[PATCH 3/7] ARM: tegra30: cpuidle: add LP2 driver for secondary CPUs
Lorenzo Pieralisi
lorenzo.pieralisi at arm.com
Mon Oct 8 12:35:04 EDT 2012
On Mon, Oct 08, 2012 at 11:26:17AM +0100, Joseph Lo wrote:
> This supports power-gated (LP2) idle on secondary CPUs for Tegra30.
> The secondary CPUs can go into LP2 state independently. When CPU goes
> into LP2 state, it saves it's state and puts itself to flow controlled
> WFI state. After that, it will been power gated.
>
> Based on the work by:
> Scott Williams <scwilliams at nvidia.com>
>
> Signed-off-by: Joseph Lo <josephl at nvidia.com>
> ---
> arch/arm/mach-tegra/Makefile | 1 +
> arch/arm/mach-tegra/cpuidle-tegra30.c | 79 +++++++++++++++++++++++++++++-
> arch/arm/mach-tegra/pm.c | 88 +++++++++++++++++++++++++++++++++
> arch/arm/mach-tegra/pm.h | 30 +++++++++++
> arch/arm/mach-tegra/reset.h | 9 +++
> arch/arm/mach-tegra/sleep-tegra30.S | 26 ++++++++++
> arch/arm/mach-tegra/sleep.S | 66 ++++++++++++++++++++++++
> arch/arm/mach-tegra/sleep.h | 2 +
> 8 files changed, 300 insertions(+), 1 deletions(-)
> create mode 100644 arch/arm/mach-tegra/pm.c
> create mode 100644 arch/arm/mach-tegra/pm.h
>
> diff --git a/arch/arm/mach-tegra/Makefile b/arch/arm/mach-tegra/Makefile
> index 9b80c1e..6f224f7 100644
> --- a/arch/arm/mach-tegra/Makefile
> +++ b/arch/arm/mach-tegra/Makefile
> @@ -8,6 +8,7 @@ obj-y += pmc.o
> obj-y += flowctrl.o
> obj-y += powergate.o
> obj-y += apbio.o
> +obj-y += pm.o
> obj-$(CONFIG_CPU_IDLE) += cpuidle.o
> obj-$(CONFIG_CPU_IDLE) += sleep.o
> obj-$(CONFIG_ARCH_TEGRA_2x_SOC) += tegra20_clocks.o
> diff --git a/arch/arm/mach-tegra/cpuidle-tegra30.c b/arch/arm/mach-tegra/cpuidle-tegra30.c
> index 0f85df8..842fef4 100644
> --- a/arch/arm/mach-tegra/cpuidle-tegra30.c
> +++ b/arch/arm/mach-tegra/cpuidle-tegra30.c
> @@ -19,21 +19,94 @@
> #include <linux/kernel.h>
> #include <linux/module.h>
> #include <linux/cpuidle.h>
> +#include <linux/cpu_pm.h>
> +#include <linux/clockchips.h>
>
> #include <asm/cpuidle.h>
> +#include <asm/proc-fns.h>
> +#include <asm/suspend.h>
> +
> +#include "pm.h"
> +#include "sleep.h"
> +
> +#ifdef CONFIG_PM_SLEEP
> +static int tegra30_idle_lp2(struct cpuidle_device *dev,
> + struct cpuidle_driver *drv,
> + int index);
> +#endif
>
> static struct cpuidle_driver tegra_idle_driver = {
> .name = "tegra_idle",
> .owner = THIS_MODULE,
> .en_core_tk_irqen = 1,
> - .state_count = 1,
> + .state_count = 2,
> .states = {
> [0] = ARM_CPUIDLE_WFI_STATE_PWR(600),
> +#ifdef CONFIG_PM_SLEEP
> + [1] = {
> + .enter = tegra30_idle_lp2,
> + .exit_latency = 2000,
> + .target_residency = 2200,
> + .power_usage = 0,
> + .flags = CPUIDLE_FLAG_TIME_VALID,
> + .name = "LP2",
> + .desc = "CPU power-gate",
> + },
> +#endif
> },
> };
>
> static DEFINE_PER_CPU(struct cpuidle_device, tegra_idle_device);
>
> +#ifdef CONFIG_PM_SLEEP
> +static bool tegra30_idle_enter_lp2_cpu_n(struct cpuidle_device *dev,
> + struct cpuidle_driver *drv,
> + int index)
> +{
> +#ifdef CONFIG_SMP
> + clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_ENTER, &dev->cpu);
> +
> + smp_wmb();
> +
> + save_cpu_arch_register();
> +
> + cpu_suspend(0, tegra30_sleep_cpu_secondary_finish);
> +
> + restore_cpu_arch_register();
> +
> + clockevents_notify(CLOCK_EVT_NOTIFY_BROADCAST_EXIT, &dev->cpu);
> +#endif
Can't you factor out this #ifdef out using an inline function ?
