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;
MIME-Version: 1.0
References: <20190904150920.13848.32271.stgit@localhost.localdomain> <20190904151030.13848.25822.stgit@localhost.localdomain>
In-Reply-To: <20190904151030.13848.25822.stgit@localhost.localdomain>
From:   Dan Williams <dan.j.williams@intel.com>
Date:   Wed, 4 Sep 2019 14:10:17 -0700
Message-ID: <CAPcyv4jZVoztoRA7hEq5xzbwg0QJ+UVASuk5XQmB5KHQrvAmfA@mail.gmail.com>
Subject: Re: [PATCH v7 1/6] mm: Adjust shuffle code to allow for future coalescing
To:     Alexander Duyck <alexander.duyck@gmail.com>
Cc:     nitesh@redhat.com, KVM list <kvm@vger.kernel.org>,
        "Michael S. Tsirkin" <mst@redhat.com>,
        David Hildenbrand <david@redhat.com>,
        Dave Hansen <dave.hansen@intel.com>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Matthew Wilcox <willy@infradead.org>,
        Michal Hocko <mhocko@kernel.org>,
        Linux MM <linux-mm@kvack.org>,
        Andrew Morton <akpm@linux-foundation.org>,
        virtio-dev@lists.oasis-open.org,
        Oscar Salvador <osalvador@suse.de>, yang.zhang.wz@gmail.com,
        Pankaj Gupta <pagupta@redhat.com>,
        Rik van Riel <riel@surriel.com>,
        Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>,
        lcapitulino@redhat.com, "Wang, Wei W" <wei.w.wang@intel.com>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Paolo Bonzini <pbonzini@redhat.com>,
        Alexander Duyck <alexander.h.duyck@linux.intel.com>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Sep 4, 2019 at 8:10 AM Alexander Duyck
<alexander.duyck@gmail.com> wrote:
>
> From: Alexander Duyck <alexander.h.duyck@linux.intel.com>
>
> Move the head/tail adding logic out of the shuffle code and into the
> __free_one_page function since ultimately that is where it is really
> needed anyway. By doing this we should be able to reduce the overhead
> and can consolidate all of the list addition bits in one spot.
>
> While changing out the code I also opted to go for a bit more thread safe
> approach to getting the boolean value. This way we can avoid possible cache
> line bouncing of the batched entropy between CPUs.

The original version of this patch just did the movement, but now the
patch also does the percpu optimization. At this point it warrants
being split into a "move" patch and then "rework". Otherwise the bulk
of the patch is not really well described by the patch title. With the
split there's a commit id for each of the performance improvement
claims.

Other than that the percpu logic changes look good to me.