Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From:   David Hildenbrand <david@redhat.com>
Mime-Version: 1.0 (1.0)
Subject: Re: [RFC v2 PATCH 4/4] mm: pre zero out free pages to speed up page allocation for __GFP_ZERO
Date:   Mon, 4 Jan 2021 21:11:07 +0100
Message-Id: <43576DAD-8A3B-4691-8808-90C5FDCF03B7@redhat.com>
References: <a5ba7bdf-8510-d0a0-9c22-ec1b81019982@intel.com>
Cc:     Matthew Wilcox <willy@infradead.org>,
        Alexander Duyck <alexander.h.duyck@linux.intel.com>,
        Mel Gorman <mgorman@techsingularity.net>,
        Andrew Morton <akpm@linux-foundation.org>,
        Andrea Arcangeli <aarcange@redhat.com>,
        Dan Williams <dan.j.williams@intel.com>,
        "Michael S. Tsirkin" <mst@redhat.com>,
        David Hildenbrand <david@redhat.com>,
        Jason Wang <jasowang@redhat.com>,
        Michal Hocko <mhocko@suse.com>,
        Liang Li <liliangleo@didiglobal.com>, linux-mm@kvack.org,
        linux-kernel@vger.kernel.org,
        virtualization@lists.linux-foundation.org
In-Reply-To: <a5ba7bdf-8510-d0a0-9c22-ec1b81019982@intel.com>
To:     Dave Hansen <dave.hansen@intel.com>
Precedence: bulk


> Am 04.01.2021 um 20:52 schrieb Dave Hansen <dave.hansen@intel.com>:
>=20
> =EF=BB=BFOn 1/4/21 11:27 AM, Matthew Wilcox wrote:
>>> On Mon, Jan 04, 2021 at 11:19:13AM -0800, Dave Hansen wrote:
>>> On 12/21/20 8:30 AM, Liang Li wrote:
>>>> --- a/include/linux/page-flags.h
>>>> +++ b/include/linux/page-flags.h
>>>> @@ -137,6 +137,9 @@ enum pageflags {
>>>> #endif
>>>> #ifdef CONFIG_64BIT
>>>>    PG_arch_2,
>>>> +#endif
>>>> +#ifdef CONFIG_PREZERO_PAGE
>>>> +    PG_zero,
>>>> #endif
>>>>    __NR_PAGEFLAGS,
>>>=20
>>> I don't think this is worth a generic page->flags bit.
>>>=20
>>> There's a ton of space in 'struct page' for pages that are in the
>>> allocator.  Can't we use some of that space?
>>=20
>> I was going to object to that too, but I think the entire approach is
>> flawed and needs to be thrown out.  It just nukes the caches in extremely=

>> subtle and hard to measure ways, lowering overall system performance.
>=20
> Yeah, it certainly can't be the default, but it *is* useful for thing
> where we know that there are no cache benefits to zeroing close to where
> the memory is allocated.
>=20
> The trick is opting into it somehow, either in a process or a VMA.
>=20

The patch set is mostly trying to optimize starting a new process. So proces=
s/vma doesn=E2=80=98t really work.

I still wonder if using tmpfs/shmem cannot somehow be used to cover the orig=
inal use case of starting a new vm fast (or rebooting an existing one involv=
ing restarting the process).=