All of lore.kernel.org
 help / color / mirror / Atom feed
From: Brian Norris <computersforpeace@gmail.com>
To: Russell King - ARM Linux <linux@arm.linux.org.uk>
Cc: Rob Herring <robherring2@gmail.com>,
	Arnd Bergmann <arnd@arndb.de>, Olof Johansson <olof@lixom.net>,
	"devicetree@vger.kernel.org" <devicetree@vger.kernel.org>,
	Florian Fainelli <f.fainelli@gmail.com>,
	Christian Daudt <bcm@fixthebug.org>,
	Linux Kernel <linux-kernel@vger.kernel.org>,
	Matt Porter <mporter@linaro.org>,
	Marc Carino <marc.ceeeee@gmail.com>,
	Gregory Fong <gregory.0xf0@gmail.com>,
	"linux-arm-kernel@lists.infradead.org" 
	<linux-arm-kernel@lists.infradead.org>
Subject: Re: [PATCH v8 01/11] ARM: brcmstb: add infrastructure for ARM-based Broadcom STB SoCs
Date: Wed, 13 Aug 2014 16:47:06 -0700	[thread overview]
Message-ID: <20140813234706.GC18411@ld-irv-0074> (raw)
In-Reply-To: <20140802092756.GZ30282@n2100.arm.linux.org.uk>

Hi Russell,

Picking up this thread again, as things are now set for dropping this
patch and resubmitting SMP support for 3.18.

On Sat, Aug 02, 2014 at 10:27:56AM +0100, Russell King wrote:
> On Thu, Jul 31, 2014 at 03:06:42PM -0700, Brian Norris wrote:
> > Yes, I noticed this. What I meant is that smp_ops.cpu_die() and
> > smp_ops.cpu_kill() are not synchronized.
> ...
> > We're not relying on the L1 cache, though. Don't sync_cache_{r,w}()
> > ensure all reads/writes reach at least the L2?
> 
> What exactly are you trying to achieve here?

Synchronization between v7_exit_coherency_flush() (on the dying CPU) and
yanking the power (brcmstb_cpu_kill(), on the murderous CPU). The core
completion-based synchronization is not sufficient, since it allows
brcmstb_smp_ops.smp_kill and brcmstb_smp_ops.smp_die to race.

Am I somehow not achieving what I intend here?

> > How does that ensure that the CPU is down by the time the work is
> > scheduled? It seems like this would just defer the work long enough that
> > it *most likely* has quiesced, but I don't see how this gives us a
> > better guarantee. Or maybe I'm missing something. (If so, please do
> > enlighten!)
> 
> Note that I said a delayed work queue.  The dying CPU runs a predictable
> sequence once cpu_die() has been entered - interrupts at the GIC have
> been programmed to be routed to other CPUs, interrupts at the CPU are
> masked, so the CPU isn't going to be doing anything else except executing
> that code path.  So, it's going to be a predictable number of CPU cycles.
> 
> That allows you to arrange a delayed workqueue (or a timer) to fire
> after a period of time that you can guarantee that the dying CPU has
> reached that wfi().

OK, that sounds workable for the active hotplug case.

But what about for the suspend case? CPUs are hot-unplugged during
disable_nonboot_cpus(), and I don't see that this would guarantee the
workqueue will complete before we enter suspend.

> Another point which raises itself in your patch is this:
> 
> +       /* Settle-time from Broadcom-internal DVT reference code */
> +       udelay(7);
> 
> 7us looks very precise, but udelay() may not be that precise.  What is
> the actual specification?  Does it say "you must wait at least 7us"?
> 
> udelay() _may_ return early, especially if it is using the CPU delay
> loop to perform the delay - I've explained why this happens previously,
> and why it isn't a bug.
> 
> If you're using a timer-based delay for udelay() (which you should be
> using if you support cpufreq) then the delay should be more accurate,
> but it's still good practise to give a little leeway on the figure.

I'm looking into this specific delay. I'd bet it's just "wait at least
7us." I could probably factor in some leeway to be safe.

Thanks,
Brian

WARNING: multiple messages have this Message-ID (diff)
From: computersforpeace@gmail.com (Brian Norris)
To: linux-arm-kernel@lists.infradead.org
Subject: [PATCH v8 01/11] ARM: brcmstb: add infrastructure for ARM-based Broadcom STB SoCs
Date: Wed, 13 Aug 2014 16:47:06 -0700	[thread overview]
Message-ID: <20140813234706.GC18411@ld-irv-0074> (raw)
In-Reply-To: <20140802092756.GZ30282@n2100.arm.linux.org.uk>

Hi Russell,

Picking up this thread again, as things are now set for dropping this
patch and resubmitting SMP support for 3.18.

