LKML Archive on lore.kernel.org
help / color / mirror / Atom feed
From: Daniel Thompson <daniel.thompson@linaro.org>
To: Nicolas Pitre <nicolas.pitre@linaro.org>
Cc: Thomas Gleixner <tglx@linutronix.de>,
	Jason Cooper <jason@lakedaemon.net>,
	Russell King <linux@arm.linux.org.uk>,
	Will Deacon <will.deacon@arm.com>,
	Catalin Marinas <catalin.marinas@arm.com>,
	Marc Zyngier <marc.zyngier@arm.com>,
	Stephen Boyd <sboyd@codeaurora.org>,
	John Stultz <john.stultz@linaro.org>,
	Steven Rostedt <rostedt@goodmis.org>,
	linux-kernel@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org, patches@linaro.org,
	linaro-kernel@lists.linaro.org,
	Sumit Semwal <sumit.semwal@linaro.org>,
	Dirk Behme <dirk.behme@de.bosch.com>,
	Daniel Drake <drake@endlessm.com>,
	Dmitry Pervushin <dpervushin@gmail.com>,
	Tim Sander <tim@krieglstein.org>
Subject: Re: [PATCH 3.19-rc6 v16 1/6] irqchip: gic: Optimize locking in gic_raise_softirq
Date: Thu, 26 Feb 2015 21:05:40 +0000	[thread overview]
Message-ID: <1424984740.21020.11.camel@linaro.org> (raw)
In-Reply-To: <alpine.LFD.2.11.1502261449380.25484@knanqh.ubzr>

On Thu, 2015-02-26 at 15:31 -0500, Nicolas Pitre wrote:
> On Tue, 3 Feb 2015, Daniel Thompson wrote:
> 
> > Currently gic_raise_softirq() is locked using upon irq_controller_lock.
> > This lock is primarily used to make register read-modify-write sequences
> > atomic but gic_raise_softirq() uses it instead to ensure that the
> > big.LITTLE migration logic can figure out when it is safe to migrate
> > interrupts between physical cores.
> > 
> > This is sub-optimal in closely related ways:
> > 
> > 1. No locking at all is required on systems where the b.L switcher is
> >    not configured.
> 
> ACK
> 
> > 2. Finer grain locking can be used on systems where the b.L switcher is
> >    present.
> 
> NAK
> 
> Consider this sequence:
> 
> 	CPU 1				CPU 2
> 	-----				-----
> 	gic_raise_softirq()		gic_migrate_target()
> 	bl_migration_lock() [OK]
> 	[...]				[...]
> 	map |= gic_cpu_map[cpu];	bl_migration_lock() [contended]
> 	bl_migration_unlock(flags);	bl_migration_lock() [OK]
> 					gic_cpu_map[cpu] = 1 << new_cpu_id;
> 					bl_migration_unlock(flags);
> 					[...]
> 					(migrate pending IPI from old CPU)
> 	writel_relaxed(map to GIC_DIST_SOFTINT);

Isn't this solved inside gic_raise_softirq? How can the writel_relaxed()
escape from the critical section and happen at the end of the sequence?


> 	[this IPI is now lost]
> 
> Granted, this race is apparently aready possible today.  We probably get 
> away with it because the locked sequence in gic_migrate_target() include 
> the retargetting of peripheral interrupts which gives plenti of time for 
> code execution in gic_raise_softirq() to post its IPI before the IPI 
> migration code is executed.  So in that sense it could be argued that 
> the reduced lock coverage from your patch doesn't make things any worse.  
> If anything it might even help by letting gic_migrate_target() complete 
> sooner.  But removing cpu_map_migration_lock altogether would improve 
> things even further by that logic.  I however don't think we should live 
> so dangerously.
> 
> Therefore, for the lock to be effective, it has to encompass the 
> changing of the CPU map _and_ migration of pending IPIs before new IPIs 
> are allowed again.  That means the locked area has to grow not shrink.
> 
> Oh, and a minor nit:
> 
> > + * This lock is used by the big.LITTLE migration code to ensure no IPIs
> > + * can be pended on the old core after the map has been updated.
> > + */
> > +#ifdef CONFIG_BL_SWITCHER
> > +static DEFINE_RAW_SPINLOCK(cpu_map_migration_lock);
> > +
> > +static inline void bl_migration_lock(unsigned long *flags)
> 
> Please name it gic_migration_lock. "bl_migration_lock" is a bit too 
> generic in this context.

I'll change this.

Daniel.



  reply	other threads:[~2015-02-26 21:05 UTC|newest]

