skb_headlen(skb);
> - dx->pa = dma_map_single(aq_nic_get_dev(self), skb->data, dx->len,
> - DMA_TO_DEVICE);
> - dx->len_pkt = skb->len;
> - dx->is_sop = 1U;
> - dx->is_mapped = 1U;
> + if (unlikely(skb_is_gso(skb))) {
> + dx_buff->flags = 0U;
> + dx_buff->len_pkt = skb->len;
> + dx_buff->len_l2 = ETH_HLEN;
> + dx_buff->len_l3 = ip_hdrlen(skb);
> + dx_buff->len_l4 = tcp_hdrlen(skb);
> + dx_buff->mss = skb_shinfo(skb)->gso_size;
> + dx_buff->is_txc = 1U;
> +
> + dx = aq_ring_next_dx(ring, dx);
> + dx_buff = &ring->buff_ring[dx];
> + ++ret;
> + }
> +
> + dx_buff->flags = 0U;
> + dx_buff->len = skb_headlen(skb);
> + dx_buff->pa = dma_map_single(aq_nic_get_dev(self),
> + skb->data,
> + dx_buff->len,
> + DMA_TO_DEVICE);
>
> + if (unlikely(dma_mapping_error(aq_nic_get_dev(self), dx_buff->pa)))
> + goto exit;
> +
> + dx_buff->len_pkt = skb->len;
> + dx_buff->is_sop = 1U;
> + dx_buff->is_mapped = 1U;
> ++ret;
>
> if (skb->ip_summed == CHECKSUM_PARTIAL) {
> - dx->is_ip_cso = (htons(ETH_P_IP) == skb->protocol) ? 1U : 0U;
> - dx->is_tcp_cso =
> + dx_buff->is_ip_cso = (htons(ETH_P_IP) == skb->protocol) ?
> + 1U : 0U;
> + dx_buff->is_tcp_cso =
> (ip_hdr(skb)->protocol == IPPROTO_TCP) ? 1U : 0U;
> - dx->is_udp_cso =
> + dx_buff->is_udp_cso =
> (ip_hdr(skb)->protocol == IPPROTO_UDP) ? 1U : 0U;
> }
>
> for (; nr_frags--; ++frag_count) {
> - unsigned int frag_len;
> + unsigned int frag_len = 0U;
> dma_addr_t frag_pa;
> skb_frag_t *frag = &skb_shinfo(skb)->frags[frag_count];
>
> frag_len = skb_frag_size(frag);
> -
> frag_pa = skb_frag_dma_map(aq_nic_get_dev(self), frag, 0,
> frag_len, DMA_TO_DEVICE);
>
> + if (unlikely(dma_mapping_error(aq_nic_get_dev(self), frag_pa)))
> + goto mapping_error;
> +
> while (frag_len > AQ_CFG_TX_FRAME_MAX) {
> - ++dx;
> - ++ret;
> - dx->flags = 0U;
> - dx->len = AQ_CFG_TX_FRAME_MAX;
> - dx->pa = frag_pa;
> - dx->is_mapped = 1U;
> + dx = aq_ring_next_dx(ring, dx);
> + dx_buff = &ring->buff_ring[dx];
> +
> + dx_buff->flags = 0U;
> + dx_buff->len = AQ_CFG_TX_FRAME_MAX;
> + dx_buff->pa = frag_pa;
> + dx_buff->is_mapped = 1U;
>
> frag_len -= AQ_CFG_TX_FRAME_MAX;
> frag_pa += AQ_CFG_TX_FRAME_MAX;
> + ++ret;
> }
>
> - ++dx;
> - ++ret;
> + dx = aq_ring_next_dx(ring, dx);
> + dx_buff = &ring->buff_ring[dx];
>
> - dx->flags = 0U;
> - dx->len = frag_len;
> - dx->pa = frag_pa;
> - dx->is_mapped = 1U;
> + dx_buff->flags = 0U;
> + dx_buff->len = frag_len;
> + dx_buff->pa = frag_pa;
> + dx_buff->is_mapped = 1U;
> + ++ret;
> }
>
> - dx->is_eop = 1U;
> - dx->skb = skb;
> -
> - return ret;
> -}
> -
> -static unsigned int aq_nic_map_skb_lso(struct aq_nic_s *self,
> - struct sk_buff *skb,
> - struct aq_ring_buff_s *dx)
> -{
> - dx->flags = 0U;
> - dx->len_pkt = skb->len;
> - dx->len_l2 = ETH_HLEN;
> - dx->len_l3 = ip_hdrlen(skb);
> - dx->len_l4 = tcp_hdrlen(skb);
> - dx->mss = skb_shinfo(skb)->gso_size;
> - dx->is_txc = 1U;
> - return 1U;
> -}
> -
> -static unsigned int aq_nic_map_skb(struct aq_nic_s *self, struct sk_buff
> *skb,
> - struct aq_ring_buff_s *dx)
> -{
> - unsigned int ret = 0U;
> -
> - if (unlikely(skb_is_gso(skb))) {
> - ret = aq_nic_map_skb_lso(self, skb, dx);
> - ++dx;
> + dx_buff->is_eop = 1U;
> + dx_buff->skb = skb;
> + goto exit;
> +
> +mapping_error:
> + for (dx = ring->sw_tail;
> + ret > 0;
> + --ret, dx = aq_ring_next_dx(ring, dx)) {
> + dx_buff = &ring->buff_ring[dx];
> +
> + if (!dx_buff->is_txc && dx_buff->pa) {
> + if (unlikely(dx_buff->is_sop)) {