On Sat, Aug 02, 2014 at 10:27:56AM +0100, Russell King wrote:
> On Thu, Jul 31, 2014 at 03:06:42PM -0700, Brian Norris wrote:
> > Yes, I noticed this. What I meant is that smp_ops.cpu_die() and
> > smp_ops.cpu_kill() are not synchronized.
> ...
> > We're not relying on the L1 cache, though. Don't sync_cache_{r,w}()
> > ensure all reads/writes reach at least the L2?
> 
> What exactly are you trying to achieve here?

Synchronization between v7_exit_coherency_flush() (on the dying CPU) and
yanking the power (brcmstb_cpu_kill(), on the murderous CPU). The core
completion-based synchronization is not sufficient, since it allows
brcmstb_smp_ops.smp_kill and brcmstb_smp_ops.smp_die to race.

Am I somehow not achieving what I intend here?

> > How does that ensure that the CPU is down by the time the work is
> > scheduled? It seems like this would just defer the work long enough that
> > it *most likely* has quiesced, but I don't see how this gives us a
> > better guarantee. Or maybe I'm missing something. (If so, please do
> > enlighten!)
> 
> Note that I said a delayed work queue.  The dying CPU runs a predictable
> sequence once cpu_die() has been entered - interrupts at the GIC have
> been programmed to be routed to other CPUs, interrupts at the CPU are
> masked, so the CPU isn't going to be doing anything else except executing
> that code path.  So, it's going to be a predictable number of CPU cycles.
> 
> That allows you to arrange a delayed workqueue (or a timer) to fire
> after a period of time that you can guarantee that the dying CPU has
> reached that wfi().

OK, that sounds workable for the active hotplug case.

But what about for the suspend case? CPUs are hot-unplugged during
disable_nonboot_cpus(), and I don't see that this would guarantee the
workqueue will complete before we enter suspend.

> Another point which raises itself in your patch is this:
> 
> +       /* Settle-time from Broadcom-internal DVT reference code */
> +       udelay(7);
> 
> 7us looks very precise, but udelay() may not be that precise.  What is
> the actual specification?  Does it say "you must wait at least 7us"?
> 
> udelay() _may_ return early, especially if it is using the CPU delay
> loop to perform the delay - I've explained why this happens previously,
> and why it isn't a bug.
> 
> If you're using a timer-based delay for udelay() (which you should be
> using if you support cpufreq) then the delay should be more accurate,
> but it's still good practise to give a little leeway on the figure.

I'm looking into this specific delay. I'd bet it's just "wait at least
7us." I could probably factor in some leeway to be safe.

Thanks,
Brian

  reply	other threads:[~2014-08-13 23:47 UTC|newest]

