Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Date:   Mon, 15 Jul 2019 17:32:22 -0700
From:   Andi Kleen <ak@linux.intel.com>
To:     Andy Lutomirski <luto@kernel.org>
Cc:     Dave Hansen <dave.hansen@intel.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Uros Bizjak <ubizjak@gmail.com>,
        LKML <linux-kernel@vger.kernel.org>, X86 ML <x86@kernel.org>
Subject: Re: [RFC PATCH, x86]: Disable CPA cache flush for selfsnoop targets
Message-ID: <20190716003222.GJ32439@tassilo.jf.intel.com>
References: <CAFULd4b=5-=WfF9OPCX+H9VDnsgbN7OBFj-XP=MZ0QqF5WpvQA@mail.gmail.com>
 <8736j7gsza.fsf@linux.intel.com>
 <alpine.DEB.2.21.1907152033020.1767@nanos.tec.linutronix.de>
 <CAFULd4bcB8tsgZuxZJm_ksp5zyDQXjO=v_Ov622Bmhx=fr7KuA@mail.gmail.com>
 <alpine.DEB.2.21.1907152129350.1767@nanos.tec.linutronix.de>
 <20190715193836.GF32439@tassilo.jf.intel.com>
 <CALCETrVonxn6tDkxZnbetM9W4Uxxm7-M-tv1e7YsieX3U5OBKA@mail.gmail.com>
 <20190715225305.GI32439@tassilo.jf.intel.com>
 <CALCETrW8yTRyP3qnOv04B2XvR5ZHDUky15CCBR2gtNVG3bea-Q@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <CALCETrW8yTRyP3qnOv04B2XvR5ZHDUky15CCBR2gtNVG3bea-Q@mail.gmail.com>
User-Agent: Mutt/1.11.3 (2019-02-01)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Jul 15, 2019 at 04:10:36PM -0700, Andy Lutomirski wrote:
> On Mon, Jul 15, 2019 at 3:53 PM Andi Kleen <ak@linux.intel.com> wrote:
> >
> > > I haven't tested on a real kernel with i915.  Does i915 really hit
> > > this code path?  Does it happen more than once or twice at boot?
> >
> > Yes some workloads allocate/free a lot of write combined memory
> > for graphics objects.
> >
> 
> But where does that memory come from?  If it's from device memory
> (i.e. memory that's not in the kernel direct map), then, unless I
> missed something, we're never changing the cache mode per se -- we're
> just ioremap_wc-ing it, which doesn't require a flush.

Integraded graphics doesn't have device memory. There's an reserved
memory area, but a lot of the buffers the GPU works with
come from main memory.


-Andi