From: Russell King - ARM Linux <linux@arm.linux.org.uk> To: Colin Cross <ccross@android.com> Cc: linux-tegra@vger.kernel.org, konkers@android.com, Gary King <gking@nvidia.com>, linux-kernel@vger.kernel.org, olof@lixom.net, linux-arm-kernel@lists.infradead.org Subject: Re: [PATCH v2 13/28] ARM: tegra: Add suspend and hotplug support Date: Mon, 24 Jan 2011 09:07:08 +0000 [thread overview] Message-ID: <20110124090708.GA16202@n2100.arm.linux.org.uk> (raw) In-Reply-To: <1295834493-5019-14-git-send-email-ccross@android.com> On Sun, Jan 23, 2011 at 06:01:18PM -0800, Colin Cross wrote: > Tegra supports three low power modes that involve powering down the CPU. It is a common misconception that on CPU hot-unplug you need to save the entire universe of the CPU going down. When CPUs are hot-plugged, they reinitialize, which includes resetting most of the CPU state. There is no point saving the user register state, svc register state, abort register state, interrupt register state, undefined mode register state, TTB base registers, TTB control register, memory type remapping registers nor the context register. These are all reinitialized when the CPU is brought up again. VFP state should already be saved - when a CPU is hot-unplugged, we can't leave a threads VFP register set stored in that hardware as it would become inaccessible, so there's no point saving VFP state. Recently code was added to invalidate the software VFP state for a CPU being hot-unplugged and reinitialize the VFP hardware on CPU hotplug. We also have the necessary hooks into the PM code to deal with VFP state saving/restore on PM suspend/resume. The breakpoint/watchdog stuff _should_ be handled by the breakpoint/ watchdog code, not by each individual platform. Please can you work with Will Deacon if this is not the case to ensure that the breakpoint/ watchdog code can handle CPU hotplug. Lastly, L2 cache controller should have suspend/resume support added. We _really_ need to stop having platforms manually saving lots of CPU state for suspend/hot unplug. That should reduce the amount of state you're saving down to something more manageable. > + > +#ifdef CONFIG_HOTPLUG_CPU > + > +int platform_cpu_kill(unsigned int cpu) > +{ > + unsigned int reg; > + > + do { > + reg = readl(CLK_RST_CONTROLLER_RST_CPU_CMPLX_SET); > + cpu_relax(); > + } while (!(reg & (1<<cpu))); > + > + spin_lock(&boot_lock); > + reg = readl(CLK_RST_CONTROLLER_CLK_CPU_CMPLX); > + writel(reg | (1<<(8+cpu)), CLK_RST_CONTROLLER_CLK_CPU_CMPLX); > + spin_unlock(&boot_lock); > + > + return 1; > +} > + > +void platform_cpu_die(unsigned int cpu) > +{ > +#ifdef DEBUG > + unsigned int this_cpu = hard_smp_processor_id(); > + > + if (cpu != this_cpu) { > + printk(KERN_CRIT "Eek! platform_cpu_die running on %u, should be %u\n", > + this_cpu, cpu); > + BUG(); > + } > +#endif > + > + gic_cpu_exit(0); > + barrier(); > + flush_cache_all(); > + barrier(); > + __cortex_a9_save(0); > + > + /* > + * __cortex_a9_save can return through __cortex_a9_restore, but that > + * should never happen for a hotplugged cpu > + */ > + BUG(); > +} > + > +int platform_cpu_disable(unsigned int cpu) > +{ > + /* > + * we don't allow CPU 0 to be shutdown (it is still too special > + * e.g. clock tick interrupts) > + */ > + if (unlikely(!tegra_context_area)) > + return -ENXIO; > + > + return cpu == 0 ? -EPERM : 0; > +} > +#endif Please leave this in hotplug.c so that it matches what other platforms do. Otherwise it's likely to be missed by any updates that happen architecture-wide.