Thread overview: 121+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-07-21 21:07 [PATCH v8 00/11] ARM: brcmstb: Add Broadcom STB SoC support Brian Norris
2014-07-21 21:07 ` Brian Norris
2014-07-21 21:07 ` Brian Norris
2014-07-21 21:07 ` [PATCH v8 01/11] ARM: brcmstb: add infrastructure for ARM-based Broadcom STB SoCs Brian Norris
2014-07-21 21:07   ` Brian Norris
2014-07-21 21:07   ` Brian Norris
2014-07-30  9:26   ` Russell King - ARM Linux
2014-07-30  9:26     ` Russell King - ARM Linux
2014-07-31  2:36     ` Brian Norris
2014-07-31  2:36       ` Brian Norris
2014-07-31  2:36       ` Brian Norris
2014-07-30 17:09   ` Rob Herring
2014-07-30 17:09     ` Rob Herring
2014-07-30 17:09     ` Rob Herring
2014-07-31  2:23     ` Brian Norris
2014-07-31  2:23       ` Brian Norris
2014-07-31  2:23       ` Brian Norris
2014-07-31  8:43       ` Russell King - ARM Linux
2014-07-31  8:43         ` Russell King - ARM Linux
2014-07-31  8:43         ` Russell King - ARM Linux
2014-07-31 22:06         ` Brian Norris
2014-07-31 22:06           ` Brian Norris
2014-07-31 22:06           ` Brian Norris
2014-08-02  9:27           ` Russell King - ARM Linux
2014-08-02  9:27             ` Russell King - ARM Linux
2014-08-02  9:27             ` Russell King - ARM Linux
2014-08-13 23:47             ` Brian Norris [this message]
2014-08-13 23:47               ` Brian Norris
2014-08-13 23:47               ` Brian Norris
2014-08-19  0:02               ` Brian Norris
2014-08-19  0:02                 ` Brian Norris
2014-08-19  0:02                 ` Brian Norris
2014-08-01 14:33       ` Rob Herring
2014-08-01 14:33         ` Rob Herring
2014-08-01 14:33         ` Rob Herring
2014-08-01 19:29         ` Florian Fainelli
2014-08-01 19:29           ` Florian Fainelli
2014-08-01 19:29           ` Florian Fainelli
2014-08-01 19:46           ` Matt Porter
2014-08-01 19:46             ` Matt Porter
2014-08-01 19:46             ` Matt Porter
2014-08-02  8:21         ` Russell King - ARM Linux
2014-08-02  8:21           ` Russell King - ARM Linux
2014-08-02  8:21           ` Russell King - ARM Linux
2014-08-02  8:19   ` Russell King - ARM Linux
2014-08-02  8:19     ` Russell King - ARM Linux
2014-08-04 17:39     ` Brian Norris
2014-08-04 17:39       ` Brian Norris
2014-07-21 21:07 ` [PATCH v8 02/11] power: reset: Add reboot driver for brcmstb Brian Norris
2014-07-21 21:07   ` Brian Norris
2014-07-21 21:07   ` Brian Norris
2014-07-22  7:28   ` Arnd Bergmann
2014-07-22  7:28     ` Arnd Bergmann
2014-07-22  7:28     ` Arnd Bergmann
2014-07-22 20:02     ` Brian Norris
2014-07-22 20:02       ` Brian Norris
2014-07-22 20:02       ` Brian Norris
2014-07-22 21:02       ` Arnd Bergmann
2014-07-22 21:02         ` Arnd Bergmann
2014-07-22 21:02         ` Arnd Bergmann
2014-07-22 22:51         ` Brian Norris
2014-07-22 22:51           ` Brian Norris
2014-07-21 21:07 ` [PATCH v8 03/11] ARM: brcmstb: add debug UART for earlyprintk support Brian Norris
2014-07-21 21:07   ` Brian Norris
2014-07-21 21:07   ` Brian Norris
2014-08-02  8:30   ` Russell King - ARM Linux
2014-08-02  8:30     ` Russell King - ARM Linux
2014-08-04 16:56     ` Brian Norris
2014-08-04 16:56       ` Brian Norris
2014-08-13 22:11       ` Brian Norris
2014-08-13 22:11         ` Brian Norris
2014-08-13 22:11         ` Brian Norris
2014-08-13 22:16         ` Olof Johansson
2014-08-13 22:16           ` Olof Johansson
2014-08-13 22:16           ` Olof Johansson
2014-09-02 22:22           ` Florian Fainelli
2014-09-02 22:22             ` Florian Fainelli
2014-09-02 22:22             ` Florian Fainelli
2014-09-02 22:44             ` Brian Norris
2014-09-02 22:44               ` Brian Norris
2014-09-02 22:44               ` Brian Norris
2014-07-21 21:07 ` [PATCH v8 04/11] ARM: do CPU-specific init for Broadcom Brahma15 cores Brian Norris
2014-07-21 21:07   ` Brian Norris
2014-07-21 21:07   ` Brian Norris
2014-07-21 21:08 ` [PATCH v8 05/11] ARM: Enable erratum 798181 for Broadcom Brahma-B15 Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08 ` [PATCH v8 06/11] ARM: brcmstb: add CPU binding for Broadcom Brahma15 Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08 ` [PATCH v8 07/11] ARM: brcmstb: add misc. DT bindings for brcmstb Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08 ` [PATCH v8 08/11] ARM: brcmstb: gic: add compatible string for Broadcom Brahma15 Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08 ` [PATCH v8 09/11] ARM: brcmstb: dts: add a reference DTS for Broadcom 7445 Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08 ` [PATCH v8 10/11] ARM: brcmstb: select GISB arbiter and interrupt drivers Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08 ` [PATCH v8 11/11] MAINTAINERS: add entry for Broadcom ARM STB architecture Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-21 21:08   ` Brian Norris
2014-07-22  7:35 ` [PATCH v8 00/11] ARM: brcmstb: Add Broadcom STB SoC support Arnd Bergmann
2014-07-22  7:35   ` Arnd Bergmann
2014-07-22  7:35   ` Arnd Bergmann
2014-07-22 20:44   ` Brian Norris
2014-07-22 20:44     ` Brian Norris
2014-07-22 20:57     ` Arnd Bergmann
2014-07-22 20:57       ` Arnd Bergmann
2014-07-22 20:57       ` Arnd Bergmann
2014-07-22 21:33       ` Matt Porter
2014-07-22 21:33         ` Matt Porter
2014-07-22 21:33         ` Matt Porter
2014-07-22 22:24         ` Arnd Bergmann
2014-07-22 22:24           ` Arnd Bergmann
2014-07-22 22:24           ` Arnd Bergmann
2014-07-22 22:30           ` Brian Norris
2014-07-22 22:30             ` Brian Norris

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=20140813234706.GC18411@ld-irv-0074 \
    --to=computersforpeace@gmail.com \
    --cc=arnd@arndb.de \
    --cc=bcm@fixthebug.org \
    --cc=devicetree@vger.kernel.org \
    --cc=f.fainelli@gmail.com \
    --cc=gregory.0xf0@gmail.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux@arm.linux.org.uk \
    --cc=marc.ceeeee@gmail.com \
    --cc=mporter@linaro.org \
    --cc=olof@lixom.net \
    --cc=robherring2@gmail.com \
    /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
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.