From mboxrd@z Thu Jan 1 00:00:00 1970 From: Florian Fainelli Subject: Re: [PATCH] bcm63xx_enet: fix poll callback. Date: Tue, 03 Mar 2015 09:42:36 -0800 Message-ID: <54F5F28C.40202@gmail.com> References: <1425383112-23851-1-git-send-email-nschichan@freebox.fr> <1425389399.5130.169.camel@edumazet-glaptop2.roam.corp.google.com> <1425390142.5130.173.camel@edumazet-glaptop2.roam.corp.google.com> <54F5BCE0.3060507@freebox.fr> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Cc: "David S. Miller" , Tobias Klauser , Felipe Balbi , Wilfried Klaebe , "Eric W. Biederman" , Alexander Duyck , netdev@vger.kernel.org, linux-kernel@vger.kernel.org, Jonas Gorski To: Nicolas Schichan , Eric Dumazet Return-path: In-Reply-To: <54F5BCE0.3060507@freebox.fr> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On 03/03/15 05:53, Nicolas Schichan wrote: > On 03/03/2015 02:42 PM, Eric Dumazet wrote: >>> To avoid that, I would take priv->tx_lock only once, or add a limit on >>> the number of skbs that can be drained per round. >> >> Something like this (untested) patch > > I'm not against testing this patch, but we do not have any SMP capable bcm63xx > board here so I don't think it will be of any use. > > bcm6358 and bcm6368 do indeed have two MIPS threads, but SMP is not possible > (due to a data cache or TLB shared across all MIPS threads , unbearably > complicating things, IIRC). 6358 does have the shared TLB (early BMIPS4350), but 6368 (later BMIPS4350) runs just fine in a regular SMP configuration, that's the default for OpenWrt actually. Maybe Jonas has something readily available he could test on? > > bcm63xx ARM SoCs look like they can support SMP though. These SoCs do not use this Ethernet controller at all. -- Florian