dri-devel Archive on lore.kernel.org
 help / color / Atom feed
From: peterz@infradead.org
To: Steven Rostedt <rostedt@goodmis.org>
Cc: Juri Lelli <juri.lelli@redhat.com>,
	David Airlie <airlied@linux.ie>,
	Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
	dri-devel <dri-devel@lists.freedesktop.org>,
	linux-mips@vger.kernel.org, Ben Segall <bsegall@google.com>,
	Max Filippov <jcmvbkbc@gmail.com>, Guo Ren <guoren@kernel.org>,
	linux-sparc <sparclinux@vger.kernel.org>,
	Vincent Chen <deanbo422@gmail.com>, Will Deacon <will@kernel.org>,
	Ard Biesheuvel <ardb@kernel.org>,
	linux-arch <linux-arch@vger.kernel.org>,
	Vincent Guittot <vincent.guittot@linaro.org>,
	Herbert Xu <herbert@gondor.apana.org.au>,
	Michael Ellerman <mpe@ellerman.id.au>,
	the arch/x86 maintainers <x86@kernel.org>,
	Russell King <linux@armlinux.org.uk>,
	linux-csky@vger.kernel.org, Mel Gorman <mgorman@suse.de>,
	"open list:SYNOPSYS ARC ARCHITECTURE"
	<linux-snps-arc@lists.infradead.org>,
	linux-xtensa@linux-xtensa.org, Paul McKenney <paulmck@kernel.org>,
	intel-gfx <intel-gfx@lists.freedesktop.org>,
	linuxppc-dev <linuxppc-dev@lists.ozlabs.org>,
	Greentime Hu <green.hu@gmail.com>,
	Rodrigo Vivi <rodrigo.vivi@intel.com>,
	Thomas Gleixner <tglx@linutronix.de>,
	Dietmar Eggemann <dietmar.eggemann@arm.com>,
	Linux ARM <linux-arm-kernel@lists.infradead.org>,
	Chris Zankel <chris@zankel.net>, Michal Simek <monstr@monstr.eu>,
	Thomas Bogendoerfer <tsbogend@alpha.franken.de>,
	Nick Hu <nickhu@andestech.com>, Linux-MM <linux-mm@kvack.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	LKML <linux-kernel@vger.kernel.org>,
	Arnd Bergmann <arnd@arndb.de>, Vineet Gupta <vgupta@synopsys.com>,
	Paul Mackerras <paulus@samba.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	Daniel Bristot de Oliveira <bristot@redhat.com>,
	"David S. Miller" <davem@davemloft.net>
Subject: Re: [patch RFC 00/15] mm/highmem: Provide a preemptible variant of kmap_atomic & friends
Date: Thu, 24 Sep 2020 10:27:17 +0200
Message-ID: <20200924082717.GA1362448@hirez.programming.kicks-ass.net> (raw)
In-Reply-To: <20200923115251.7cc63a7e@oasis.local.home>

On Wed, Sep 23, 2020 at 11:52:51AM -0400, Steven Rostedt wrote:
> On Wed, 23 Sep 2020 10:40:32 +0200
> peterz@infradead.org wrote:
> 
> > However, with migrate_disable() we can have each task preempted in a
> > migrate_disable() region, worse we can stack them all on the _same_ CPU
> > (super ridiculous odds, sure). And then we end up only able to run one
> > task, with the rest of the CPUs picking their nose.
> 
> What if we just made migrate_disable() a local_lock() available for !RT?

Can't, neiter migrate_disable() nor migrate_enable() are allowed to
block -- which is what makes their implementation so 'interesting'.

> This should lower the SHC in theory, if you can't have stacked migrate
> disables on the same CPU.

See this email in that other thread (you're on Cc too IIRC):

  https://lkml.kernel.org/r/20200923170809.GY1362448@hirez.programming.kicks-ass.net

I think that is we 'frob' the balance PULL, we'll end up with something
similar.

Whichever way around we turn this thing, the migrate_disable() runtime
(we'll have to add a tracer for that), will be an interference term on
the lower priority task, exactly like preempt_disable() would be. We'll
just not exclude a higher priority task from running.


