ARM: vexpress/TC2: basic PM support
authorNicolas Pitre <nico@linaro.org>
Tue, 6 Aug 2013 18:10:08 +0000 (19:10 +0100)
committerPawel Moll <pawel.moll@arm.com>
Wed, 7 Aug 2013 13:55:52 +0000 (14:55 +0100)
This is the MCPM backend for the Virtual Express A15x2 A7x3 CoreTile
aka TC2.  This provides cluster management for SMP secondary boot and
CPU hotplug.

Signed-off-by: Nicolas Pitre <nico@linaro.org>
Acked-by: Pawel Moll <pawel.moll@arm.com>
Reviewed-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
[PM: made it drive SCC registers directly and provide base for SPC]
Signed-off-by: Pawel Moll <pawel.moll@arm.com>
arch/arm/mach-vexpress/Kconfig
arch/arm/mach-vexpress/Makefile
arch/arm/mach-vexpress/tc2_pm.c [new file with mode: 0644]

index b8bbabe..c700e62 100644 (file)
@@ -66,4 +66,12 @@ config ARCH_VEXPRESS_DCSCB
          This is needed to provide CPU and cluster power management
          on RTSM implementing big.LITTLE.
 
+config ARCH_VEXPRESS_TC2_PM
+       bool "Versatile Express TC2 power management"
+       depends on MCPM
+       select ARM_CCI
+       help
+         Support for CPU and cluster power management on Versatile Express
+         with a TC2 (A15x2 A7x3) big.LITTLE core tile.
+
 endmenu
index 48ba89a..36ea824 100644 (file)
@@ -7,5 +7,6 @@ ccflags-$(CONFIG_ARCH_MULTIPLATFORM) := -I$(srctree)/$(src)/include \
 obj-y                                  := v2m.o
 obj-$(CONFIG_ARCH_VEXPRESS_CA9X4)      += ct-ca9x4.o
 obj-$(CONFIG_ARCH_VEXPRESS_DCSCB)      += dcscb.o      dcscb_setup.o
+obj-$(CONFIG_ARCH_VEXPRESS_TC2_PM)     += tc2_pm.o spc.o
 obj-$(CONFIG_SMP)                      += platsmp.o
 obj-$(CONFIG_HOTPLUG_CPU)              += hotplug.o
diff --git a/arch/arm/mach-vexpress/tc2_pm.c b/arch/arm/mach-vexpress/tc2_pm.c
new file mode 100644 (file)
index 0000000..cdc6068
--- /dev/null
@@ -0,0 +1,327 @@
+/*
+ * arch/arm/mach-vexpress/tc2_pm.c - TC2 power management support
+ *
+ * Created by: Nicolas Pitre, October 2012
+ * Copyright:  (C) 2012-2013  Linaro Limited
+ *
+ * Some portions of this file were originally written by Achin Gupta
+ * Copyright:   (C) 2012  ARM Limited
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/kernel.h>
+#include <linux/of_address.h>
+#include <linux/spinlock.h>
+#include <linux/errno.h>
+
+#include <asm/mcpm.h>
+#include <asm/proc-fns.h>
+#include <asm/cacheflush.h>
+#include <asm/cputype.h>
+#include <asm/cp15.h>
+
+#include <linux/arm-cci.h>
+
+#include "spc.h"
+
+/* SCC conf registers */
+#define A15_CONF               0x400
+#define A7_CONF                        0x500
+#define SYS_INFO               0x700
+#define SPC_BASE               0xb00
+
+/*
+ * We can't use regular spinlocks. In the switcher case, it is possible
+ * for an outbound CPU to call power_down() after its inbound counterpart
+ * is already live using the same logical CPU number which trips lockdep
+ * debugging.
