netdev - Re: [PATCH net-next v4 4/6] net: tn40xx: add basic Rx handling

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: Windows password security audit tool. GUI, reports in PDF.

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

Date: Wed, 08 May 2024 16:26:18 +0900 (JST)
From: FUJITA Tomonori <fujita.tomonori@...il.com>
To: pabeni@...hat.com
Cc: fujita.tomonori@...il.com, netdev@...r.kernel.org, andrew@...n.ch,
 kuba@...nel.org, jiri@...nulli.us, horms@...nel.org
Subject: Re: [PATCH net-next v4 4/6] net: tn40xx: add basic Rx handling

Hi,

Thanks for reviewing the patch!

On Mon, 06 May 2024 11:20:55 +0200
Paolo Abeni <pabeni@...hat.com> wrote:

> 
> On Thu, 2024-05-02 at 08:05 +0900, FUJITA Tomonori wrote:
>> +static struct tn40_rx_page *tn40_rx_page_alloc(struct tn40_priv *priv)
>> +{
>> +	struct tn40_rx_page *rx_page = &priv->rx_page_table.rx_pages;
>> +	int page_size = priv->rx_page_table.page_size;
>> +	struct page *page;
>> +	gfp_t gfp_mask;
>> +	dma_addr_t dma;
>> +
>> +	gfp_mask = GFP_ATOMIC | __GFP_NOWARN;
>> +	if (page_size > PAGE_SIZE)
>> +		gfp_mask |= __GFP_COMP;
>> +
>> +	page = alloc_pages(gfp_mask, get_order(page_size));
>> +	if (likely(page)) {
> 
> Note that this allocation schema can be problematic when the NIC will
> receive traffic from many different streams/connection: a single packet
> can keep a full order 4 page in use leading to overall memory usage
> much greater the what truesize will report.
> 
> See commit 3226b158e67c. Here the under-estimation could fair worse.
> 
> Drivers usually use order-0 or order-1 pages.

Understood. I fixed the driver to use only order-0 or order-1 pages.

> [...]
>> +static void tn40_recycle_skb(struct tn40_priv *priv, struct tn40_rxd_desc *rxdd)
>> +{
> 
> Minor nit: the function name is confusing, at it does recycle in
> internal buffer, not a skbuff.

Sure, I'll go with recycle_rx_buffer instead.