AFAICT; the best case is a single migrate_disable() nesting, where a
higher priority task preempts in a migrate_disable() section -- this is
per design.

When this preempted task becomes elegible to run under the ideal model
(IOW it becomes one of the M highest priority tasks), it might still
have to wait for the preemptee's migrate_disable() section to complete.
Thereby suffering interference in the exact duration of
migrate_disable() section.

Per this argument, the change from preempt_disable() to
migrate_disable() gets us:

 - higher priority tasks gain reduced wake-up latency
 - lower priority tasks are unchanged and are subject to the exact same
   interference term as if the higher priority task were using
   preempt_disable().

Since we've already established this term is unbounded, any task but the
highest priority task is basically buggered.

TL;DR, if we get balancing fixed and achieve (near) the optimal case
above, migrate_disable() is an over-all win. But it's provably
non-deterministic as long as the migrate_disable() sections are
non-deterministic.


The reason this all mostly works in practise is (I think) because:

 - People care most about the higher prio RT tasks and craft them to
   mostly avoid the migrate_disable() infected code.

 - The preemption scenario is most pronounced at higher utilization
   scenarios, and I suspect this is fairly rare to begin with.

 - And while many of these migrate_disable() regions are unbound in
   theory, in practise they're often fairly reasonable.


So my current todo list is:

 - Change RT PULL
 - Change DL PULL
 - Add migrate_disable() tracer; exactly like preempt/irqoff, except
   measuring task-runtime instead of cpu-time.
 - Add a mode that measures actual interference.
 - Add a traceevent to detect preemption in migrate_disable().


And then I suppose I should twist Daniel's arm to update his model to
include these scenarios and numbers.
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/dri-devel

  parent reply index

Thread overview: 52+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-09-19  9:17 Thomas Gleixner
2020-09-19  9:17 ` [patch RFC 01/15] mm/highmem: Un-EXPORT __kmap_atomic_idx() Thomas Gleixner
2020-09-19  9:17 ` [patch RFC 02/15] highmem: Provide generic variant of kmap_atomic* Thomas Gleixner
2020-09-19  9:17 ` [patch RFC 03/15] x86/mm/highmem: Use generic kmap atomic implementation Thomas Gleixner
2020-09-19  9:17 ` [patch RFC 04/15] arc/mm/highmem: " Thomas Gleixner
2020-09-19  9:17 ` [patch RFC 05/15] ARM: highmem: Switch to generic kmap atomic Thomas Gleixner
2020-09-19  9:17 ` [patch RFC 06/15] csky/mm/highmem: " Thomas Gleixner
2020-09-23  0:05   ` Guo Ren
2020-09-19  9:17 ` [patch RFC 07/15] microblaze/mm/highmem: " Thomas Gleixner
2020-09-19  9:17 ` [patch RFC 08/15] mips/mm/highmem: " Thomas Gleixner
2020-09-19  9:18 ` [patch RFC 09/15] nds32/mm/highmem: " Thomas Gleixner
2020-09-19  9:18 ` [patch RFC 10/15] powerpc/mm/highmem: " Thomas Gleixner
2020-09-19  9:18 ` [patch RFC 11/15] sparc/mm/highmem: " Thomas Gleixner
2020-09-19  9:18 ` [patch RFC 12/15] xtensa/mm/highmem: " Thomas Gleixner
2020-09-19  9:18 ` [patch RFC 13/15] mm/highmem: Remove the old kmap_atomic cruft Thomas Gleixner
2020-09-19  9:18 ` [patch RFC 14/15] sched: highmem: Store temporary kmaps in task struct Thomas Gleixner
2020-09-19  9:18 ` [patch RFC 15/15] mm/highmem: Provide kmap_temporary* Thomas Gleixner
2020-09-19 10:35 ` [patch RFC 00/15] mm/highmem: Provide a preemptible variant of kmap_atomic & friends Daniel Vetter
2020-09-19 10:37   ` Daniel Vetter
2020-09-20  6:23     ` Thomas Gleixner
2020-09-20  8:23       ` Daniel Vetter
2020-09-20 17:24         ` Thomas Gleixner
2020-09-19 17:18 ` Linus Torvalds
2020-09-19 17:39   ` Matthew Wilcox
2020-09-19 19:13     ` Linus Torvalds
2020-09-21 19:58     ` Ira Weiny
2020-09-20  6:41   ` Thomas Gleixner
2020-09-20  8:49     ` Thomas Gleixner
2020-09-20 16:57       ` Linus Torvalds
2020-09-20 17:40         ` Thomas Gleixner
2020-09-20 17:42           ` Linus Torvalds
2020-09-20 17:58             ` Linus Torvalds
2020-09-21  7:39             ` Thomas Gleixner
2020-09-21 16:24               ` Linus Torvalds
2020-09-21 19:27                 ` Thomas Gleixner
2020-09-23  8:40                   ` peterz
2020-09-23 13:35                     ` Thomas Gleixner
2020-09-23 15:52                     ` Steven Rostedt
2020-09-23 20:55                       ` Thomas Gleixner
2020-09-23 21:12                         ` Steven Rostedt
2020-09-24  6:57                           ` Thomas Gleixner
2020-09-24 12:32                             ` Steven Rostedt
2020-09-24 12:42                               ` Peter Zijlstra
2020-09-24 13:51                                 ` Steven Rostedt
2020-09-24 13:58                                   ` Peter Zijlstra
2020-09-24 17:55                               ` Thomas Gleixner
2020-09-24 18:58                                 ` Steven Rostedt
2020-09-24  8:27                       ` peterz [this message]
2020-09-24 19:36                         ` Daniel Bristot de Oliveira
2020-09-23 10:19                   ` peterz
2020-09-23 12:33                     ` Thomas Gleixner
2020-09-23 14:33                   ` Thomas Gleixner

