From: Stephen Boyd <sboyd@codeaurora.org> To: Russell King - ARM Linux <linux@arm.linux.org.uk> Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>, Krzysztof Kozlowski <k.kozlowski@samsung.com>, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, Arnd Bergmann <arnd@arndb.de>, Mark Rutland <mark.rutland@arm.com>, Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>, Marek Szyprowski <m.szyprowski@samsung.com>, Catalin Marinas <catalin.marinas@arm.com>, Will Deacon <will.deacon@arm.com> Subject: Re: [PATCH v2] ARM: Don't use complete() during __cpu_die Date: Tue, 10 Feb 2015 10:33:59 -0800 [thread overview] Message-ID: <54DA4F17.2040705@codeaurora.org> (raw) In-Reply-To: <20150210154157.GL8656@n2100.arm.linux.org.uk> On 02/10/15 07:41, Russell King - ARM Linux wrote: > On Mon, Feb 09, 2015 at 05:24:08PM -0800, Stephen Boyd wrote: > > >> Maybe we can do the same thing here by using a >> spinlock for synchronization between the IPI handler and the dying CPU? >> So lock/unlock around the IPI sending from the dying CPU and then do a >> lock/unlock on the killing CPU before continuing. > It would be nice, but it means exporting irq_controller_lock from irq_gic.c. > It's doable, but it's really not nice - it creates a layering issue, buy > making arch/arm/kernel/smp.c depend on symbols exported from > drivers/irqchip/irq-gic.c. I wasn't talking about the irq_controller_lock. I was saying we should add another spinlock for synchronization purposes in arm/kernel/smp.c ---8<---- diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c index 02c5da16c7ed..fe0386c751b2 100644 --- a/arch/arm/kernel/smp.c +++ b/arch/arm/kernel/smp.c @@ -225,6 +225,7 @@ int __cpu_disable(void) } static DECLARE_COMPLETION(cpu_died); +static DEFINE_RAW_SPINLOCK(stop_lock); /* * called on the thread which is asking for a CPU to be shutdown - @@ -232,10 +233,13 @@ static DECLARE_COMPLETION(cpu_died); */ void __cpu_die(unsigned int cpu) { + unsigned long flags; if (!wait_for_completion_timeout(&cpu_died, msecs_to_jiffies(5000))) { pr_err("CPU%u: cpu didn't die\n", cpu); return; } + raw_spin_lock_irqsave(&stop_lock, flags); + raw_spin_unlock_irqrestore(&stop_lock, flags); pr_notice("CPU%u: shutdown\n", cpu); /* @@ -280,7 +284,17 @@ void __ref cpu_die(void) * has been sent, power and/or clocks can be removed at any point * from this CPU and its cache by platform_cpu_kill(). */ + raw_spin_lock(&stop_lock); __smp_cross_call(cpumask_of(cpumask_any(cpu_online_mask)), IPI_CPU_DEAD); + raw_spin_unlock(&stop_lock); + + /* + * Ensure that the cache lines associated with the stop_lock are + * written out. This covers the case where _this_ CPU is doing the + * powering down, to ensure that the lock is visible to the + * CPU waiting for this one. + */ + flush_cache_louis(); /* * The actual CPU shutdown procedure is at least platform (if not @@ -517,8 +531,6 @@ void tick_broadcast(const struct cpumask *mask) } #endif -static DEFINE_RAW_SPINLOCK(stop_lock); - /* * ipi_cpu_stop - handle IPI from smp_send_stop() */ -- Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
WARNING: multiple messages have this Message-ID (diff)
