[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <7ba34f16-dff9-12bb-889e-b9542740f19a@pensando.io>
Date: Mon, 31 Aug 2020 21:19:01 -0700
From: Shannon Nelson <snelson@...sando.io>
To: David Miller <davem@...emloft.net>
Cc: netdev@...r.kernel.org, neel@...sando.io
Subject: Re: [PATCH net-next 1/5] ionic: clean up page handling code
On 8/31/20 5:14 PM, David Miller wrote:
> From: Shannon Nelson <snelson@...sando.io>
> Date: Mon, 31 Aug 2020 16:35:54 -0700
>
>> @@ -100,6 +100,8 @@ static struct sk_buff *ionic_rx_frags(struct ionic_queue *q,
>> frag_len = min(len, (u16)PAGE_SIZE);
>> len -= frag_len;
>>
>> + dma_sync_single_for_cpu(dev, dma_unmap_addr(page_info, dma_addr),
>> + len, DMA_FROM_DEVICE);
>> dma_unmap_page(dev, dma_unmap_addr(page_info, dma_addr),
>> PAGE_SIZE, DMA_FROM_DEVICE);
>> skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags,
> The unmap operation performs a sync, if necessary, for you.
>
> That's the pattern of usage:
>
> map();
> device read/write memory
> unmap();
>
> That's it, no more, no less.
>
> The time to use sync is when you want to maintain the mapping and keep
> using it.
Thanks, I'll drop that part.
sln
Powered by blists - more mailing lists