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20200924082717.GA1362448@hirez.programming.kicks-ass.net \
    --to=peterz@infradead.org \
    --cc=airlied@linux.ie \
    --cc=akpm@linux-foundation.org \
    --cc=ardb@kernel.org \
    --cc=arnd@arndb.de \
    --cc=bigeasy@linutronix.de \
    --cc=bristot@redhat.com \
    --cc=bsegall@google.com \
    --cc=chris@zankel.net \
    --cc=davem@davemloft.net \
    --cc=deanbo422@gmail.com \
    --cc=dietmar.eggemann@arm.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=green.hu@gmail.com \
    --cc=guoren@kernel.org \
    --cc=herbert@gondor.apana.org.au \
    --cc=intel-gfx@lists.freedesktop.org \
    --cc=jcmvbkbc@gmail.com \
    --cc=juri.lelli@redhat.com \
    --cc=linux-arch@vger.kernel.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-csky@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-mips@vger.kernel.org \
    --cc=linux-mm@kvack.org \
    --cc=linux-snps-arc@lists.infradead.org \
    --cc=linux-xtensa@linux-xtensa.org \
    --cc=linux@armlinux.org.uk \
    --cc=linuxppc-dev@lists.ozlabs.org \
    --cc=mgorman@suse.de \
    --cc=monstr@monstr.eu \
    --cc=mpe@ellerman.id.au \
    --cc=nickhu@andestech.com \
    --cc=paulmck@kernel.org \
    --cc=paulus@samba.org \
    --cc=rodrigo.vivi@intel.com \
    --cc=rostedt@goodmis.org \
    --cc=sparclinux@vger.kernel.org \
    --cc=tglx@linutronix.de \
    --cc=torvalds@linux-foundation.org \
    --cc=tsbogend@alpha.franken.de \
    --cc=vgupta@synopsys.com \
    --cc=vincent.guittot@linaro.org \
    --cc=will@kernel.org \
    --cc=x86@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

dri-devel Archive on lore.kernel.org

Archives are clonable:
	git clone --mirror https://lore.kernel.org/dri-devel/0 dri-devel/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dri-devel dri-devel/ https://lore.kernel.org/dri-devel \
		dri-devel@lists.freedesktop.org
	public-inbox-index dri-devel

Example config snippet for mirrors

Newsgroup available over NNTP:
	nntp://nntp.lore.kernel.org/org.freedesktop.lists.dri-devel


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git