linux-kernel - Re: [net-next PATCH v6 6/6] octeontx2-pf: AF

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <Z7SEleIJ636O+XZI@boxer>
Date: Tue, 18 Feb 2025 14:01:09 +0100
From: Maciej Fijalkowski <maciej.fijalkowski@...el.com>
To: Suman Ghosh <sumang@...vell.com>
CC: <horms@...nel.org>, <sgoutham@...vell.com>, <gakula@...vell.com>,
	<sbhatta@...vell.com>, <hkelam@...vell.com>, <davem@...emloft.net>,
	<edumazet@...gle.com>, <kuba@...nel.org>, <pabeni@...hat.com>,
	<netdev@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
	<lcherian@...vell.com>, <jerinj@...vell.com>, <john.fastabend@...il.com>,
	<bbhushan2@...vell.com>, <hawk@...nel.org>, <andrew+netdev@...n.ch>,
	<ast@...nel.org>, <daniel@...earbox.net>, <bpf@...r.kernel.org>,
	<larysa.zaremba@...el.com>
Subject: Re: [net-next PATCH v6 6/6] octeontx2-pf: AF_XDP zero copy transmit
 support

On Thu, Feb 13, 2025 at 11:01:41AM +0530, Suman Ghosh wrote:
> This patch implements below changes,
> 
> 1. To avoid concurrency with normal traffic uses
>    XDP queues.
> 2. Since there are chances that XDP and AF_XDP can
>    fall under same queue uses separate flags to handle
>    dma buffers.
> 
> Signed-off-by: Hariprasad Kelam <hkelam@...vell.com>
> Signed-off-by: Suman Ghosh <sumang@...vell.com>
> ---

[...]

> +void otx2_zc_napi_handler(struct otx2_nic *pfvf, struct xsk_buff_pool *pool,
> +			  int queue, int budget)
> +{
> +	struct xdp_desc *xdp_desc = pool->tx_descs;
> +	int err, i, work_done = 0, batch;
> +
> +	budget = min(budget, otx2_read_free_sqe(pfvf, queue));
> +	batch = xsk_tx_peek_release_desc_batch(pool, budget);
> +	if (!batch)
> +		return;
> +
> +	for (i = 0; i < batch; i++) {
> +		dma_addr_t dma_addr;
> +
> +		dma_addr = xsk_buff_raw_get_dma(pool, xdp_desc[i].addr);
> +		err = otx2_xdp_sq_append_pkt(pfvf, NULL, dma_addr, xdp_desc[i].len,
> +					     queue, OTX2_AF_XDP_FRAME);
> +		if (!err) {
> +			netdev_err(pfvf->netdev, "AF_XDP: Unable to transfer packet err%d\n", err);
> +			break;
> +		}
> +		work_done++;
> +	}
> +
> +	if (work_done)
> +		xsk_tx_release(pool);

this is broken actually. the batch api you're using above is doing tx
release internally for you.

Sorry for not catching this earlier but i was never CCed in this series.

> +}
> diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.h b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.h
> index 022b3433edbb..8047fafee8fe 100644
> --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.h
> +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_xsk.h
> @@ -17,5 +17,8 @@ int otx2_xsk_pool_disable(struct otx2_nic *pf, u16 qid);
>  int otx2_xsk_pool_alloc_buf(struct otx2_nic *pfvf, struct otx2_pool *pool,
>  			    dma_addr_t *dma, int idx);
>  int otx2_xsk_wakeup(struct net_device *dev, u32 queue_id, u32 flags);
> +void otx2_zc_napi_handler(struct otx2_nic *pfvf, struct xsk_buff_pool *pool,
> +			  int queue, int budget);
> +void otx2_attach_xsk_buff(struct otx2_nic *pfvf, struct otx2_snd_queue *sq, int qidx);
>  
>  #endif /* OTX2_XSK_H */
> -- 
> 2.25.1
> 
>