Message-ID: <1512855422.25033.34.camel@gmail.com>
Subject: Re: NFS corruption, fixed by echo 1 > /proc/sys/vm/drop_caches --
 next debugging steps?
From: Eric Dumazet <eric.dumazet@gmail.com>
To: Matt Turner <mattst88@gmail.com>
Cc: Eric Dumazet <edumazet@google.com>,
        "linux-mips@linux-mips.org" <linux-mips@linux-mips.org>,
        linux-nfs@vger.kernel.org, Paolo Abeni <pabeni@redhat.com>,
        Hannes Frederic Sowa <hannes@stressinduktion.org>,
        "Peter Zijlstra (Intel)" <peterz@infradead.org>,
        Thomas Gleixner <tglx@linutronix.de>, Ingo Molnar <mingo@kernel.org>,
        Manuel Lauss <manuel.lauss@gmail.com>,
        LKML <linux-kernel@vger.kernel.org>, netdev <netdev@vger.kernel.org>
Date: Sat, 09 Dec 2017 13:37:02 -0800
In-Reply-To: <CAEdQ38H+jUF3OXpe13Vfm=QZE3iHa=B7PpXkpbek1PnY2E1u5w@mail.gmail.com>
References: <CAEdQ38HcOgAT6wJWWKY3P0hzYwkBGSQkRSQ2a=eaGmD6c6rwXA@mail.gmail.com>
         <CAEdQ38G4VTXDGOarmmTac=hP92VJbQHRFxQTaSWQ3j4d63pogg@mail.gmail.com>
         <CAEdQ38HcPswBk3pUHzQerFZ=4KjPc5nVYTqNnGQNMk7QbPXuOQ@mail.gmail.com>
         <CANn89iJKGRLVNAE99JWiyXcOXveytkjbQAiZ9XPiJc6fyEdFVA@mail.gmail.com>
         <1512741164.25033.28.camel@gmail.com>
         <CAEdQ38HEduSTY38Noj4peaMN_G++5sLJfqzCMkd3M4pPNTpU_Q@mail.gmail.com>
         <1512767781.25033.30.camel@gmail.com>
         <CAEdQ38H+jUF3OXpe13Vfm=QZE3iHa=B7PpXkpbek1PnY2E1u5w@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 777
Lines: 30

On Sat, 2017-12-09 at 13:03 -0800, Matt Turner wrote:
> On Fri, Dec 8, 2017 at 1:16 PM, Eric Dumazet <eric.dumazet@gmail.com>
> wrote:
> > On Fri, 2017-12-08 at 12:26 -0800, Matt Turner wrote:
> > > 
> > > Thanks for the quick reply!
> > > 
> > > I tried the patch on top of master, but unfortunately the
> > > corruption
> > > still occurs.
> > 
> > You might try replacing in sbdma_add_rcvbuffer()
> > 
> > sb_new = netdev_alloc_skb(dev, size);
> > 
> > by
> > 
> > sb_new = alloc_skb(size, GFP_ATOMIC);
> > 
> > Maybe the device does not like having a frame spanning 2 pages.
> 
> No such luck. I also gave changing the page size from 16K to 4K a
> shot
> without success.


If your hist is SMP, could you try running it with one CPU only ?

Sorry, I have no more ideas :/