WARNING: multiple messages have this Message-ID (diff)
From: linux@arm.linux.org.uk (Russell King - ARM Linux) To: linux-arm-kernel@lists.infradead.org Subject: [PATCH v2 13/28] ARM: tegra: Add suspend and hotplug support Date: Mon, 24 Jan 2011 09:07:08 +0000 [thread overview] Message-ID: <20110124090708.GA16202@n2100.arm.linux.org.uk> (raw) In-Reply-To: <1295834493-5019-14-git-send-email-ccross@android.com> On Sun, Jan 23, 2011 at 06:01:18PM -0800, Colin Cross wrote: > Tegra supports three low power modes that involve powering down the CPU. It is a common misconception that on CPU hot-unplug you need to save the entire universe of the CPU going down. When CPUs are hot-plugged, they reinitialize, which includes resetting most of the CPU state. There is no point saving the user register state, svc register state, abort register state, interrupt register state, undefined mode register state, TTB base registers, TTB control register, memory type remapping registers nor the context register. These are all reinitialized when the CPU is brought up again. VFP state should already be saved - when a CPU is hot-unplugged, we can't leave a threads VFP register set stored in that hardware as it would become inaccessible, so there's no point saving VFP state. Recently code was added to invalidate the software VFP state for a CPU being hot-unplugged and reinitialize the VFP hardware on CPU hotplug. We also have the necessary hooks into the PM code to deal with VFP state saving/restore on PM suspend/resume. The breakpoint/watchdog stuff _should_ be handled by the breakpoint/ watchdog code, not by each individual platform. Please can you work with Will Deacon if this is not the case to ensure that the breakpoint/ watchdog code can handle CPU hotplug. Lastly, L2 cache controller should have suspend/resume support added. We _really_ need to stop having platforms manually saving lots of CPU state for suspend/hot unplug. That should reduce the amount of state you're saving down to something more manageable. > + > +#ifdef CONFIG_HOTPLUG_CPU > + > +int platform_cpu_kill(unsigned int cpu) > +{ > + unsigned int reg; > + > + do { > + reg = readl(CLK_RST_CONTROLLER_RST_CPU_CMPLX_SET); > + cpu_relax(); > + } while (!(reg & (1<<cpu))); > + > + spin_lock(&boot_lock); > + reg = readl(CLK_RST_CONTROLLER_CLK_CPU_CMPLX); > + writel(reg | (1<<(8+cpu)), CLK_RST_CONTROLLER_CLK_CPU_CMPLX); > + spin_unlock(&boot_lock); > + > + return 1; > +} > + > +void platform_cpu_die(unsigned int cpu) > +{ > +#ifdef DEBUG > + unsigned int this_cpu = hard_smp_processor_id(); > + > + if (cpu != this_cpu) { > + printk(KERN_CRIT "Eek! platform_cpu_die running on %u, should be %u\n", > + this_cpu, cpu); > + BUG(); > + } > +#endif > + > + gic_cpu_exit(0); > + barrier(); > + flush_cache_all(); > + barrier(); > + __cortex_a9_save(0); > + > + /* > + * __cortex_a9_save can return through __cortex_a9_restore, but that > + * should never happen for a hotplugged cpu > + */ > + BUG(); > +} > + > +int platform_cpu_disable(unsigned int cpu) > +{ > + /* > + * we don't allow CPU 0 to be shutdown (it is still too special > + * e.g. clock tick interrupts) > + */ > + if (unlikely(!tegra_context_area)) > + return -ENXIO; > + > + return cpu == 0 ? -EPERM : 0; > +} > +#endif Please leave this in hotplug.c so that it matches what other platforms do. Otherwise it's likely to be missed by any updates that happen architecture-wide.
