[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <481149de-1212-4a43-a7cb-52351a0e29ad@linux.dev>
Date: Tue, 6 Aug 2024 23:59:01 +0800
From: Zhu Yanjun <yanjun.zhu@...ux.dev>
To: longli@...rosoft.com, "K. Y. Srinivasan" <kys@...rosoft.com>,
Haiyang Zhang <haiyangz@...rosoft.com>, Wei Liu <wei.liu@...nel.org>,
Dexuan Cui <decui@...rosoft.com>, "David S. Miller" <davem@...emloft.net>,
Eric Dumazet <edumazet@...gle.com>, Jakub Kicinski <kuba@...nel.org>,
Paolo Abeni <pabeni@...hat.com>,
Shradha Gupta <shradhagupta@...ux.microsoft.com>,
Simon Horman <horms@...nel.org>, Konstantin Taranov
<kotaranov@...rosoft.com>,
Souradeep Chakrabarti <schakrabarti@...ux.microsoft.com>,
Erick Archer <erick.archer@...look.com>, linux-hyperv@...r.kernel.org,
netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
linux-rdma@...r.kernel.org
Cc: stable@...r.kernel.org
Subject: Re: [PATCH net] net: mana: Fix doorbell out of order violation and
avoid unnecessary doorbell rings
在 2024/8/6 7:38, longli@...uxonhyperv.com 写道:
> From: Long Li <longli@...rosoft.com>
>
> After napi_complete_done() is called, another NAPI may be running on
> another CPU and ring the doorbell before the current CPU does. When
> combined with unnecessary rings when there is no need to ARM the CQ, this
> triggers error paths in the hardware.
>
> Fix this by always ring the doorbell in sequence and avoid unnecessary
> rings.
Trivial problem^_^
s/ring/ringing ?
Zhu Yanjun
>
> Cc: stable@...r.kernel.org
> Fixes: e1b5683ff62e ("net: mana: Move NAPI from EQ to CQ")
> Signed-off-by: Long Li <longli@...rosoft.com>
> ---
> drivers/net/ethernet/microsoft/mana/mana_en.c | 24 ++++++++++++-------
> include/net/mana/mana.h | 1 +
> 2 files changed, 16 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
> index d2f07e179e86..7d08e23c6749 100644
> --- a/drivers/net/ethernet/microsoft/mana/mana_en.c
> +++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
> @@ -1788,7 +1788,6 @@ static void mana_poll_rx_cq(struct mana_cq *cq)
> static int mana_cq_handler(void *context, struct gdma_queue *gdma_queue)
> {
> struct mana_cq *cq = context;
> - u8 arm_bit;
> int w;
>
> WARN_ON_ONCE(cq->gdma_cq != gdma_queue);
> @@ -1799,16 +1798,23 @@ static int mana_cq_handler(void *context, struct gdma_queue *gdma_queue)
> mana_poll_tx_cq(cq);
>
> w = cq->work_done;
> -
> - if (w < cq->budget &&
> - napi_complete_done(&cq->napi, w)) {
> - arm_bit = SET_ARM_BIT;
> - } else {
> - arm_bit = 0;
> + cq->work_done_since_doorbell += w;
> +
> + if (w < cq->budget) {
> + mana_gd_ring_cq(gdma_queue, SET_ARM_BIT);
> + cq->work_done_since_doorbell = 0;
> + napi_complete_done(&cq->napi, w);
> + } else if (cq->work_done_since_doorbell >
> + cq->gdma_cq->queue_size / COMP_ENTRY_SIZE * 4) {
> + /* MANA hardware requires at least one doorbell ring every 8
> + * wraparounds of CQ even there is no need to ARM. This driver
> + * rings the doorbell as soon as we have execceded 4
> + * wraparounds.
> + */
> + mana_gd_ring_cq(gdma_queue, 0);
> + cq->work_done_since_doorbell = 0;
> }
>
> - mana_gd_ring_cq(gdma_queue, arm_bit);
> -
> return w;
> }
>
> diff --git a/include/net/mana/mana.h b/include/net/mana/mana.h
> index 6439fd8b437b..7caa334f4888 100644
> --- a/include/net/mana/mana.h
> +++ b/include/net/mana/mana.h
> @@ -275,6 +275,7 @@ struct mana_cq {
> /* NAPI data */
> struct napi_struct napi;
> int work_done;
> + int work_done_since_doorbell;
> int budget;
> };
>
Powered by blists - more mailing lists