Thread overview: 47+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-01-23 14:22 [PATCH 3.19-rc2 v15 0/8] irq/arm: Implement arch_trigger_all_cpu_backtrace Daniel Thompson
2015-01-23 14:22 ` [PATCH 3.19-rc2 v15 1/8] irqchip: gic: Optimize locking in gic_raise_softirq Daniel Thompson
2015-01-23 14:22 ` [PATCH 3.19-rc2 v15 2/8] irqchip: gic: Make gic_raise_softirq FIQ-safe Daniel Thompson
2015-01-23 14:22 ` [PATCH 3.19-rc2 v15 3/8] irqchip: gic: Introduce plumbing for IPI FIQ Daniel Thompson
2015-01-23 14:22 ` [PATCH 3.19-rc2 v15 4/8] sched_clock: Avoid deadlock during read from NMI Daniel Thompson
2015-01-24 22:40   ` Thomas Gleixner
2015-01-26 20:28     ` Daniel Thompson
2015-01-23 14:22 ` [PATCH 3.19-rc2 v15 5/8] printk: Simple implementation for NMI backtracing Daniel Thompson
2015-01-24 21:44   ` Thomas Gleixner
2015-01-26 17:21     ` Daniel Thompson
2015-01-23 14:22 ` [PATCH 3.19-rc2 v15 6/8] x86/nmi: Use common printk functions Daniel Thompson
2015-01-23 14:22 ` [PATCH 3.19-rc2 v15 7/8] ARM: Add support for on-demand backtrace of other CPUs Daniel Thompson
2015-01-23 14:22 ` [PATCH 3.19-rc2 v15 8/8] ARM: Fix on-demand backtrace triggered by IRQ Daniel Thompson
2015-02-03 19:06 ` [PATCH 3.19-rc6 v16 0/6] irq/arm: Implement arch_trigger_all_cpu_backtrace Daniel Thompson
2015-02-03 19:06   ` [PATCH 3.19-rc6 v16 1/6] irqchip: gic: Optimize locking in gic_raise_softirq Daniel Thompson
2015-02-26 20:31     ` Nicolas Pitre
2015-02-26 21:05       ` Daniel Thompson [this message]
2015-02-26 21:33         ` Nicolas Pitre
2015-02-03 19:06   ` [PATCH 3.19-rc6 v16 2/6] irqchip: gic: Make gic_raise_softirq FIQ-safe Daniel Thompson
2015-02-26 20:33     ` Nicolas Pitre
2015-02-03 19:06   ` [PATCH 3.19-rc6 v16 3/6] irqchip: gic: Introduce plumbing for IPI FIQ Daniel Thompson
2015-02-03 19:06   ` [PATCH 3.19-rc6 v16 4/6] printk: Simple implementation for NMI backtracing Daniel Thompson
2015-02-03 19:06   ` [PATCH 3.19-rc6 v16 5/6] x86/nmi: Use common printk functions Daniel Thompson
2015-02-03 19:06   ` [PATCH 3.19-rc6 v16 6/6] ARM: Add support for on-demand backtrace of other CPUs Daniel Thompson
2015-03-04 10:12 ` [PATCH 4.0-rc1 v17 0/6] irq/arm: Implement arch_trigger_all_cpu_backtrace Daniel Thompson
2015-03-04 10:12   ` [PATCH 4.0-rc1 v17 1/6] irqchip: gic: Optimize locking in gic_raise_softirq Daniel Thompson
2015-03-04 10:12   ` [PATCH 4.0-rc1 v17 2/6] irqchip: gic: Make gic_raise_softirq FIQ-safe Daniel Thompson
2015-03-04 10:12   ` [PATCH 4.0-rc1 v17 3/6] irqchip: gic: Introduce plumbing for IPI FIQ Daniel Thompson
2015-03-04 10:12   ` [PATCH 4.0-rc1 v17 4/6] printk: Simple implementation for NMI backtracing Daniel Thompson
2015-03-04 16:13     ` Joe Perches
2015-03-04 16:20       ` Steven Rostedt
2015-03-04 16:33         ` Daniel Thompson
2015-03-04 17:21           ` Joe Perches
2015-03-05 12:11             ` Daniel Thompson
2015-03-04 10:12   ` [PATCH 4.0-rc1 v17 5/6] x86/nmi: Use common printk functions Daniel Thompson
2015-03-05  0:54     ` Ingo Molnar
2015-03-05 12:29       ` Daniel Thompson
2015-03-05 19:46         ` Ingo Molnar
2015-03-06 19:02           ` Daniel Thompson
2015-03-04 10:12   ` [PATCH 4.0-rc1 v17 6/6] ARM: Add support for on-demand backtrace of other CPUs Daniel Thompson
2015-03-12 13:39 ` [PATCH 4.0-rc2 v18 0/6] irq/arm: Implement arch_trigger_all_cpu_backtrace Daniel Thompson
2015-03-12 13:39   ` [PATCH 4.0-rc2 v18 1/6] irqchip: gic: Optimize locking in gic_raise_softirq Daniel Thompson
2015-03-12 13:39   ` [PATCH 4.0-rc2 v18 2/6] irqchip: gic: Make gic_raise_softirq FIQ-safe Daniel Thompson
2015-03-12 13:39   ` [PATCH 4.0-rc2 v18 3/6] irqchip: gic: Introduce plumbing for IPI FIQ Daniel Thompson
2015-03-12 13:39   ` [PATCH 4.0-rc2 v18 4/6] printk: Simple implementation for NMI backtracing Daniel Thompson
2015-03-12 13:39   ` [PATCH 4.0-rc2 v18 5/6] x86/nmi: Use common printk functions Daniel Thompson
2015-03-12 13:39   ` [PATCH 4.0-rc2 v18 6/6] ARM: Add support for on-demand backtrace of other CPUs Daniel Thompson

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=1424984740.21020.11.camel@linaro.org \
    --to=daniel.thompson@linaro.org \
    --cc=catalin.marinas@arm.com \
    --cc=dirk.behme@de.bosch.com \
    --cc=dpervushin@gmail.com \
    --cc=drake@endlessm.com \
    --cc=jason@lakedaemon.net \
    --cc=john.stultz@linaro.org \
    --cc=linaro-kernel@lists.linaro.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux@arm.linux.org.uk \
    --cc=marc.zyngier@arm.com \
    --cc=nicolas.pitre@linaro.org \
    --cc=patches@linaro.org \
    --cc=rostedt@goodmis.org \
    --cc=sboyd@codeaurora.org \
    --cc=sumit.semwal@linaro.org \
    --cc=tglx@linutronix.de \
    --cc=tim@krieglstein.org \
    --cc=will.deacon@arm.com \
    --subject='Re: [PATCH 3.19-rc6 v16 1/6] irqchip: gic: Optimize locking in gic_raise_softirq' \
    /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

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).