Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <69098dcb-c184-7d93-4045-7ac1bc0ac6d0@linux.microsoft.com>
Date:   Mon, 26 Jun 2023 19:50:44 +0530
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:102.0) Gecko/20100101
 Thunderbird/102.12.0
Subject: Re: [PATCH 1/2 V3 net] net: mana: Fix MANA VF unload when host is
 unresponsive
Content-Language: en-US
To:     souradeep chakrabarti <schakrabarti@linux.microsoft.com>,
        kys@microsoft.com, haiyangz@microsoft.com, wei.liu@kernel.org,
        decui@microsoft.com, davem@davemloft.net, edumazet@google.com,
        kuba@kernel.org, pabeni@redhat.com, longli@microsoft.com,
        sharmaajay@microsoft.com, leon@kernel.org, cai.huoqing@linux.dev,
        ssengar@linux.microsoft.com, vkuznets@redhat.com,
        tglx@linutronix.de, linux-hyperv@vger.kernel.org,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
        linux-rdma@vger.kernel.org
Cc:     stable@vger.kernel.org, schakrabarti@microsoft.com
References: <1687771098-26775-1-git-send-email-schakrabarti@linux.microsoft.com>
 <1687771137-26911-1-git-send-email-schakrabarti@linux.microsoft.com>
From:   Praveen Kumar <kumarpraveen@linux.microsoft.com>
In-Reply-To: <1687771137-26911-1-git-send-email-schakrabarti@linux.microsoft.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit
Precedence: bulk

On 6/26/2023 2:48 PM, souradeep chakrabarti wrote:
> From: Souradeep Chakrabarti <schakrabarti@linux.microsoft.com>
> 
> This patch addresses the VF unload issue, where mana_dealloc_queues()
> gets stuck in infinite while loop, because of host unresponsiveness.
> It adds a timeout in the while loop, to fix it.
> 
> Fixes: ca9c54d2d6a5ab2430c4eda364c77125d62e5e0f (net: mana: Add a driver for
> Microsoft Azure Network Adapter)
> Signed-off-by: Souradeep Chakrabarti <schakrabarti@linux.microsoft.com>
> ---
> V2 -> V3:
> * Splitted the patch in two parts.
> * Removed the unnecessary braces from mana_dealloc_queues().
> ---
>  drivers/net/ethernet/microsoft/mana/mana_en.c | 19 +++++++++++++++++--
>  1 file changed, 17 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/microsoft/mana/mana_en.c b/drivers/net/ethernet/microsoft/mana/mana_en.c
> index d907727c7b7a..cb5c43c3c47e 100644
> --- a/drivers/net/ethernet/microsoft/mana/mana_en.c
> +++ b/drivers/net/ethernet/microsoft/mana/mana_en.c
> @@ -2329,7 +2329,10 @@ static int mana_dealloc_queues(struct net_device *ndev)
>  {
>  	struct mana_port_context *apc = netdev_priv(ndev);
>  	struct gdma_dev *gd = apc->ac->gdma_dev;
> +	unsigned long timeout;
>  	struct mana_txq *txq;
> +	struct sk_buff *skb;
> +	struct mana_cq *cq;
>  	int i, err;
>  
>  	if (apc->port_is_up)
> @@ -2348,13 +2351,25 @@ static int mana_dealloc_queues(struct net_device *ndev)
>  	 *
>  	 * Drain all the in-flight TX packets
>  	 */
> +
> +	timeout = jiffies + 120 * HZ;
>  	for (i = 0; i < apc->num_queues; i++) {
>  		txq = &apc->tx_qp[i].txq;
> -
> -		while (atomic_read(&txq->pending_sends) > 0)
> +		while (atomic_read(&txq->pending_sends) > 0 &&
> +		       time_before(jiffies, timeout))
>  			usleep_range(1000, 2000);
>  	}
>  
> +	for (i = 0; i < apc->num_queues; i++) {
> +		txq = &apc->tx_qp[i].txq;
> +		cq = &apc->tx_qp[i].tx_cq;
> +		while (atomic_read(&txq->pending_sends)) {
> +			skb = skb_dequeue(&txq->pending_skbs);
> +			mana_unmap_skb(skb, apc);
> +			napi_consume_skb(skb, cq->budget);
> +			atomic_sub(1, &txq->pending_sends);
> +		}
> +	}

Can we combine these 2 loops into 1 something like this ?

	for (i = 0; i < apc->num_queues; i++) {
		txq = &apc->tx_qp[i].txq;
		cq = &apc->tx_qp[i].tx_cq;
		while (atomic_read(&txq->pending_sends)) {
			if (time_before(jiffies, timeout)) {
				usleep_range(1000, 2000);
			} else {
				skb = skb_dequeue(&txq->pending_skbs);
				mana_unmap_skb(skb, apc);
				napi_consume_skb(skb, cq->budget);
				atomic_sub(1, &txq->pending_sends);
			}
		}
	}
>  	/* We're 100% sure the queues can no longer be woken up, because
>  	 * we're sure now mana_poll_tx_cq() can't be running.
>  	 */