RE: [EXT] Re: [PATCH net-next] net: mvpp2: Use relaxed I/O in data path

[Yan Markman <ymarkman@marvell.com>]

Thanks David
We are aware and it is working properly
Yan Markman


-----Original Message-----
From: David Miller [mailto:davem@davemloft.net] 
Sent: Tuesday, March 27, 2018 8:16 PM
To: maxime.chevallier@bootlin.com
Cc: Yan Markman <ymarkman@marvell.com>; netdev@vger.kernel.org; linux-kernel@vger.kernel.org; antoine.tenart@bootlin.com; thomas.petazzoni@bootlin.com; gregory.clement@bootlin.com; miquel.raynal@bootlin.com; Nadav Haklai <nadavh@marvell.com>; Stefan Chulski <stefanc@marvell.com>; mw@semihalf.com
Subject: [EXT] Re: [PATCH net-next] net: mvpp2: Use relaxed I/O in data path

External Email

----------------------------------------------------------------------
From: Maxime Chevallier <maxime.chevallier@bootlin.com>
Date: Tue, 27 Mar 2018 16:49:05 +0200

> From: Yan Markman <ymarkman@marvell.com>
> 
> Use relaxed I/O on the hot path. This achieves significant performance 
> improvements. On a 10G link, this makes a basic iperf TCP test go from 
> an average of 4.5 Gbits/sec to about 9.40 Gbits/sec.
> 
> Signed-off-by: Yan Markman <ymarkman@marvell.com>
> [Maxime: Commit message, cosmetic changes]
> Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
> ---
> David, this patch should not conflict with other the pending PPv2 
> series I sent earlier ("[PATCH net-next 0/2] net: mvpp2: Remove 
> unnecessary dynamic allocs")

Applied, but it is up to you to make sure the correct barriers are in place such that undesirable memory and IOMEM operation reordering does not occur.

Said another way, you get to keep the pieces if this breaks things :-)