[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAHk-=wgAH-yXX=1T_hqenVVmQPVKP-dkRKKkHYcWAXK0tRzz6Q@mail.gmail.com>
Date: Mon, 15 Jun 2020 12:09:58 -0700
From: Linus Torvalds <torvalds@...ux-foundation.org>
To: Jann Horn <jannh@...gle.com>
Cc: kernel test robot <rong.a.chen@...el.com>,
Christoph Hellwig <hch@....de>,
Oleg Nesterov <oleg@...hat.com>,
Kirill Shutemov <kirill@...temov.name>,
Jan Kara <jack@...e.cz>,
Andrea Arcangeli <aarcange@...hat.com>,
Matthew Wilcox <willy@...radead.org>,
LKML <linux-kernel@...r.kernel.org>, lkp@...ts.01.org
Subject: Re: [gup] 17839856fd: stress-ng.vm-splice.ops_per_sec 2158.6% improvement
On Mon, Jun 15, 2020 at 12:33 AM Jann Horn <jannh@...gle.com> wrote:
>
> So in summary I guess the test was just really slow up until now
> because it was hitting a slowpath that you wouldn't hit during normal
> usage? At least for vmsplice(), writing uninitialized pages doesn't
> really make a whole lot of sense...
Heh. My main worry with that commit was that we'd see the reverse
effect: somebody implicitly depending on the continued sharing of
pages in some odd splice situation, and where breaking the COW would
result in a performance degradation.
The fact that it went the other way instead is somewhat ironic.
But good.
I guess I'll just ping the stable people to say "yeah, put that in
stable too". It's a fix, and apparently it _helps_ performance for the
degenerate cases rather than hurting anything.
Linus
Powered by blists - more mailing lists