From: sboyd@codeaurora.org (Stephen Boyd) To: linux-arm-kernel@lists.infradead.org Subject: [PATCH v2] ARM: Don't use complete() during __cpu_die Date: Tue, 10 Feb 2015 10:33:59 -0800 [thread overview] Message-ID: <54DA4F17.2040705@codeaurora.org> (raw) In-Reply-To: <20150210154157.GL8656@n2100.arm.linux.org.uk> On 02/10/15 07:41, Russell King - ARM Linux wrote: > On Mon, Feb 09, 2015 at 05:24:08PM -0800, Stephen Boyd wrote: > > >> Maybe we can do the same thing here by using a >> spinlock for synchronization between the IPI handler and the dying CPU? >> So lock/unlock around the IPI sending from the dying CPU and then do a >> lock/unlock on the killing CPU before continuing. > It would be nice, but it means exporting irq_controller_lock from irq_gic.c. > It's doable, but it's really not nice - it creates a layering issue, buy > making arch/arm/kernel/smp.c depend on symbols exported from > drivers/irqchip/irq-gic.c. I wasn't talking about the irq_controller_lock. I was saying we should add another spinlock for synchronization purposes in arm/kernel/smp.c ---8<---- diff --git a/arch/arm/kernel/smp.c b/arch/arm/kernel/smp.c index 02c5da16c7ed..fe0386c751b2 100644 --- a/arch/arm/kernel/smp.c +++ b/arch/arm/kernel/smp.c @@ -225,6 +225,7 @@ int __cpu_disable(void) } static DECLARE_COMPLETION(cpu_died); +static DEFINE_RAW_SPINLOCK(stop_lock); /* * called on the thread which is asking for a CPU to be shutdown - @@ -232,10 +233,13 @@ static DECLARE_COMPLETION(cpu_died); */ void __cpu_die(unsigned int cpu) { + unsigned long flags; if (!wait_for_completion_timeout(&cpu_died, msecs_to_jiffies(5000))) { pr_err("CPU%u: cpu didn't die\n", cpu); return; } + raw_spin_lock_irqsave(&stop_lock, flags); + raw_spin_unlock_irqrestore(&stop_lock, flags); pr_notice("CPU%u: shutdown\n", cpu); /* @@ -280,7 +284,17 @@ void __ref cpu_die(void) * has been sent, power and/or clocks can be removed at any point * from this CPU and its cache by platform_cpu_kill(). */ + raw_spin_lock(&stop_lock); __smp_cross_call(cpumask_of(cpumask_any(cpu_online_mask)), IPI_CPU_DEAD); + raw_spin_unlock(&stop_lock); + + /* + * Ensure that the cache lines associated with the stop_lock are + * written out. This covers the case where _this_ CPU is doing the + * powering down, to ensure that the lock is visible to the + * CPU waiting for this one. + */ + flush_cache_louis(); /* * The actual CPU shutdown procedure is@least platform (if not @@ -517,8 +531,6 @@ void tick_broadcast(const struct cpumask *mask) } #endif -static DEFINE_RAW_SPINLOCK(stop_lock); - /* * ipi_cpu_stop - handle IPI from smp_send_stop() */ -- Qualcomm Innovation Center, Inc. is a member of Code Aurora Forum, a Linux Foundation Collaborative Project
next prev parent reply other threads:[~2015-02-10 18:34 UTC|newest] Thread overview: 92+ messages / expand[flat|nested] mbox.gz Atom feed top 2015-02-05 10:14 [PATCH v2] ARM: Don't use complete() during __cpu_die Krzysztof Kozlowski 2015-02-05 10:14 ` Krzysztof Kozlowski 2015-02-05 10:50 ` Russell King - ARM Linux 2015-02-05 10:50 ` Russell King - ARM Linux 2015-02-05 11:00 ` Krzysztof Kozlowski 2015-02-05 11:00 ` Krzysztof Kozlowski 2015-02-05 11:08 ` Russell King - ARM Linux 2015-02-05 11:08 ` Russell King - ARM Linux 2015-02-05 11:28 ` Mark Rutland 2015-02-05 11:28 ` Mark Rutland 2015-02-05 11:30 ` Russell King - ARM Linux 2015-02-05 11:30 ` Russell King - ARM Linux 2015-02-05 14:29 ` Paul E. McKenney 2015-02-05 14:29 ` Paul E. McKenney 2015-02-05 16:11 ` Russell King - ARM Linux 2015-02-05 16:11 ` Russell King - ARM Linux 