+ */
+static arch_spinlock_t tc2_pm_lock = __ARCH_SPIN_LOCK_UNLOCKED;
+
+#define TC2_CLUSTERS                   2
+#define TC2_MAX_CPUS_PER_CLUSTER       3
+
+static unsigned int tc2_nr_cpus[TC2_CLUSTERS];
+
+/* Keep per-cpu usage count to cope with unordered up/down requests */
+static int tc2_pm_use_count[TC2_MAX_CPUS_PER_CLUSTER][TC2_CLUSTERS];
+
+#define tc2_cluster_unused(cluster) \
+       (!tc2_pm_use_count[0][cluster] && \
+        !tc2_pm_use_count[1][cluster] && \
+        !tc2_pm_use_count[2][cluster])
+
+static int tc2_pm_power_up(unsigned int cpu, unsigned int cluster)
+{
+       pr_debug("%s: cpu %u cluster %u\n", __func__, cpu, cluster);
+       if (cluster >= TC2_CLUSTERS || cpu >= tc2_nr_cpus[cluster])
+               return -EINVAL;
+
+       /*
+        * Since this is called with IRQs enabled, and no arch_spin_lock_irq
+        * variant exists, we need to disable IRQs manually here.
+        */
+       local_irq_disable();
+       arch_spin_lock(&tc2_pm_lock);
+
+       if (tc2_cluster_unused(cluster))
+               ve_spc_powerdown(cluster, false);
+
+       tc2_pm_use_count[cpu][cluster]++;
+       if (tc2_pm_use_count[cpu][cluster] == 1) {
+               ve_spc_set_resume_addr(cluster, cpu,
+                                      virt_to_phys(mcpm_entry_point));
+               ve_spc_cpu_wakeup_irq(cluster, cpu, true);
+       } else if (tc2_pm_use_count[cpu][cluster] != 2) {
+               /*
+                * The only possible values are:
+                * 0 = CPU down
+                * 1 = CPU (still) up
+                * 2 = CPU requested to be up before it had a chance
+                *     to actually make itself down.
+                * Any other value is a bug.
+                */
+               BUG();
+       }
+
+       arch_spin_unlock(&tc2_pm_lock);
+       local_irq_enable();
+
+       return 0;
+}
+
+static void tc2_pm_power_down(void)
+{
+       unsigned int mpidr, cpu, cluster;
+       bool last_man = false, skip_wfi = false;
+
+       mpidr = read_cpuid_mpidr();
+       cpu = MPIDR_AFFINITY_LEVEL(mpidr, 0);
+       cluster = MPIDR_AFFINITY_LEVEL(mpidr, 1);
+
+       pr_debug("%s: cpu %u cluster %u\n", __func__, cpu, cluster);
+       BUG_ON(cluster >= TC2_CLUSTERS || cpu >= TC2_MAX_CPUS_PER_CLUSTER);
+
+       __mcpm_cpu_going_down(cpu, cluster);
+
+       arch_spin_lock(&tc2_pm_lock);
+       BUG_ON(__mcpm_cluster_state(cluster) != CLUSTER_UP);
+       tc2_pm_use_count[cpu][cluster]--;
+       if (tc2_pm_use_count[cpu][cluster] == 0) {
+               ve_spc_cpu_wakeup_irq(cluster, cpu, true);
+               if (tc2_cluster_unused(cluster)) {
+                       ve_spc_powerdown(cluster, true);
+                       ve_spc_global_wakeup_irq(true);
+                       last_man = true;
+               }
+       } else if (tc2_pm_use_count[cpu][cluster] == 1) {
+               /*
+                * A power_up request went ahead of us.
+                * Even if we do not want to shut this CPU down,
+                * the caller expects a certain state as if the WFI
+                * was aborted.  So let's continue with cache cleaning.
+                */
+               skip_wfi = true;
+       } else
+               BUG();
+
+       if (last_man && __mcpm_outbound_enter_critical(cpu, cluster)) {
+               arch_spin_unlock(&tc2_pm_lock);
+
+               if (read_cpuid_part_number() == ARM_CPU_PART_CORTEX_A15) {
+                       /*
+                        * On the Cortex-A15 we need to disable
+                        * L2 prefetching before flushing the cache.
