| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <CAOUHufZxguv_m3Td7e5Qt-yKpV7rmWyv_m_UFS9n19K=_=xLcA@mail.gmail.com>
Date: Wed, 23 Nov 2022 18:18:06 -0700
From: Yu Zhao <yuzhao@...gle.com>
To: Johannes Weiner <hannes@...xchg.org>,
Ivan Babrou <ivan@...udflare.com>
Cc: Linux MM <linux-mm@...ck.org>,
Linux Kernel Network Developers <netdev@...r.kernel.org>,
linux-kernel <linux-kernel@...r.kernel.org>,
Michal Hocko <mhocko@...nel.org>,
Roman Gushchin <roman.gushchin@...ux.dev>,
Shakeel Butt <shakeelb@...gle.com>,
Muchun Song <songmuchun@...edance.com>,
Andrew Morton <akpm@...ux-foundation.org>,
Eric Dumazet <edumazet@...gle.com>,
"David S. Miller" <davem@...emloft.net>,
Hideaki YOSHIFUJI <yoshfuji@...ux-ipv6.org>,
David Ahern <dsahern@...nel.org>,
Jakub Kicinski <kuba@...nel.org>,
Paolo Abeni <pabeni@...hat.com>, cgroups@...r.kernel.org,
kernel-team <kernel-team@...udflare.com>
Subject: Re: Low TCP throughput due to vmpressure with swap enabled
On Wed, Nov 23, 2022 at 2:22 PM Johannes Weiner <hannes@...xchg.org> wrote:
>
> On Tue, Nov 22, 2022 at 05:44:44PM -0700, Yu Zhao wrote:
> > Hi Johannes,
> >
> > Do you think it makes sense to have the below for both the baseline and
> > MGLRU or it's some behavior change that the baseline doesn't want to
> > risk?
>
> It looks good to me. Besides the new FMODE_NOREUSE, it's also a nice
> cleanup on the rmap side!
>
> It would just be good to keep the comment from folio_referenced_one() and
> move it to the vma_has_locality() check in invalid_folio_referenced_vma().
>
> Otherwise,
>
> Acked-by: Johannes Weiner <hannes@...xchg.org>
Thanks.
I've added Ivan's test case to my collection. Interestingly, after
this patch, the download speed increased while fio was running (my
guess is that fio pushed more cold anon into swap):
$ uname
Linux test127 6.1.0-rc6-dirty #2 SMP PREEMPT_DYNAMIC Wed Nov 23
16:51:20 MST 2022 x86_64 x86_64 x86_64 GNU/Linux
$ go version
go version go1.18.1 linux/amd64
$ fio -v
fio-3.28
$ curl --version
curl 7.81.0 (x86_64-pc-linux-gnu) libcurl/7.81.0 OpenSSL/3.0.2
zlib/1.2.11 brotli/1.0.9 zstd/1.4.8 libidn2/2.3.2 libpsl/0.21.0
(+libidn2/2.3.2) libssh/0.9.6/openssl/zlib nghttp2/1.43.0 librtmp/2.3
OpenLDAP/2.5.13
Release-Date: 2022-01-05
Protocols: dict file ftp ftps gopher gophers http https imap imaps
ldap ldaps mqtt pop3 pop3s rtmp rtsp scp sftp smb smbs smtp smtps
telnet tftp
Features: alt-svc AsynchDNS brotli GSS-API HSTS HTTP2 HTTPS-proxy IDN
IPv6 Kerberos Largefile libz NTLM NTLM_WB PSL SPNEGO SSL TLS-SRP
UnixSockets zstd
fio NOT running:
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 83.6M 0 --:--:-- 0:00:57 --:--:-- 87.0M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 82.8M 0 --:--:-- 0:00:57 --:--:-- 79.1M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 82.7M 0 --:--:-- 0:00:57 --:--:-- 89.7M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 87.4M 0 --:--:-- 0:00:54 --:--:-- 94.3M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 88.1M 0 --:--:-- 0:00:54 --:--:-- 94.7M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 82.6M 0 --:--:-- 0:00:57 --:--:-- 83.9M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 86.4M 0 --:--:-- 0:00:55 --:--:-- 90.1M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 82.8M 0 --:--:-- 0:00:57 --:--:-- 67.5M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 83.4M 0 --:--:-- 0:00:57 --:--:-- 78.7M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 84.0M 0 --:--:-- 0:00:56 --:--:-- 87.4M
fio running:
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 86.7M 0 --:--:-- 0:01:11 --:--:-- 88.7M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 87.7M 0 --:--:-- 0:00:54 --:--:-- 93.5M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 88.5M 0 --:--:-- 0:00:53 --:--:-- 95.1M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 91.6M 0 --:--:-- 0:00:52 --:--:-- 94.4M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 89.4M 0 --:--:-- 0:00:53 --:--:-- 86.6M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 88.6M 0 --:--:-- 0:00:53 --:--:-- 84.8M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 84.6M 0 --:--:-- 0:00:56 --:--:-- 87.5M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 86.9M 0 --:--:-- 0:00:54 --:--:-- 81.4M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 89.0M 0 --:--:-- 0:00:53 --:--:-- 86.4M
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
100 4768M 0 4768M 0 0 91.1M 0 --:--:-- 0:00:52 --:--:-- 90.6M
Powered by blists - more mailing lists