> + dma_unmap_single(aq_nic_get_dev(self),
> + dx_buff->pa,
> + dx_buff->len,
> + DMA_TO_DEVICE);
> + } else {
> + dma_unmap_page(aq_nic_get_dev(self),
> + dx_buff->pa,
> + dx_buff->len,
> + DMA_TO_DEVICE);
> + }
> + }
> }
>
> - ret += aq_nic_map_skb_frag(self, skb, dx);
> -
> +exit:
> return ret;
> }
>
> @@ -571,7 +592,6 @@ __acquires(&ring->lock)
> unsigned int trys = AQ_CFG_LOCK_TRYS;
> int err = NETDEV_TX_OK;
> bool is_nic_in_bad_state;
> - struct aq_ring_buff_s buffers[AQ_CFG_SKB_FRAGS_MAX];
>
> frags = skb_shinfo(skb)->nr_frags + 1;
>
> @@ -595,23 +615,27 @@ __acquires(&ring->lock)
>
> do {
> if (spin_trylock(&ring->header.lock)) {
> - frags = aq_nic_map_skb(self, skb, &buffers[0]);
> -
> - aq_ring_tx_append_buffs(ring, &buffers[0], frags);
> -
> - err = self->aq_hw_ops.hw_ring_tx_xmit(self->aq_hw,
> - ring, frags);
> - if (err >= 0) {
> - if (aq_ring_avail_dx(ring) <
> - AQ_CFG_SKB_FRAGS_MAX + 1)
> - aq_nic_ndev_queue_stop(self, ring->idx);
> + frags = aq_nic_map_skb(self, skb, ring);
> +
> + if (likely(frags)) {
> + err = self->aq_hw_ops.hw_ring_tx_xmit(
> + self->aq_hw,
> + ring, frags);
> + if (err >= 0) {
> + if (aq_ring_avail_dx(ring) <
> + AQ_CFG_SKB_FRAGS_MAX + 1)
> + aq_nic_ndev_queue_stop(
> + self,
> + ring->idx);
> +
> + ++ring->stats.tx.packets;
> + ring->stats.tx.bytes += skb->len;
> + }
> + } else {
> + err = NETDEV_TX_BUSY;
> }
> - spin_unlock(&ring->header.lock);
>
> - if (err >= 0) {
> - ++ring->stats.tx.packets;
> - ring->stats.tx.bytes += skb->len;
> - }
> + spin_unlock(&ring->header.lock);
> break;
> }
> } while (--trys);
> diff --git a/drivers/net/ethernet/aquantia/atlantic/aq_ring.c
> b/drivers/net/ethernet/aquantia/atlantic/aq_ring.c
> index 51f4e7f..0358e607 100644
> --- a/drivers/net/ethernet/aquantia/atlantic/aq_ring.c
> +++ b/drivers/net/ethernet/aquantia/atlantic/aq_ring.c
> @@ -104,25 +104,6 @@ int aq_ring_init(struct aq_ring_s *self)
> return 0;
> }
>
> -void aq_ring_tx_append_buffs(struct aq_ring_s *self,
> - struct aq_ring_buff_s *buffer,
> - unsigned int buffers)
> -{
> - if (likely(self->sw_tail + buffers < self->size)) {
> - memcpy(&self->buff_ring[self->sw_tail], buffer,
> - sizeof(buffer[0]) * buffers);
> - } else {
> - unsigned int first_part = self->size - self->sw_tail;
> - unsigned int second_part = buffers - first_part;
> -
> - memcpy(&self->buff_ring[self->sw_tail], buffer,
> - sizeof(buffer[0]) * first_part);
> -
> - memcpy(&self->buff_ring[0], &buffer[first_part],
> - sizeof(buffer[0]) * second_part);
> - }
> -}
> -
> void aq_ring_tx_clean(struct aq_ring_s *self)
> {
> struct device *dev = aq_nic_get_dev(self->aq_nic);
> diff --git a/drivers/net/ethernet/aquantia/atlantic/aq_ring.h
> b/drivers/net/ethernet/aquantia/atlantic/aq_ring.h
> index fb296b3..2572546 100644
> --- a/drivers/net/ethernet/aquantia/atlantic/aq_ring.h
> +++ b/drivers/net/ethernet/aquantia/atlantic/aq_ring.h
> @@ -146,9 +146,6 @@ struct aq_ring_s *aq_ring_rx_alloc(struct aq_ring_s *self,
> int aq_ring_init(struct aq_ring_s *self);
> void aq_ring_rx_deinit(struct aq_ring_s *self);
> void aq_ring_free(struct aq_ring_s *self);
> -void aq_ring_tx_append_buffs(struct aq_ring_s *ring,
> - struct aq_ring_buff_s *buffer,
> - unsigned int buffers);
> void aq_ring_tx_clean(struct aq_ring_s *self);
> int aq_ring_rx_clean(struct aq_ring_s *self, int *work_done, int budget);
> int aq_ring_rx_fill(struct aq_ring_s *self);
>
Looks good.
Reviewed-by: Lino Sanfilippo <linosanfili...@gmx.de>
Regards,
Lino