> +
> + return true;
> +}
> +
> +static int __cpuinit tegra30_idle_lp2(struct cpuidle_device *dev,
> + struct cpuidle_driver *drv,
> + int index)
> +{
> + bool entered_lp2 = false;
> +
> + local_fiq_disable();
> +
> + tegra_set_cpu_in_lp2(dev->cpu);
> + cpu_pm_enter();
> +
> + if (dev->cpu == 0)
Logical cpu 0 ? Or you need a HW cpu 0 check here ? If you boot on a CPU
that is different from HW CPU 0 (do not know if that's possible) you
might have a problem.
[...]
> +bool __cpuinit tegra_set_cpu_in_lp2(int cpu)
> +{
> + bool last_cpu = false;
> +
> + spin_lock(&tegra_lp2_lock);
> + BUG_ON(cpumask_test_cpu(cpu, &tegra_in_lp2));
> + cpumask_set_cpu(cpu, &tegra_in_lp2);
> +
> + /*
> + * Update the IRAM copy used by the reset handler. The IRAM copy
> + * can't use used directly by cpumask_set_cpu() because it uses
> + * LDREX/STREX which requires the addressed location to be inner
> + * cacheable and sharable which IRAM isn't.
> + */
> + writel(tegra_in_lp2.bits[0], tegra_cpu_lp2_mask);
> + dsb();
> +
> + if ((cpu == 0) && cpumask_equal(&tegra_in_lp2, cpu_online_mask))
> + last_cpu = true;
For cpu == 0, see above.
[...]
> +ENTRY(tegra_flush_l1_cache)
> + stmfd sp!, {r4-r5, r7, r9-r11, lr}
> + dmb @ ensure ordering
> +
> + /* Disable the data cache */
> + mrc p15, 0, r2, c1, c0, 0
> + bic r2, r2, #CR_C
> + dsb
> + mcr p15, 0, r2, c1, c0, 0
> +
> + /* Flush data cache */
> + mov r10, #0
> +#ifdef CONFIG_PREEMPT
> + save_and_disable_irqs_notrace r9 @ make cssr&csidr read atomic
> +#endif
> + mcr p15, 2, r10, c0, c0, 0 @ select current cache level in cssr
> + isb @ isb to sych the new cssr&csidr
> + mrc p15, 1, r1, c0, c0, 0 @ read the new csidr
> +#ifdef CONFIG_PREEMPT
> + restore_irqs_notrace r9
> +#endif
> + and r2, r1, #7 @ extract the length of the cache lines
> + add r2, r2, #4 @ add 4 (line length offset)
> + ldr r4, =0x3ff
> + ands r4, r4, r1, lsr #3 @ find maximum number on the way size
> + clz r5, r4 @ find bit position of way size increment
> + ldr r7, =0x7fff
> + ands r7, r7, r1, lsr #13 @ extract max number of the index size
> +loop2:
> + mov r9, r4 @ create working copy of max way size
> +loop3:
> + orr r11, r10, r9, lsl r5 @ factor way and cache number into r11
> + orr r11, r11, r7, lsl r2 @ factor index number into r11
> + mcr p15, 0, r11, c7, c14, 2 @ clean & invalidate by set/way
> + subs r9, r9, #1 @ decrement the way
> + bge loop3
> + subs r7, r7, #1 @ decrement the index
> + bge loop2
> +finished:
> + mov r10, #0 @ swith back to cache level 0
> + mcr p15, 2, r10, c0, c0, 0 @ select current cache level in cssr
> + dsb
> + isb
This code is already in the kernel in cache-v7.S, please use that.
We are just adding the new LoUIS API that probably does what you
want, even though for Tegra, that is an A9 based platform I fail to
understand why Level of Coherency differs from L1.
Can you explain to me please why Level of Coherency (LoC) is != from L1
on Tegra ?
Thanks,
Lorenzo
More information about the linux-arm-kernel
mailing list