2015-02-05 17:02 ` Paul E. McKenney 2015-02-05 17:02 ` Paul E. McKenney 2015-02-05 17:34 ` Russell King - ARM Linux 2015-02-05 17:34 ` Russell King - ARM Linux 2015-02-05 17:54 ` Paul E. McKenney 2015-02-05 17:54 ` Paul E. McKenney 2015-02-10 1:24 ` Stephen Boyd 2015-02-10 1:24 ` Stephen Boyd 2015-02-10 1:37 ` Paul E. McKenney 2015-02-10 1:37 ` Paul E. McKenney 2015-02-10 2:05 ` Stephen Boyd 2015-02-10 2:05 ` Stephen Boyd 2015-02-10 3:05 ` Paul E. McKenney 2015-02-10 3:05 ` Paul E. McKenney 2015-02-10 15:14 ` Mark Rutland 2015-02-10 15:14 ` Mark Rutland 2015-02-10 20:48 ` Stephen Boyd 2015-02-10 20:48 ` Stephen Boyd 2015-02-10 21:04 ` Stephen Boyd 2015-02-10 21:04 ` Stephen Boyd 2015-02-10 21:15 ` Russell King - ARM Linux 2015-02-10 21:15 ` Russell King - ARM Linux 2015-02-10 21:49 ` Stephen Boyd 2015-02-10 21:49 ` Stephen Boyd 2015-02-10 22:05 ` Stephen Boyd 2015-02-10 22:05 ` Stephen Boyd 2015-02-13 15:52 ` Mark Rutland 2015-02-13 15:52 ` Mark Rutland 2015-02-13 16:27 ` Russell King - ARM Linux 2015-02-13 16:27 ` Russell King - ARM Linux 2015-02-13 17:21 ` Mark Rutland 2015-02-13 17:21 ` Mark Rutland 2015-02-13 17:30 ` Russell King - ARM Linux 2015-02-13 17:30 ` Russell King - ARM Linux 2015-02-13 16:28 ` Stephen Boyd 2015-02-13 16:28 ` Stephen Boyd 2015-02-13 15:38 ` Mark Rutland 2015-02-13 15:38 ` Mark Rutland 2015-02-10 20:58 ` Russell King - ARM Linux 2015-02-10 20:58 ` Russell King - ARM Linux 2015-02-10 15:41 ` Russell King - ARM Linux 2015-02-10 15:41 ` Russell King - ARM Linux 2015-02-10 18:33 ` Stephen Boyd [this message] 2015-02-10 18:33 ` Stephen Boyd 2015-02-25 12:56 ` Russell King - ARM Linux 2015-02-25 12:56 ` Russell King - ARM Linux 2015-02-25 16:47 ` Nicolas Pitre 2015-02-25 16:47 ` Nicolas Pitre 2015-02-25 17:00 ` Russell King - ARM Linux 2015-02-25 17:00 ` Russell King - ARM Linux 2015-02-25 18:13 ` Nicolas Pitre 2015-02-25 18:13 ` Nicolas Pitre 2015-02-25 20:16 ` Nicolas Pitre 2015-02-25 20:16 ` Nicolas Pitre 2015-02-26 1:05 ` Paul E. McKenney 2015-02-26 1:05 ` Paul E. McKenney 2015-03-22 23:30 ` Paul E. McKenney 2015-03-22 23:30 ` Paul E. McKenney 2015-03-23 12:55 ` Russell King - ARM Linux 2015-03-23 12:55 ` Russell King - ARM Linux 2015-03-23 13:21 ` Paul E. McKenney 2015-03-23 13:21 ` Paul E. McKenney 2015-03-23 14:00 ` Russell King - ARM Linux 2015-03-23 14:00 ` Russell King - ARM Linux 2015-03-23 15:37 ` Paul E. McKenney 2015-03-23 15:37 ` Paul E. McKenney 2015-03-23 16:56 ` Paul E. McKenney 2015-03-23 16:56 ` Paul E. McKenney 2015-02-26 19:14 ` Daniel Thompson 2015-02-26 19:14 ` Daniel Thompson 2015-02-26 19:47 ` Nicolas Pitre 2015-02-26 19:47 ` Nicolas Pitre 2015-02-05 10:53 ` Mark Rutland 2015-02-05 10:53 ` Mark Rutland 2015-02-05 10:59 ` Krzysztof Kozlowski 2015-02-05 10:59 ` Krzysztof Kozlowski
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=54DA4F17.2040705@codeaurora.org \ --to=sboyd@codeaurora.org \ --cc=arnd@arndb.de \ --cc=b.zolnierkie@samsung.com \ --cc=catalin.marinas@arm.com \ --cc=k.kozlowski@samsung.com \ --cc=linux-arm-kernel@lists.infradead.org \ --cc=linux-kernel@vger.kernel.org \ --cc=linux@arm.linux.org.uk \ --cc=m.szyprowski@samsung.com \ --cc=mark.rutland@arm.com \ --cc=paulmck@linux.vnet.ibm.com \ --cc=will.deacon@arm.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: linkBe 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.