From: Long Li longli@microsoft.com
It's inefficient to ring the doorbell page every time a WQE is posted to the received queue. Excessive MMIO writes result in CPU spending more time waiting on LOCK instructions (atomic operations), resulting in poor scaling performance.
Move the code for ringing doorbell page to where after we have posted all WQEs to the receive queue during a callback from napi_poll().
With this change, tests showed an improvement from 120G/s to 160G/s on a 200G physical link, with 16 or 32 hardware queues.
Tests showed no regression in network latency benchmarks on single connection.
While we are making changes in this code path, change the code for ringing doorbell to set the WQE_COUNT to 0 for Receive Queue. The hardware specification specifies that it should set to 0. Although currently the hardware doesn't enforce the check, in the future releases it may do.
Cc: stable@vger.kernel.org Fixes: ca9c54d2d6a5 ("net: mana: Add a driver for Microsoft Azure Network Adapter (MANA)")
Reviewed-by: Haiyang Zhang haiyangz@microsoft.com Reviewed-by: Dexuan Cui decui@microsoft.com Signed-off-by: Long Li longli@microsoft.com --- Change log: v2: Check for comp_read > 0 as it might be negative on completion error. Set rq.wqe_cnt to 0 according to BNIC spec.
v3: Add details in the commit on the reason of performance increase and test numbers. Add details in the commit on why rq.wqe_cnt should be set to 0 according to hardware spec. Add "Reviewed-by" from Haiyang and Dexuan.
drivers/net/ethernet/microsoft/mana/gdma_main.c | 5 ++++- drivers/net/ethernet/microsoft/mana/mana_en.c | 10 ++++++++-- 2 files changed, 12 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ethernet/microsoft/mana/gdma_main.c b/drivers/net/ethernet/microsoft/mana/gdma_main.c index 8f3f78b68592..3765d3389a9a 100644 --- a/drivers/net/ethernet/microsoft/mana/gdma_main.c +++ b/drivers/net/ethernet/microsoft/mana/gdma_main.c @@ -300,8 +300,11 @@ static void mana_gd_ring_doorbell(struct gdma_context *gc, u32 db_index,
void mana_gd_wq_ring_doorbell(struct gdma_context *gc, struct gdma_queue *queue) { + /* Hardware Spec specifies that software client should set 0 for + * wqe_cnt for Receive Queues. This value is not used in Send Queues. + */ mana_gd_ring_doorbell(gc, queue->gdma_dev->doorbell, queue->type, - queue->id, queue->head * GDMA_WQE_BU_SIZE, 1); + queue->id, queue->head * GDMA_WQE_BU_SIZE, 0); }
void mana_gd_ring_cq(struct gdma_queue *cq, u8 arm_bit) diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c index cd4d5ceb9f2d..1d8abe63fcb8 100644 --- a/drivers/net/ethernet/microsoft/mana/mana_en.c +++ b/drivers/net/ethernet/microsoft/mana/mana_en.c @@ -1383,8 +1383,8 @@ static void mana_post_pkt_rxq(struct mana_rxq *rxq)
recv_buf_oob = &rxq->rx_oobs[curr_index];
- err = mana_gd_post_and_ring(rxq->gdma_rq, &recv_buf_oob->wqe_req, - &recv_buf_oob->wqe_inf); + err = mana_gd_post_work_request(rxq->gdma_rq, &recv_buf_oob->wqe_req, + &recv_buf_oob->wqe_inf); if (WARN_ON_ONCE(err)) return;
@@ -1654,6 +1654,12 @@ static void mana_poll_rx_cq(struct mana_cq *cq) mana_process_rx_cqe(rxq, cq, &comp[i]); }
+ if (comp_read > 0) { + struct gdma_context *gc = rxq->gdma_rq->gdma_dev->gdma_context; + + mana_gd_wq_ring_doorbell(gc, rxq->gdma_rq); + } + if (rxq->xdp_flush) xdp_do_flush(); }