[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <6b0cf200-c946-ff7d-e80a-920ab0770b67@huawei.com>
Date: Sat, 2 Apr 2022 16:09:30 +0800
From: "wangjie (L)" <wangjie125@...wei.com>
To: Jakub Kicinski <kuba@...nel.org>
CC: <mkubecek@...e.cz>, <davem@...emloft.net>,
<netdev@...r.kernel.org>, <huangguangbin2@...wei.com>,
<lipeng321@...wei.com>, <shenjian15@...wei.com>,
<moyufeng@...wei.com>, <linyunsheng@...wei.com>,
<salil.mehta@...wei.com>, <chenhao288@...ilicon.com>
Subject: Re: [RFCv4 PATCH net-next 1/3] net-next: ethtool: extend ringparam
set/get APIs for tx_push
On 2022/4/1 11:22, Jakub Kicinski wrote:
> On Thu, 31 Mar 2022 16:43:40 +0800 Jie Wang wrote:
>> +``ETHTOOL_A_RINGS_TX_PUSH`` controls flag to choose the fast path or the
>> +normal path to send packets. Setting tx push attribute "on" will enable tx
>> +push mode and send packets in fast path. For those not supported hardwares,
>> +this attributes is "off" by default settings.
>
> You need to be more specific what a "fast path" is, that term is
> too general.
>
> I presume you want to say something about the descriptor being written
> directly into device address space... and that it lowers the latency
> but increases the cost...
>
OK, I will add tx push description in detail in next version.
> .
>
Powered by blists - more mailing lists