+                        */
+                       asm volatile(
+                       "mcr    p15, 1, %0, c15, c0, 3 \n\t"
+                       "isb    \n\t"
+                       "dsb    "
+                       : : "r" (0x400) );
+               }
+
+               /*
+                * We need to disable and flush the whole (L1 and L2) cache.
+                * Let's do it in the safest possible way i.e. with
+                * no memory access within the following sequence
+                * including the stack.
+                */
+               asm volatile(
+               "mrc    p15, 0, r0, c1, c0, 0   @ get CR \n\t"
+               "bic    r0, r0, #"__stringify(CR_C)" \n\t"
+               "mcr    p15, 0, r0, c1, c0, 0   @ set CR \n\t"
+               "isb    \n\t"
+               "bl     v7_flush_dcache_all \n\t"
+               "clrex  \n\t"
+               "mrc    p15, 0, r0, c1, c0, 1   @ get AUXCR \n\t"
+               "bic    r0, r0, #(1 << 6)       @ disable local coherency \n\t"
+               "mcr    p15, 0, r0, c1, c0, 1   @ set AUXCR \n\t"
+               "isb    \n\t"
+               "dsb    "
+               : : : "r0","r1","r2","r3","r4","r5","r6","r7",
+                     "r9","r10","r11","lr","memory");
+
+               cci_disable_port_by_cpu(mpidr);
+
+               __mcpm_outbound_leave_critical(cluster, CLUSTER_DOWN);
+       } else {
+               /*
+                * If last man then undo any setup done previously.
+                */
+               if (last_man) {
+                       ve_spc_powerdown(cluster, false);
+                       ve_spc_global_wakeup_irq(false);
+               }
+
+               arch_spin_unlock(&tc2_pm_lock);
+
+               /*
+                * We need to disable and flush only the L1 cache.
+                * Let's do it in the safest possible way as above.
+                */
+               asm volatile(
+               "mrc    p15, 0, r0, c1, c0, 0   @ get CR \n\t"
+               "bic    r0, r0, #"__stringify(CR_C)" \n\t"
+               "mcr    p15, 0, r0, c1, c0, 0   @ set CR \n\t"
+               "isb    \n\t"
+               "bl     v7_flush_dcache_louis \n\t"
+               "clrex  \n\t"
+               "mrc    p15, 0, r0, c1, c0, 1   @ get AUXCR \n\t"
+               "bic    r0, r0, #(1 << 6)       @ disable local coherency \n\t"
+               "mcr    p15, 0, r0, c1, c0, 1   @ set AUXCR \n\t"
+               "isb    \n\t"
+               "dsb    "
+               : : : "r0","r1","r2","r3","r4","r5","r6","r7",
+                     "r9","r10","r11","lr","memory");
+       }
+
+       __mcpm_cpu_down(cpu, cluster);
+
+       /* Now we are prepared for power-down, do it: */
+       if (!skip_wfi)
+               wfi();
+
+       /* Not dead at this point?  Let our caller cope. */
+}
+
+static void tc2_pm_powered_up(void)
+{
+       unsigned int mpidr, cpu, cluster;
+       unsigned long flags;
+
+       mpidr = read_cpuid_mpidr();
+       cpu = MPIDR_AFFINITY_LEVEL(mpidr, 0);
+       cluster = MPIDR_AFFINITY_LEVEL(mpidr, 1);
+
+       pr_debug("%s: cpu %u cluster %u\n", __func__, cpu, cluster);
+       BUG_ON(cluster >= TC2_CLUSTERS || cpu >= TC2_MAX_CPUS_PER_CLUSTER);
+
+       local_irq_save(flags);
