Re: [RFC][CFT][PATCHSET v1] uaccess unification

From	Linus Torvalds <torvalds@linux-foundation.org>
Newsgroups	linux.kernel
Subject	Re: [RFC][CFT][PATCHSET v1] uaccess unification
Date	2017-03-30 21:00 +0200
Message-ID	<tqNsB-2b6-1@gated-at.bofh.it> (permalink)
References	<tqeOd-2vj-13@gated-at.bofh.it> <tqL7r-AJ-3@gated-at.bofh.it> <tqLqN-KI-7@gated-at.bofh.it> <tqLTQ-1a6-7@gated-at.bofh.it> <tqNiW-27G-13@gated-at.bofh.it>
Organization	linux.* mail to news gateway

Show all headers | View raw

On Thu, Mar 30, 2017 at 11:48 AM, Al Viro <viro@zeniv.linux.org.uk> wrote:
>
> This is not going into the tree - it's just a "let's check your
> theory about might_fault() overhead being the source of slowdown
> you are seeing" quick-and-dirty patch.

Note that for cached hdparm reads, I suspect a *much* bigger effects
than the fairly cheap might_fault() tests is just the random layout of
the data in the page cache.

Memory is just more expensive than CPU is.

The precise physical address that gets allocated for the page cache
entries ends up mattering, and is obviously fairly "sticky" within one
reboot (unless you have a huge working set and that flushes it, or you
use something like

    echo 3 > /proc/sys/vm/drop_caches

to flush filesystem caches manually).

The reason things like page allocation matter for performance testing
is simply that the CPU caches are physically indexed (the L1 might not
be, but outer levels definitely are), and so page allocation ends up
impacting caching unless you have very high associativity.

And even if your workload doesn't fit in your CPU caches (I'd hope
that the "cached" hdparm is still doing a fairly big area), you'll
still see memory performance depend on physical addresses.

Doing kernel performance testing without rebooting several times is
generally very hard.

                   Linus

Thread

[RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-29 08:00 +0200
  Re: [RFC][CFT][PATCHSET v1] uaccess unification Vineet Gupta <Vineet.Gupta1@synopsys.com> - 2017-03-29 22:20 +0200
    Re: [RFC][CFT][PATCHSET v1] uaccess unification Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-29 22:40 +0200
      Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-29 23:10 +0200
        Re: [RFC][CFT][PATCHSET v1] uaccess unification Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-29 23:30 +0200
          Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-30 01:20 +0200
            Re: [RFC][CFT][PATCHSET v1] uaccess unification Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-30 01:50 +0200
              Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-30 17:40 +0200
    Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-29 22:40 +0200
      Re: [RFC][CFT][PATCHSET v1] uaccess unification Vineet Gupta <Vineet.Gupta1@synopsys.com> - 2017-03-29 23:20 +0200
        Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-30 01:50 +0200
          Re: [RFC][CFT][PATCHSET v1] uaccess unification Vineet Gupta <Vineet.Gupta1@synopsys.com> - 2017-03-30 02:10 +0200
            Re: [RFC][CFT][PATCHSET v1] uaccess unification Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-30 02:40 +0200
              Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-30 03:20 +0200
              Re: [RFC][CFT][PATCHSET v1] uaccess unification Vineet Gupta <Vineet.Gupta1@synopsys.com> - 2017-03-30 22:50 +0200
                Re: [RFC][CFT][PATCHSET v1] uaccess unification Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-30 23:10 +0200
                Re: [RFC][CFT][PATCHSET v1] uaccess unification Russell King - ARM Linux <linux@armlinux.org.uk> - 2017-03-31 01:30 +0200
  Re: [RFC][CFT][PATCHSET v1] uaccess unification Martin Schwidefsky <schwidefsky@de.ibm.com> - 2017-03-30 14:40 +0200
    Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-30 16:50 +0200
  Re: [RFC][CFT][PATCHSET v1] uaccess unification Russell King - ARM Linux <linux@armlinux.org.uk> - 2017-03-30 18:30 +0200
    Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-30 18:50 +0200
      Re: [RFC][CFT][PATCHSET v1] uaccess unification Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-30 19:20 +0200
        Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-30 20:50 +0200
          Re: [RFC][CFT][PATCHSET v1] uaccess unification Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-30 21:00 +0200
          Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-30 21:00 +0200
            Re: [RFC][CFT][PATCHSET v1] uaccess unification Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-30 21:00 +0200
              Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-30 21:20 +0200
                Re: [RFC][CFT][PATCHSET v1] uaccess unification Linus Torvalds <torvalds@linux-foundation.org> - 2017-03-30 21:30 +0200
                Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-03-30 23:20 +0200
  Re: [RFC][CFT][PATCHSET v1] uaccess unification Kees Cook <keescook@chromium.org> - 2017-03-31 02:30 +0200
    Re: [RFC][CFT][PATCHSET v1] uaccess unification James Hogan <james.hogan@imgtec.com> - 2017-03-31 15:40 +0200
  Re: [RFC][CFT][PATCHSET v1] uaccess unification James Morse <james.morse@arm.com> - 2017-04-03 18:30 +0200
  Re: [RFC][CFT][PATCHSET v1] uaccess unification Max Filippov <jcmvbkbc@gmail.com> - 2017-04-04 22:30 +0200
    Re: [RFC][CFT][PATCHSET v1] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-04-04 23:00 +0200
  ia64 exceptions (Re: [RFC][CFT][PATCHSET v1] uaccess unification) Al Viro <viro@ZenIV.linux.org.uk> - 2017-04-05 07:10 +0200
    Re: ia64 exceptions (Re: [RFC][CFT][PATCHSET v1] uaccess unification) Al Viro <viro@ZenIV.linux.org.uk> - 2017-04-05 10:10 +0200
      Re: ia64 exceptions (Re: [RFC][CFT][PATCHSET v1] uaccess unification) Tony Luck <tony.luck@gmail.com> - 2017-04-05 20:50 +0200
        Re: ia64 exceptions (Re: [RFC][CFT][PATCHSET v1] uaccess unification) Al Viro <viro@ZenIV.linux.org.uk> - 2017-04-05 22:40 +0200
  [RFC][CFT][PATCHSET v2] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-04-07 02:30 +0200
    Re: [RFC][CFT][PATCHSET v2] uaccess unification Al Viro <viro@ZenIV.linux.org.uk> - 2017-04-07 02:40 +0200

csiph-web