next prev parent reply other threads:[~2011-01-24 9:07 UTC|newest] Thread overview: 137+ messages / expand[flat|nested] mbox.gz Atom feed top 2011-01-24 2:01 [PATCH v2 00/28] Updates for Tegra support in 2.6.39 Colin Cross 2011-01-24 2:01 ` [PATCH v2 01/28] ARM: tegra: irq: Rename gic pointers to avoid conflicts Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-26 7:10 ` Colin Cross 2011-01-26 7:10 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 02/28] ARM: gic: Add functions to save and restore gic state Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-02-01 13:03 ` Russell King - ARM Linux 2011-02-01 13:03 ` Russell King - ARM Linux 2011-01-24 2:01 ` [PATCH v2 03/28] ARM: gic: Export irq chip functions Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 6:39 ` Santosh Shilimkar 2011-01-24 6:39 ` Santosh Shilimkar 2011-01-24 2:01 ` [PATCH v2 04/28] ARM: mm: cache-l2x0: Add support for re-enabling l2x0 Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-25 15:14 ` Catalin Marinas 2011-01-25 15:14 ` Catalin Marinas 2011-01-25 15:41 ` Russell King - ARM Linux 2011-01-25 15:41 ` Russell King - ARM Linux 2011-01-25 18:14 ` Catalin Marinas 2011-01-25 18:14 ` Catalin Marinas 2011-01-25 18:32 ` Santosh Shilimkar 2011-01-25 18:32 ` Santosh Shilimkar 2011-01-25 18:39 ` Will Deacon [not found] ` <-8932138696981683633@unknownmsgid> 2011-02-04 23:32 ` Colin Cross 2011-02-04 23:32 ` Colin Cross 2011-02-04 23:32 ` Colin Cross 2011-02-04 23:43 ` Russell King - ARM Linux 2011-02-04 23:43 ` Russell King - ARM Linux [not found] ` <20110204234331.GF8732-l+eeeJia6m9vn6HldHNs0ANdhmdF6hFW@public.gmane.org> 2011-02-05 1:44 ` Colin Cross 2011-02-05 1:44 ` Colin Cross 2011-02-05 1:44 ` Colin Cross [not found] ` <AANLkTi=fHnivHXHnYrQvdP6JWbEA3t1X3DuBxj5gN3H0-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> 2011-02-05 7:51 ` Santosh Shilimkar 2011-02-05 7:51 ` Santosh Shilimkar 2011-02-05 7:51 ` Santosh Shilimkar [not found] ` <1bebe4b5c8590059b70a146d5486fa6a-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> 2011-02-05 9:47 ` Russell King - ARM Linux 2011-02-05 9:47 ` Russell King - ARM Linux 2011-02-05 9:47 ` Russell King - ARM Linux 2011-02-05 10:41 ` [PATCH v2 04/28] ARM: mm: cache-l2x0: Add support forre-enabling l2x0 Santosh Shilimkar 2011-02-05 10:41 ` Santosh Shilimkar 2011-02-05 16:36 ` Colin Cross 2011-02-05 16:36 ` Colin Cross [not found] ` <AANLkTik_r4k_5o+F47vRbGPcWLwfgHgWqhym49XfhBZ9-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org> 2011-02-07 6:13 ` Santosh Shilimkar 2011-02-07 6:13 ` Santosh Shilimkar 2011-02-07 6:13 ` Santosh Shilimkar 2011-01-24 2:01 ` [PATCH v2 05/28] ARM: tegra: Centralize macros to define debug uart base Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 06/28] ARM: tegra: Add api to control internal powergating Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 07/28] ARM: tegra: irqs: Update irq list Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 08/28] ARM: tegra: Add prototypes for subsystem suspend functions Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 09/28] ARM: tegra: clock: Suspend fixes, and add new clocks Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 10/28] ARM: tegra: pinmux: Add missing drive pingroups and fix suspend Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 11/28] ARM: tegra: timer: Add idle and suspend support to timers Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 12/28] ARM: tegra: irq: Add support for suspend wake sources Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 13/28] ARM: tegra: Add suspend and hotplug support Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 7:31 ` Colin Cross 2011-01-24 7:31 ` Colin Cross 2011-01-24 9:07 ` Russell King - ARM Linux [this message] 2011-01-24 9:07 ` Russell King - ARM Linux 2011-01-24 9:26 ` Colin Cross 2011-01-24 9:26 ` Colin Cross 2011-01-24 10:15 ` Russell King - ARM Linux 2011-01-24 10:15 ` Russell King - ARM Linux 2011-01-24 2:01 ` [PATCH v2 14/28] ARM: tegra: irq: Add set_wake and set_type support for suspend Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 15/28] ARM: tegra: irq: Add debugfs file to show wake irqs Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-27 19:45 ` Stephen Boyd 2011-01-27 19:45 ` Stephen Boyd 2011-01-24 2:01 ` [PATCH v2 16/28] ARM: tegra: irq: Implement retrigger Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 17/28] ARM: tegra: gpio: Add support for waking from suspend Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 18/28] ARM: tegra: add CPU_IDLE driver Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-27 20:38 ` Stephen Boyd 2011-01-27 20:38 ` Stephen Boyd 2011-01-27 21:51 ` Colin Cross 2011-01-27 21:51 ` Colin Cross 2011-01-27 22:07 ` Colin Cross 2011-01-27 22:07 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 19/28] ARM: tegra: iomap: Add missing devices, fix use of SZ_8, SZ_64 Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 20/28] ARM: tegra: cpufreq: Disable cpufreq during suspend Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 14:41 ` Mark Brown 2011-01-24 14:41 ` Mark Brown 2011-01-24 18:50 ` Colin Cross 2011-01-24 18:50 ` Colin Cross 2011-01-24 19:35 ` Mark Brown 2011-01-24 19:35 ` Mark Brown 2011-01-24 19:52 ` Colin Cross 2011-01-24 19:52 ` Colin Cross 2011-01-24 20:26 ` Mark Brown 2011-01-24 20:26 ` Mark Brown 2011-01-24 20:52 ` Colin Cross 2011-01-24 20:52 ` Colin Cross 2011-01-24 21:08 ` Mark Brown 2011-01-24 21:08 ` Mark Brown 2011-01-24 21:24 ` Colin Cross 2011-01-24 21:24 ` Colin Cross 2011-01-25 4:26 ` Kyungmin Park 2011-01-25 4:26 ` Kyungmin Park 2011-01-24 2:01 ` [PATCH v2 21/28] ARM: tegra: Allow overriding arch_reset Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 22/28] ARM: tegra: dma: Fix critical data corruption bugs Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 23/28] ARM: tegra: add tegra_defconfig Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 24/28] ARM: tegra: Use writel_relaxed in tegra_init_cache Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 25/28] ARM: tegra: suspend: Save protected aperture across LP0 Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 26/28] ARM: tegra: suspend: protect suspend functions with CONFIG_PM Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 27/28] ARM: tegra: enable emc clock updates after lp0 Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 2:01 ` [PATCH v2 28/28] ARM: tegra: clock: Add forward reference to struct clk Colin Cross 2011-01-24 2:01 ` Colin Cross 2011-01-24 10:26 ` Russell King - ARM Linux 2011-01-24 10:26 ` Russell King - ARM Linux 2011-01-25 1:23 ` Colin Cross 2011-01-25 1:23 ` Colin Cross 2011-01-26 3:19 ` Grant Likely 2011-01-26 3:19 ` Grant Likely 2011-01-26 3:21 ` Colin Cross 2011-01-26 3:21 ` Colin Cross 2011-01-26 4:16 ` Grant Likely 2011-01-26 4:16 ` Grant Likely
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=20110124090708.GA16202@n2100.arm.linux.org.uk \ --to=linux@arm.linux.org.uk \ --cc=ccross@android.com \ --cc=gking@nvidia.com \ --cc=konkers@android.com \ --cc=linux-arm-kernel@lists.infradead.org \ --cc=linux-kernel@vger.kernel.org \ --cc=linux-tegra@vger.kernel.org \ --cc=olof@lixom.net \ /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.