+       arch_spin_lock(&tc2_pm_lock);
+
+       if (tc2_cluster_unused(cluster)) {
+               ve_spc_powerdown(cluster, false);
+               ve_spc_global_wakeup_irq(false);
+       }
+
+       if (!tc2_pm_use_count[cpu][cluster])
+               tc2_pm_use_count[cpu][cluster] = 1;
+
+       ve_spc_cpu_wakeup_irq(cluster, cpu, false);
+       ve_spc_set_resume_addr(cluster, cpu, 0);
+
+       arch_spin_unlock(&tc2_pm_lock);
+       local_irq_restore(flags);
+}
+
+static const struct mcpm_platform_ops tc2_pm_power_ops = {
+       .power_up       = tc2_pm_power_up,
+       .power_down     = tc2_pm_power_down,
+       .powered_up     = tc2_pm_powered_up,
+};
+
+static bool __init tc2_pm_usage_count_init(void)
+{
+       unsigned int mpidr, cpu, cluster;
+
+       mpidr = read_cpuid_mpidr();
+       cpu = MPIDR_AFFINITY_LEVEL(mpidr, 0);
+       cluster = MPIDR_AFFINITY_LEVEL(mpidr, 1);
+
+       pr_debug("%s: cpu %u cluster %u\n", __func__, cpu, cluster);
+       if (cluster >= TC2_CLUSTERS || cpu >= tc2_nr_cpus[cluster]) {
+               pr_err("%s: boot CPU is out of bound!\n", __func__);
+               return false;
+       }
+       tc2_pm_use_count[cpu][cluster] = 1;
+       return true;
+}
+
+/*
+ * Enable cluster-level coherency, in preparation for turning on the MMU.
+ */
+static void __naked tc2_pm_power_up_setup(unsigned int affinity_level)
+{
+       asm volatile (" \n"
+"      cmp     r0, #1 \n"
+"      bxne    lr \n"
+"      b       cci_enable_port_for_self ");
+}
+
+static int __init tc2_pm_init(void)
+{
+       int ret;
+       void __iomem *scc;
+       u32 a15_cluster_id, a7_cluster_id, sys_info;
+       struct device_node *np;
+
+       /*
+        * The power management-related features are hidden behind
+        * SCC registers. We need to extract runtime information like
+        * cluster ids and number of CPUs really available in clusters.
+        */
+       np = of_find_compatible_node(NULL, NULL,
+                       "arm,vexpress-scc,v2p-ca15_a7");
+       scc = of_iomap(np, 0);
+       if (!scc)
+               return -ENODEV;
+
+       a15_cluster_id = readl_relaxed(scc + A15_CONF) & 0xf;
+       a7_cluster_id = readl_relaxed(scc + A7_CONF) & 0xf;
+       if (a15_cluster_id >= TC2_CLUSTERS || a7_cluster_id >= TC2_CLUSTERS)
+               return -EINVAL;
+
+       sys_info = readl_relaxed(scc + SYS_INFO);
+       tc2_nr_cpus[a15_cluster_id] = (sys_info >> 16) & 0xf;
+       tc2_nr_cpus[a7_cluster_id] = (sys_info >> 20) & 0xf;
+
+       /*
+        * A subset of the SCC registers is also used to communicate
+        * with the SPC (power controller). We need to be able to
+        * drive it very early in the boot process to power up
+        * processors, so we initialize the SPC driver here.
+        */
+       ret = ve_spc_init(scc + SPC_BASE, a15_cluster_id);
+       if (ret)
+               return ret;
+
+       if (!cci_probed())
+               return -ENODEV;
+
+       if (!tc2_pm_usage_count_init())
+               return -EINVAL;
+
+       ret = mcpm_platform_register(&tc2_pm_power_ops);
+       if (!ret) {
+               mcpm_sync_init(tc2_pm_power_up_setup);
+               pr_info("TC2 power management initialized\n");
+       }
+       return ret;
+}
+
+early_initcall(tc2_pm_init);