clear_user (was: [patch 02/14] tmpfs: fix regressions from wider use of ZERO_PAGE)

From: Borislav Petkov <bp@alien8.de>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Mark Hemment <markhemm@googlemail.com>,
	Andrew Morton <akpm@linux-foundation.org>,
	the arch/x86 maintainers <x86@kernel.org>,
	Peter Zijlstra <peterz@infradead.org>,
	patrice.chotard@foss.st.com,
	Mikulas Patocka <mpatocka@redhat.com>,
	Lukas Czerner <lczerner@redhat.com>,
	Christoph Hellwig <hch@lst.de>,
	"Darrick J. Wong" <djwong@kernel.org>,
	Chuck Lever <chuck.lever@oracle.com>,
	Hugh Dickins <hughd@google.com>,
	patches@lists.linux.dev, Linux-MM <linux-mm@kvack.org>,
	mm-commits@vger.kernel.org, Mel Gorman <mgorman@suse.de>
Subject: clear_user (was: [patch 02/14] tmpfs: fix regressions from wider use of ZERO_PAGE)
Date: Tue, 10 May 2022 11:31:28 +0200	[thread overview]
Message-ID: <Ynow8F3G8Kl6V3gu@zn.tnic> (raw)
In-Reply-To: <CAHk-=wjUX5DGNSiBYvPC8fQJGRe5_RWR8NW=gYF4=UpPiwCE8A@mail.gmail.com>

Lemme fix that subject so that I can find it easier in my avalanche mbox...

On Wed, May 04, 2022 at 02:09:52PM -0700, Linus Torvalds wrote:
> I don't tend to particularly care about "how many times has this been
> called" kind of trace profiles. It's the actual expense in CPU cycles
> I tend to care about.

Yeah, but, I wanted to measure how much perf improvement that would
bring with the git test suite and wanted to know how often clear_user()
is called in conjunction with it. Because the benchmarks I ran would
show very small improvements and a PF benchmark would even show weird
things like slowdowns with higher core counts.

So for a ~6m running test suite, the function gets called under 700K
times, all from padzero:

           <...>-2536    [006] .....   261.208801: padzero: to: 0x55b0663ed214, size: 3564, cycles: 21900
           <...>-2536    [006] .....   261.208819: padzero: to: 0x7f061adca078, size: 3976, cycles: 17160
           <...>-2537    [008] .....   261.211027: padzero: to: 0x5572d019e240, size: 3520, cycles: 23850
           <...>-2537    [008] .....   261.211049: padzero: to: 0x7f1288dc9078, size: 3976, cycles: 15900
	   ...

which is around 1%-ish of the total time and which is consistent with
the benchmark numbers.

So Mel gave me the idea to simply measure how fast the function becomes. I.e.:

                start = rdtsc_ordered();
                ret = __clear_user(to, n);
                end = rdtsc_ordered();

Computing the mean average of all the samples collected during the test
suite run then shows some improvement:

clear_user_original:
Amean: 9219.71 (Sum: 6340154910, samples: 687674)

fsrm:
Amean: 8030.63 (Sum: 5522277720, samples: 687652)

That's on Zen3.

I'll run this on Icelake now too.

> I haven't really done serious profiling work for a while (which is
> just as well, because it's one of the things that went backwards when
> I switch to the Zen 2 threadripper for my main machine)

Because of the not as advanced perf support there? Any pain points I can
forward?

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette