From 57f0f512b273f60d52568b8c6b77e17f5636edc0 Mon Sep 17 00:00:00 2001 From: André Fabian Silva Delgado Date: Wed, 5 Aug 2015 17:04:01 -0300 Subject: Initial import --- drivers/bus/Kconfig | 116 ++++ drivers/bus/Makefile | 19 + drivers/bus/arm-cci.c | 1535 +++++++++++++++++++++++++++++++++++++++++ drivers/bus/arm-ccn.c | 1393 +++++++++++++++++++++++++++++++++++++ drivers/bus/brcmstb_gisb.c | 403 +++++++++++ drivers/bus/imx-weim.c | 217 ++++++ drivers/bus/mips_cdmm.c | 716 +++++++++++++++++++ drivers/bus/mvebu-mbus.c | 1209 ++++++++++++++++++++++++++++++++ drivers/bus/omap-ocp2scp.c | 123 ++++ drivers/bus/omap_l3_noc.c | 377 ++++++++++ drivers/bus/omap_l3_noc.h | 501 ++++++++++++++ drivers/bus/omap_l3_smx.c | 311 +++++++++ drivers/bus/omap_l3_smx.h | 338 +++++++++ drivers/bus/simple-pm-bus.c | 58 ++ drivers/bus/vexpress-config.c | 202 ++++++ 15 files changed, 7518 insertions(+) create mode 100644 drivers/bus/Kconfig create mode 100644 drivers/bus/Makefile create mode 100644 drivers/bus/arm-cci.c create mode 100644 drivers/bus/arm-ccn.c create mode 100644 drivers/bus/brcmstb_gisb.c create mode 100644 drivers/bus/imx-weim.c create mode 100644 drivers/bus/mips_cdmm.c create mode 100644 drivers/bus/mvebu-mbus.c create mode 100644 drivers/bus/omap-ocp2scp.c create mode 100644 drivers/bus/omap_l3_noc.c create mode 100644 drivers/bus/omap_l3_noc.h create mode 100644 drivers/bus/omap_l3_smx.c create mode 100644 drivers/bus/omap_l3_smx.h create mode 100644 drivers/bus/simple-pm-bus.c create mode 100644 drivers/bus/vexpress-config.c (limited to 'drivers/bus') diff --git a/drivers/bus/Kconfig b/drivers/bus/Kconfig new file mode 100644 index 000000000..a1d4af6df --- /dev/null +++ b/drivers/bus/Kconfig @@ -0,0 +1,116 @@ +# +# Bus Devices +# + +menu "Bus devices" + +config ARM_CCI + bool + +config ARM_CCI400_COMMON + bool + select ARM_CCI + +config ARM_CCI400_PMU + bool "ARM CCI400 PMU support" + default y + depends on ARM || ARM64 + depends on HW_PERF_EVENTS + select ARM_CCI400_COMMON + help + Support for PMU events monitoring on the ARM CCI cache coherent + interconnect. + + If unsure, say Y + +config ARM_CCI400_PORT_CTRL + bool + depends on ARM && OF && CPU_V7 + select ARM_CCI400_COMMON + help + Low level power management driver for CCI400 cache coherent + interconnect for ARM platforms. + +config ARM_CCN + bool "ARM CCN driver support" + depends on ARM || ARM64 + depends on PERF_EVENTS + help + PMU (perf) driver supporting the ARM CCN (Cache Coherent Network) + interconnect. + +config BRCMSTB_GISB_ARB + bool "Broadcom STB GISB bus arbiter" + depends on ARM || MIPS + help + Driver for the Broadcom Set Top Box System-on-a-chip internal bus + arbiter. This driver provides timeout and target abort error handling + and internal bus master decoding. + +config IMX_WEIM + bool "Freescale EIM DRIVER" + depends on ARCH_MXC + help + Driver for i.MX WEIM controller. + The WEIM(Wireless External Interface Module) works like a bus. + You can attach many different devices on it, such as NOR, onenand. + +config MIPS_CDMM + bool "MIPS Common Device Memory Map (CDMM) Driver" + depends on CPU_MIPSR2 + help + Driver needed for the MIPS Common Device Memory Map bus in MIPS + cores. This bus is for per-CPU tightly coupled devices such as the + Fast Debug Channel (FDC). + + For this to work, either your bootloader needs to enable the CDMM + region at an unused physical address on the boot CPU, or else your + platform code needs to implement mips_cdmm_phys_base() (see + asm/cdmm.h). + +config MVEBU_MBUS + bool + depends on PLAT_ORION + help + Driver needed for the MBus configuration on Marvell EBU SoCs + (Kirkwood, Dove, Orion5x, MV78XX0 and Armada 370/XP). + +config OMAP_INTERCONNECT + tristate "OMAP INTERCONNECT DRIVER" + depends on ARCH_OMAP2PLUS + + help + Driver to enable OMAP interconnect error handling driver. + +config OMAP_OCP2SCP + tristate "OMAP OCP2SCP DRIVER" + depends on ARCH_OMAP2PLUS + help + Driver to enable ocp2scp module which transforms ocp interface + protocol to scp protocol. In OMAP4, USB PHY is connected via + OCP2SCP and in OMAP5, both USB PHY and SATA PHY is connected via + OCP2SCP. + +config SIMPLE_PM_BUS + bool "Simple Power-Managed Bus Driver" + depends on OF && PM + depends on ARCH_SHMOBILE || COMPILE_TEST + help + Driver for transparent busses that don't need a real driver, but + where the bus controller is part of a PM domain, or under the control + of a functional clock, and thus relies on runtime PM for managing + this PM domain and/or clock. + An example of such a bus controller is the Renesas Bus State + Controller (BSC, sometimes called "LBSC within Bus Bridge", or + "External Bus Interface") as found on several Renesas ARM SoCs. + +config VEXPRESS_CONFIG + bool "Versatile Express configuration bus" + default y if ARCH_VEXPRESS + depends on ARM || ARM64 + depends on OF + select REGMAP + help + Platform configuration infrastructure for the ARM Ltd. + Versatile Express. +endmenu diff --git a/drivers/bus/Makefile b/drivers/bus/Makefile new file mode 100644 index 000000000..790e7b933 --- /dev/null +++ b/drivers/bus/Makefile @@ -0,0 +1,19 @@ +# +# Makefile for the bus drivers. +# + +# Interconnect bus drivers for ARM platforms +obj-$(CONFIG_ARM_CCI) += arm-cci.o +obj-$(CONFIG_ARM_CCN) += arm-ccn.o + +obj-$(CONFIG_BRCMSTB_GISB_ARB) += brcmstb_gisb.o +obj-$(CONFIG_IMX_WEIM) += imx-weim.o +obj-$(CONFIG_MIPS_CDMM) += mips_cdmm.o +obj-$(CONFIG_MVEBU_MBUS) += mvebu-mbus.o + +# Interconnect bus driver for OMAP SoCs. +obj-$(CONFIG_OMAP_INTERCONNECT) += omap_l3_smx.o omap_l3_noc.o + +obj-$(CONFIG_OMAP_OCP2SCP) += omap-ocp2scp.o +obj-$(CONFIG_SIMPLE_PM_BUS) += simple-pm-bus.o +obj-$(CONFIG_VEXPRESS_CONFIG) += vexpress-config.o diff --git a/drivers/bus/arm-cci.c b/drivers/bus/arm-cci.c new file mode 100644 index 000000000..5340604b2 --- /dev/null +++ b/drivers/bus/arm-cci.c @@ -0,0 +1,1535 @@ +/* + * CCI cache coherent interconnect driver + * + * Copyright (C) 2013 ARM Ltd. + * Author: Lorenzo Pieralisi + * + * 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. + * + * This program is distributed "as is" WITHOUT ANY WARRANTY of any + * kind, whether express or implied; without even the implied warranty + * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +static void __iomem *cci_ctrl_base; +static unsigned long cci_ctrl_phys; + +#ifdef CONFIG_ARM_CCI400_PORT_CTRL +struct cci_nb_ports { + unsigned int nb_ace; + unsigned int nb_ace_lite; +}; + +static const struct cci_nb_ports cci400_ports = { + .nb_ace = 2, + .nb_ace_lite = 3 +}; + +#define CCI400_PORTS_DATA (&cci400_ports) +#else +#define CCI400_PORTS_DATA (NULL) +#endif + +static const struct of_device_id arm_cci_matches[] = { +#ifdef CONFIG_ARM_CCI400_COMMON + {.compatible = "arm,cci-400", .data = CCI400_PORTS_DATA }, +#endif + {}, +}; + +#ifdef CONFIG_ARM_CCI400_PMU + +#define DRIVER_NAME "CCI-400" +#define DRIVER_NAME_PMU DRIVER_NAME " PMU" + +#define CCI_PMCR 0x0100 +#define CCI_PID2 0x0fe8 + +#define CCI_PMCR_CEN 0x00000001 +#define CCI_PMCR_NCNT_MASK 0x0000f800 +#define CCI_PMCR_NCNT_SHIFT 11 + +#define CCI_PID2_REV_MASK 0xf0 +#define CCI_PID2_REV_SHIFT 4 + +#define CCI_PMU_EVT_SEL 0x000 +#define CCI_PMU_CNTR 0x004 +#define CCI_PMU_CNTR_CTRL 0x008 +#define CCI_PMU_OVRFLW 0x00c + +#define CCI_PMU_OVRFLW_FLAG 1 + +#define CCI_PMU_CNTR_BASE(idx) ((idx) * SZ_4K) + +#define CCI_PMU_CNTR_MASK ((1ULL << 32) -1) + +#define CCI_PMU_EVENT_MASK 0xffUL +#define CCI_PMU_EVENT_SOURCE(event) ((event >> 5) & 0x7) +#define CCI_PMU_EVENT_CODE(event) (event & 0x1f) + +#define CCI_PMU_MAX_HW_EVENTS 5 /* CCI PMU has 4 counters + 1 cycle counter */ + +/* Types of interfaces that can generate events */ +enum { + CCI_IF_SLAVE, + CCI_IF_MASTER, + CCI_IF_MAX, +}; + +struct event_range { + u32 min; + u32 max; +}; + +struct cci_pmu_hw_events { + struct perf_event *events[CCI_PMU_MAX_HW_EVENTS]; + unsigned long used_mask[BITS_TO_LONGS(CCI_PMU_MAX_HW_EVENTS)]; + raw_spinlock_t pmu_lock; +}; + +struct cci_pmu_model { + char *name; + struct event_range event_ranges[CCI_IF_MAX]; +}; + +static struct cci_pmu_model cci_pmu_models[]; + +struct cci_pmu { + void __iomem *base; + struct pmu pmu; + int nr_irqs; + int irqs[CCI_PMU_MAX_HW_EVENTS]; + unsigned long active_irqs; + const struct cci_pmu_model *model; + struct cci_pmu_hw_events hw_events; + struct platform_device *plat_device; + int num_events; + atomic_t active_events; + struct mutex reserve_mutex; + cpumask_t cpus; +}; +static struct cci_pmu *pmu; + +#define to_cci_pmu(c) (container_of(c, struct cci_pmu, pmu)) + +/* Port ids */ +#define CCI_PORT_S0 0 +#define CCI_PORT_S1 1 +#define CCI_PORT_S2 2 +#define CCI_PORT_S3 3 +#define CCI_PORT_S4 4 +#define CCI_PORT_M0 5 +#define CCI_PORT_M1 6 +#define CCI_PORT_M2 7 + +#define CCI_REV_R0 0 +#define CCI_REV_R1 1 +#define CCI_REV_R1_PX 5 + +/* + * Instead of an event id to monitor CCI cycles, a dedicated counter is + * provided. Use 0xff to represent CCI cycles and hope that no future revisions + * make use of this event in hardware. + */ +enum cci400_perf_events { + CCI_PMU_CYCLES = 0xff +}; + +#define CCI_PMU_CYCLE_CNTR_IDX 0 +#define CCI_PMU_CNTR0_IDX 1 +#define CCI_PMU_CNTR_LAST(cci_pmu) (CCI_PMU_CYCLE_CNTR_IDX + cci_pmu->num_events - 1) + +/* + * CCI PMU event id is an 8-bit value made of two parts - bits 7:5 for one of 8 + * ports and bits 4:0 are event codes. There are different event codes + * associated with each port type. + * + * Additionally, the range of events associated with the port types changed + * between Rev0 and Rev1. + * + * The constants below define the range of valid codes for each port type for + * the different revisions and are used to validate the event to be monitored. + */ + +#define CCI_REV_R0_SLAVE_PORT_MIN_EV 0x00 +#define CCI_REV_R0_SLAVE_PORT_MAX_EV 0x13 +#define CCI_REV_R0_MASTER_PORT_MIN_EV 0x14 +#define CCI_REV_R0_MASTER_PORT_MAX_EV 0x1a + +#define CCI_REV_R1_SLAVE_PORT_MIN_EV 0x00 +#define CCI_REV_R1_SLAVE_PORT_MAX_EV 0x14 +#define CCI_REV_R1_MASTER_PORT_MIN_EV 0x00 +#define CCI_REV_R1_MASTER_PORT_MAX_EV 0x11 + +static int pmu_validate_hw_event(unsigned long hw_event) +{ + u8 ev_source = CCI_PMU_EVENT_SOURCE(hw_event); + u8 ev_code = CCI_PMU_EVENT_CODE(hw_event); + int if_type; + + if (hw_event & ~CCI_PMU_EVENT_MASK) + return -ENOENT; + + switch (ev_source) { + case CCI_PORT_S0: + case CCI_PORT_S1: + case CCI_PORT_S2: + case CCI_PORT_S3: + case CCI_PORT_S4: + /* Slave Interface */ + if_type = CCI_IF_SLAVE; + break; + case CCI_PORT_M0: + case CCI_PORT_M1: + case CCI_PORT_M2: + /* Master Interface */ + if_type = CCI_IF_MASTER; + break; + default: + return -ENOENT; + } + + if (ev_code >= pmu->model->event_ranges[if_type].min && + ev_code <= pmu->model->event_ranges[if_type].max) + return hw_event; + + return -ENOENT; +} + +static int probe_cci_revision(void) +{ + int rev; + rev = readl_relaxed(cci_ctrl_base + CCI_PID2) & CCI_PID2_REV_MASK; + rev >>= CCI_PID2_REV_SHIFT; + + if (rev < CCI_REV_R1_PX) + return CCI_REV_R0; + else + return CCI_REV_R1; +} + +static const struct cci_pmu_model *probe_cci_model(struct platform_device *pdev) +{ + if (platform_has_secure_cci_access()) + return &cci_pmu_models[probe_cci_revision()]; + return NULL; +} + +static int pmu_is_valid_counter(struct cci_pmu *cci_pmu, int idx) +{ + return CCI_PMU_CYCLE_CNTR_IDX <= idx && + idx <= CCI_PMU_CNTR_LAST(cci_pmu); +} + +static u32 pmu_read_register(int idx, unsigned int offset) +{ + return readl_relaxed(pmu->base + CCI_PMU_CNTR_BASE(idx) + offset); +} + +static void pmu_write_register(u32 value, int idx, unsigned int offset) +{ + return writel_relaxed(value, pmu->base + CCI_PMU_CNTR_BASE(idx) + offset); +} + +static void pmu_disable_counter(int idx) +{ + pmu_write_register(0, idx, CCI_PMU_CNTR_CTRL); +} + +static void pmu_enable_counter(int idx) +{ + pmu_write_register(1, idx, CCI_PMU_CNTR_CTRL); +} + +static void pmu_set_event(int idx, unsigned long event) +{ + pmu_write_register(event, idx, CCI_PMU_EVT_SEL); +} + +static u32 pmu_get_max_counters(void) +{ + u32 n_cnts = (readl_relaxed(cci_ctrl_base + CCI_PMCR) & + CCI_PMCR_NCNT_MASK) >> CCI_PMCR_NCNT_SHIFT; + + /* add 1 for cycle counter */ + return n_cnts + 1; +} + +static int pmu_get_event_idx(struct cci_pmu_hw_events *hw, struct perf_event *event) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu); + struct hw_perf_event *hw_event = &event->hw; + unsigned long cci_event = hw_event->config_base; + int idx; + + if (cci_event == CCI_PMU_CYCLES) { + if (test_and_set_bit(CCI_PMU_CYCLE_CNTR_IDX, hw->used_mask)) + return -EAGAIN; + + return CCI_PMU_CYCLE_CNTR_IDX; + } + + for (idx = CCI_PMU_CNTR0_IDX; idx <= CCI_PMU_CNTR_LAST(cci_pmu); ++idx) + if (!test_and_set_bit(idx, hw->used_mask)) + return idx; + + /* No counters available */ + return -EAGAIN; +} + +static int pmu_map_event(struct perf_event *event) +{ + int mapping; + unsigned long config = event->attr.config; + + if (event->attr.type < PERF_TYPE_MAX) + return -ENOENT; + + if (config == CCI_PMU_CYCLES) + mapping = config; + else + mapping = pmu_validate_hw_event(config); + + return mapping; +} + +static int pmu_request_irq(struct cci_pmu *cci_pmu, irq_handler_t handler) +{ + int i; + struct platform_device *pmu_device = cci_pmu->plat_device; + + if (unlikely(!pmu_device)) + return -ENODEV; + + if (pmu->nr_irqs < 1) { + dev_err(&pmu_device->dev, "no irqs for CCI PMUs defined\n"); + return -ENODEV; + } + + /* + * Register all available CCI PMU interrupts. In the interrupt handler + * we iterate over the counters checking for interrupt source (the + * overflowing counter) and clear it. + * + * This should allow handling of non-unique interrupt for the counters. + */ + for (i = 0; i < pmu->nr_irqs; i++) { + int err = request_irq(pmu->irqs[i], handler, IRQF_SHARED, + "arm-cci-pmu", cci_pmu); + if (err) { + dev_err(&pmu_device->dev, "unable to request IRQ%d for ARM CCI PMU counters\n", + pmu->irqs[i]); + return err; + } + + set_bit(i, &pmu->active_irqs); + } + + return 0; +} + +static void pmu_free_irq(struct cci_pmu *cci_pmu) +{ + int i; + + for (i = 0; i < pmu->nr_irqs; i++) { + if (!test_and_clear_bit(i, &pmu->active_irqs)) + continue; + + free_irq(pmu->irqs[i], cci_pmu); + } +} + +static u32 pmu_read_counter(struct perf_event *event) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu); + struct hw_perf_event *hw_counter = &event->hw; + int idx = hw_counter->idx; + u32 value; + + if (unlikely(!pmu_is_valid_counter(cci_pmu, idx))) { + dev_err(&cci_pmu->plat_device->dev, "Invalid CCI PMU counter %d\n", idx); + return 0; + } + value = pmu_read_register(idx, CCI_PMU_CNTR); + + return value; +} + +static void pmu_write_counter(struct perf_event *event, u32 value) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu); + struct hw_perf_event *hw_counter = &event->hw; + int idx = hw_counter->idx; + + if (unlikely(!pmu_is_valid_counter(cci_pmu, idx))) + dev_err(&cci_pmu->plat_device->dev, "Invalid CCI PMU counter %d\n", idx); + else + pmu_write_register(value, idx, CCI_PMU_CNTR); +} + +static u64 pmu_event_update(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + u64 delta, prev_raw_count, new_raw_count; + + do { + prev_raw_count = local64_read(&hwc->prev_count); + new_raw_count = pmu_read_counter(event); + } while (local64_cmpxchg(&hwc->prev_count, prev_raw_count, + new_raw_count) != prev_raw_count); + + delta = (new_raw_count - prev_raw_count) & CCI_PMU_CNTR_MASK; + + local64_add(delta, &event->count); + + return new_raw_count; +} + +static void pmu_read(struct perf_event *event) +{ + pmu_event_update(event); +} + +void pmu_event_set_period(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + /* + * The CCI PMU counters have a period of 2^32. To account for the + * possiblity of extreme interrupt latency we program for a period of + * half that. Hopefully we can handle the interrupt before another 2^31 + * events occur and the counter overtakes its previous value. + */ + u64 val = 1ULL << 31; + local64_set(&hwc->prev_count, val); + pmu_write_counter(event, val); +} + +static irqreturn_t pmu_handle_irq(int irq_num, void *dev) +{ + unsigned long flags; + struct cci_pmu *cci_pmu = dev; + struct cci_pmu_hw_events *events = &pmu->hw_events; + int idx, handled = IRQ_NONE; + + raw_spin_lock_irqsave(&events->pmu_lock, flags); + /* + * Iterate over counters and update the corresponding perf events. + * This should work regardless of whether we have per-counter overflow + * interrupt or a combined overflow interrupt. + */ + for (idx = CCI_PMU_CYCLE_CNTR_IDX; idx <= CCI_PMU_CNTR_LAST(cci_pmu); idx++) { + struct perf_event *event = events->events[idx]; + struct hw_perf_event *hw_counter; + + if (!event) + continue; + + hw_counter = &event->hw; + + /* Did this counter overflow? */ + if (!(pmu_read_register(idx, CCI_PMU_OVRFLW) & + CCI_PMU_OVRFLW_FLAG)) + continue; + + pmu_write_register(CCI_PMU_OVRFLW_FLAG, idx, CCI_PMU_OVRFLW); + + pmu_event_update(event); + pmu_event_set_period(event); + handled = IRQ_HANDLED; + } + raw_spin_unlock_irqrestore(&events->pmu_lock, flags); + + return IRQ_RETVAL(handled); +} + +static int cci_pmu_get_hw(struct cci_pmu *cci_pmu) +{ + int ret = pmu_request_irq(cci_pmu, pmu_handle_irq); + if (ret) { + pmu_free_irq(cci_pmu); + return ret; + } + return 0; +} + +static void cci_pmu_put_hw(struct cci_pmu *cci_pmu) +{ + pmu_free_irq(cci_pmu); +} + +static void hw_perf_event_destroy(struct perf_event *event) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu); + atomic_t *active_events = &cci_pmu->active_events; + struct mutex *reserve_mutex = &cci_pmu->reserve_mutex; + + if (atomic_dec_and_mutex_lock(active_events, reserve_mutex)) { + cci_pmu_put_hw(cci_pmu); + mutex_unlock(reserve_mutex); + } +} + +static void cci_pmu_enable(struct pmu *pmu) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(pmu); + struct cci_pmu_hw_events *hw_events = &cci_pmu->hw_events; + int enabled = bitmap_weight(hw_events->used_mask, cci_pmu->num_events); + unsigned long flags; + u32 val; + + if (!enabled) + return; + + raw_spin_lock_irqsave(&hw_events->pmu_lock, flags); + + /* Enable all the PMU counters. */ + val = readl_relaxed(cci_ctrl_base + CCI_PMCR) | CCI_PMCR_CEN; + writel(val, cci_ctrl_base + CCI_PMCR); + raw_spin_unlock_irqrestore(&hw_events->pmu_lock, flags); + +} + +static void cci_pmu_disable(struct pmu *pmu) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(pmu); + struct cci_pmu_hw_events *hw_events = &cci_pmu->hw_events; + unsigned long flags; + u32 val; + + raw_spin_lock_irqsave(&hw_events->pmu_lock, flags); + + /* Disable all the PMU counters. */ + val = readl_relaxed(cci_ctrl_base + CCI_PMCR) & ~CCI_PMCR_CEN; + writel(val, cci_ctrl_base + CCI_PMCR); + raw_spin_unlock_irqrestore(&hw_events->pmu_lock, flags); +} + +static void cci_pmu_start(struct perf_event *event, int pmu_flags) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu); + struct cci_pmu_hw_events *hw_events = &cci_pmu->hw_events; + struct hw_perf_event *hwc = &event->hw; + int idx = hwc->idx; + unsigned long flags; + + /* + * To handle interrupt latency, we always reprogram the period + * regardlesss of PERF_EF_RELOAD. + */ + if (pmu_flags & PERF_EF_RELOAD) + WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE)); + + hwc->state = 0; + + if (unlikely(!pmu_is_valid_counter(cci_pmu, idx))) { + dev_err(&cci_pmu->plat_device->dev, "Invalid CCI PMU counter %d\n", idx); + return; + } + + raw_spin_lock_irqsave(&hw_events->pmu_lock, flags); + + /* Configure the event to count, unless you are counting cycles */ + if (idx != CCI_PMU_CYCLE_CNTR_IDX) + pmu_set_event(idx, hwc->config_base); + + pmu_event_set_period(event); + pmu_enable_counter(idx); + + raw_spin_unlock_irqrestore(&hw_events->pmu_lock, flags); +} + +static void cci_pmu_stop(struct perf_event *event, int pmu_flags) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu); + struct hw_perf_event *hwc = &event->hw; + int idx = hwc->idx; + + if (hwc->state & PERF_HES_STOPPED) + return; + + if (unlikely(!pmu_is_valid_counter(cci_pmu, idx))) { + dev_err(&cci_pmu->plat_device->dev, "Invalid CCI PMU counter %d\n", idx); + return; + } + + /* + * We always reprogram the counter, so ignore PERF_EF_UPDATE. See + * cci_pmu_start() + */ + pmu_disable_counter(idx); + pmu_event_update(event); + hwc->state |= PERF_HES_STOPPED | PERF_HES_UPTODATE; +} + +static int cci_pmu_add(struct perf_event *event, int flags) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu); + struct cci_pmu_hw_events *hw_events = &cci_pmu->hw_events; + struct hw_perf_event *hwc = &event->hw; + int idx; + int err = 0; + + perf_pmu_disable(event->pmu); + + /* If we don't have a space for the counter then finish early. */ + idx = pmu_get_event_idx(hw_events, event); + if (idx < 0) { + err = idx; + goto out; + } + + event->hw.idx = idx; + hw_events->events[idx] = event; + + hwc->state = PERF_HES_STOPPED | PERF_HES_UPTODATE; + if (flags & PERF_EF_START) + cci_pmu_start(event, PERF_EF_RELOAD); + + /* Propagate our changes to the userspace mapping. */ + perf_event_update_userpage(event); + +out: + perf_pmu_enable(event->pmu); + return err; +} + +static void cci_pmu_del(struct perf_event *event, int flags) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu); + struct cci_pmu_hw_events *hw_events = &cci_pmu->hw_events; + struct hw_perf_event *hwc = &event->hw; + int idx = hwc->idx; + + cci_pmu_stop(event, PERF_EF_UPDATE); + hw_events->events[idx] = NULL; + clear_bit(idx, hw_events->used_mask); + + perf_event_update_userpage(event); +} + +static int +validate_event(struct pmu *cci_pmu, + struct cci_pmu_hw_events *hw_events, + struct perf_event *event) +{ + if (is_software_event(event)) + return 1; + + /* + * Reject groups spanning multiple HW PMUs (e.g. CPU + CCI). The + * core perf code won't check that the pmu->ctx == leader->ctx + * until after pmu->event_init(event). + */ + if (event->pmu != cci_pmu) + return 0; + + if (event->state < PERF_EVENT_STATE_OFF) + return 1; + + if (event->state == PERF_EVENT_STATE_OFF && !event->attr.enable_on_exec) + return 1; + + return pmu_get_event_idx(hw_events, event) >= 0; +} + +static int +validate_group(struct perf_event *event) +{ + struct perf_event *sibling, *leader = event->group_leader; + struct cci_pmu_hw_events fake_pmu = { + /* + * Initialise the fake PMU. We only need to populate the + * used_mask for the purposes of validation. + */ + .used_mask = { 0 }, + }; + + if (!validate_event(event->pmu, &fake_pmu, leader)) + return -EINVAL; + + list_for_each_entry(sibling, &leader->sibling_list, group_entry) { + if (!validate_event(event->pmu, &fake_pmu, sibling)) + return -EINVAL; + } + + if (!validate_event(event->pmu, &fake_pmu, event)) + return -EINVAL; + + return 0; +} + +static int +__hw_perf_event_init(struct perf_event *event) +{ + struct hw_perf_event *hwc = &event->hw; + int mapping; + + mapping = pmu_map_event(event); + + if (mapping < 0) { + pr_debug("event %x:%llx not supported\n", event->attr.type, + event->attr.config); + return mapping; + } + + /* + * We don't assign an index until we actually place the event onto + * hardware. Use -1 to signify that we haven't decided where to put it + * yet. + */ + hwc->idx = -1; + hwc->config_base = 0; + hwc->config = 0; + hwc->event_base = 0; + + /* + * Store the event encoding into the config_base field. + */ + hwc->config_base |= (unsigned long)mapping; + + /* + * Limit the sample_period to half of the counter width. That way, the + * new counter value is far less likely to overtake the previous one + * unless you have some serious IRQ latency issues. + */ + hwc->sample_period = CCI_PMU_CNTR_MASK >> 1; + hwc->last_period = hwc->sample_period; + local64_set(&hwc->period_left, hwc->sample_period); + + if (event->group_leader != event) { + if (validate_group(event) != 0) + return -EINVAL; + } + + return 0; +} + +static int cci_pmu_event_init(struct perf_event *event) +{ + struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu); + atomic_t *active_events = &cci_pmu->active_events; + int err = 0; + int cpu; + + if (event->attr.type != event->pmu->type) + return -ENOENT; + + /* Shared by all CPUs, no meaningful state to sample */ + if (is_sampling_event(event) || event->attach_state & PERF_ATTACH_TASK) + return -EOPNOTSUPP; + + /* We have no filtering of any kind */ + if (event->attr.exclude_user || + event->attr.exclude_kernel || + event->attr.exclude_hv || + event->attr.exclude_idle || + event->attr.exclude_host || + event->attr.exclude_guest) + return -EINVAL; + + /* + * Following the example set by other "uncore" PMUs, we accept any CPU + * and rewrite its affinity dynamically rather than having perf core + * handle cpu == -1 and pid == -1 for this case. + * + * The perf core will pin online CPUs for the duration of this call and + * the event being installed into its context, so the PMU's CPU can't + * change under our feet. + */ + cpu = cpumask_first(&cci_pmu->cpus); + if (event->cpu < 0 || cpu < 0) + return -EINVAL; + event->cpu = cpu; + + event->destroy = hw_perf_event_destroy; + if (!atomic_inc_not_zero(active_events)) { + mutex_lock(&cci_pmu->reserve_mutex); + if (atomic_read(active_events) == 0) + err = cci_pmu_get_hw(cci_pmu); + if (!err) + atomic_inc(active_events); + mutex_unlock(&cci_pmu->reserve_mutex); + } + if (err) + return err; + + err = __hw_perf_event_init(event); + if (err) + hw_perf_event_destroy(event); + + return err; +} + +static ssize_t pmu_attr_cpumask_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + int n = scnprintf(buf, PAGE_SIZE - 1, "%*pbl", + cpumask_pr_args(&pmu->cpus)); + buf[n++] = '\n'; + buf[n] = '\0'; + return n; +} + +static DEVICE_ATTR(cpumask, S_IRUGO, pmu_attr_cpumask_show, NULL); + +static struct attribute *pmu_attrs[] = { + &dev_attr_cpumask.attr, + NULL, +}; + +static struct attribute_group pmu_attr_group = { + .attrs = pmu_attrs, +}; + +static const struct attribute_group *pmu_attr_groups[] = { + &pmu_attr_group, + NULL +}; + +static int cci_pmu_init(struct cci_pmu *cci_pmu, struct platform_device *pdev) +{ + char *name = cci_pmu->model->name; + cci_pmu->pmu = (struct pmu) { + .name = cci_pmu->model->name, + .task_ctx_nr = perf_invalid_context, + .pmu_enable = cci_pmu_enable, + .pmu_disable = cci_pmu_disable, + .event_init = cci_pmu_event_init, + .add = cci_pmu_add, + .del = cci_pmu_del, + .start = cci_pmu_start, + .stop = cci_pmu_stop, + .read = pmu_read, + .attr_groups = pmu_attr_groups, + }; + + cci_pmu->plat_device = pdev; + cci_pmu->num_events = pmu_get_max_counters(); + + return perf_pmu_register(&cci_pmu->pmu, name, -1); +} + +static int cci_pmu_cpu_notifier(struct notifier_block *self, + unsigned long action, void *hcpu) +{ + unsigned int cpu = (long)hcpu; + unsigned int target; + + switch (action & ~CPU_TASKS_FROZEN) { + case CPU_DOWN_PREPARE: + if (!cpumask_test_and_clear_cpu(cpu, &pmu->cpus)) + break; + target = cpumask_any_but(cpu_online_mask, cpu); + if (target < 0) // UP, last CPU + break; + /* + * TODO: migrate context once core races on event->ctx have + * been fixed. + */ + cpumask_set_cpu(target, &pmu->cpus); + default: + break; + } + + return NOTIFY_OK; +} + +static struct notifier_block cci_pmu_cpu_nb = { + .notifier_call = cci_pmu_cpu_notifier, + /* + * to migrate uncore events, our notifier should be executed + * before perf core's notifier. + */ + .priority = CPU_PRI_PERF + 1, +}; + +static struct cci_pmu_model cci_pmu_models[] = { + [CCI_REV_R0] = { + .name = "CCI_400", + .event_ranges = { + [CCI_IF_SLAVE] = { + CCI_REV_R0_SLAVE_PORT_MIN_EV, + CCI_REV_R0_SLAVE_PORT_MAX_EV, + }, + [CCI_IF_MASTER] = { + CCI_REV_R0_MASTER_PORT_MIN_EV, + CCI_REV_R0_MASTER_PORT_MAX_EV, + }, + }, + }, + [CCI_REV_R1] = { + .name = "CCI_400_r1", + .event_ranges = { + [CCI_IF_SLAVE] = { + CCI_REV_R1_SLAVE_PORT_MIN_EV, + CCI_REV_R1_SLAVE_PORT_MAX_EV, + }, + [CCI_IF_MASTER] = { + CCI_REV_R1_MASTER_PORT_MIN_EV, + CCI_REV_R1_MASTER_PORT_MAX_EV, + }, + }, + }, +}; + +static const struct of_device_id arm_cci_pmu_matches[] = { + { + .compatible = "arm,cci-400-pmu", + .data = NULL, + }, + { + .compatible = "arm,cci-400-pmu,r0", + .data = &cci_pmu_models[CCI_REV_R0], + }, + { + .compatible = "arm,cci-400-pmu,r1", + .data = &cci_pmu_models[CCI_REV_R1], + }, + {}, +}; + +static inline const struct cci_pmu_model *get_cci_model(struct platform_device *pdev) +{ + const struct of_device_id *match = of_match_node(arm_cci_pmu_matches, + pdev->dev.of_node); + if (!match) + return NULL; + if (match->data) + return match->data; + + dev_warn(&pdev->dev, "DEPRECATED compatible property," + "requires secure access to CCI registers"); + return probe_cci_model(pdev); +} + +static bool is_duplicate_irq(int irq, int *irqs, int nr_irqs) +{ + int i; + + for (i = 0; i < nr_irqs; i++) + if (irq == irqs[i]) + return true; + + return false; +} + +static int cci_pmu_probe(struct platform_device *pdev) +{ + struct resource *res; + int i, ret, irq; + const struct cci_pmu_model *model; + + model = get_cci_model(pdev); + if (!model) { + dev_warn(&pdev->dev, "CCI PMU version not supported\n"); + return -ENODEV; + } + + pmu = devm_kzalloc(&pdev->dev, sizeof(*pmu), GFP_KERNEL); + if (!pmu) + return -ENOMEM; + + pmu->model = model; + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + pmu->base = devm_ioremap_resource(&pdev->dev, res); + if (IS_ERR(pmu->base)) + return -ENOMEM; + + /* + * CCI PMU has 5 overflow signals - one per counter; but some may be tied + * together to a common interrupt. + */ + pmu->nr_irqs = 0; + for (i = 0; i < CCI_PMU_MAX_HW_EVENTS; i++) { + irq = platform_get_irq(pdev, i); + if (irq < 0) + break; + + if (is_duplicate_irq(irq, pmu->irqs, pmu->nr_irqs)) + continue; + + pmu->irqs[pmu->nr_irqs++] = irq; + } + + /* + * Ensure that the device tree has as many interrupts as the number + * of counters. + */ + if (i < CCI_PMU_MAX_HW_EVENTS) { + dev_warn(&pdev->dev, "In-correct number of interrupts: %d, should be %d\n", + i, CCI_PMU_MAX_HW_EVENTS); + return -EINVAL; + } + + raw_spin_lock_init(&pmu->hw_events.pmu_lock); + mutex_init(&pmu->reserve_mutex); + atomic_set(&pmu->active_events, 0); + cpumask_set_cpu(smp_processor_id(), &pmu->cpus); + + ret = register_cpu_notifier(&cci_pmu_cpu_nb); + if (ret) + return ret; + + ret = cci_pmu_init(pmu, pdev); + if (ret) + return ret; + + pr_info("ARM %s PMU driver probed", pmu->model->name); + return 0; +} + +static int cci_platform_probe(struct platform_device *pdev) +{ + if (!cci_probed()) + return -ENODEV; + + return of_platform_populate(pdev->dev.of_node, NULL, NULL, &pdev->dev); +} + +static struct platform_driver cci_pmu_driver = { + .driver = { + .name = DRIVER_NAME_PMU, + .of_match_table = arm_cci_pmu_matches, + }, + .probe = cci_pmu_probe, +}; + +static struct platform_driver cci_platform_driver = { + .driver = { + .name = DRIVER_NAME, + .of_match_table = arm_cci_matches, + }, + .probe = cci_platform_probe, +}; + +static int __init cci_platform_init(void) +{ + int ret; + + ret = platform_driver_register(&cci_pmu_driver); + if (ret) + return ret; + + return platform_driver_register(&cci_platform_driver); +} + +#else /* !CONFIG_ARM_CCI400_PMU */ + +static int __init cci_platform_init(void) +{ + return 0; +} + +#endif /* CONFIG_ARM_CCI400_PMU */ + +#ifdef CONFIG_ARM_CCI400_PORT_CTRL + +#define CCI_PORT_CTRL 0x0 +#define CCI_CTRL_STATUS 0xc + +#define CCI_ENABLE_SNOOP_REQ 0x1 +#define CCI_ENABLE_DVM_REQ 0x2 +#define CCI_ENABLE_REQ (CCI_ENABLE_SNOOP_REQ | CCI_ENABLE_DVM_REQ) + +enum cci_ace_port_type { + ACE_INVALID_PORT = 0x0, + ACE_PORT, + ACE_LITE_PORT, +}; + +struct cci_ace_port { + void __iomem *base; + unsigned long phys; + enum cci_ace_port_type type; + struct device_node *dn; +}; + +static struct cci_ace_port *ports; +static unsigned int nb_cci_ports; + +struct cpu_port { + u64 mpidr; + u32 port; +}; + +/* + * Use the port MSB as valid flag, shift can be made dynamic + * by computing number of bits required for port indexes. + * Code disabling CCI cpu ports runs with D-cache invalidated + * and SCTLR bit clear so data accesses must be kept to a minimum + * to improve performance; for now shift is left static to + * avoid one more data access while disabling the CCI port. + */ +#define PORT_VALID_SHIFT 31 +#define PORT_VALID (0x1 << PORT_VALID_SHIFT) + +static inline void init_cpu_port(struct cpu_port *port, u32 index, u64 mpidr) +{ + port->port = PORT_VALID | index; + port->mpidr = mpidr; +} + +static inline bool cpu_port_is_valid(struct cpu_port *port) +{ + return !!(port->port & PORT_VALID); +} + +static inline bool cpu_port_match(struct cpu_port *port, u64 mpidr) +{ + return port->mpidr == (mpidr & MPIDR_HWID_BITMASK); +} + +static struct cpu_port cpu_port[NR_CPUS]; + +/** + * __cci_ace_get_port - Function to retrieve the port index connected to + * a cpu or device. + * + * @dn: device node of the device to look-up + * @type: port type + * + * Return value: + * - CCI port index if success + * - -ENODEV if failure + */ +static int __cci_ace_get_port(struct device_node *dn, int type) +{ + int i; + bool ace_match; + struct device_node *cci_portn; + + cci_portn = of_parse_phandle(dn, "cci-control-port", 0); + for (i = 0; i < nb_cci_ports; i++) { + ace_match = ports[i].type == type; + if (ace_match && cci_portn == ports[i].dn) + return i; + } + return -ENODEV; +} + +int cci_ace_get_port(struct device_node *dn) +{ + return __cci_ace_get_port(dn, ACE_LITE_PORT); +} +EXPORT_SYMBOL_GPL(cci_ace_get_port); + +static void cci_ace_init_ports(void) +{ + int port, cpu; + struct device_node *cpun; + + /* + * Port index look-up speeds up the function disabling ports by CPU, + * since the logical to port index mapping is done once and does + * not change after system boot. + * The stashed index array is initialized for all possible CPUs + * at probe time. + */ + for_each_possible_cpu(cpu) { + /* too early to use cpu->of_node */ + cpun = of_get_cpu_node(cpu, NULL); + + if (WARN(!cpun, "Missing cpu device node\n")) + continue; + + port = __cci_ace_get_port(cpun, ACE_PORT); + if (port < 0) + continue; + + init_cpu_port(&cpu_port[cpu], port, cpu_logical_map(cpu)); + } + + for_each_possible_cpu(cpu) { + WARN(!cpu_port_is_valid(&cpu_port[cpu]), + "CPU %u does not have an associated CCI port\n", + cpu); + } +} +/* + * Functions to enable/disable a CCI interconnect slave port + * + * They are called by low-level power management code to disable slave + * interfaces snoops and DVM broadcast. + * Since they may execute with cache data allocation disabled and + * after the caches have been cleaned and invalidated the functions provide + * no explicit locking since they may run with D-cache disabled, so normal + * cacheable kernel locks based on ldrex/strex may not work. + * Locking has to be provided by BSP implementations to ensure proper + * operations. + */ + +/** + * cci_port_control() - function to control a CCI port + * + * @port: index of the port to setup + * @enable: if true enables the port, if false disables it + */ +static void notrace cci_port_control(unsigned int port, bool enable) +{ + void __iomem *base = ports[port].base; + + writel_relaxed(enable ? CCI_ENABLE_REQ : 0, base + CCI_PORT_CTRL); + /* + * This function is called from power down procedures + * and must not execute any instruction that might + * cause the processor to be put in a quiescent state + * (eg wfi). Hence, cpu_relax() can not be added to this + * read loop to optimize power, since it might hide possibly + * disruptive operations. + */ + while (readl_relaxed(cci_ctrl_base + CCI_CTRL_STATUS) & 0x1) + ; +} + +/** + * cci_disable_port_by_cpu() - function to disable a CCI port by CPU + * reference + * + * @mpidr: mpidr of the CPU whose CCI port should be disabled + * + * Disabling a CCI port for a CPU implies disabling the CCI port + * controlling that CPU cluster. Code disabling CPU CCI ports + * must make sure that the CPU running the code is the last active CPU + * in the cluster ie all other CPUs are quiescent in a low power state. + * + * Return: + * 0 on success + * -ENODEV on port look-up failure + */ +int notrace cci_disable_port_by_cpu(u64 mpidr) +{ + int cpu; + bool is_valid; + for (cpu = 0; cpu < nr_cpu_ids; cpu++) { + is_valid = cpu_port_is_valid(&cpu_port[cpu]); + if (is_valid && cpu_port_match(&cpu_port[cpu], mpidr)) { + cci_port_control(cpu_port[cpu].port, false); + return 0; + } + } + return -ENODEV; +} +EXPORT_SYMBOL_GPL(cci_disable_port_by_cpu); + +/** + * cci_enable_port_for_self() - enable a CCI port for calling CPU + * + * Enabling a CCI port for the calling CPU implies enabling the CCI + * port controlling that CPU's cluster. Caller must make sure that the + * CPU running the code is the first active CPU in the cluster and all + * other CPUs are quiescent in a low power state or waiting for this CPU + * to complete the CCI initialization. + * + * Because this is called when the MMU is still off and with no stack, + * the code must be position independent and ideally rely on callee + * clobbered registers only. To achieve this we must code this function + * entirely in assembler. + * + * On success this returns with the proper CCI port enabled. In case of + * any failure this never returns as the inability to enable the CCI is + * fatal and there is no possible recovery at this stage. + */ +asmlinkage void __naked cci_enable_port_for_self(void) +{ + asm volatile ("\n" +" .arch armv7-a\n" +" mrc p15, 0, r0, c0, c0, 5 @ get MPIDR value \n" +" and r0, r0, #"__stringify(MPIDR_HWID_BITMASK)" \n" +" adr r1, 5f \n" +" ldr r2, [r1] \n" +" add r1, r1, r2 @ &cpu_port \n" +" add ip, r1, %[sizeof_cpu_port] \n" + + /* Loop over the cpu_port array looking for a matching MPIDR */ +"1: ldr r2, [r1, %[offsetof_cpu_port_mpidr_lsb]] \n" +" cmp r2, r0 @ compare MPIDR \n" +" bne 2f \n" + + /* Found a match, now test port validity */ +" ldr r3, [r1, %[offsetof_cpu_port_port]] \n" +" tst r3, #"__stringify(PORT_VALID)" \n" +" bne 3f \n" + + /* no match, loop with the next cpu_port entry */ +"2: add r1, r1, %[sizeof_struct_cpu_port] \n" +" cmp r1, ip @ done? \n" +" blo 1b \n" + + /* CCI port not found -- cheaply try to stall this CPU */ +"cci_port_not_found: \n" +" wfi \n" +" wfe \n" +" b cci_port_not_found \n" + + /* Use matched port index to look up the corresponding ports entry */ +"3: bic r3, r3, #"__stringify(PORT_VALID)" \n" +" adr r0, 6f \n" +" ldmia r0, {r1, r2} \n" +" sub r1, r1, r0 @ virt - phys \n" +" ldr r0, [r0, r2] @ *(&ports) \n" +" mov r2, %[sizeof_struct_ace_port] \n" +" mla r0, r2, r3, r0 @ &ports[index] \n" +" sub r0, r0, r1 @ virt_to_phys() \n" + + /* Enable the CCI port */ +" ldr r0, [r0, %[offsetof_port_phys]] \n" +" mov r3, %[cci_enable_req]\n" +" str r3, [r0, #"__stringify(CCI_PORT_CTRL)"] \n" + + /* poll the status reg for completion */ +" adr r1, 7f \n" +" ldr r0, [r1] \n" +" ldr r0, [r0, r1] @ cci_ctrl_base \n" +"4: ldr r1, [r0, #"__stringify(CCI_CTRL_STATUS)"] \n" +" tst r1, %[cci_control_status_bits] \n" +" bne 4b \n" + +" mov r0, #0 \n" +" bx lr \n" + +" .align 2 \n" +"5: .word cpu_port - . \n" +"6: .word . \n" +" .word ports - 6b \n" +"7: .word cci_ctrl_phys - . \n" + : : + [sizeof_cpu_port] "i" (sizeof(cpu_port)), + [cci_enable_req] "i" cpu_to_le32(CCI_ENABLE_REQ), + [cci_control_status_bits] "i" cpu_to_le32(1), +#ifndef __ARMEB__ + [offsetof_cpu_port_mpidr_lsb] "i" (offsetof(struct cpu_port, mpidr)), +#else + [offsetof_cpu_port_mpidr_lsb] "i" (offsetof(struct cpu_port, mpidr)+4), +#endif + [offsetof_cpu_port_port] "i" (offsetof(struct cpu_port, port)), + [sizeof_struct_cpu_port] "i" (sizeof(struct cpu_port)), + [sizeof_struct_ace_port] "i" (sizeof(struct cci_ace_port)), + [offsetof_port_phys] "i" (offsetof(struct cci_ace_port, phys)) ); + + unreachable(); +} + +/** + * __cci_control_port_by_device() - function to control a CCI port by device + * reference + * + * @dn: device node pointer of the device whose CCI port should be + * controlled + * @enable: if true enables the port, if false disables it + * + * Return: + * 0 on success + * -ENODEV on port look-up failure + */ +int notrace __cci_control_port_by_device(struct device_node *dn, bool enable) +{ + int port; + + if (!dn) + return -ENODEV; + + port = __cci_ace_get_port(dn, ACE_LITE_PORT); + if (WARN_ONCE(port < 0, "node %s ACE lite port look-up failure\n", + dn->full_name)) + return -ENODEV; + cci_port_control(port, enable); + return 0; +} +EXPORT_SYMBOL_GPL(__cci_control_port_by_device); + +/** + * __cci_control_port_by_index() - function to control a CCI port by port index + * + * @port: port index previously retrieved with cci_ace_get_port() + * @enable: if true enables the port, if false disables it + * + * Return: + * 0 on success + * -ENODEV on port index out of range + * -EPERM if operation carried out on an ACE PORT + */ +int notrace __cci_control_port_by_index(u32 port, bool enable) +{ + if (port >= nb_cci_ports || ports[port].type == ACE_INVALID_PORT) + return -ENODEV; + /* + * CCI control for ports connected to CPUS is extremely fragile + * and must be made to go through a specific and controlled + * interface (ie cci_disable_port_by_cpu(); control by general purpose + * indexing is therefore disabled for ACE ports. + */ + if (ports[port].type == ACE_PORT) + return -EPERM; + + cci_port_control(port, enable); + return 0; +} +EXPORT_SYMBOL_GPL(__cci_control_port_by_index); + +static const struct of_device_id arm_cci_ctrl_if_matches[] = { + {.compatible = "arm,cci-400-ctrl-if", }, + {}, +}; + +static int cci_probe_ports(struct device_node *np) +{ + struct cci_nb_ports const *cci_config; + int ret, i, nb_ace = 0, nb_ace_lite = 0; + struct device_node *cp; + struct resource res; + const char *match_str; + bool is_ace; + + + cci_config = of_match_node(arm_cci_matches, np)->data; + if (!cci_config) + return -ENODEV; + + nb_cci_ports = cci_config->nb_ace + cci_config->nb_ace_lite; + + ports = kcalloc(nb_cci_ports, sizeof(*ports), GFP_KERNEL); + if (!ports) + return -ENOMEM; + + for_each_child_of_node(np, cp) { + if (!of_match_node(arm_cci_ctrl_if_matches, cp)) + continue; + + i = nb_ace + nb_ace_lite; + + if (i >= nb_cci_ports) + break; + + if (of_property_read_string(cp, "interface-type", + &match_str)) { + WARN(1, "node %s missing interface-type property\n", + cp->full_name); + continue; + } + is_ace = strcmp(match_str, "ace") == 0; + if (!is_ace && strcmp(match_str, "ace-lite")) { + WARN(1, "node %s containing invalid interface-type property, skipping it\n", + cp->full_name); + continue; + } + + ret = of_address_to_resource(cp, 0, &res); + if (!ret) { + ports[i].base = ioremap(res.start, resource_size(&res)); + ports[i].phys = res.start; + } + if (ret || !ports[i].base) { + WARN(1, "unable to ioremap CCI port %d\n", i); + continue; + } + + if (is_ace) { + if (WARN_ON(nb_ace >= cci_config->nb_ace)) + continue; + ports[i].type = ACE_PORT; + ++nb_ace; + } else { + if (WARN_ON(nb_ace_lite >= cci_config->nb_ace_lite)) + continue; + ports[i].type = ACE_LITE_PORT; + ++nb_ace_lite; + } + ports[i].dn = cp; + } + + /* initialize a stashed array of ACE ports to speed-up look-up */ + cci_ace_init_ports(); + + /* + * Multi-cluster systems may need this data when non-coherent, during + * cluster power-up/power-down. Make sure it reaches main memory. + */ + sync_cache_w(&cci_ctrl_base); + sync_cache_w(&cci_ctrl_phys); + sync_cache_w(&ports); + sync_cache_w(&cpu_port); + __sync_cache_range_w(ports, sizeof(*ports) * nb_cci_ports); + pr_info("ARM CCI driver probed\n"); + + return 0; +} +#else /* !CONFIG_ARM_CCI400_PORT_CTRL */ +static inline int cci_probe_ports(struct device_node *np) +{ + return 0; +} +#endif /* CONFIG_ARM_CCI400_PORT_CTRL */ + +static int cci_probe(void) +{ + int ret; + struct device_node *np; + struct resource res; + + np = of_find_matching_node(NULL, arm_cci_matches); + if(!np || !of_device_is_available(np)) + return -ENODEV; + + ret = of_address_to_resource(np, 0, &res); + if (!ret) { + cci_ctrl_base = ioremap(res.start, resource_size(&res)); + cci_ctrl_phys = res.start; + } + if (ret || !cci_ctrl_base) { + WARN(1, "unable to ioremap CCI ctrl\n"); + return -ENXIO; + } + + return cci_probe_ports(np); +} + +static int cci_init_status = -EAGAIN; +static DEFINE_MUTEX(cci_probing); + +static int cci_init(void) +{ + if (cci_init_status != -EAGAIN) + return cci_init_status; + + mutex_lock(&cci_probing); + if (cci_init_status == -EAGAIN) + cci_init_status = cci_probe(); + mutex_unlock(&cci_probing); + return cci_init_status; +} + +/* + * To sort out early init calls ordering a helper function is provided to + * check if the CCI driver has beed initialized. Function check if the driver + * has been initialized, if not it calls the init function that probes + * the driver and updates the return value. + */ +bool cci_probed(void) +{ + return cci_init() == 0; +} +EXPORT_SYMBOL_GPL(cci_probed); + +early_initcall(cci_init); +core_initcall(cci_platform_init); +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("ARM CCI support"); diff --git a/drivers/bus/arm-ccn.c b/drivers/bus/arm-ccn.c new file mode 100644 index 000000000..60397ec77 --- /dev/null +++ b/drivers/bus/arm-ccn.c @@ -0,0 +1,1393 @@ +/* + * 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. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * Copyright (C) 2014 ARM Limited + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#define CCN_NUM_XP_PORTS 2 +#define CCN_NUM_VCS 4 +#define CCN_NUM_REGIONS 256 +#define CCN_REGION_SIZE 0x10000 + +#define CCN_ALL_OLY_ID 0xff00 +#define CCN_ALL_OLY_ID__OLY_ID__SHIFT 0 +#define CCN_ALL_OLY_ID__OLY_ID__MASK 0x1f +#define CCN_ALL_OLY_ID__NODE_ID__SHIFT 8 +#define CCN_ALL_OLY_ID__NODE_ID__MASK 0x3f + +#define CCN_MN_ERRINT_STATUS 0x0008 +#define CCN_MN_ERRINT_STATUS__INTREQ__DESSERT 0x11 +#define CCN_MN_ERRINT_STATUS__ALL_ERRORS__ENABLE 0x02 +#define CCN_MN_ERRINT_STATUS__ALL_ERRORS__DISABLED 0x20 +#define CCN_MN_ERRINT_STATUS__ALL_ERRORS__DISABLE 0x22 +#define CCN_MN_ERRINT_STATUS__CORRECTED_ERRORS_ENABLE 0x04 +#define CCN_MN_ERRINT_STATUS__CORRECTED_ERRORS_DISABLED 0x40 +#define CCN_MN_ERRINT_STATUS__CORRECTED_ERRORS_DISABLE 0x44 +#define CCN_MN_ERRINT_STATUS__PMU_EVENTS__ENABLE 0x08 +#define CCN_MN_ERRINT_STATUS__PMU_EVENTS__DISABLED 0x80 +#define CCN_MN_ERRINT_STATUS__PMU_EVENTS__DISABLE 0x88 +#define CCN_MN_OLY_COMP_LIST_63_0 0x01e0 +#define CCN_MN_ERR_SIG_VAL_63_0 0x0300 +#define CCN_MN_ERR_SIG_VAL_63_0__DT (1 << 1) + +#define CCN_DT_ACTIVE_DSM 0x0000 +#define CCN_DT_ACTIVE_DSM__DSM_ID__SHIFT(n) ((n) * 8) +#define CCN_DT_ACTIVE_DSM__DSM_ID__MASK 0xff +#define CCN_DT_CTL 0x0028 +#define CCN_DT_CTL__DT_EN (1 << 0) +#define CCN_DT_PMEVCNT(n) (0x0100 + (n) * 0x8) +#define CCN_DT_PMCCNTR 0x0140 +#define CCN_DT_PMCCNTRSR 0x0190 +#define CCN_DT_PMOVSR 0x0198 +#define CCN_DT_PMOVSR_CLR 0x01a0 +#define CCN_DT_PMOVSR_CLR__MASK 0x1f +#define CCN_DT_PMCR 0x01a8 +#define CCN_DT_PMCR__OVFL_INTR_EN (1 << 6) +#define CCN_DT_PMCR__PMU_EN (1 << 0) +#define CCN_DT_PMSR 0x01b0 +#define CCN_DT_PMSR_REQ 0x01b8 +#define CCN_DT_PMSR_CLR 0x01c0 + +#define CCN_HNF_PMU_EVENT_SEL 0x0600 +#define CCN_HNF_PMU_EVENT_SEL__ID__SHIFT(n) ((n) * 4) +#define CCN_HNF_PMU_EVENT_SEL__ID__MASK 0xf + +#define CCN_XP_DT_CONFIG 0x0300 +#define CCN_XP_DT_CONFIG__DT_CFG__SHIFT(n) ((n) * 4) +#define CCN_XP_DT_CONFIG__DT_CFG__MASK 0xf +#define CCN_XP_DT_CONFIG__DT_CFG__PASS_THROUGH 0x0 +#define CCN_XP_DT_CONFIG__DT_CFG__WATCHPOINT_0_OR_1 0x1 +#define CCN_XP_DT_CONFIG__DT_CFG__WATCHPOINT(n) (0x2 + (n)) +#define CCN_XP_DT_CONFIG__DT_CFG__XP_PMU_EVENT(n) (0x4 + (n)) +#define CCN_XP_DT_CONFIG__DT_CFG__DEVICE_PMU_EVENT(d, n) (0x8 + (d) * 4 + (n)) +#define CCN_XP_DT_INTERFACE_SEL 0x0308 +#define CCN_XP_DT_INTERFACE_SEL__DT_IO_SEL__SHIFT(n) (0 + (n) * 8) +#define CCN_XP_DT_INTERFACE_SEL__DT_IO_SEL__MASK 0x1 +#define CCN_XP_DT_INTERFACE_SEL__DT_DEV_SEL__SHIFT(n) (1 + (n) * 8) +#define CCN_XP_DT_INTERFACE_SEL__DT_DEV_SEL__MASK 0x1 +#define CCN_XP_DT_INTERFACE_SEL__DT_VC_SEL__SHIFT(n) (2 + (n) * 8) +#define CCN_XP_DT_INTERFACE_SEL__DT_VC_SEL__MASK 0x3 +#define CCN_XP_DT_CMP_VAL_L(n) (0x0310 + (n) * 0x40) +#define CCN_XP_DT_CMP_VAL_H(n) (0x0318 + (n) * 0x40) +#define CCN_XP_DT_CMP_MASK_L(n) (0x0320 + (n) * 0x40) +#define CCN_XP_DT_CMP_MASK_H(n) (0x0328 + (n) * 0x40) +#define CCN_XP_DT_CONTROL 0x0370 +#define CCN_XP_DT_CONTROL__DT_ENABLE (1 << 0) +#define CCN_XP_DT_CONTROL__WP_ARM_SEL__SHIFT(n) (12 + (n) * 4) +#define CCN_XP_DT_CONTROL__WP_ARM_SEL__MASK 0xf +#define CCN_XP_DT_CONTROL__WP_ARM_SEL__ALWAYS 0xf +#define CCN_XP_PMU_EVENT_SEL 0x0600 +#define CCN_XP_PMU_EVENT_SEL__ID__SHIFT(n) ((n) * 7) +#define CCN_XP_PMU_EVENT_SEL__ID__MASK 0x3f + +#define CCN_SBAS_PMU_EVENT_SEL 0x0600 +#define CCN_SBAS_PMU_EVENT_SEL__ID__SHIFT(n) ((n) * 4) +#define CCN_SBAS_PMU_EVENT_SEL__ID__MASK 0xf + +#define CCN_RNI_PMU_EVENT_SEL 0x0600 +#define CCN_RNI_PMU_EVENT_SEL__ID__SHIFT(n) ((n) * 4) +#define CCN_RNI_PMU_EVENT_SEL__ID__MASK 0xf + +#define CCN_TYPE_MN 0x01 +#define CCN_TYPE_DT 0x02 +#define CCN_TYPE_HNF 0x04 +#define CCN_TYPE_HNI 0x05 +#define CCN_TYPE_XP 0x08 +#define CCN_TYPE_SBSX 0x0c +#define CCN_TYPE_SBAS 0x10 +#define CCN_TYPE_RNI_1P 0x14 +#define CCN_TYPE_RNI_2P 0x15 +#define CCN_TYPE_RNI_3P 0x16 +#define CCN_TYPE_RND_1P 0x18 /* RN-D = RN-I + DVM */ +#define CCN_TYPE_RND_2P 0x19 +#define CCN_TYPE_RND_3P 0x1a +#define CCN_TYPE_CYCLES 0xff /* Pseudotype */ + +#define CCN_EVENT_WATCHPOINT 0xfe /* Pseudoevent */ + +#define CCN_NUM_PMU_EVENTS 4 +#define CCN_NUM_XP_WATCHPOINTS 2 /* See DT.dbg_id.num_watchpoints */ +#define CCN_NUM_PMU_EVENT_COUNTERS 8 /* See DT.dbg_id.num_pmucntr */ +#define CCN_IDX_PMU_CYCLE_COUNTER CCN_NUM_PMU_EVENT_COUNTERS + +#define CCN_NUM_PREDEFINED_MASKS 4 +#define CCN_IDX_MASK_ANY (CCN_NUM_PMU_EVENT_COUNTERS + 0) +#define CCN_IDX_MASK_EXACT (CCN_NUM_PMU_EVENT_COUNTERS + 1) +#define CCN_IDX_MASK_ORDER (CCN_NUM_PMU_EVENT_COUNTERS + 2) +#define CCN_IDX_MASK_OPCODE (CCN_NUM_PMU_EVENT_COUNTERS + 3) + +struct arm_ccn_component { + void __iomem *base; + u32 type; + + DECLARE_BITMAP(pmu_events_mask, CCN_NUM_PMU_EVENTS); + union { + struct { + DECLARE_BITMAP(dt_cmp_mask, CCN_NUM_XP_WATCHPOINTS); + } xp; + }; +}; + +#define pmu_to_arm_ccn(_pmu) container_of(container_of(_pmu, \ + struct arm_ccn_dt, pmu), struct arm_ccn, dt) + +struct arm_ccn_dt { + int id; + void __iomem *base; + + spinlock_t config_lock; + + DECLARE_BITMAP(pmu_counters_mask, CCN_NUM_PMU_EVENT_COUNTERS + 1); + struct { + struct arm_ccn_component *source; + struct perf_event *event; + } pmu_counters[CCN_NUM_PMU_EVENT_COUNTERS + 1]; + + struct { + u64 l, h; + } cmp_mask[CCN_NUM_PMU_EVENT_COUNTERS + CCN_NUM_PREDEFINED_MASKS]; + + struct hrtimer hrtimer; + + struct pmu pmu; +}; + +struct arm_ccn { + struct device *dev; + void __iomem *base; + unsigned irq_used:1; + unsigned sbas_present:1; + unsigned sbsx_present:1; + + int num_nodes; + struct arm_ccn_component *node; + + int num_xps; + struct arm_ccn_component *xp; + + struct arm_ccn_dt dt; +}; + + +static int arm_ccn_node_to_xp(int node) +{ + return node / CCN_NUM_XP_PORTS; +} + +static int arm_ccn_node_to_xp_port(int node) +{ + return node % CCN_NUM_XP_PORTS; +} + + +/* + * Bit shifts and masks in these defines must be kept in sync with + * arm_ccn_pmu_config_set() and CCN_FORMAT_ATTRs below! + */ +#define CCN_CONFIG_NODE(_config) (((_config) >> 0) & 0xff) +#define CCN_CONFIG_XP(_config) (((_config) >> 0) & 0xff) +#define CCN_CONFIG_TYPE(_config) (((_config) >> 8) & 0xff) +#define CCN_CONFIG_EVENT(_config) (((_config) >> 16) & 0xff) +#define CCN_CONFIG_PORT(_config) (((_config) >> 24) & 0x3) +#define CCN_CONFIG_VC(_config) (((_config) >> 26) & 0x7) +#define CCN_CONFIG_DIR(_config) (((_config) >> 29) & 0x1) +#define CCN_CONFIG_MASK(_config) (((_config) >> 30) & 0xf) + +static void arm_ccn_pmu_config_set(u64 *config, u32 node_xp, u32 type, u32 port) +{ + *config &= ~((0xff << 0) | (0xff << 8) | (0x3 << 24)); + *config |= (node_xp << 0) | (type << 8) | (port << 24); +} + +static ssize_t arm_ccn_pmu_format_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct dev_ext_attribute *ea = container_of(attr, + struct dev_ext_attribute, attr); + + return snprintf(buf, PAGE_SIZE, "%s\n", (char *)ea->var); +} + +#define CCN_FORMAT_ATTR(_name, _config) \ + struct dev_ext_attribute arm_ccn_pmu_format_attr_##_name = \ + { __ATTR(_name, S_IRUGO, arm_ccn_pmu_format_show, \ + NULL), _config } + +static CCN_FORMAT_ATTR(node, "config:0-7"); +static CCN_FORMAT_ATTR(xp, "config:0-7"); +static CCN_FORMAT_ATTR(type, "config:8-15"); +static CCN_FORMAT_ATTR(event, "config:16-23"); +static CCN_FORMAT_ATTR(port, "config:24-25"); +static CCN_FORMAT_ATTR(vc, "config:26-28"); +static CCN_FORMAT_ATTR(dir, "config:29-29"); +static CCN_FORMAT_ATTR(mask, "config:30-33"); +static CCN_FORMAT_ATTR(cmp_l, "config1:0-62"); +static CCN_FORMAT_ATTR(cmp_h, "config2:0-59"); + +static struct attribute *arm_ccn_pmu_format_attrs[] = { + &arm_ccn_pmu_format_attr_node.attr.attr, + &arm_ccn_pmu_format_attr_xp.attr.attr, + &arm_ccn_pmu_format_attr_type.attr.attr, + &arm_ccn_pmu_format_attr_event.attr.attr, + &arm_ccn_pmu_format_attr_port.attr.attr, + &arm_ccn_pmu_format_attr_vc.attr.attr, + &arm_ccn_pmu_format_attr_dir.attr.attr, + &arm_ccn_pmu_format_attr_mask.attr.attr, + &arm_ccn_pmu_format_attr_cmp_l.attr.attr, + &arm_ccn_pmu_format_attr_cmp_h.attr.attr, + NULL +}; + +static struct attribute_group arm_ccn_pmu_format_attr_group = { + .name = "format", + .attrs = arm_ccn_pmu_format_attrs, +}; + + +struct arm_ccn_pmu_event { + struct device_attribute attr; + u32 type; + u32 event; + int num_ports; + int num_vcs; + const char *def; + int mask; +}; + +#define CCN_EVENT_ATTR(_name) \ + __ATTR(_name, S_IRUGO, arm_ccn_pmu_event_show, NULL) + +/* + * Events defined in TRM for MN, HN-I and SBSX are actually watchpoints set on + * their ports in XP they are connected to. For the sake of usability they are + * explicitly defined here (and translated into a relevant watchpoint in + * arm_ccn_pmu_event_init()) so the user can easily request them without deep + * knowledge of the flit format. + */ + +#define CCN_EVENT_MN(_name, _def, _mask) { .attr = CCN_EVENT_ATTR(mn_##_name), \ + .type = CCN_TYPE_MN, .event = CCN_EVENT_WATCHPOINT, \ + .num_ports = CCN_NUM_XP_PORTS, .num_vcs = CCN_NUM_VCS, \ + .def = _def, .mask = _mask, } + +#define CCN_EVENT_HNI(_name, _def, _mask) { \ + .attr = CCN_EVENT_ATTR(hni_##_name), .type = CCN_TYPE_HNI, \ + .event = CCN_EVENT_WATCHPOINT, .num_ports = CCN_NUM_XP_PORTS, \ + .num_vcs = CCN_NUM_VCS, .def = _def, .mask = _mask, } + +#define CCN_EVENT_SBSX(_name, _def, _mask) { \ + .attr = CCN_EVENT_ATTR(sbsx_##_name), .type = CCN_TYPE_SBSX, \ + .event = CCN_EVENT_WATCHPOINT, .num_ports = CCN_NUM_XP_PORTS, \ + .num_vcs = CCN_NUM_VCS, .def = _def, .mask = _mask, } + +#define CCN_EVENT_HNF(_name, _event) { .attr = CCN_EVENT_ATTR(hnf_##_name), \ + .type = CCN_TYPE_HNF, .event = _event, } + +#define CCN_EVENT_XP(_name, _event) { .attr = CCN_EVENT_ATTR(xp_##_name), \ + .type = CCN_TYPE_XP, .event = _event, \ + .num_ports = CCN_NUM_XP_PORTS, .num_vcs = CCN_NUM_VCS, } + +/* + * RN-I & RN-D (RN-D = RN-I + DVM) nodes have different type ID depending + * on configuration. One of them is picked to represent the whole group, + * as they all share the same event types. + */ +#define CCN_EVENT_RNI(_name, _event) { .attr = CCN_EVENT_ATTR(rni_##_name), \ + .type = CCN_TYPE_RNI_3P, .event = _event, } + +#define CCN_EVENT_SBAS(_name, _event) { .attr = CCN_EVENT_ATTR(sbas_##_name), \ + .type = CCN_TYPE_SBAS, .event = _event, } + +#define CCN_EVENT_CYCLES(_name) { .attr = CCN_EVENT_ATTR(_name), \ + .type = CCN_TYPE_CYCLES } + + +static ssize_t arm_ccn_pmu_event_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct arm_ccn_pmu_event *event = container_of(attr, + struct arm_ccn_pmu_event, attr); + ssize_t res; + + res = snprintf(buf, PAGE_SIZE, "type=0x%x", event->type); + if (event->event) + res += snprintf(buf + res, PAGE_SIZE - res, ",event=0x%x", + event->event); + if (event->def) + res += snprintf(buf + res, PAGE_SIZE - res, ",%s", + event->def); + if (event->mask) + res += snprintf(buf + res, PAGE_SIZE - res, ",mask=0x%x", + event->mask); + res += snprintf(buf + res, PAGE_SIZE - res, "\n"); + + return res; +} + +static umode_t arm_ccn_pmu_events_is_visible(struct kobject *kobj, + struct attribute *attr, int index) +{ + struct device *dev = kobj_to_dev(kobj); + struct arm_ccn *ccn = pmu_to_arm_ccn(dev_get_drvdata(dev)); + struct device_attribute *dev_attr = container_of(attr, + struct device_attribute, attr); + struct arm_ccn_pmu_event *event = container_of(dev_attr, + struct arm_ccn_pmu_event, attr); + + if (event->type == CCN_TYPE_SBAS && !ccn->sbas_present) + return 0; + if (event->type == CCN_TYPE_SBSX && !ccn->sbsx_present) + return 0; + + return attr->mode; +} + +static struct arm_ccn_pmu_event arm_ccn_pmu_events[] = { + CCN_EVENT_MN(eobarrier, "dir=0,vc=0,cmp_h=0x1c00", CCN_IDX_MASK_OPCODE), + CCN_EVENT_MN(ecbarrier, "dir=0,vc=0,cmp_h=0x1e00", CCN_IDX_MASK_OPCODE), + CCN_EVENT_MN(dvmop, "dir=0,vc=0,cmp_h=0x2800", CCN_IDX_MASK_OPCODE), + CCN_EVENT_HNI(txdatflits, "dir=1,vc=3", CCN_IDX_MASK_ANY), + CCN_EVENT_HNI(rxdatflits, "dir=0,vc=3", CCN_IDX_MASK_ANY), + CCN_EVENT_HNI(txreqflits, "dir=1,vc=0", CCN_IDX_MASK_ANY), + CCN_EVENT_HNI(rxreqflits, "dir=0,vc=0", CCN_IDX_MASK_ANY), + CCN_EVENT_HNI(rxreqflits_order, "dir=0,vc=0,cmp_h=0x8000", + CCN_IDX_MASK_ORDER), + CCN_EVENT_SBSX(txdatflits, "dir=1,vc=3", CCN_IDX_MASK_ANY), + CCN_EVENT_SBSX(rxdatflits, "dir=0,vc=3", CCN_IDX_MASK_ANY), + CCN_EVENT_SBSX(txreqflits, "dir=1,vc=0", CCN_IDX_MASK_ANY), + CCN_EVENT_SBSX(rxreqflits, "dir=0,vc=0", CCN_IDX_MASK_ANY), + CCN_EVENT_SBSX(rxreqflits_order, "dir=0,vc=0,cmp_h=0x8000", + CCN_IDX_MASK_ORDER), + CCN_EVENT_HNF(cache_miss, 0x1), + CCN_EVENT_HNF(l3_sf_cache_access, 0x02), + CCN_EVENT_HNF(cache_fill, 0x3), + CCN_EVENT_HNF(pocq_retry, 0x4), + CCN_EVENT_HNF(pocq_reqs_recvd, 0x5), + CCN_EVENT_HNF(sf_hit, 0x6), + CCN_EVENT_HNF(sf_evictions, 0x7), + CCN_EVENT_HNF(snoops_sent, 0x8), + CCN_EVENT_HNF(snoops_broadcast, 0x9), + CCN_EVENT_HNF(l3_eviction, 0xa), + CCN_EVENT_HNF(l3_fill_invalid_way, 0xb), + CCN_EVENT_HNF(mc_retries, 0xc), + CCN_EVENT_HNF(mc_reqs, 0xd), + CCN_EVENT_HNF(qos_hh_retry, 0xe), + CCN_EVENT_RNI(rdata_beats_p0, 0x1), + CCN_EVENT_RNI(rdata_beats_p1, 0x2), + CCN_EVENT_RNI(rdata_beats_p2, 0x3), + CCN_EVENT_RNI(rxdat_flits, 0x4), + CCN_EVENT_RNI(txdat_flits, 0x5), + CCN_EVENT_RNI(txreq_flits, 0x6), + CCN_EVENT_RNI(txreq_flits_retried, 0x7), + CCN_EVENT_RNI(rrt_full, 0x8), + CCN_EVENT_RNI(wrt_full, 0x9), + CCN_EVENT_RNI(txreq_flits_replayed, 0xa), + CCN_EVENT_XP(upload_starvation, 0x1), + CCN_EVENT_XP(download_starvation, 0x2), + CCN_EVENT_XP(respin, 0x3), + CCN_EVENT_XP(valid_flit, 0x4), + CCN_EVENT_XP(watchpoint, CCN_EVENT_WATCHPOINT), + CCN_EVENT_SBAS(rdata_beats_p0, 0x1), + CCN_EVENT_SBAS(rxdat_flits, 0x4), + CCN_EVENT_SBAS(txdat_flits, 0x5), + CCN_EVENT_SBAS(txreq_flits, 0x6), + CCN_EVENT_SBAS(txreq_flits_retried, 0x7), + CCN_EVENT_SBAS(rrt_full, 0x8), + CCN_EVENT_SBAS(wrt_full, 0x9), + CCN_EVENT_SBAS(txreq_flits_replayed, 0xa), + CCN_EVENT_CYCLES(cycles), +}; + +/* Populated in arm_ccn_init() */ +static struct attribute + *arm_ccn_pmu_events_attrs[ARRAY_SIZE(arm_ccn_pmu_events) + 1]; + +static struct attribute_group arm_ccn_pmu_events_attr_group = { + .name = "events", + .is_visible = arm_ccn_pmu_events_is_visible, + .attrs = arm_ccn_pmu_events_attrs, +}; + + +static u64 *arm_ccn_pmu_get_cmp_mask(struct arm_ccn *ccn, const char *name) +{ + unsigned long i; + + if (WARN_ON(!name || !name[0] || !isxdigit(name[0]) || !name[1])) + return NULL; + i = isdigit(name[0]) ? name[0] - '0' : 0xa + tolower(name[0]) - 'a'; + + switch (name[1]) { + case 'l': + return &ccn->dt.cmp_mask[i].l; + case 'h': + return &ccn->dt.cmp_mask[i].h; + default: + return NULL; + } +} + +static ssize_t arm_ccn_pmu_cmp_mask_show(struct device *dev, + struct device_attribute *attr, char *buf) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(dev_get_drvdata(dev)); + u64 *mask = arm_ccn_pmu_get_cmp_mask(ccn, attr->attr.name); + + return mask ? snprintf(buf, PAGE_SIZE, "0x%016llx\n", *mask) : -EINVAL; +} + +static ssize_t arm_ccn_pmu_cmp_mask_store(struct device *dev, + struct device_attribute *attr, const char *buf, size_t count) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(dev_get_drvdata(dev)); + u64 *mask = arm_ccn_pmu_get_cmp_mask(ccn, attr->attr.name); + int err = -EINVAL; + + if (mask) + err = kstrtoull(buf, 0, mask); + + return err ? err : count; +} + +#define CCN_CMP_MASK_ATTR(_name) \ + struct device_attribute arm_ccn_pmu_cmp_mask_attr_##_name = \ + __ATTR(_name, S_IRUGO | S_IWUSR, \ + arm_ccn_pmu_cmp_mask_show, arm_ccn_pmu_cmp_mask_store) + +#define CCN_CMP_MASK_ATTR_RO(_name) \ + struct device_attribute arm_ccn_pmu_cmp_mask_attr_##_name = \ + __ATTR(_name, S_IRUGO, arm_ccn_pmu_cmp_mask_show, NULL) + +static CCN_CMP_MASK_ATTR(0l); +static CCN_CMP_MASK_ATTR(0h); +static CCN_CMP_MASK_ATTR(1l); +static CCN_CMP_MASK_ATTR(1h); +static CCN_CMP_MASK_ATTR(2l); +static CCN_CMP_MASK_ATTR(2h); +static CCN_CMP_MASK_ATTR(3l); +static CCN_CMP_MASK_ATTR(3h); +static CCN_CMP_MASK_ATTR(4l); +static CCN_CMP_MASK_ATTR(4h); +static CCN_CMP_MASK_ATTR(5l); +static CCN_CMP_MASK_ATTR(5h); +static CCN_CMP_MASK_ATTR(6l); +static CCN_CMP_MASK_ATTR(6h); +static CCN_CMP_MASK_ATTR(7l); +static CCN_CMP_MASK_ATTR(7h); +static CCN_CMP_MASK_ATTR_RO(8l); +static CCN_CMP_MASK_ATTR_RO(8h); +static CCN_CMP_MASK_ATTR_RO(9l); +static CCN_CMP_MASK_ATTR_RO(9h); +static CCN_CMP_MASK_ATTR_RO(al); +static CCN_CMP_MASK_ATTR_RO(ah); +static CCN_CMP_MASK_ATTR_RO(bl); +static CCN_CMP_MASK_ATTR_RO(bh); + +static struct attribute *arm_ccn_pmu_cmp_mask_attrs[] = { + &arm_ccn_pmu_cmp_mask_attr_0l.attr, &arm_ccn_pmu_cmp_mask_attr_0h.attr, + &arm_ccn_pmu_cmp_mask_attr_1l.attr, &arm_ccn_pmu_cmp_mask_attr_1h.attr, + &arm_ccn_pmu_cmp_mask_attr_2l.attr, &arm_ccn_pmu_cmp_mask_attr_2h.attr, + &arm_ccn_pmu_cmp_mask_attr_3l.attr, &arm_ccn_pmu_cmp_mask_attr_3h.attr, + &arm_ccn_pmu_cmp_mask_attr_4l.attr, &arm_ccn_pmu_cmp_mask_attr_4h.attr, + &arm_ccn_pmu_cmp_mask_attr_5l.attr, &arm_ccn_pmu_cmp_mask_attr_5h.attr, + &arm_ccn_pmu_cmp_mask_attr_6l.attr, &arm_ccn_pmu_cmp_mask_attr_6h.attr, + &arm_ccn_pmu_cmp_mask_attr_7l.attr, &arm_ccn_pmu_cmp_mask_attr_7h.attr, + &arm_ccn_pmu_cmp_mask_attr_8l.attr, &arm_ccn_pmu_cmp_mask_attr_8h.attr, + &arm_ccn_pmu_cmp_mask_attr_9l.attr, &arm_ccn_pmu_cmp_mask_attr_9h.attr, + &arm_ccn_pmu_cmp_mask_attr_al.attr, &arm_ccn_pmu_cmp_mask_attr_ah.attr, + &arm_ccn_pmu_cmp_mask_attr_bl.attr, &arm_ccn_pmu_cmp_mask_attr_bh.attr, + NULL +}; + +static struct attribute_group arm_ccn_pmu_cmp_mask_attr_group = { + .name = "cmp_mask", + .attrs = arm_ccn_pmu_cmp_mask_attrs, +}; + + +/* + * Default poll period is 10ms, which is way over the top anyway, + * as in the worst case scenario (an event every cycle), with 1GHz + * clocked bus, the smallest, 32 bit counter will overflow in + * more than 4s. + */ +static unsigned int arm_ccn_pmu_poll_period_us = 10000; +module_param_named(pmu_poll_period_us, arm_ccn_pmu_poll_period_us, uint, + S_IRUGO | S_IWUSR); + +static ktime_t arm_ccn_pmu_timer_period(void) +{ + return ns_to_ktime((u64)arm_ccn_pmu_poll_period_us * 1000); +} + + +static const struct attribute_group *arm_ccn_pmu_attr_groups[] = { + &arm_ccn_pmu_events_attr_group, + &arm_ccn_pmu_format_attr_group, + &arm_ccn_pmu_cmp_mask_attr_group, + NULL +}; + + +static int arm_ccn_pmu_alloc_bit(unsigned long *bitmap, unsigned long size) +{ + int bit; + + do { + bit = find_first_zero_bit(bitmap, size); + if (bit >= size) + return -EAGAIN; + } while (test_and_set_bit(bit, bitmap)); + + return bit; +} + +/* All RN-I and RN-D nodes have identical PMUs */ +static int arm_ccn_pmu_type_eq(u32 a, u32 b) +{ + if (a == b) + return 1; + + switch (a) { + case CCN_TYPE_RNI_1P: + case CCN_TYPE_RNI_2P: + case CCN_TYPE_RNI_3P: + case CCN_TYPE_RND_1P: + case CCN_TYPE_RND_2P: + case CCN_TYPE_RND_3P: + switch (b) { + case CCN_TYPE_RNI_1P: + case CCN_TYPE_RNI_2P: + case CCN_TYPE_RNI_3P: + case CCN_TYPE_RND_1P: + case CCN_TYPE_RND_2P: + case CCN_TYPE_RND_3P: + return 1; + } + break; + } + + return 0; +} + +static void arm_ccn_pmu_event_destroy(struct perf_event *event) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(event->pmu); + struct hw_perf_event *hw = &event->hw; + + if (hw->idx == CCN_IDX_PMU_CYCLE_COUNTER) { + clear_bit(CCN_IDX_PMU_CYCLE_COUNTER, ccn->dt.pmu_counters_mask); + } else { + struct arm_ccn_component *source = + ccn->dt.pmu_counters[hw->idx].source; + + if (CCN_CONFIG_TYPE(event->attr.config) == CCN_TYPE_XP && + CCN_CONFIG_EVENT(event->attr.config) == + CCN_EVENT_WATCHPOINT) + clear_bit(hw->config_base, source->xp.dt_cmp_mask); + else + clear_bit(hw->config_base, source->pmu_events_mask); + clear_bit(hw->idx, ccn->dt.pmu_counters_mask); + } + + ccn->dt.pmu_counters[hw->idx].source = NULL; + ccn->dt.pmu_counters[hw->idx].event = NULL; +} + +static int arm_ccn_pmu_event_init(struct perf_event *event) +{ + struct arm_ccn *ccn; + struct hw_perf_event *hw = &event->hw; + u32 node_xp, type, event_id; + int valid, bit; + struct arm_ccn_component *source; + int i; + + if (event->attr.type != event->pmu->type) + return -ENOENT; + + ccn = pmu_to_arm_ccn(event->pmu); + event->destroy = arm_ccn_pmu_event_destroy; + + if (hw->sample_period) { + dev_warn(ccn->dev, "Sampling not supported!\n"); + return -EOPNOTSUPP; + } + + if (has_branch_stack(event) || event->attr.exclude_user || + event->attr.exclude_kernel || event->attr.exclude_hv || + event->attr.exclude_idle) { + dev_warn(ccn->dev, "Can't exclude execution levels!\n"); + return -EOPNOTSUPP; + } + + if (event->cpu < 0) { + dev_warn(ccn->dev, "Can't provide per-task data!\n"); + return -EOPNOTSUPP; + } + + node_xp = CCN_CONFIG_NODE(event->attr.config); + type = CCN_CONFIG_TYPE(event->attr.config); + event_id = CCN_CONFIG_EVENT(event->attr.config); + + /* Validate node/xp vs topology */ + switch (type) { + case CCN_TYPE_XP: + if (node_xp >= ccn->num_xps) { + dev_warn(ccn->dev, "Invalid XP ID %d!\n", node_xp); + return -EINVAL; + } + break; + case CCN_TYPE_CYCLES: + break; + default: + if (node_xp >= ccn->num_nodes) { + dev_warn(ccn->dev, "Invalid node ID %d!\n", node_xp); + return -EINVAL; + } + if (!arm_ccn_pmu_type_eq(type, ccn->node[node_xp].type)) { + dev_warn(ccn->dev, "Invalid type 0x%x for node %d!\n", + type, node_xp); + return -EINVAL; + } + break; + } + + /* Validate event ID vs available for the type */ + for (i = 0, valid = 0; i < ARRAY_SIZE(arm_ccn_pmu_events) && !valid; + i++) { + struct arm_ccn_pmu_event *e = &arm_ccn_pmu_events[i]; + u32 port = CCN_CONFIG_PORT(event->attr.config); + u32 vc = CCN_CONFIG_VC(event->attr.config); + + if (!arm_ccn_pmu_type_eq(type, e->type)) + continue; + if (event_id != e->event) + continue; + if (e->num_ports && port >= e->num_ports) { + dev_warn(ccn->dev, "Invalid port %d for node/XP %d!\n", + port, node_xp); + return -EINVAL; + } + if (e->num_vcs && vc >= e->num_vcs) { + dev_warn(ccn->dev, "Invalid vc %d for node/XP %d!\n", + vc, node_xp); + return -EINVAL; + } + valid = 1; + } + if (!valid) { + dev_warn(ccn->dev, "Invalid event 0x%x for node/XP %d!\n", + event_id, node_xp); + return -EINVAL; + } + + /* Watchpoint-based event for a node is actually set on XP */ + if (event_id == CCN_EVENT_WATCHPOINT && type != CCN_TYPE_XP) { + u32 port; + + type = CCN_TYPE_XP; + port = arm_ccn_node_to_xp_port(node_xp); + node_xp = arm_ccn_node_to_xp(node_xp); + + arm_ccn_pmu_config_set(&event->attr.config, + node_xp, type, port); + } + + /* Allocate the cycle counter */ + if (type == CCN_TYPE_CYCLES) { + if (test_and_set_bit(CCN_IDX_PMU_CYCLE_COUNTER, + ccn->dt.pmu_counters_mask)) + return -EAGAIN; + + hw->idx = CCN_IDX_PMU_CYCLE_COUNTER; + ccn->dt.pmu_counters[CCN_IDX_PMU_CYCLE_COUNTER].event = event; + + return 0; + } + + /* Allocate an event counter */ + hw->idx = arm_ccn_pmu_alloc_bit(ccn->dt.pmu_counters_mask, + CCN_NUM_PMU_EVENT_COUNTERS); + if (hw->idx < 0) { + dev_warn(ccn->dev, "No more counters available!\n"); + return -EAGAIN; + } + + if (type == CCN_TYPE_XP) + source = &ccn->xp[node_xp]; + else + source = &ccn->node[node_xp]; + ccn->dt.pmu_counters[hw->idx].source = source; + + /* Allocate an event source or a watchpoint */ + if (type == CCN_TYPE_XP && event_id == CCN_EVENT_WATCHPOINT) + bit = arm_ccn_pmu_alloc_bit(source->xp.dt_cmp_mask, + CCN_NUM_XP_WATCHPOINTS); + else + bit = arm_ccn_pmu_alloc_bit(source->pmu_events_mask, + CCN_NUM_PMU_EVENTS); + if (bit < 0) { + dev_warn(ccn->dev, "No more event sources/watchpoints on node/XP %d!\n", + node_xp); + clear_bit(hw->idx, ccn->dt.pmu_counters_mask); + return -EAGAIN; + } + hw->config_base = bit; + + ccn->dt.pmu_counters[hw->idx].event = event; + + return 0; +} + +static u64 arm_ccn_pmu_read_counter(struct arm_ccn *ccn, int idx) +{ + u64 res; + + if (idx == CCN_IDX_PMU_CYCLE_COUNTER) { +#ifdef readq + res = readq(ccn->dt.base + CCN_DT_PMCCNTR); +#else + /* 40 bit counter, can do snapshot and read in two parts */ + writel(0x1, ccn->dt.base + CCN_DT_PMSR_REQ); + while (!(readl(ccn->dt.base + CCN_DT_PMSR) & 0x1)) + ; + writel(0x1, ccn->dt.base + CCN_DT_PMSR_CLR); + res = readl(ccn->dt.base + CCN_DT_PMCCNTRSR + 4) & 0xff; + res <<= 32; + res |= readl(ccn->dt.base + CCN_DT_PMCCNTRSR); +#endif + } else { + res = readl(ccn->dt.base + CCN_DT_PMEVCNT(idx)); + } + + return res; +} + +static void arm_ccn_pmu_event_update(struct perf_event *event) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(event->pmu); + struct hw_perf_event *hw = &event->hw; + u64 prev_count, new_count, mask; + + do { + prev_count = local64_read(&hw->prev_count); + new_count = arm_ccn_pmu_read_counter(ccn, hw->idx); + } while (local64_xchg(&hw->prev_count, new_count) != prev_count); + + mask = (1LLU << (hw->idx == CCN_IDX_PMU_CYCLE_COUNTER ? 40 : 32)) - 1; + + local64_add((new_count - prev_count) & mask, &event->count); +} + +static void arm_ccn_pmu_xp_dt_config(struct perf_event *event, int enable) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(event->pmu); + struct hw_perf_event *hw = &event->hw; + struct arm_ccn_component *xp; + u32 val, dt_cfg; + + if (CCN_CONFIG_TYPE(event->attr.config) == CCN_TYPE_XP) + xp = &ccn->xp[CCN_CONFIG_XP(event->attr.config)]; + else + xp = &ccn->xp[arm_ccn_node_to_xp( + CCN_CONFIG_NODE(event->attr.config))]; + + if (enable) + dt_cfg = hw->event_base; + else + dt_cfg = CCN_XP_DT_CONFIG__DT_CFG__PASS_THROUGH; + + spin_lock(&ccn->dt.config_lock); + + val = readl(xp->base + CCN_XP_DT_CONFIG); + val &= ~(CCN_XP_DT_CONFIG__DT_CFG__MASK << + CCN_XP_DT_CONFIG__DT_CFG__SHIFT(hw->idx)); + val |= dt_cfg << CCN_XP_DT_CONFIG__DT_CFG__SHIFT(hw->idx); + writel(val, xp->base + CCN_XP_DT_CONFIG); + + spin_unlock(&ccn->dt.config_lock); +} + +static void arm_ccn_pmu_event_start(struct perf_event *event, int flags) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(event->pmu); + struct hw_perf_event *hw = &event->hw; + + local64_set(&event->hw.prev_count, + arm_ccn_pmu_read_counter(ccn, hw->idx)); + hw->state = 0; + + if (!ccn->irq_used) + hrtimer_start(&ccn->dt.hrtimer, arm_ccn_pmu_timer_period(), + HRTIMER_MODE_REL); + + /* Set the DT bus input, engaging the counter */ + arm_ccn_pmu_xp_dt_config(event, 1); +} + +static void arm_ccn_pmu_event_stop(struct perf_event *event, int flags) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(event->pmu); + struct hw_perf_event *hw = &event->hw; + u64 timeout; + + /* Disable counting, setting the DT bus to pass-through mode */ + arm_ccn_pmu_xp_dt_config(event, 0); + + if (!ccn->irq_used) + hrtimer_cancel(&ccn->dt.hrtimer); + + /* Let the DT bus drain */ + timeout = arm_ccn_pmu_read_counter(ccn, CCN_IDX_PMU_CYCLE_COUNTER) + + ccn->num_xps; + while (arm_ccn_pmu_read_counter(ccn, CCN_IDX_PMU_CYCLE_COUNTER) < + timeout) + cpu_relax(); + + if (flags & PERF_EF_UPDATE) + arm_ccn_pmu_event_update(event); + + hw->state |= PERF_HES_STOPPED; +} + +static void arm_ccn_pmu_xp_watchpoint_config(struct perf_event *event) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(event->pmu); + struct hw_perf_event *hw = &event->hw; + struct arm_ccn_component *source = + ccn->dt.pmu_counters[hw->idx].source; + unsigned long wp = hw->config_base; + u32 val; + u64 cmp_l = event->attr.config1; + u64 cmp_h = event->attr.config2; + u64 mask_l = ccn->dt.cmp_mask[CCN_CONFIG_MASK(event->attr.config)].l; + u64 mask_h = ccn->dt.cmp_mask[CCN_CONFIG_MASK(event->attr.config)].h; + + hw->event_base = CCN_XP_DT_CONFIG__DT_CFG__WATCHPOINT(wp); + + /* Direction (RX/TX), device (port) & virtual channel */ + val = readl(source->base + CCN_XP_DT_INTERFACE_SEL); + val &= ~(CCN_XP_DT_INTERFACE_SEL__DT_IO_SEL__MASK << + CCN_XP_DT_INTERFACE_SEL__DT_IO_SEL__SHIFT(wp)); + val |= CCN_CONFIG_DIR(event->attr.config) << + CCN_XP_DT_INTERFACE_SEL__DT_IO_SEL__SHIFT(wp); + val &= ~(CCN_XP_DT_INTERFACE_SEL__DT_DEV_SEL__MASK << + CCN_XP_DT_INTERFACE_SEL__DT_DEV_SEL__SHIFT(wp)); + val |= CCN_CONFIG_PORT(event->attr.config) << + CCN_XP_DT_INTERFACE_SEL__DT_DEV_SEL__SHIFT(wp); + val &= ~(CCN_XP_DT_INTERFACE_SEL__DT_VC_SEL__MASK << + CCN_XP_DT_INTERFACE_SEL__DT_VC_SEL__SHIFT(wp)); + val |= CCN_CONFIG_VC(event->attr.config) << + CCN_XP_DT_INTERFACE_SEL__DT_VC_SEL__SHIFT(wp); + writel(val, source->base + CCN_XP_DT_INTERFACE_SEL); + + /* Comparison values */ + writel(cmp_l & 0xffffffff, source->base + CCN_XP_DT_CMP_VAL_L(wp)); + writel((cmp_l >> 32) & 0xefffffff, + source->base + CCN_XP_DT_CMP_VAL_L(wp) + 4); + writel(cmp_h & 0xffffffff, source->base + CCN_XP_DT_CMP_VAL_H(wp)); + writel((cmp_h >> 32) & 0x0fffffff, + source->base + CCN_XP_DT_CMP_VAL_H(wp) + 4); + + /* Mask */ + writel(mask_l & 0xffffffff, source->base + CCN_XP_DT_CMP_MASK_L(wp)); + writel((mask_l >> 32) & 0xefffffff, + source->base + CCN_XP_DT_CMP_MASK_L(wp) + 4); + writel(mask_h & 0xffffffff, source->base + CCN_XP_DT_CMP_MASK_H(wp)); + writel((mask_h >> 32) & 0x0fffffff, + source->base + CCN_XP_DT_CMP_MASK_H(wp) + 4); +} + +static void arm_ccn_pmu_xp_event_config(struct perf_event *event) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(event->pmu); + struct hw_perf_event *hw = &event->hw; + struct arm_ccn_component *source = + ccn->dt.pmu_counters[hw->idx].source; + u32 val, id; + + hw->event_base = CCN_XP_DT_CONFIG__DT_CFG__XP_PMU_EVENT(hw->config_base); + + id = (CCN_CONFIG_VC(event->attr.config) << 4) | + (CCN_CONFIG_PORT(event->attr.config) << 3) | + (CCN_CONFIG_EVENT(event->attr.config) << 0); + + val = readl(source->base + CCN_XP_PMU_EVENT_SEL); + val &= ~(CCN_XP_PMU_EVENT_SEL__ID__MASK << + CCN_XP_PMU_EVENT_SEL__ID__SHIFT(hw->config_base)); + val |= id << CCN_XP_PMU_EVENT_SEL__ID__SHIFT(hw->config_base); + writel(val, source->base + CCN_XP_PMU_EVENT_SEL); +} + +static void arm_ccn_pmu_node_event_config(struct perf_event *event) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(event->pmu); + struct hw_perf_event *hw = &event->hw; + struct arm_ccn_component *source = + ccn->dt.pmu_counters[hw->idx].source; + u32 type = CCN_CONFIG_TYPE(event->attr.config); + u32 val, port; + + port = arm_ccn_node_to_xp_port(CCN_CONFIG_NODE(event->attr.config)); + hw->event_base = CCN_XP_DT_CONFIG__DT_CFG__DEVICE_PMU_EVENT(port, + hw->config_base); + + /* These *_event_sel regs should be identical, but let's make sure... */ + BUILD_BUG_ON(CCN_HNF_PMU_EVENT_SEL != CCN_SBAS_PMU_EVENT_SEL); + BUILD_BUG_ON(CCN_SBAS_PMU_EVENT_SEL != CCN_RNI_PMU_EVENT_SEL); + BUILD_BUG_ON(CCN_HNF_PMU_EVENT_SEL__ID__SHIFT(1) != + CCN_SBAS_PMU_EVENT_SEL__ID__SHIFT(1)); + BUILD_BUG_ON(CCN_SBAS_PMU_EVENT_SEL__ID__SHIFT(1) != + CCN_RNI_PMU_EVENT_SEL__ID__SHIFT(1)); + BUILD_BUG_ON(CCN_HNF_PMU_EVENT_SEL__ID__MASK != + CCN_SBAS_PMU_EVENT_SEL__ID__MASK); + BUILD_BUG_ON(CCN_SBAS_PMU_EVENT_SEL__ID__MASK != + CCN_RNI_PMU_EVENT_SEL__ID__MASK); + if (WARN_ON(type != CCN_TYPE_HNF && type != CCN_TYPE_SBAS && + !arm_ccn_pmu_type_eq(type, CCN_TYPE_RNI_3P))) + return; + + /* Set the event id for the pre-allocated counter */ + val = readl(source->base + CCN_HNF_PMU_EVENT_SEL); + val &= ~(CCN_HNF_PMU_EVENT_SEL__ID__MASK << + CCN_HNF_PMU_EVENT_SEL__ID__SHIFT(hw->config_base)); + val |= CCN_CONFIG_EVENT(event->attr.config) << + CCN_HNF_PMU_EVENT_SEL__ID__SHIFT(hw->config_base); + writel(val, source->base + CCN_HNF_PMU_EVENT_SEL); +} + +static void arm_ccn_pmu_event_config(struct perf_event *event) +{ + struct arm_ccn *ccn = pmu_to_arm_ccn(event->pmu); + struct hw_perf_event *hw = &event->hw; + u32 xp, offset, val; + + /* Cycle counter requires no setup */ + if (hw->idx == CCN_IDX_PMU_CYCLE_COUNTER) + return; + + if (CCN_CONFIG_TYPE(event->attr.config) == CCN_TYPE_XP) + xp = CCN_CONFIG_XP(event->attr.config); + else + xp = arm_ccn_node_to_xp(CCN_CONFIG_NODE(event->attr.config)); + + spin_lock(&ccn->dt.config_lock); + + /* Set the DT bus "distance" register */ + offset = (hw->idx / 4) * 4; + val = readl(ccn->dt.base + CCN_DT_ACTIVE_DSM + offset); + val &= ~(CCN_DT_ACTIVE_DSM__DSM_ID__MASK << + CCN_DT_ACTIVE_DSM__DSM_ID__SHIFT(hw->idx % 4)); + val |= xp << CCN_DT_ACTIVE_DSM__DSM_ID__SHIFT(hw->idx % 4); + writel(val, ccn->dt.base + CCN_DT_ACTIVE_DSM + offset); + + if (CCN_CONFIG_TYPE(event->attr.config) == CCN_TYPE_XP) { + if (CCN_CONFIG_EVENT(event->attr.config) == + CCN_EVENT_WATCHPOINT) + arm_ccn_pmu_xp_watchpoint_config(event); + else + arm_ccn_pmu_xp_event_config(event); + } else { + arm_ccn_pmu_node_event_config(event); + } + + spin_unlock(&ccn->dt.config_lock); +} + +static int arm_ccn_pmu_event_add(struct perf_event *event, int flags) +{ + struct hw_perf_event *hw = &event->hw; + + arm_ccn_pmu_event_config(event); + + hw->state = PERF_HES_STOPPED; + + if (flags & PERF_EF_START) + arm_ccn_pmu_event_start(event, PERF_EF_UPDATE); + + return 0; +} + +static void arm_ccn_pmu_event_del(struct perf_event *event, int flags) +{ + arm_ccn_pmu_event_stop(event, PERF_EF_UPDATE); +} + +static void arm_ccn_pmu_event_read(struct perf_event *event) +{ + arm_ccn_pmu_event_update(event); +} + +static irqreturn_t arm_ccn_pmu_overflow_handler(struct arm_ccn_dt *dt) +{ + u32 pmovsr = readl(dt->base + CCN_DT_PMOVSR); + int idx; + + if (!pmovsr) + return IRQ_NONE; + + writel(pmovsr, dt->base + CCN_DT_PMOVSR_CLR); + + BUILD_BUG_ON(CCN_IDX_PMU_CYCLE_COUNTER != CCN_NUM_PMU_EVENT_COUNTERS); + + for (idx = 0; idx < CCN_NUM_PMU_EVENT_COUNTERS + 1; idx++) { + struct perf_event *event = dt->pmu_counters[idx].event; + int overflowed = pmovsr & BIT(idx); + + WARN_ON_ONCE(overflowed && !event && + idx != CCN_IDX_PMU_CYCLE_COUNTER); + + if (!event || !overflowed) + continue; + + arm_ccn_pmu_event_update(event); + } + + return IRQ_HANDLED; +} + +static enum hrtimer_restart arm_ccn_pmu_timer_handler(struct hrtimer *hrtimer) +{ + struct arm_ccn_dt *dt = container_of(hrtimer, struct arm_ccn_dt, + hrtimer); + unsigned long flags; + + local_irq_save(flags); + arm_ccn_pmu_overflow_handler(dt); + local_irq_restore(flags); + + hrtimer_forward_now(hrtimer, arm_ccn_pmu_timer_period()); + return HRTIMER_RESTART; +} + + +static DEFINE_IDA(arm_ccn_pmu_ida); + +static int arm_ccn_pmu_init(struct arm_ccn *ccn) +{ + int i; + char *name; + + /* Initialize DT subsystem */ + ccn->dt.base = ccn->base + CCN_REGION_SIZE; + spin_lock_init(&ccn->dt.config_lock); + writel(CCN_DT_PMOVSR_CLR__MASK, ccn->dt.base + CCN_DT_PMOVSR_CLR); + writel(CCN_DT_CTL__DT_EN, ccn->dt.base + CCN_DT_CTL); + writel(CCN_DT_PMCR__OVFL_INTR_EN | CCN_DT_PMCR__PMU_EN, + ccn->dt.base + CCN_DT_PMCR); + writel(0x1, ccn->dt.base + CCN_DT_PMSR_CLR); + for (i = 0; i < ccn->num_xps; i++) { + writel(0, ccn->xp[i].base + CCN_XP_DT_CONFIG); + writel((CCN_XP_DT_CONTROL__WP_ARM_SEL__ALWAYS << + CCN_XP_DT_CONTROL__WP_ARM_SEL__SHIFT(0)) | + (CCN_XP_DT_CONTROL__WP_ARM_SEL__ALWAYS << + CCN_XP_DT_CONTROL__WP_ARM_SEL__SHIFT(1)) | + CCN_XP_DT_CONTROL__DT_ENABLE, + ccn->xp[i].base + CCN_XP_DT_CONTROL); + } + ccn->dt.cmp_mask[CCN_IDX_MASK_ANY].l = ~0; + ccn->dt.cmp_mask[CCN_IDX_MASK_ANY].h = ~0; + ccn->dt.cmp_mask[CCN_IDX_MASK_EXACT].l = 0; + ccn->dt.cmp_mask[CCN_IDX_MASK_EXACT].h = 0; + ccn->dt.cmp_mask[CCN_IDX_MASK_ORDER].l = ~0; + ccn->dt.cmp_mask[CCN_IDX_MASK_ORDER].h = ~(0x1 << 15); + ccn->dt.cmp_mask[CCN_IDX_MASK_OPCODE].l = ~0; + ccn->dt.cmp_mask[CCN_IDX_MASK_OPCODE].h = ~(0x1f << 9); + + /* Get a convenient /sys/event_source/devices/ name */ + ccn->dt.id = ida_simple_get(&arm_ccn_pmu_ida, 0, 0, GFP_KERNEL); + if (ccn->dt.id == 0) { + name = "ccn"; + } else { + int len = snprintf(NULL, 0, "ccn_%d", ccn->dt.id); + + name = devm_kzalloc(ccn->dev, len + 1, GFP_KERNEL); + snprintf(name, len + 1, "ccn_%d", ccn->dt.id); + } + + /* Perf driver registration */ + ccn->dt.pmu = (struct pmu) { + .attr_groups = arm_ccn_pmu_attr_groups, + .task_ctx_nr = perf_invalid_context, + .event_init = arm_ccn_pmu_event_init, + .add = arm_ccn_pmu_event_add, + .del = arm_ccn_pmu_event_del, + .start = arm_ccn_pmu_event_start, + .stop = arm_ccn_pmu_event_stop, + .read = arm_ccn_pmu_event_read, + }; + + /* No overflow interrupt? Have to use a timer instead. */ + if (!ccn->irq_used) { + dev_info(ccn->dev, "No access to interrupts, using timer.\n"); + hrtimer_init(&ccn->dt.hrtimer, CLOCK_MONOTONIC, + HRTIMER_MODE_REL); + ccn->dt.hrtimer.function = arm_ccn_pmu_timer_handler; + } + + return perf_pmu_register(&ccn->dt.pmu, name, -1); +} + +static void arm_ccn_pmu_cleanup(struct arm_ccn *ccn) +{ + int i; + + for (i = 0; i < ccn->num_xps; i++) + writel(0, ccn->xp[i].base + CCN_XP_DT_CONTROL); + writel(0, ccn->dt.base + CCN_DT_PMCR); + perf_pmu_unregister(&ccn->dt.pmu); + ida_simple_remove(&arm_ccn_pmu_ida, ccn->dt.id); +} + + +static int arm_ccn_for_each_valid_region(struct arm_ccn *ccn, + int (*callback)(struct arm_ccn *ccn, int region, + void __iomem *base, u32 type, u32 id)) +{ + int region; + + for (region = 0; region < CCN_NUM_REGIONS; region++) { + u32 val, type, id; + void __iomem *base; + int err; + + val = readl(ccn->base + CCN_MN_OLY_COMP_LIST_63_0 + + 4 * (region / 32)); + if (!(val & (1 << (region % 32)))) + continue; + + base = ccn->base + region * CCN_REGION_SIZE; + val = readl(base + CCN_ALL_OLY_ID); + type = (val >> CCN_ALL_OLY_ID__OLY_ID__SHIFT) & + CCN_ALL_OLY_ID__OLY_ID__MASK; + id = (val >> CCN_ALL_OLY_ID__NODE_ID__SHIFT) & + CCN_ALL_OLY_ID__NODE_ID__MASK; + + err = callback(ccn, region, base, type, id); + if (err) + return err; + } + + return 0; +} + +static int arm_ccn_get_nodes_num(struct arm_ccn *ccn, int region, + void __iomem *base, u32 type, u32 id) +{ + + if (type == CCN_TYPE_XP && id >= ccn->num_xps) + ccn->num_xps = id + 1; + else if (id >= ccn->num_nodes) + ccn->num_nodes = id + 1; + + return 0; +} + +static int arm_ccn_init_nodes(struct arm_ccn *ccn, int region, + void __iomem *base, u32 type, u32 id) +{ + struct arm_ccn_component *component; + + dev_dbg(ccn->dev, "Region %d: id=%u, type=0x%02x\n", region, id, type); + + switch (type) { + case CCN_TYPE_MN: + case CCN_TYPE_DT: + return 0; + case CCN_TYPE_XP: + component = &ccn->xp[id]; + break; + case CCN_TYPE_SBSX: + ccn->sbsx_present = 1; + component = &ccn->node[id]; + break; + case CCN_TYPE_SBAS: + ccn->sbas_present = 1; + /* Fall-through */ + default: + component = &ccn->node[id]; + break; + } + + component->base = base; + component->type = type; + + return 0; +} + + +static irqreturn_t arm_ccn_error_handler(struct arm_ccn *ccn, + const u32 *err_sig_val) +{ + /* This should be really handled by firmware... */ + dev_err(ccn->dev, "Error reported in %08x%08x%08x%08x%08x%08x.\n", + err_sig_val[5], err_sig_val[4], err_sig_val[3], + err_sig_val[2], err_sig_val[1], err_sig_val[0]); + dev_err(ccn->dev, "Disabling interrupt generation for all errors.\n"); + writel(CCN_MN_ERRINT_STATUS__ALL_ERRORS__DISABLE, + ccn->base + CCN_MN_ERRINT_STATUS); + + return IRQ_HANDLED; +} + + +static irqreturn_t arm_ccn_irq_handler(int irq, void *dev_id) +{ + irqreturn_t res = IRQ_NONE; + struct arm_ccn *ccn = dev_id; + u32 err_sig_val[6]; + u32 err_or; + int i; + + /* PMU overflow is a special case */ + err_or = err_sig_val[0] = readl(ccn->base + CCN_MN_ERR_SIG_VAL_63_0); + if (err_or & CCN_MN_ERR_SIG_VAL_63_0__DT) { + err_or &= ~CCN_MN_ERR_SIG_VAL_63_0__DT; + res = arm_ccn_pmu_overflow_handler(&ccn->dt); + } + + /* Have to read all err_sig_vals to clear them */ + for (i = 1; i < ARRAY_SIZE(err_sig_val); i++) { + err_sig_val[i] = readl(ccn->base + + CCN_MN_ERR_SIG_VAL_63_0 + i * 4); + err_or |= err_sig_val[i]; + } + if (err_or) + res |= arm_ccn_error_handler(ccn, err_sig_val); + + if (res != IRQ_NONE) + writel(CCN_MN_ERRINT_STATUS__INTREQ__DESSERT, + ccn->base + CCN_MN_ERRINT_STATUS); + + return res; +} + + +static int arm_ccn_probe(struct platform_device *pdev) +{ + struct arm_ccn *ccn; + struct resource *res; + int err; + + ccn = devm_kzalloc(&pdev->dev, sizeof(*ccn), GFP_KERNEL); + if (!ccn) + return -ENOMEM; + ccn->dev = &pdev->dev; + platform_set_drvdata(pdev, ccn); + + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!res) + return -EINVAL; + + if (!devm_request_mem_region(ccn->dev, res->start, + resource_size(res), pdev->name)) + return -EBUSY; + + ccn->base = devm_ioremap(ccn->dev, res->start, + resource_size(res)); + if (!ccn->base) + return -EFAULT; + + res = platform_get_resource(pdev, IORESOURCE_IRQ, 0); + if (!res) + return -EINVAL; + + /* Check if we can use the interrupt */ + writel(CCN_MN_ERRINT_STATUS__PMU_EVENTS__DISABLE, + ccn->base + CCN_MN_ERRINT_STATUS); + if (readl(ccn->base + CCN_MN_ERRINT_STATUS) & + CCN_MN_ERRINT_STATUS__PMU_EVENTS__DISABLED) { + /* Can set 'disable' bits, so can acknowledge interrupts */ + writel(CCN_MN_ERRINT_STATUS__PMU_EVENTS__ENABLE, + ccn->base + CCN_MN_ERRINT_STATUS); + err = devm_request_irq(ccn->dev, res->start, + arm_ccn_irq_handler, 0, dev_name(ccn->dev), + ccn); + if (err) + return err; + + ccn->irq_used = 1; + } + + + /* Build topology */ + + err = arm_ccn_for_each_valid_region(ccn, arm_ccn_get_nodes_num); + if (err) + return err; + + ccn->node = devm_kzalloc(ccn->dev, sizeof(*ccn->node) * ccn->num_nodes, + GFP_KERNEL); + ccn->xp = devm_kzalloc(ccn->dev, sizeof(*ccn->node) * ccn->num_xps, + GFP_KERNEL); + if (!ccn->node || !ccn->xp) + return -ENOMEM; + + err = arm_ccn_for_each_valid_region(ccn, arm_ccn_init_nodes); + if (err) + return err; + + return arm_ccn_pmu_init(ccn); +} + +static int arm_ccn_remove(struct platform_device *pdev) +{ + struct arm_ccn *ccn = platform_get_drvdata(pdev); + + arm_ccn_pmu_cleanup(ccn); + + return 0; +} + +static const struct of_device_id arm_ccn_match[] = { + { .compatible = "arm,ccn-504", }, + {}, +}; + +static struct platform_driver arm_ccn_driver = { + .driver = { + .name = "arm-ccn", + .of_match_table = arm_ccn_match, + }, + .probe = arm_ccn_probe, + .remove = arm_ccn_remove, +}; + +static int __init arm_ccn_init(void) +{ + int i; + + for (i = 0; i < ARRAY_SIZE(arm_ccn_pmu_events); i++) + arm_ccn_pmu_events_attrs[i] = &arm_ccn_pmu_events[i].attr.attr; + + return platform_driver_register(&arm_ccn_driver); +} + +static void __exit arm_ccn_exit(void) +{ + platform_driver_unregister(&arm_ccn_driver); +} + +module_init(arm_ccn_init); +module_exit(arm_ccn_exit); + +MODULE_AUTHOR("Pawel Moll "); +MODULE_LICENSE("GPL"); diff --git a/drivers/bus/brcmstb_gisb.c b/drivers/bus/brcmstb_gisb.c new file mode 100644 index 000000000..738612c45 --- /dev/null +++ b/drivers/bus/brcmstb_gisb.c @@ -0,0 +1,403 @@ +/* + * Copyright (C) 2014 Broadcom Corporation + * + * 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. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#ifdef CONFIG_ARM +#include +#include +#endif + +#define ARB_ERR_CAP_CLEAR (1 << 0) +#define ARB_ERR_CAP_STATUS_TIMEOUT (1 << 12) +#define ARB_ERR_CAP_STATUS_TEA (1 << 11) +#define ARB_ERR_CAP_STATUS_BS_SHIFT (1 << 2) +#define ARB_ERR_CAP_STATUS_BS_MASK 0x3c +#define ARB_ERR_CAP_STATUS_WRITE (1 << 1) +#define ARB_ERR_CAP_STATUS_VALID (1 << 0) + +enum { + ARB_TIMER, + ARB_ERR_CAP_CLR, + ARB_ERR_CAP_HI_ADDR, + ARB_ERR_CAP_ADDR, + ARB_ERR_CAP_DATA, + ARB_ERR_CAP_STATUS, + ARB_ERR_CAP_MASTER, +}; + +static const int gisb_offsets_bcm7038[] = { + [ARB_TIMER] = 0x00c, + [ARB_ERR_CAP_CLR] = 0x0c4, + [ARB_ERR_CAP_HI_ADDR] = -1, + [ARB_ERR_CAP_ADDR] = 0x0c8, + [ARB_ERR_CAP_DATA] = 0x0cc, + [ARB_ERR_CAP_STATUS] = 0x0d0, + [ARB_ERR_CAP_MASTER] = -1, +}; + +static const int gisb_offsets_bcm7400[] = { + [ARB_TIMER] = 0x00c, + [ARB_ERR_CAP_CLR] = 0x0c8, + [ARB_ERR_CAP_HI_ADDR] = -1, + [ARB_ERR_CAP_ADDR] = 0x0cc, + [ARB_ERR_CAP_DATA] = 0x0d0, + [ARB_ERR_CAP_STATUS] = 0x0d4, + [ARB_ERR_CAP_MASTER] = 0x0d8, +}; + +static const int gisb_offsets_bcm7435[] = { + [ARB_TIMER] = 0x00c, + [ARB_ERR_CAP_CLR] = 0x168, + [ARB_ERR_CAP_HI_ADDR] = -1, + [ARB_ERR_CAP_ADDR] = 0x16c, + [ARB_ERR_CAP_DATA] = 0x170, + [ARB_ERR_CAP_STATUS] = 0x174, + [ARB_ERR_CAP_MASTER] = 0x178, +}; + +static const int gisb_offsets_bcm7445[] = { + [ARB_TIMER] = 0x008, + [ARB_ERR_CAP_CLR] = 0x7e4, + [ARB_ERR_CAP_HI_ADDR] = 0x7e8, + [ARB_ERR_CAP_ADDR] = 0x7ec, + [ARB_ERR_CAP_DATA] = 0x7f0, + [ARB_ERR_CAP_STATUS] = 0x7f4, + [ARB_ERR_CAP_MASTER] = 0x7f8, +}; + +struct brcmstb_gisb_arb_device { + void __iomem *base; + const int *gisb_offsets; + struct mutex lock; + struct list_head next; + u32 valid_mask; + const char *master_names[sizeof(u32) * BITS_PER_BYTE]; + u32 saved_timeout; +}; + +static LIST_HEAD(brcmstb_gisb_arb_device_list); + +static u32 gisb_read(struct brcmstb_gisb_arb_device *gdev, int reg) +{ + int offset = gdev->gisb_offsets[reg]; + + /* return 1 if the hardware doesn't have ARB_ERR_CAP_MASTER */ + if (offset == -1) + return 1; + + return ioread32(gdev->base + offset); +} + +static void gisb_write(struct brcmstb_gisb_arb_device *gdev, u32 val, int reg) +{ + int offset = gdev->gisb_offsets[reg]; + + if (offset == -1) + return; + iowrite32(val, gdev->base + reg); +} + +static ssize_t gisb_arb_get_timeout(struct device *dev, + struct device_attribute *attr, + char *buf) +{ + struct platform_device *pdev = to_platform_device(dev); + struct brcmstb_gisb_arb_device *gdev = platform_get_drvdata(pdev); + u32 timeout; + + mutex_lock(&gdev->lock); + timeout = gisb_read(gdev, ARB_TIMER); + mutex_unlock(&gdev->lock); + + return sprintf(buf, "%d", timeout); +} + +static ssize_t gisb_arb_set_timeout(struct device *dev, + struct device_attribute *attr, + const char *buf, size_t count) +{ + struct platform_device *pdev = to_platform_device(dev); + struct brcmstb_gisb_arb_device *gdev = platform_get_drvdata(pdev); + int val, ret; + + ret = kstrtoint(buf, 10, &val); + if (ret < 0) + return ret; + + if (val == 0 || val >= 0xffffffff) + return -EINVAL; + + mutex_lock(&gdev->lock); + gisb_write(gdev, val, ARB_TIMER); + mutex_unlock(&gdev->lock); + + return count; +} + +static const char * +brcmstb_gisb_master_to_str(struct brcmstb_gisb_arb_device *gdev, + u32 masters) +{ + u32 mask = gdev->valid_mask & masters; + + if (hweight_long(mask) != 1) + return NULL; + + return gdev->master_names[ffs(mask) - 1]; +} + +static int brcmstb_gisb_arb_decode_addr(struct brcmstb_gisb_arb_device *gdev, + const char *reason) +{ + u32 cap_status; + unsigned long arb_addr; + u32 master; + const char *m_name; + char m_fmt[11]; + + cap_status = gisb_read(gdev, ARB_ERR_CAP_STATUS); + + /* Invalid captured address, bail out */ + if (!(cap_status & ARB_ERR_CAP_STATUS_VALID)) + return 1; + + /* Read the address and master */ + arb_addr = gisb_read(gdev, ARB_ERR_CAP_ADDR) & 0xffffffff; +#if (IS_ENABLED(CONFIG_PHYS_ADDR_T_64BIT)) + arb_addr |= (u64)gisb_read(gdev, ARB_ERR_CAP_HI_ADDR) << 32; +#endif + master = gisb_read(gdev, ARB_ERR_CAP_MASTER); + + m_name = brcmstb_gisb_master_to_str(gdev, master); + if (!m_name) { + snprintf(m_fmt, sizeof(m_fmt), "0x%08x", master); + m_name = m_fmt; + } + + pr_crit("%s: %s at 0x%lx [%c %s], core: %s\n", + __func__, reason, arb_addr, + cap_status & ARB_ERR_CAP_STATUS_WRITE ? 'W' : 'R', + cap_status & ARB_ERR_CAP_STATUS_TIMEOUT ? "timeout" : "", + m_name); + + /* clear the GISB error */ + gisb_write(gdev, ARB_ERR_CAP_CLEAR, ARB_ERR_CAP_CLR); + + return 0; +} + +#ifdef CONFIG_ARM +static int brcmstb_bus_error_handler(unsigned long addr, unsigned int fsr, + struct pt_regs *regs) +{ + int ret = 0; + struct brcmstb_gisb_arb_device *gdev; + + /* iterate over each GISB arb registered handlers */ + list_for_each_entry(gdev, &brcmstb_gisb_arb_device_list, next) + ret |= brcmstb_gisb_arb_decode_addr(gdev, "bus error"); + /* + * If it was an imprecise abort, then we need to correct the + * return address to be _after_ the instruction. + */ + if (fsr & (1 << 10)) + regs->ARM_pc += 4; + + return ret; +} +#endif + +static irqreturn_t brcmstb_gisb_timeout_handler(int irq, void *dev_id) +{ + brcmstb_gisb_arb_decode_addr(dev_id, "timeout"); + + return IRQ_HANDLED; +} + +static irqreturn_t brcmstb_gisb_tea_handler(int irq, void *dev_id) +{ + brcmstb_gisb_arb_decode_addr(dev_id, "target abort"); + + return IRQ_HANDLED; +} + +static DEVICE_ATTR(gisb_arb_timeout, S_IWUSR | S_IRUGO, + gisb_arb_get_timeout, gisb_arb_set_timeout); + +static struct attribute *gisb_arb_sysfs_attrs[] = { + &dev_attr_gisb_arb_timeout.attr, + NULL, +}; + +static struct attribute_group gisb_arb_sysfs_attr_group = { + .attrs = gisb_arb_sysfs_attrs, +}; + +static const struct of_device_id brcmstb_gisb_arb_of_match[] = { + { .compatible = "brcm,gisb-arb", .data = gisb_offsets_bcm7445 }, + { .compatible = "brcm,bcm7445-gisb-arb", .data = gisb_offsets_bcm7445 }, + { .compatible = "brcm,bcm7435-gisb-arb", .data = gisb_offsets_bcm7435 }, + { .compatible = "brcm,bcm7400-gisb-arb", .data = gisb_offsets_bcm7400 }, + { .compatible = "brcm,bcm7038-gisb-arb", .data = gisb_offsets_bcm7038 }, + { }, +}; + +static int __init brcmstb_gisb_arb_probe(struct platform_device *pdev) +{ + struct device_node *dn = pdev->dev.of_node; + struct brcmstb_gisb_arb_device *gdev; + const struct of_device_id *of_id; + struct resource *r; + int err, timeout_irq, tea_irq; + unsigned int num_masters, j = 0; + int i, first, last; + + r = platform_get_resource(pdev, IORESOURCE_MEM, 0); + timeout_irq = platform_get_irq(pdev, 0); + tea_irq = platform_get_irq(pdev, 1); + + gdev = devm_kzalloc(&pdev->dev, sizeof(*gdev), GFP_KERNEL); + if (!gdev) + return -ENOMEM; + + mutex_init(&gdev->lock); + INIT_LIST_HEAD(&gdev->next); + + gdev->base = devm_ioremap_resource(&pdev->dev, r); + if (IS_ERR(gdev->base)) + return PTR_ERR(gdev->base); + + of_id = of_match_node(brcmstb_gisb_arb_of_match, dn); + if (!of_id) { + pr_err("failed to look up compatible string\n"); + return -EINVAL; + } + gdev->gisb_offsets = of_id->data; + + err = devm_request_irq(&pdev->dev, timeout_irq, + brcmstb_gisb_timeout_handler, 0, pdev->name, + gdev); + if (err < 0) + return err; + + err = devm_request_irq(&pdev->dev, tea_irq, + brcmstb_gisb_tea_handler, 0, pdev->name, + gdev); + if (err < 0) + return err; + + /* If we do not have a valid mask, assume all masters are enabled */ + if (of_property_read_u32(dn, "brcm,gisb-arb-master-mask", + &gdev->valid_mask)) + gdev->valid_mask = 0xffffffff; + + /* Proceed with reading the litteral names if we agree on the + * number of masters + */ + num_masters = of_property_count_strings(dn, + "brcm,gisb-arb-master-names"); + if (hweight_long(gdev->valid_mask) == num_masters) { + first = ffs(gdev->valid_mask) - 1; + last = fls(gdev->valid_mask) - 1; + + for (i = first; i < last; i++) { + if (!(gdev->valid_mask & BIT(i))) + continue; + + of_property_read_string_index(dn, + "brcm,gisb-arb-master-names", j, + &gdev->master_names[i]); + j++; + } + } + + err = sysfs_create_group(&pdev->dev.kobj, &gisb_arb_sysfs_attr_group); + if (err) + return err; + + platform_set_drvdata(pdev, gdev); + + list_add_tail(&gdev->next, &brcmstb_gisb_arb_device_list); + +#ifdef CONFIG_ARM + hook_fault_code(22, brcmstb_bus_error_handler, SIGBUS, 0, + "imprecise external abort"); +#endif + + dev_info(&pdev->dev, "registered mem: %p, irqs: %d, %d\n", + gdev->base, timeout_irq, tea_irq); + + return 0; +} + +#ifdef CONFIG_PM_SLEEP +static int brcmstb_gisb_arb_suspend(struct device *dev) +{ + struct platform_device *pdev = to_platform_device(dev); + struct brcmstb_gisb_arb_device *gdev = platform_get_drvdata(pdev); + + gdev->saved_timeout = gisb_read(gdev, ARB_TIMER); + + return 0; +} + +/* Make sure we provide the same timeout value that was configured before, and + * do this before the GISB timeout interrupt handler has any chance to run. + */ +static int brcmstb_gisb_arb_resume_noirq(struct device *dev) +{ + struct platform_device *pdev = to_platform_device(dev); + struct brcmstb_gisb_arb_device *gdev = platform_get_drvdata(pdev); + + gisb_write(gdev, gdev->saved_timeout, ARB_TIMER); + + return 0; +} +#else +#define brcmstb_gisb_arb_suspend NULL +#define brcmstb_gisb_arb_resume_noirq NULL +#endif + +static const struct dev_pm_ops brcmstb_gisb_arb_pm_ops = { + .suspend = brcmstb_gisb_arb_suspend, + .resume_noirq = brcmstb_gisb_arb_resume_noirq, +}; + +static struct platform_driver brcmstb_gisb_arb_driver = { + .driver = { + .name = "brcm-gisb-arb", + .of_match_table = brcmstb_gisb_arb_of_match, + .pm = &brcmstb_gisb_arb_pm_ops, + }, +}; + +static int __init brcm_gisb_driver_init(void) +{ + return platform_driver_probe(&brcmstb_gisb_arb_driver, + brcmstb_gisb_arb_probe); +} + +module_init(brcm_gisb_driver_init); diff --git a/drivers/bus/imx-weim.c b/drivers/bus/imx-weim.c new file mode 100644 index 000000000..e98d15eaa --- /dev/null +++ b/drivers/bus/imx-weim.c @@ -0,0 +1,217 @@ +/* + * EIM driver for Freescale's i.MX chips + * + * Copyright (C) 2013 Freescale Semiconductor, Inc. + * + * This file is licensed under the terms of the GNU General Public + * License version 2. This program is licensed "as is" without any + * warranty of any kind, whether express or implied. + */ +#include +#include +#include +#include +#include +#include +#include + +struct imx_weim_devtype { + unsigned int cs_count; + unsigned int cs_regs_count; + unsigned int cs_stride; +}; + +static const struct imx_weim_devtype imx1_weim_devtype = { + .cs_count = 6, + .cs_regs_count = 2, + .cs_stride = 0x08, +}; + +static const struct imx_weim_devtype imx27_weim_devtype = { + .cs_count = 6, + .cs_regs_count = 3, + .cs_stride = 0x10, +}; + +static const struct imx_weim_devtype imx50_weim_devtype = { + .cs_count = 4, + .cs_regs_count = 6, + .cs_stride = 0x18, +}; + +static const struct imx_weim_devtype imx51_weim_devtype = { + .cs_count = 6, + .cs_regs_count = 6, + .cs_stride = 0x18, +}; + +static const struct of_device_id weim_id_table[] = { + /* i.MX1/21 */ + { .compatible = "fsl,imx1-weim", .data = &imx1_weim_devtype, }, + /* i.MX25/27/31/35 */ + { .compatible = "fsl,imx27-weim", .data = &imx27_weim_devtype, }, + /* i.MX50/53/6Q */ + { .compatible = "fsl,imx50-weim", .data = &imx50_weim_devtype, }, + { .compatible = "fsl,imx6q-weim", .data = &imx50_weim_devtype, }, + /* i.MX51 */ + { .compatible = "fsl,imx51-weim", .data = &imx51_weim_devtype, }, + { } +}; +MODULE_DEVICE_TABLE(of, weim_id_table); + +static int __init imx_weim_gpr_setup(struct platform_device *pdev) +{ + struct device_node *np = pdev->dev.of_node; + struct property *prop; + const __be32 *p; + struct regmap *gpr; + u32 gprvals[4] = { + 05, /* CS0(128M) CS1(0M) CS2(0M) CS3(0M) */ + 033, /* CS0(64M) CS1(64M) CS2(0M) CS3(0M) */ + 0113, /* CS0(64M) CS1(32M) CS2(32M) CS3(0M) */ + 01111, /* CS0(32M) CS1(32M) CS2(32M) CS3(32M) */ + }; + u32 gprval = 0; + u32 val; + int cs = 0; + int i = 0; + + gpr = syscon_regmap_lookup_by_phandle(np, "fsl,weim-cs-gpr"); + if (IS_ERR(gpr)) { + dev_dbg(&pdev->dev, "failed to find weim-cs-gpr\n"); + return 0; + } + + of_property_for_each_u32(np, "ranges", prop, p, val) { + if (i % 4 == 0) { + cs = val; + } else if (i % 4 == 3 && val) { + val = (val / SZ_32M) | 1; + gprval |= val << cs * 3; + } + i++; + } + + if (i == 0 || i % 4) + goto err; + + for (i = 0; i < ARRAY_SIZE(gprvals); i++) { + if (gprval == gprvals[i]) { + /* Found it. Set up IOMUXC_GPR1[11:0] with it. */ + regmap_update_bits(gpr, IOMUXC_GPR1, 0xfff, gprval); + return 0; + } + } + +err: + dev_err(&pdev->dev, "Invalid 'ranges' configuration\n"); + return -EINVAL; +} + +/* Parse and set the timing for this device. */ +static int __init weim_timing_setup(struct device_node *np, void __iomem *base, + const struct imx_weim_devtype *devtype) +{ + u32 cs_idx, value[devtype->cs_regs_count]; + int i, ret; + + /* get the CS index from this child node's "reg" property. */ + ret = of_property_read_u32(np, "reg", &cs_idx); + if (ret) + return ret; + + if (cs_idx >= devtype->cs_count) + return -EINVAL; + + ret = of_property_read_u32_array(np, "fsl,weim-cs-timing", + value, devtype->cs_regs_count); + if (ret) + return ret; + + /* set the timing for WEIM */ + for (i = 0; i < devtype->cs_regs_count; i++) + writel(value[i], base + cs_idx * devtype->cs_stride + i * 4); + + return 0; +} + +static int __init weim_parse_dt(struct platform_device *pdev, + void __iomem *base) +{ + const struct of_device_id *of_id = of_match_device(weim_id_table, + &pdev->dev); + const struct imx_weim_devtype *devtype = of_id->data; + struct device_node *child; + int ret, have_child = 0; + + if (devtype == &imx50_weim_devtype) { + ret = imx_weim_gpr_setup(pdev); + if (ret) + return ret; + } + + for_each_child_of_node(pdev->dev.of_node, child) { + if (!child->name) + continue; + + ret = weim_timing_setup(child, base, devtype); + if (ret) + dev_warn(&pdev->dev, "%s set timing failed.\n", + child->full_name); + else + have_child = 1; + } + + if (have_child) + ret = of_platform_populate(pdev->dev.of_node, + of_default_bus_match_table, + NULL, &pdev->dev); + if (ret) + dev_err(&pdev->dev, "%s fail to create devices.\n", + pdev->dev.of_node->full_name); + return ret; +} + +static int __init weim_probe(struct platform_device *pdev) +{ + struct resource *res; + struct clk *clk; + void __iomem *base; + int ret; + + /* get the resource */ + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + base = devm_ioremap_resource(&pdev->dev, res); + if (IS_ERR(base)) + return PTR_ERR(base); + + /* get the clock */ + clk = devm_clk_get(&pdev->dev, NULL); + if (IS_ERR(clk)) + return PTR_ERR(clk); + + ret = clk_prepare_enable(clk); + if (ret) + return ret; + + /* parse the device node */ + ret = weim_parse_dt(pdev, base); + if (ret) + clk_disable_unprepare(clk); + else + dev_info(&pdev->dev, "Driver registered.\n"); + + return ret; +} + +static struct platform_driver weim_driver = { + .driver = { + .name = "imx-weim", + .of_match_table = weim_id_table, + }, +}; +module_platform_driver_probe(weim_driver, weim_probe); + +MODULE_AUTHOR("Freescale Semiconductor Inc."); +MODULE_DESCRIPTION("i.MX EIM Controller Driver"); +MODULE_LICENSE("GPL"); diff --git a/drivers/bus/mips_cdmm.c b/drivers/bus/mips_cdmm.c new file mode 100644 index 000000000..ab3bde16e --- /dev/null +++ b/drivers/bus/mips_cdmm.c @@ -0,0 +1,716 @@ +/* + * Bus driver for MIPS Common Device Memory Map (CDMM). + * + * Copyright (C) 2014-2015 Imagination Technologies Ltd. + * + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +/* Access control and status register fields */ +#define CDMM_ACSR_DEVTYPE_SHIFT 24 +#define CDMM_ACSR_DEVTYPE (255ul << CDMM_ACSR_DEVTYPE_SHIFT) +#define CDMM_ACSR_DEVSIZE_SHIFT 16 +#define CDMM_ACSR_DEVSIZE (31ul << CDMM_ACSR_DEVSIZE_SHIFT) +#define CDMM_ACSR_DEVREV_SHIFT 12 +#define CDMM_ACSR_DEVREV (15ul << CDMM_ACSR_DEVREV_SHIFT) +#define CDMM_ACSR_UW (1ul << 3) +#define CDMM_ACSR_UR (1ul << 2) +#define CDMM_ACSR_SW (1ul << 1) +#define CDMM_ACSR_SR (1ul << 0) + +/* Each block of device registers is 64 bytes */ +#define CDMM_DRB_SIZE 64 + +#define to_mips_cdmm_driver(d) container_of(d, struct mips_cdmm_driver, drv) + +/* Default physical base address */ +static phys_addr_t mips_cdmm_default_base; + +/* Bus operations */ + +static const struct mips_cdmm_device_id * +mips_cdmm_lookup(const struct mips_cdmm_device_id *table, + struct mips_cdmm_device *dev) +{ + int ret = 0; + + for (; table->type; ++table) { + ret = (dev->type == table->type); + if (ret) + break; + } + + return ret ? table : NULL; +} + +static int mips_cdmm_match(struct device *dev, struct device_driver *drv) +{ + struct mips_cdmm_device *cdev = to_mips_cdmm_device(dev); + struct mips_cdmm_driver *cdrv = to_mips_cdmm_driver(drv); + + return mips_cdmm_lookup(cdrv->id_table, cdev) != NULL; +} + +static int mips_cdmm_uevent(struct device *dev, struct kobj_uevent_env *env) +{ + struct mips_cdmm_device *cdev = to_mips_cdmm_device(dev); + int retval = 0; + + retval = add_uevent_var(env, "CDMM_CPU=%u", cdev->cpu); + if (retval) + return retval; + + retval = add_uevent_var(env, "CDMM_TYPE=0x%02x", cdev->type); + if (retval) + return retval; + + retval = add_uevent_var(env, "CDMM_REV=%u", cdev->rev); + if (retval) + return retval; + + retval = add_uevent_var(env, "MODALIAS=mipscdmm:t%02X", cdev->type); + return retval; +} + +/* Device attributes */ + +#define CDMM_ATTR(name, fmt, arg...) \ +static ssize_t name##_show(struct device *_dev, \ + struct device_attribute *attr, char *buf) \ +{ \ + struct mips_cdmm_device *dev = to_mips_cdmm_device(_dev); \ + return sprintf(buf, fmt, arg); \ +} \ +static DEVICE_ATTR_RO(name); + +CDMM_ATTR(cpu, "%u\n", dev->cpu); +CDMM_ATTR(type, "0x%02x\n", dev->type); +CDMM_ATTR(revision, "%u\n", dev->rev); +CDMM_ATTR(modalias, "mipscdmm:t%02X\n", dev->type); +CDMM_ATTR(resource, "\t%016llx\t%016llx\t%016lx\n", + (unsigned long long)dev->res.start, + (unsigned long long)dev->res.end, + dev->res.flags); + +static struct attribute *mips_cdmm_dev_attrs[] = { + &dev_attr_cpu.attr, + &dev_attr_type.attr, + &dev_attr_revision.attr, + &dev_attr_modalias.attr, + &dev_attr_resource.attr, + NULL, +}; +ATTRIBUTE_GROUPS(mips_cdmm_dev); + +struct bus_type mips_cdmm_bustype = { + .name = "cdmm", + .dev_groups = mips_cdmm_dev_groups, + .match = mips_cdmm_match, + .uevent = mips_cdmm_uevent, +}; +EXPORT_SYMBOL_GPL(mips_cdmm_bustype); + +/* + * Standard driver callback helpers. + * + * All the CDMM driver callbacks need to be executed on the appropriate CPU from + * workqueues. For the standard driver callbacks we need a work function + * (mips_cdmm_{void,int}_work()) to do the actual call from the right CPU, and a + * wrapper function (generated with BUILD_PERCPU_HELPER) to arrange for the work + * function to be called on that CPU. + */ + +/** + * struct mips_cdmm_work_dev - Data for per-device call work. + * @fn: CDMM driver callback function to call for the device. + * @dev: CDMM device to pass to @fn. + */ +struct mips_cdmm_work_dev { + void *fn; + struct mips_cdmm_device *dev; +}; + +/** + * mips_cdmm_void_work() - Call a void returning CDMM driver callback. + * @data: struct mips_cdmm_work_dev pointer. + * + * A work_on_cpu() callback function to call an arbitrary CDMM driver callback + * function which doesn't return a value. + */ +static long mips_cdmm_void_work(void *data) +{ + struct mips_cdmm_work_dev *work = data; + void (*fn)(struct mips_cdmm_device *) = work->fn; + + fn(work->dev); + return 0; +} + +/** + * mips_cdmm_int_work() - Call an int returning CDMM driver callback. + * @data: struct mips_cdmm_work_dev pointer. + * + * A work_on_cpu() callback function to call an arbitrary CDMM driver callback + * function which returns an int. + */ +static long mips_cdmm_int_work(void *data) +{ + struct mips_cdmm_work_dev *work = data; + int (*fn)(struct mips_cdmm_device *) = work->fn; + + return fn(work->dev); +} + +#define _BUILD_RET_void +#define _BUILD_RET_int return + +/** + * BUILD_PERCPU_HELPER() - Helper to call a CDMM driver callback on right CPU. + * @_ret: Return type (void or int). + * @_name: Name of CDMM driver callback function. + * + * Generates a specific device callback function to call a CDMM driver callback + * function on the appropriate CPU for the device, and if applicable return the + * result. + */ +#define BUILD_PERCPU_HELPER(_ret, _name) \ +static _ret mips_cdmm_##_name(struct device *dev) \ +{ \ + struct mips_cdmm_device *cdev = to_mips_cdmm_device(dev); \ + struct mips_cdmm_driver *cdrv = to_mips_cdmm_driver(dev->driver); \ + struct mips_cdmm_work_dev work = { \ + .fn = cdrv->_name, \ + .dev = cdev, \ + }; \ + \ + _BUILD_RET_##_ret work_on_cpu(cdev->cpu, \ + mips_cdmm_##_ret##_work, &work); \ +} + +/* Driver callback functions */ +BUILD_PERCPU_HELPER(int, probe) /* int mips_cdmm_probe(struct device) */ +BUILD_PERCPU_HELPER(int, remove) /* int mips_cdmm_remove(struct device) */ +BUILD_PERCPU_HELPER(void, shutdown) /* void mips_cdmm_shutdown(struct device) */ + + +/* Driver registration */ + +/** + * mips_cdmm_driver_register() - Register a CDMM driver. + * @drv: CDMM driver information. + * + * Register a CDMM driver with the CDMM subsystem. The driver will be informed + * of matching devices which are discovered. + * + * Returns: 0 on success. + */ +int mips_cdmm_driver_register(struct mips_cdmm_driver *drv) +{ + drv->drv.bus = &mips_cdmm_bustype; + + if (drv->probe) + drv->drv.probe = mips_cdmm_probe; + if (drv->remove) + drv->drv.remove = mips_cdmm_remove; + if (drv->shutdown) + drv->drv.shutdown = mips_cdmm_shutdown; + + return driver_register(&drv->drv); +} +EXPORT_SYMBOL_GPL(mips_cdmm_driver_register); + +/** + * mips_cdmm_driver_unregister() - Unregister a CDMM driver. + * @drv: CDMM driver information. + * + * Unregister a CDMM driver from the CDMM subsystem. + */ +void mips_cdmm_driver_unregister(struct mips_cdmm_driver *drv) +{ + driver_unregister(&drv->drv); +} +EXPORT_SYMBOL_GPL(mips_cdmm_driver_unregister); + + +/* CDMM initialisation and bus discovery */ + +/** + * struct mips_cdmm_bus - Info about CDMM bus. + * @phys: Physical address at which it is mapped. + * @regs: Virtual address where registers can be accessed. + * @drbs: Total number of DRBs. + * @drbs_reserved: Number of DRBs reserved. + * @discovered: Whether the devices on the bus have been discovered yet. + * @offline: Whether the CDMM bus is going offline (or very early + * coming back online), in which case it should be + * reconfigured each time. + */ +struct mips_cdmm_bus { + phys_addr_t phys; + void __iomem *regs; + unsigned int drbs; + unsigned int drbs_reserved; + bool discovered; + bool offline; +}; + +static struct mips_cdmm_bus mips_cdmm_boot_bus; +static DEFINE_PER_CPU(struct mips_cdmm_bus *, mips_cdmm_buses); +static atomic_t mips_cdmm_next_id = ATOMIC_INIT(-1); + +/** + * mips_cdmm_get_bus() - Get the per-CPU CDMM bus information. + * + * Get information about the per-CPU CDMM bus, if the bus is present. + * + * The caller must prevent migration to another CPU, either by disabling + * pre-emption or by running from a pinned kernel thread. + * + * Returns: Pointer to CDMM bus information for the current CPU. + * May return ERR_PTR(-errno) in case of error, so check with + * IS_ERR(). + */ +static struct mips_cdmm_bus *mips_cdmm_get_bus(void) +{ + struct mips_cdmm_bus *bus, **bus_p; + unsigned long flags; + unsigned int cpu; + + if (!cpu_has_cdmm) + return ERR_PTR(-ENODEV); + + cpu = smp_processor_id(); + /* Avoid early use of per-cpu primitives before initialised */ + if (cpu == 0) + return &mips_cdmm_boot_bus; + + /* Get bus pointer */ + bus_p = per_cpu_ptr(&mips_cdmm_buses, cpu); + local_irq_save(flags); + bus = *bus_p; + /* Attempt allocation if NULL */ + if (unlikely(!bus)) { + bus = kzalloc(sizeof(*bus), GFP_ATOMIC); + if (unlikely(!bus)) + bus = ERR_PTR(-ENOMEM); + else + *bus_p = bus; + } + local_irq_restore(flags); + return bus; +} + +/** + * mips_cdmm_cur_base() - Find current physical base address of CDMM region. + * + * Returns: Physical base address of CDMM region according to cdmmbase CP0 + * register, or 0 if the CDMM region is disabled. + */ +static phys_addr_t mips_cdmm_cur_base(void) +{ + unsigned long cdmmbase = read_c0_cdmmbase(); + + if (!(cdmmbase & MIPS_CDMMBASE_EN)) + return 0; + + return (cdmmbase >> MIPS_CDMMBASE_ADDR_SHIFT) + << MIPS_CDMMBASE_ADDR_START; +} + +/** + * mips_cdmm_setup() - Ensure the CDMM bus is initialised and usable. + * @bus: Pointer to bus information for current CPU. + * IS_ERR(bus) is checked, so no need for caller to check. + * + * The caller must prevent migration to another CPU, either by disabling + * pre-emption or by running from a pinned kernel thread. + * + * Returns 0 on success, -errno on failure. + */ +static int mips_cdmm_setup(struct mips_cdmm_bus *bus) +{ + unsigned long cdmmbase, flags; + int ret = 0; + + if (IS_ERR(bus)) + return PTR_ERR(bus); + + local_irq_save(flags); + /* Don't set up bus a second time unless marked offline */ + if (bus->offline) { + /* If CDMM region is still set up, nothing to do */ + if (bus->phys == mips_cdmm_cur_base()) + goto out; + /* + * The CDMM region isn't set up as expected, so it needs + * reconfiguring, but then we can stop checking it. + */ + bus->offline = false; + } else if (bus->phys > 1) { + goto out; + } + + /* If the CDMM region is already configured, inherit that setup */ + if (!bus->phys) + bus->phys = mips_cdmm_cur_base(); + /* Otherwise, ask platform code for suggestions */ + if (!bus->phys && mips_cdmm_phys_base) + bus->phys = mips_cdmm_phys_base(); + /* Otherwise, copy what other CPUs have done */ + if (!bus->phys) + bus->phys = mips_cdmm_default_base; + /* Otherwise, complain once */ + if (!bus->phys) { + bus->phys = 1; + /* + * If you hit this, either your bootloader needs to set up the + * CDMM on the boot CPU, or else you need to implement + * mips_cdmm_phys_base() for your platform (see asm/cdmm.h). + */ + pr_err("cdmm%u: Failed to choose a physical base\n", + smp_processor_id()); + } + /* Already complained? */ + if (bus->phys == 1) { + ret = -ENOMEM; + goto out; + } + /* Record our success for other CPUs to copy */ + mips_cdmm_default_base = bus->phys; + + pr_debug("cdmm%u: Enabling CDMM region at %pa\n", + smp_processor_id(), &bus->phys); + + /* Enable CDMM */ + cdmmbase = read_c0_cdmmbase(); + cdmmbase &= (1ul << MIPS_CDMMBASE_ADDR_SHIFT) - 1; + cdmmbase |= (bus->phys >> MIPS_CDMMBASE_ADDR_START) + << MIPS_CDMMBASE_ADDR_SHIFT; + cdmmbase |= MIPS_CDMMBASE_EN; + write_c0_cdmmbase(cdmmbase); + tlbw_use_hazard(); + + bus->regs = (void __iomem *)CKSEG1ADDR(bus->phys); + bus->drbs = 1 + ((cdmmbase & MIPS_CDMMBASE_SIZE) >> + MIPS_CDMMBASE_SIZE_SHIFT); + bus->drbs_reserved = !!(cdmmbase & MIPS_CDMMBASE_CI); + +out: + local_irq_restore(flags); + return ret; +} + +/** + * mips_cdmm_early_probe() - Minimally probe for a specific device on CDMM. + * @dev_type: CDMM type code to look for. + * + * Minimally configure the in-CPU Common Device Memory Map (CDMM) and look for a + * specific device. This can be used to find a device very early in boot for + * example to configure an early FDC console device. + * + * The caller must prevent migration to another CPU, either by disabling + * pre-emption or by running from a pinned kernel thread. + * + * Returns: MMIO pointer to device memory. The caller can read the ACSR + * register to find more information about the device (such as the + * version number or the number of blocks). + * May return IOMEM_ERR_PTR(-errno) in case of error, so check with + * IS_ERR(). + */ +void __iomem *mips_cdmm_early_probe(unsigned int dev_type) +{ + struct mips_cdmm_bus *bus; + void __iomem *cdmm; + u32 acsr; + unsigned int drb, type, size; + int err; + + if (WARN_ON(!dev_type)) + return IOMEM_ERR_PTR(-ENODEV); + + bus = mips_cdmm_get_bus(); + err = mips_cdmm_setup(bus); + if (err) + return IOMEM_ERR_PTR(err); + + /* Skip the first block if it's reserved for more registers */ + drb = bus->drbs_reserved; + cdmm = bus->regs; + + /* Look for a specific device type */ + for (; drb < bus->drbs; drb += size + 1) { + acsr = __raw_readl(cdmm + drb * CDMM_DRB_SIZE); + type = (acsr & CDMM_ACSR_DEVTYPE) >> CDMM_ACSR_DEVTYPE_SHIFT; + if (type == dev_type) + return cdmm + drb * CDMM_DRB_SIZE; + size = (acsr & CDMM_ACSR_DEVSIZE) >> CDMM_ACSR_DEVSIZE_SHIFT; + } + + return IOMEM_ERR_PTR(-ENODEV); +} +EXPORT_SYMBOL_GPL(mips_cdmm_early_probe); + +/** + * mips_cdmm_release() - Release a removed CDMM device. + * @dev: Device object + * + * Clean up the struct mips_cdmm_device for an unused CDMM device. This is + * called automatically by the driver core when a device is removed. + */ +static void mips_cdmm_release(struct device *dev) +{ + struct mips_cdmm_device *cdev = to_mips_cdmm_device(dev); + + kfree(cdev); +} + +/** + * mips_cdmm_bus_discover() - Discover the devices on the CDMM bus. + * @bus: CDMM bus information, must already be set up. + */ +static void mips_cdmm_bus_discover(struct mips_cdmm_bus *bus) +{ + void __iomem *cdmm; + u32 acsr; + unsigned int drb, type, size, rev; + struct mips_cdmm_device *dev; + unsigned int cpu = smp_processor_id(); + int ret = 0; + int id = 0; + + /* Skip the first block if it's reserved for more registers */ + drb = bus->drbs_reserved; + cdmm = bus->regs; + + /* Discover devices */ + bus->discovered = true; + pr_info("cdmm%u discovery (%u blocks)\n", cpu, bus->drbs); + for (; drb < bus->drbs; drb += size + 1) { + acsr = __raw_readl(cdmm + drb * CDMM_DRB_SIZE); + type = (acsr & CDMM_ACSR_DEVTYPE) >> CDMM_ACSR_DEVTYPE_SHIFT; + size = (acsr & CDMM_ACSR_DEVSIZE) >> CDMM_ACSR_DEVSIZE_SHIFT; + rev = (acsr & CDMM_ACSR_DEVREV) >> CDMM_ACSR_DEVREV_SHIFT; + + if (!type) + continue; + + pr_info("cdmm%u-%u: @%u (%#x..%#x), type 0x%02x, rev %u\n", + cpu, id, drb, drb * CDMM_DRB_SIZE, + (drb + size + 1) * CDMM_DRB_SIZE - 1, + type, rev); + + dev = kzalloc(sizeof(*dev), GFP_KERNEL); + if (!dev) + break; + + dev->cpu = cpu; + dev->res.start = bus->phys + drb * CDMM_DRB_SIZE; + dev->res.end = bus->phys + + (drb + size + 1) * CDMM_DRB_SIZE - 1; + dev->res.flags = IORESOURCE_MEM; + dev->type = type; + dev->rev = rev; + dev->dev.parent = get_cpu_device(cpu); + dev->dev.bus = &mips_cdmm_bustype; + dev->dev.id = atomic_inc_return(&mips_cdmm_next_id); + dev->dev.release = mips_cdmm_release; + + dev_set_name(&dev->dev, "cdmm%u-%u", cpu, id); + ++id; + ret = device_register(&dev->dev); + if (ret) { + put_device(&dev->dev); + kfree(dev); + } + } +} + + +/* + * CPU hotplug and initialisation + * + * All the CDMM driver callbacks need to be executed on the appropriate CPU from + * workqueues. For the CPU callbacks, they need to be called for all devices on + * that CPU, so the work function calls bus_for_each_dev, using a helper + * (generated with BUILD_PERDEV_HELPER) to call the driver callback if the + * device's CPU matches. + */ + +/** + * BUILD_PERDEV_HELPER() - Helper to call a CDMM driver callback if CPU matches. + * @_name: Name of CDMM driver callback function. + * + * Generates a bus_for_each_dev callback function to call a specific CDMM driver + * callback function for the device if the device's CPU matches that pointed to + * by the data argument. + * + * This is used for informing drivers for all devices on a given CPU of some + * event (such as the CPU going online/offline). + * + * It is expected to already be called from the appropriate CPU. + */ +#define BUILD_PERDEV_HELPER(_name) \ +static int mips_cdmm_##_name##_helper(struct device *dev, void *data) \ +{ \ + struct mips_cdmm_device *cdev = to_mips_cdmm_device(dev); \ + struct mips_cdmm_driver *cdrv; \ + unsigned int cpu = *(unsigned int *)data; \ + \ + if (cdev->cpu != cpu || !dev->driver) \ + return 0; \ + \ + cdrv = to_mips_cdmm_driver(dev->driver); \ + if (!cdrv->_name) \ + return 0; \ + return cdrv->_name(cdev); \ +} + +/* bus_for_each_dev callback helper functions */ +BUILD_PERDEV_HELPER(cpu_down) /* int mips_cdmm_cpu_down_helper(...) */ +BUILD_PERDEV_HELPER(cpu_up) /* int mips_cdmm_cpu_up_helper(...) */ + +/** + * mips_cdmm_bus_down() - Tear down the CDMM bus. + * @data: Pointer to unsigned int CPU number. + * + * This work_on_cpu callback function is executed on a given CPU to call the + * CDMM driver cpu_down callback for all devices on that CPU. + */ +static long mips_cdmm_bus_down(void *data) +{ + struct mips_cdmm_bus *bus; + long ret; + + /* Inform all the devices on the bus */ + ret = bus_for_each_dev(&mips_cdmm_bustype, NULL, data, + mips_cdmm_cpu_down_helper); + + /* + * While bus is offline, each use of it should reconfigure it just in + * case it is first use when coming back online again. + */ + bus = mips_cdmm_get_bus(); + if (!IS_ERR(bus)) + bus->offline = true; + + return ret; +} + +/** + * mips_cdmm_bus_up() - Bring up the CDMM bus. + * @data: Pointer to unsigned int CPU number. + * + * This work_on_cpu callback function is executed on a given CPU to discover + * CDMM devices on that CPU, or to call the CDMM driver cpu_up callback for all + * devices already discovered on that CPU. + * + * It is used during initialisation and when CPUs are brought online. + */ +static long mips_cdmm_bus_up(void *data) +{ + struct mips_cdmm_bus *bus; + long ret; + + bus = mips_cdmm_get_bus(); + ret = mips_cdmm_setup(bus); + if (ret) + return ret; + + /* Bus now set up, so we can drop the offline flag if still set */ + bus->offline = false; + + if (!bus->discovered) + mips_cdmm_bus_discover(bus); + else + /* Inform all the devices on the bus */ + ret = bus_for_each_dev(&mips_cdmm_bustype, NULL, data, + mips_cdmm_cpu_up_helper); + + return ret; +} + +/** + * mips_cdmm_cpu_notify() - Take action when a CPU is going online or offline. + * @nb: CPU notifier block . + * @action: Event that has taken place (CPU_*). + * @data: CPU number. + * + * This notifier is used to keep the CDMM buses updated as CPUs are offlined and + * onlined. When CPUs go offline or come back online, so does their CDMM bus, so + * devices must be informed. Also when CPUs come online for the first time the + * devices on the CDMM bus need discovering. + * + * Returns: NOTIFY_OK if event was used. + * NOTIFY_DONE if we didn't care. + */ +static int mips_cdmm_cpu_notify(struct notifier_block *nb, + unsigned long action, void *data) +{ + unsigned int cpu = (unsigned int)data; + + switch (action & ~CPU_TASKS_FROZEN) { + case CPU_ONLINE: + case CPU_DOWN_FAILED: + work_on_cpu(cpu, mips_cdmm_bus_up, &cpu); + break; + case CPU_DOWN_PREPARE: + work_on_cpu(cpu, mips_cdmm_bus_down, &cpu); + break; + default: + return NOTIFY_DONE; + } + + return NOTIFY_OK; +} + +static struct notifier_block mips_cdmm_cpu_nb = { + .notifier_call = mips_cdmm_cpu_notify, +}; + +/** + * mips_cdmm_init() - Initialise CDMM bus. + * + * Initialise CDMM bus, discover CDMM devices for online CPUs, and arrange for + * hotplug notifications so the CDMM drivers can be kept up to date. + */ +static int __init mips_cdmm_init(void) +{ + unsigned int cpu; + int ret; + + /* Register the bus */ + ret = bus_register(&mips_cdmm_bustype); + if (ret) + return ret; + + /* We want to be notified about new CPUs */ + ret = register_cpu_notifier(&mips_cdmm_cpu_nb); + if (ret) { + pr_warn("cdmm: Failed to register CPU notifier\n"); + goto out; + } + + /* Discover devices on CDMM of online CPUs */ + for_each_online_cpu(cpu) + work_on_cpu(cpu, mips_cdmm_bus_up, &cpu); + + return 0; +out: + bus_unregister(&mips_cdmm_bustype); + return ret; +} +subsys_initcall(mips_cdmm_init); diff --git a/drivers/bus/mvebu-mbus.c b/drivers/bus/mvebu-mbus.c new file mode 100644 index 000000000..6f047dcb9 --- /dev/null +++ b/drivers/bus/mvebu-mbus.c @@ -0,0 +1,1209 @@ +/* + * Address map functions for Marvell EBU SoCs (Kirkwood, Armada + * 370/XP, Dove, Orion5x and MV78xx0) + * + * This file is licensed under the terms of the GNU General Public + * License version 2. This program is licensed "as is" without any + * warranty of any kind, whether express or implied. + * + * The Marvell EBU SoCs have a configurable physical address space: + * the physical address at which certain devices (PCIe, NOR, NAND, + * etc.) sit can be configured. The configuration takes place through + * two sets of registers: + * + * - One to configure the access of the CPU to the devices. Depending + * on the families, there are between 8 and 20 configurable windows, + * each can be use to create a physical memory window that maps to a + * specific device. Devices are identified by a tuple (target, + * attribute). + * + * - One to configure the access to the CPU to the SDRAM. There are + * either 2 (for Dove) or 4 (for other families) windows to map the + * SDRAM into the physical address space. + * + * This driver: + * + * - Reads out the SDRAM address decoding windows at initialization + * time, and fills the mvebu_mbus_dram_info structure with these + * informations. The exported function mv_mbus_dram_info() allow + * device drivers to get those informations related to the SDRAM + * address decoding windows. This is because devices also have their + * own windows (configured through registers that are part of each + * device register space), and therefore the drivers for Marvell + * devices have to configure those device -> SDRAM windows to ensure + * that DMA works properly. + * + * - Provides an API for platform code or device drivers to + * dynamically add or remove address decoding windows for the CPU -> + * device accesses. This API is mvebu_mbus_add_window_by_id(), + * mvebu_mbus_add_window_remap_by_id() and + * mvebu_mbus_del_window(). + * + * - Provides a debugfs interface in /sys/kernel/debug/mvebu-mbus/ to + * see the list of CPU -> SDRAM windows and their configuration + * (file 'sdram') and the list of CPU -> devices windows and their + * configuration (file 'devices'). + */ + +#define pr_fmt(fmt) KBUILD_MODNAME ": " fmt + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +/* + * DDR target is the same on all platforms. + */ +#define TARGET_DDR 0 + +/* + * CPU Address Decode Windows registers + */ +#define WIN_CTRL_OFF 0x0000 +#define WIN_CTRL_ENABLE BIT(0) +/* Only on HW I/O coherency capable platforms */ +#define WIN_CTRL_SYNCBARRIER BIT(1) +#define WIN_CTRL_TGT_MASK 0xf0 +#define WIN_CTRL_TGT_SHIFT 4 +#define WIN_CTRL_ATTR_MASK 0xff00 +#define WIN_CTRL_ATTR_SHIFT 8 +#define WIN_CTRL_SIZE_MASK 0xffff0000 +#define WIN_CTRL_SIZE_SHIFT 16 +#define WIN_BASE_OFF 0x0004 +#define WIN_BASE_LOW 0xffff0000 +#define WIN_BASE_HIGH 0xf +#define WIN_REMAP_LO_OFF 0x0008 +#define WIN_REMAP_LOW 0xffff0000 +#define WIN_REMAP_HI_OFF 0x000c + +#define UNIT_SYNC_BARRIER_OFF 0x84 +#define UNIT_SYNC_BARRIER_ALL 0xFFFF + +#define ATTR_HW_COHERENCY (0x1 << 4) + +#define DDR_BASE_CS_OFF(n) (0x0000 + ((n) << 3)) +#define DDR_BASE_CS_HIGH_MASK 0xf +#define DDR_BASE_CS_LOW_MASK 0xff000000 +#define DDR_SIZE_CS_OFF(n) (0x0004 + ((n) << 3)) +#define DDR_SIZE_ENABLED BIT(0) +#define DDR_SIZE_CS_MASK 0x1c +#define DDR_SIZE_CS_SHIFT 2 +#define DDR_SIZE_MASK 0xff000000 + +#define DOVE_DDR_BASE_CS_OFF(n) ((n) << 4) + +/* Relative to mbusbridge_base */ +#define MBUS_BRIDGE_CTRL_OFF 0x0 +#define MBUS_BRIDGE_BASE_OFF 0x4 + +/* Maximum number of windows, for all known platforms */ +#define MBUS_WINS_MAX 20 + +struct mvebu_mbus_state; + +struct mvebu_mbus_soc_data { + unsigned int num_wins; + bool has_mbus_bridge; + unsigned int (*win_cfg_offset)(const int win); + unsigned int (*win_remap_offset)(const int win); + void (*setup_cpu_target)(struct mvebu_mbus_state *s); + int (*save_cpu_target)(struct mvebu_mbus_state *s, + u32 *store_addr); + int (*show_cpu_target)(struct mvebu_mbus_state *s, + struct seq_file *seq, void *v); +}; + +/* + * Used to store the state of one MBus window accross suspend/resume. + */ +struct mvebu_mbus_win_data { + u32 ctrl; + u32 base; + u32 remap_lo; + u32 remap_hi; +}; + +struct mvebu_mbus_state { + void __iomem *mbuswins_base; + void __iomem *sdramwins_base; + void __iomem *mbusbridge_base; + phys_addr_t sdramwins_phys_base; + struct dentry *debugfs_root; + struct dentry *debugfs_sdram; + struct dentry *debugfs_devs; + struct resource pcie_mem_aperture; + struct resource pcie_io_aperture; + const struct mvebu_mbus_soc_data *soc; + int hw_io_coherency; + + /* Used during suspend/resume */ + u32 mbus_bridge_ctrl; + u32 mbus_bridge_base; + struct mvebu_mbus_win_data wins[MBUS_WINS_MAX]; +}; + +static struct mvebu_mbus_state mbus_state; + +static struct mbus_dram_target_info mvebu_mbus_dram_info; +const struct mbus_dram_target_info *mv_mbus_dram_info(void) +{ + return &mvebu_mbus_dram_info; +} +EXPORT_SYMBOL_GPL(mv_mbus_dram_info); + +/* Checks whether the given window has remap capability */ +static bool mvebu_mbus_window_is_remappable(struct mvebu_mbus_state *mbus, + const int win) +{ + return mbus->soc->win_remap_offset(win) != MVEBU_MBUS_NO_REMAP; +} + +/* + * Functions to manipulate the address decoding windows + */ + +static void mvebu_mbus_read_window(struct mvebu_mbus_state *mbus, + int win, int *enabled, u64 *base, + u32 *size, u8 *target, u8 *attr, + u64 *remap) +{ + void __iomem *addr = mbus->mbuswins_base + + mbus->soc->win_cfg_offset(win); + u32 basereg = readl(addr + WIN_BASE_OFF); + u32 ctrlreg = readl(addr + WIN_CTRL_OFF); + + if (!(ctrlreg & WIN_CTRL_ENABLE)) { + *enabled = 0; + return; + } + + *enabled = 1; + *base = ((u64)basereg & WIN_BASE_HIGH) << 32; + *base |= (basereg & WIN_BASE_LOW); + *size = (ctrlreg | ~WIN_CTRL_SIZE_MASK) + 1; + + if (target) + *target = (ctrlreg & WIN_CTRL_TGT_MASK) >> WIN_CTRL_TGT_SHIFT; + + if (attr) + *attr = (ctrlreg & WIN_CTRL_ATTR_MASK) >> WIN_CTRL_ATTR_SHIFT; + + if (remap) { + if (mvebu_mbus_window_is_remappable(mbus, win)) { + u32 remap_low, remap_hi; + void __iomem *addr_rmp = mbus->mbuswins_base + + mbus->soc->win_remap_offset(win); + remap_low = readl(addr_rmp + WIN_REMAP_LO_OFF); + remap_hi = readl(addr_rmp + WIN_REMAP_HI_OFF); + *remap = ((u64)remap_hi << 32) | remap_low; + } else + *remap = 0; + } +} + +static void mvebu_mbus_disable_window(struct mvebu_mbus_state *mbus, + int win) +{ + void __iomem *addr; + + addr = mbus->mbuswins_base + mbus->soc->win_cfg_offset(win); + writel(0, addr + WIN_BASE_OFF); + writel(0, addr + WIN_CTRL_OFF); + + if (mvebu_mbus_window_is_remappable(mbus, win)) { + addr = mbus->mbuswins_base + mbus->soc->win_remap_offset(win); + writel(0, addr + WIN_REMAP_LO_OFF); + writel(0, addr + WIN_REMAP_HI_OFF); + } +} + +/* Checks whether the given window number is available */ + +static int mvebu_mbus_window_is_free(struct mvebu_mbus_state *mbus, + const int win) +{ + void __iomem *addr = mbus->mbuswins_base + + mbus->soc->win_cfg_offset(win); + u32 ctrl = readl(addr + WIN_CTRL_OFF); + + return !(ctrl & WIN_CTRL_ENABLE); +} + +/* + * Checks whether the given (base, base+size) area doesn't overlap an + * existing region + */ +static int mvebu_mbus_window_conflicts(struct mvebu_mbus_state *mbus, + phys_addr_t base, size_t size, + u8 target, u8 attr) +{ + u64 end = (u64)base + size; + int win; + + for (win = 0; win < mbus->soc->num_wins; win++) { + u64 wbase, wend; + u32 wsize; + u8 wtarget, wattr; + int enabled; + + mvebu_mbus_read_window(mbus, win, + &enabled, &wbase, &wsize, + &wtarget, &wattr, NULL); + + if (!enabled) + continue; + + wend = wbase + wsize; + + /* + * Check if the current window overlaps with the + * proposed physical range + */ + if ((u64)base < wend && end > wbase) + return 0; + } + + return 1; +} + +static int mvebu_mbus_find_window(struct mvebu_mbus_state *mbus, + phys_addr_t base, size_t size) +{ + int win; + + for (win = 0; win < mbus->soc->num_wins; win++) { + u64 wbase; + u32 wsize; + int enabled; + + mvebu_mbus_read_window(mbus, win, + &enabled, &wbase, &wsize, + NULL, NULL, NULL); + + if (!enabled) + continue; + + if (base == wbase && size == wsize) + return win; + } + + return -ENODEV; +} + +static int mvebu_mbus_setup_window(struct mvebu_mbus_state *mbus, + int win, phys_addr_t base, size_t size, + phys_addr_t remap, u8 target, + u8 attr) +{ + void __iomem *addr = mbus->mbuswins_base + + mbus->soc->win_cfg_offset(win); + u32 ctrl, remap_addr; + + if (!is_power_of_2(size)) { + WARN(true, "Invalid MBus window size: 0x%zx\n", size); + return -EINVAL; + } + + if ((base & (phys_addr_t)(size - 1)) != 0) { + WARN(true, "Invalid MBus base/size: %pa len 0x%zx\n", &base, + size); + return -EINVAL; + } + + ctrl = ((size - 1) & WIN_CTRL_SIZE_MASK) | + (attr << WIN_CTRL_ATTR_SHIFT) | + (target << WIN_CTRL_TGT_SHIFT) | + WIN_CTRL_ENABLE; + if (mbus->hw_io_coherency) + ctrl |= WIN_CTRL_SYNCBARRIER; + + writel(base & WIN_BASE_LOW, addr + WIN_BASE_OFF); + writel(ctrl, addr + WIN_CTRL_OFF); + + if (mvebu_mbus_window_is_remappable(mbus, win)) { + void __iomem *addr_rmp = mbus->mbuswins_base + + mbus->soc->win_remap_offset(win); + + if (remap == MVEBU_MBUS_NO_REMAP) + remap_addr = base; + else + remap_addr = remap; + writel(remap_addr & WIN_REMAP_LOW, addr_rmp + WIN_REMAP_LO_OFF); + writel(0, addr_rmp + WIN_REMAP_HI_OFF); + } + + return 0; +} + +static int mvebu_mbus_alloc_window(struct mvebu_mbus_state *mbus, + phys_addr_t base, size_t size, + phys_addr_t remap, u8 target, + u8 attr) +{ + int win; + + if (remap == MVEBU_MBUS_NO_REMAP) { + for (win = 0; win < mbus->soc->num_wins; win++) { + if (mvebu_mbus_window_is_remappable(mbus, win)) + continue; + + if (mvebu_mbus_window_is_free(mbus, win)) + return mvebu_mbus_setup_window(mbus, win, base, + size, remap, + target, attr); + } + } + + for (win = 0; win < mbus->soc->num_wins; win++) { + /* Skip window if need remap but is not supported */ + if ((remap != MVEBU_MBUS_NO_REMAP) && + !mvebu_mbus_window_is_remappable(mbus, win)) + continue; + + if (mvebu_mbus_window_is_free(mbus, win)) + return mvebu_mbus_setup_window(mbus, win, base, size, + remap, target, attr); + } + + return -ENOMEM; +} + +/* + * Debugfs debugging + */ + +/* Common function used for Dove, Kirkwood, Armada 370/XP and Orion 5x */ +static int mvebu_sdram_debug_show_orion(struct mvebu_mbus_state *mbus, + struct seq_file *seq, void *v) +{ + int i; + + for (i = 0; i < 4; i++) { + u32 basereg = readl(mbus->sdramwins_base + DDR_BASE_CS_OFF(i)); + u32 sizereg = readl(mbus->sdramwins_base + DDR_SIZE_CS_OFF(i)); + u64 base; + u32 size; + + if (!(sizereg & DDR_SIZE_ENABLED)) { + seq_printf(seq, "[%d] disabled\n", i); + continue; + } + + base = ((u64)basereg & DDR_BASE_CS_HIGH_MASK) << 32; + base |= basereg & DDR_BASE_CS_LOW_MASK; + size = (sizereg | ~DDR_SIZE_MASK); + + seq_printf(seq, "[%d] %016llx - %016llx : cs%d\n", + i, (unsigned long long)base, + (unsigned long long)base + size + 1, + (sizereg & DDR_SIZE_CS_MASK) >> DDR_SIZE_CS_SHIFT); + } + + return 0; +} + +/* Special function for Dove */ +static int mvebu_sdram_debug_show_dove(struct mvebu_mbus_state *mbus, + struct seq_file *seq, void *v) +{ + int i; + + for (i = 0; i < 2; i++) { + u32 map = readl(mbus->sdramwins_base + DOVE_DDR_BASE_CS_OFF(i)); + u64 base; + u32 size; + + if (!(map & 1)) { + seq_printf(seq, "[%d] disabled\n", i); + continue; + } + + base = map & 0xff800000; + size = 0x100000 << (((map & 0x000f0000) >> 16) - 4); + + seq_printf(seq, "[%d] %016llx - %016llx : cs%d\n", + i, (unsigned long long)base, + (unsigned long long)base + size, i); + } + + return 0; +} + +static int mvebu_sdram_debug_show(struct seq_file *seq, void *v) +{ + struct mvebu_mbus_state *mbus = &mbus_state; + return mbus->soc->show_cpu_target(mbus, seq, v); +} + +static int mvebu_sdram_debug_open(struct inode *inode, struct file *file) +{ + return single_open(file, mvebu_sdram_debug_show, inode->i_private); +} + +static const struct file_operations mvebu_sdram_debug_fops = { + .open = mvebu_sdram_debug_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +static int mvebu_devs_debug_show(struct seq_file *seq, void *v) +{ + struct mvebu_mbus_state *mbus = &mbus_state; + int win; + + for (win = 0; win < mbus->soc->num_wins; win++) { + u64 wbase, wremap; + u32 wsize; + u8 wtarget, wattr; + int enabled; + + mvebu_mbus_read_window(mbus, win, + &enabled, &wbase, &wsize, + &wtarget, &wattr, &wremap); + + if (!enabled) { + seq_printf(seq, "[%02d] disabled\n", win); + continue; + } + + seq_printf(seq, "[%02d] %016llx - %016llx : %04x:%04x", + win, (unsigned long long)wbase, + (unsigned long long)(wbase + wsize), wtarget, wattr); + + if (!is_power_of_2(wsize) || + ((wbase & (u64)(wsize - 1)) != 0)) + seq_puts(seq, " (Invalid base/size!!)"); + + if (mvebu_mbus_window_is_remappable(mbus, win)) { + seq_printf(seq, " (remap %016llx)\n", + (unsigned long long)wremap); + } else + seq_printf(seq, "\n"); + } + + return 0; +} + +static int mvebu_devs_debug_open(struct inode *inode, struct file *file) +{ + return single_open(file, mvebu_devs_debug_show, inode->i_private); +} + +static const struct file_operations mvebu_devs_debug_fops = { + .open = mvebu_devs_debug_open, + .read = seq_read, + .llseek = seq_lseek, + .release = single_release, +}; + +/* + * SoC-specific functions and definitions + */ + +static unsigned int generic_mbus_win_cfg_offset(int win) +{ + return win << 4; +} + +static unsigned int armada_370_xp_mbus_win_cfg_offset(int win) +{ + /* The register layout is a bit annoying and the below code + * tries to cope with it. + * - At offset 0x0, there are the registers for the first 8 + * windows, with 4 registers of 32 bits per window (ctrl, + * base, remap low, remap high) + * - Then at offset 0x80, there is a hole of 0x10 bytes for + * the internal registers base address and internal units + * sync barrier register. + * - Then at offset 0x90, there the registers for 12 + * windows, with only 2 registers of 32 bits per window + * (ctrl, base). + */ + if (win < 8) + return win << 4; + else + return 0x90 + ((win - 8) << 3); +} + +static unsigned int mv78xx0_mbus_win_cfg_offset(int win) +{ + if (win < 8) + return win << 4; + else + return 0x900 + ((win - 8) << 4); +} + +static unsigned int generic_mbus_win_remap_2_offset(int win) +{ + if (win < 2) + return generic_mbus_win_cfg_offset(win); + else + return MVEBU_MBUS_NO_REMAP; +} + +static unsigned int generic_mbus_win_remap_4_offset(int win) +{ + if (win < 4) + return generic_mbus_win_cfg_offset(win); + else + return MVEBU_MBUS_NO_REMAP; +} + +static unsigned int generic_mbus_win_remap_8_offset(int win) +{ + if (win < 8) + return generic_mbus_win_cfg_offset(win); + else + return MVEBU_MBUS_NO_REMAP; +} + +static unsigned int armada_xp_mbus_win_remap_offset(int win) +{ + if (win < 8) + return generic_mbus_win_cfg_offset(win); + else if (win == 13) + return 0xF0 - WIN_REMAP_LO_OFF; + else + return MVEBU_MBUS_NO_REMAP; +} + +static void __init +mvebu_mbus_default_setup_cpu_target(struct mvebu_mbus_state *mbus) +{ + int i; + int cs; + + mvebu_mbus_dram_info.mbus_dram_target_id = TARGET_DDR; + + for (i = 0, cs = 0; i < 4; i++) { + u32 base = readl(mbus->sdramwins_base + DDR_BASE_CS_OFF(i)); + u32 size = readl(mbus->sdramwins_base + DDR_SIZE_CS_OFF(i)); + + /* + * We only take care of entries for which the chip + * select is enabled, and that don't have high base + * address bits set (devices can only access the first + * 32 bits of the memory). + */ + if ((size & DDR_SIZE_ENABLED) && + !(base & DDR_BASE_CS_HIGH_MASK)) { + struct mbus_dram_window *w; + + w = &mvebu_mbus_dram_info.cs[cs++]; + w->cs_index = i; + w->mbus_attr = 0xf & ~(1 << i); + if (mbus->hw_io_coherency) + w->mbus_attr |= ATTR_HW_COHERENCY; + w->base = base & DDR_BASE_CS_LOW_MASK; + w->size = (size | ~DDR_SIZE_MASK) + 1; + } + } + mvebu_mbus_dram_info.num_cs = cs; +} + +static int +mvebu_mbus_default_save_cpu_target(struct mvebu_mbus_state *mbus, + u32 *store_addr) +{ + int i; + + for (i = 0; i < 4; i++) { + u32 base = readl(mbus->sdramwins_base + DDR_BASE_CS_OFF(i)); + u32 size = readl(mbus->sdramwins_base + DDR_SIZE_CS_OFF(i)); + + writel(mbus->sdramwins_phys_base + DDR_BASE_CS_OFF(i), + store_addr++); + writel(base, store_addr++); + writel(mbus->sdramwins_phys_base + DDR_SIZE_CS_OFF(i), + store_addr++); + writel(size, store_addr++); + } + + /* We've written 16 words to the store address */ + return 16; +} + +static void __init +mvebu_mbus_dove_setup_cpu_target(struct mvebu_mbus_state *mbus) +{ + int i; + int cs; + + mvebu_mbus_dram_info.mbus_dram_target_id = TARGET_DDR; + + for (i = 0, cs = 0; i < 2; i++) { + u32 map = readl(mbus->sdramwins_base + DOVE_DDR_BASE_CS_OFF(i)); + + /* + * Chip select enabled? + */ + if (map & 1) { + struct mbus_dram_window *w; + + w = &mvebu_mbus_dram_info.cs[cs++]; + w->cs_index = i; + w->mbus_attr = 0; /* CS address decoding done inside */ + /* the DDR controller, no need to */ + /* provide attributes */ + w->base = map & 0xff800000; + w->size = 0x100000 << (((map & 0x000f0000) >> 16) - 4); + } + } + + mvebu_mbus_dram_info.num_cs = cs; +} + +static int +mvebu_mbus_dove_save_cpu_target(struct mvebu_mbus_state *mbus, + u32 *store_addr) +{ + int i; + + for (i = 0; i < 2; i++) { + u32 map = readl(mbus->sdramwins_base + DOVE_DDR_BASE_CS_OFF(i)); + + writel(mbus->sdramwins_phys_base + DOVE_DDR_BASE_CS_OFF(i), + store_addr++); + writel(map, store_addr++); + } + + /* We've written 4 words to the store address */ + return 4; +} + +int mvebu_mbus_save_cpu_target(u32 *store_addr) +{ + return mbus_state.soc->save_cpu_target(&mbus_state, store_addr); +} + +static const struct mvebu_mbus_soc_data armada_370_mbus_data = { + .num_wins = 20, + .has_mbus_bridge = true, + .win_cfg_offset = armada_370_xp_mbus_win_cfg_offset, + .win_remap_offset = generic_mbus_win_remap_8_offset, + .setup_cpu_target = mvebu_mbus_default_setup_cpu_target, + .show_cpu_target = mvebu_sdram_debug_show_orion, + .save_cpu_target = mvebu_mbus_default_save_cpu_target, +}; + +static const struct mvebu_mbus_soc_data armada_xp_mbus_data = { + .num_wins = 20, + .has_mbus_bridge = true, + .win_cfg_offset = armada_370_xp_mbus_win_cfg_offset, + .win_remap_offset = armada_xp_mbus_win_remap_offset, + .setup_cpu_target = mvebu_mbus_default_setup_cpu_target, + .show_cpu_target = mvebu_sdram_debug_show_orion, + .save_cpu_target = mvebu_mbus_default_save_cpu_target, +}; + +static const struct mvebu_mbus_soc_data kirkwood_mbus_data = { + .num_wins = 8, + .win_cfg_offset = generic_mbus_win_cfg_offset, + .save_cpu_target = mvebu_mbus_default_save_cpu_target, + .win_remap_offset = generic_mbus_win_remap_4_offset, + .setup_cpu_target = mvebu_mbus_default_setup_cpu_target, + .show_cpu_target = mvebu_sdram_debug_show_orion, +}; + +static const struct mvebu_mbus_soc_data dove_mbus_data = { + .num_wins = 8, + .win_cfg_offset = generic_mbus_win_cfg_offset, + .save_cpu_target = mvebu_mbus_dove_save_cpu_target, + .win_remap_offset = generic_mbus_win_remap_4_offset, + .setup_cpu_target = mvebu_mbus_dove_setup_cpu_target, + .show_cpu_target = mvebu_sdram_debug_show_dove, +}; + +/* + * Some variants of Orion5x have 4 remappable windows, some other have + * only two of them. + */ +static const struct mvebu_mbus_soc_data orion5x_4win_mbus_data = { + .num_wins = 8, + .win_cfg_offset = generic_mbus_win_cfg_offset, + .save_cpu_target = mvebu_mbus_default_save_cpu_target, + .win_remap_offset = generic_mbus_win_remap_4_offset, + .setup_cpu_target = mvebu_mbus_default_setup_cpu_target, + .show_cpu_target = mvebu_sdram_debug_show_orion, +}; + +static const struct mvebu_mbus_soc_data orion5x_2win_mbus_data = { + .num_wins = 8, + .win_cfg_offset = generic_mbus_win_cfg_offset, + .save_cpu_target = mvebu_mbus_default_save_cpu_target, + .win_remap_offset = generic_mbus_win_remap_2_offset, + .setup_cpu_target = mvebu_mbus_default_setup_cpu_target, + .show_cpu_target = mvebu_sdram_debug_show_orion, +}; + +static const struct mvebu_mbus_soc_data mv78xx0_mbus_data = { + .num_wins = 14, + .win_cfg_offset = mv78xx0_mbus_win_cfg_offset, + .save_cpu_target = mvebu_mbus_default_save_cpu_target, + .win_remap_offset = generic_mbus_win_remap_8_offset, + .setup_cpu_target = mvebu_mbus_default_setup_cpu_target, + .show_cpu_target = mvebu_sdram_debug_show_orion, +}; + +static const struct of_device_id of_mvebu_mbus_ids[] = { + { .compatible = "marvell,armada370-mbus", + .data = &armada_370_mbus_data, }, + { .compatible = "marvell,armada375-mbus", + .data = &armada_xp_mbus_data, }, + { .compatible = "marvell,armada380-mbus", + .data = &armada_xp_mbus_data, }, + { .compatible = "marvell,armadaxp-mbus", + .data = &armada_xp_mbus_data, }, + { .compatible = "marvell,kirkwood-mbus", + .data = &kirkwood_mbus_data, }, + { .compatible = "marvell,dove-mbus", + .data = &dove_mbus_data, }, + { .compatible = "marvell,orion5x-88f5281-mbus", + .data = &orion5x_4win_mbus_data, }, + { .compatible = "marvell,orion5x-88f5182-mbus", + .data = &orion5x_2win_mbus_data, }, + { .compatible = "marvell,orion5x-88f5181-mbus", + .data = &orion5x_2win_mbus_data, }, + { .compatible = "marvell,orion5x-88f6183-mbus", + .data = &orion5x_4win_mbus_data, }, + { .compatible = "marvell,mv78xx0-mbus", + .data = &mv78xx0_mbus_data, }, + { }, +}; + +/* + * Public API of the driver + */ +int mvebu_mbus_add_window_remap_by_id(unsigned int target, + unsigned int attribute, + phys_addr_t base, size_t size, + phys_addr_t remap) +{ + struct mvebu_mbus_state *s = &mbus_state; + + if (!mvebu_mbus_window_conflicts(s, base, size, target, attribute)) { + pr_err("cannot add window '%x:%x', conflicts with another window\n", + target, attribute); + return -EINVAL; + } + + return mvebu_mbus_alloc_window(s, base, size, remap, target, attribute); +} + +int mvebu_mbus_add_window_by_id(unsigned int target, unsigned int attribute, + phys_addr_t base, size_t size) +{ + return mvebu_mbus_add_window_remap_by_id(target, attribute, base, + size, MVEBU_MBUS_NO_REMAP); +} + +int mvebu_mbus_del_window(phys_addr_t base, size_t size) +{ + int win; + + win = mvebu_mbus_find_window(&mbus_state, base, size); + if (win < 0) + return win; + + mvebu_mbus_disable_window(&mbus_state, win); + return 0; +} + +void mvebu_mbus_get_pcie_mem_aperture(struct resource *res) +{ + if (!res) + return; + *res = mbus_state.pcie_mem_aperture; +} + +void mvebu_mbus_get_pcie_io_aperture(struct resource *res) +{ + if (!res) + return; + *res = mbus_state.pcie_io_aperture; +} + +static __init int mvebu_mbus_debugfs_init(void) +{ + struct mvebu_mbus_state *s = &mbus_state; + + /* + * If no base has been initialized, doesn't make sense to + * register the debugfs entries. We may be on a multiplatform + * kernel that isn't running a Marvell EBU SoC. + */ + if (!s->mbuswins_base) + return 0; + + s->debugfs_root = debugfs_create_dir("mvebu-mbus", NULL); + if (s->debugfs_root) { + s->debugfs_sdram = debugfs_create_file("sdram", S_IRUGO, + s->debugfs_root, NULL, + &mvebu_sdram_debug_fops); + s->debugfs_devs = debugfs_create_file("devices", S_IRUGO, + s->debugfs_root, NULL, + &mvebu_devs_debug_fops); + } + + return 0; +} +fs_initcall(mvebu_mbus_debugfs_init); + +static int mvebu_mbus_suspend(void) +{ + struct mvebu_mbus_state *s = &mbus_state; + int win; + + if (!s->mbusbridge_base) + return -ENODEV; + + for (win = 0; win < s->soc->num_wins; win++) { + void __iomem *addr = s->mbuswins_base + + s->soc->win_cfg_offset(win); + void __iomem *addr_rmp; + + s->wins[win].base = readl(addr + WIN_BASE_OFF); + s->wins[win].ctrl = readl(addr + WIN_CTRL_OFF); + + if (!mvebu_mbus_window_is_remappable(s, win)) + continue; + + addr_rmp = s->mbuswins_base + + s->soc->win_remap_offset(win); + + s->wins[win].remap_lo = readl(addr_rmp + WIN_REMAP_LO_OFF); + s->wins[win].remap_hi = readl(addr_rmp + WIN_REMAP_HI_OFF); + } + + s->mbus_bridge_ctrl = readl(s->mbusbridge_base + + MBUS_BRIDGE_CTRL_OFF); + s->mbus_bridge_base = readl(s->mbusbridge_base + + MBUS_BRIDGE_BASE_OFF); + + return 0; +} + +static void mvebu_mbus_resume(void) +{ + struct mvebu_mbus_state *s = &mbus_state; + int win; + + writel(s->mbus_bridge_ctrl, + s->mbusbridge_base + MBUS_BRIDGE_CTRL_OFF); + writel(s->mbus_bridge_base, + s->mbusbridge_base + MBUS_BRIDGE_BASE_OFF); + + for (win = 0; win < s->soc->num_wins; win++) { + void __iomem *addr = s->mbuswins_base + + s->soc->win_cfg_offset(win); + void __iomem *addr_rmp; + + writel(s->wins[win].base, addr + WIN_BASE_OFF); + writel(s->wins[win].ctrl, addr + WIN_CTRL_OFF); + + if (!mvebu_mbus_window_is_remappable(s, win)) + continue; + + addr_rmp = s->mbuswins_base + + s->soc->win_remap_offset(win); + + writel(s->wins[win].remap_lo, addr_rmp + WIN_REMAP_LO_OFF); + writel(s->wins[win].remap_hi, addr_rmp + WIN_REMAP_HI_OFF); + } +} + +struct syscore_ops mvebu_mbus_syscore_ops = { + .suspend = mvebu_mbus_suspend, + .resume = mvebu_mbus_resume, +}; + +static int __init mvebu_mbus_common_init(struct mvebu_mbus_state *mbus, + phys_addr_t mbuswins_phys_base, + size_t mbuswins_size, + phys_addr_t sdramwins_phys_base, + size_t sdramwins_size, + phys_addr_t mbusbridge_phys_base, + size_t mbusbridge_size, + bool is_coherent) +{ + int win; + + mbus->mbuswins_base = ioremap(mbuswins_phys_base, mbuswins_size); + if (!mbus->mbuswins_base) + return -ENOMEM; + + mbus->sdramwins_base = ioremap(sdramwins_phys_base, sdramwins_size); + if (!mbus->sdramwins_base) { + iounmap(mbus_state.mbuswins_base); + return -ENOMEM; + } + + mbus->sdramwins_phys_base = sdramwins_phys_base; + + if (mbusbridge_phys_base) { + mbus->mbusbridge_base = ioremap(mbusbridge_phys_base, + mbusbridge_size); + if (!mbus->mbusbridge_base) { + iounmap(mbus->sdramwins_base); + iounmap(mbus->mbuswins_base); + return -ENOMEM; + } + } else + mbus->mbusbridge_base = NULL; + + for (win = 0; win < mbus->soc->num_wins; win++) + mvebu_mbus_disable_window(mbus, win); + + mbus->soc->setup_cpu_target(mbus); + + if (is_coherent) + writel(UNIT_SYNC_BARRIER_ALL, + mbus->mbuswins_base + UNIT_SYNC_BARRIER_OFF); + + register_syscore_ops(&mvebu_mbus_syscore_ops); + + return 0; +} + +int __init mvebu_mbus_init(const char *soc, phys_addr_t mbuswins_phys_base, + size_t mbuswins_size, + phys_addr_t sdramwins_phys_base, + size_t sdramwins_size) +{ + const struct of_device_id *of_id; + + for (of_id = of_mvebu_mbus_ids; of_id->compatible[0]; of_id++) + if (!strcmp(of_id->compatible, soc)) + break; + + if (!of_id->compatible[0]) { + pr_err("could not find a matching SoC family\n"); + return -ENODEV; + } + + mbus_state.soc = of_id->data; + + return mvebu_mbus_common_init(&mbus_state, + mbuswins_phys_base, + mbuswins_size, + sdramwins_phys_base, + sdramwins_size, 0, 0, false); +} + +#ifdef CONFIG_OF +/* + * The window IDs in the ranges DT property have the following format: + * - bits 28 to 31: MBus custom field + * - bits 24 to 27: window target ID + * - bits 16 to 23: window attribute ID + * - bits 0 to 15: unused + */ +#define CUSTOM(id) (((id) & 0xF0000000) >> 24) +#define TARGET(id) (((id) & 0x0F000000) >> 24) +#define ATTR(id) (((id) & 0x00FF0000) >> 16) + +static int __init mbus_dt_setup_win(struct mvebu_mbus_state *mbus, + u32 base, u32 size, + u8 target, u8 attr) +{ + if (!mvebu_mbus_window_conflicts(mbus, base, size, target, attr)) { + pr_err("cannot add window '%04x:%04x', conflicts with another window\n", + target, attr); + return -EBUSY; + } + + if (mvebu_mbus_alloc_window(mbus, base, size, MVEBU_MBUS_NO_REMAP, + target, attr)) { + pr_err("cannot add window '%04x:%04x', too many windows\n", + target, attr); + return -ENOMEM; + } + return 0; +} + +static int __init +mbus_parse_ranges(struct device_node *node, + int *addr_cells, int *c_addr_cells, int *c_size_cells, + int *cell_count, const __be32 **ranges_start, + const __be32 **ranges_end) +{ + const __be32 *prop; + int ranges_len, tuple_len; + + /* Allow a node with no 'ranges' property */ + *ranges_start = of_get_property(node, "ranges", &ranges_len); + if (*ranges_start == NULL) { + *addr_cells = *c_addr_cells = *c_size_cells = *cell_count = 0; + *ranges_start = *ranges_end = NULL; + return 0; + } + *ranges_end = *ranges_start + ranges_len / sizeof(__be32); + + *addr_cells = of_n_addr_cells(node); + + prop = of_get_property(node, "#address-cells", NULL); + *c_addr_cells = be32_to_cpup(prop); + + prop = of_get_property(node, "#size-cells", NULL); + *c_size_cells = be32_to_cpup(prop); + + *cell_count = *addr_cells + *c_addr_cells + *c_size_cells; + tuple_len = (*cell_count) * sizeof(__be32); + + if (ranges_len % tuple_len) { + pr_warn("malformed ranges entry '%s'\n", node->name); + return -EINVAL; + } + return 0; +} + +static int __init mbus_dt_setup(struct mvebu_mbus_state *mbus, + struct device_node *np) +{ + int addr_cells, c_addr_cells, c_size_cells; + int i, ret, cell_count; + const __be32 *r, *ranges_start, *ranges_end; + + ret = mbus_parse_ranges(np, &addr_cells, &c_addr_cells, + &c_size_cells, &cell_count, + &ranges_start, &ranges_end); + if (ret < 0) + return ret; + + for (i = 0, r = ranges_start; r < ranges_end; r += cell_count, i++) { + u32 windowid, base, size; + u8 target, attr; + + /* + * An entry with a non-zero custom field do not + * correspond to a static window, so skip it. + */ + windowid = of_read_number(r, 1); + if (CUSTOM(windowid)) + continue; + + target = TARGET(windowid); + attr = ATTR(windowid); + + base = of_read_number(r + c_addr_cells, addr_cells); + size = of_read_number(r + c_addr_cells + addr_cells, + c_size_cells); + ret = mbus_dt_setup_win(mbus, base, size, target, attr); + if (ret < 0) + return ret; + } + return 0; +} + +static void __init mvebu_mbus_get_pcie_resources(struct device_node *np, + struct resource *mem, + struct resource *io) +{ + u32 reg[2]; + int ret; + + /* + * These are optional, so we make sure that resource_size(x) will + * return 0. + */ + memset(mem, 0, sizeof(struct resource)); + mem->end = -1; + memset(io, 0, sizeof(struct resource)); + io->end = -1; + + ret = of_property_read_u32_array(np, "pcie-mem-aperture", reg, ARRAY_SIZE(reg)); + if (!ret) { + mem->start = reg[0]; + mem->end = mem->start + reg[1] - 1; + mem->flags = IORESOURCE_MEM; + } + + ret = of_property_read_u32_array(np, "pcie-io-aperture", reg, ARRAY_SIZE(reg)); + if (!ret) { + io->start = reg[0]; + io->end = io->start + reg[1] - 1; + io->flags = IORESOURCE_IO; + } +} + +int __init mvebu_mbus_dt_init(bool is_coherent) +{ + struct resource mbuswins_res, sdramwins_res, mbusbridge_res; + struct device_node *np, *controller; + const struct of_device_id *of_id; + const __be32 *prop; + int ret; + + np = of_find_matching_node_and_match(NULL, of_mvebu_mbus_ids, &of_id); + if (!np) { + pr_err("could not find a matching SoC family\n"); + return -ENODEV; + } + + mbus_state.soc = of_id->data; + + prop = of_get_property(np, "controller", NULL); + if (!prop) { + pr_err("required 'controller' property missing\n"); + return -EINVAL; + } + + controller = of_find_node_by_phandle(be32_to_cpup(prop)); + if (!controller) { + pr_err("could not find an 'mbus-controller' node\n"); + return -ENODEV; + } + + if (of_address_to_resource(controller, 0, &mbuswins_res)) { + pr_err("cannot get MBUS register address\n"); + return -EINVAL; + } + + if (of_address_to_resource(controller, 1, &sdramwins_res)) { + pr_err("cannot get SDRAM register address\n"); + return -EINVAL; + } + + /* + * Set the resource to 0 so that it can be left unmapped by + * mvebu_mbus_common_init() if the DT doesn't carry the + * necessary information. This is needed to preserve backward + * compatibility. + */ + memset(&mbusbridge_res, 0, sizeof(mbusbridge_res)); + + if (mbus_state.soc->has_mbus_bridge) { + if (of_address_to_resource(controller, 2, &mbusbridge_res)) + pr_warn(FW_WARN "deprecated mbus-mvebu Device Tree, suspend/resume will not work\n"); + } + + mbus_state.hw_io_coherency = is_coherent; + + /* Get optional pcie-{mem,io}-aperture properties */ + mvebu_mbus_get_pcie_resources(np, &mbus_state.pcie_mem_aperture, + &mbus_state.pcie_io_aperture); + + ret = mvebu_mbus_common_init(&mbus_state, + mbuswins_res.start, + resource_size(&mbuswins_res), + sdramwins_res.start, + resource_size(&sdramwins_res), + mbusbridge_res.start, + resource_size(&mbusbridge_res), + is_coherent); + if (ret) + return ret; + + /* Setup statically declared windows in the DT */ + return mbus_dt_setup(&mbus_state, np); +} +#endif diff --git a/drivers/bus/omap-ocp2scp.c b/drivers/bus/omap-ocp2scp.c new file mode 100644 index 000000000..9f1856948 --- /dev/null +++ b/drivers/bus/omap-ocp2scp.c @@ -0,0 +1,123 @@ +/* + * omap-ocp2scp.c - transform ocp interface protocol to scp protocol + * + * Copyright (C) 2012 Texas Instruments Incorporated - http://www.ti.com + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * Author: Kishon Vijay Abraham I + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + */ + +#include +#include +#include +#include +#include +#include +#include + +#define OCP2SCP_TIMING 0x18 +#define SYNC2_MASK 0xf + +static int ocp2scp_remove_devices(struct device *dev, void *c) +{ + struct platform_device *pdev = to_platform_device(dev); + + platform_device_unregister(pdev); + + return 0; +} + +static int omap_ocp2scp_probe(struct platform_device *pdev) +{ + int ret; + u32 reg; + void __iomem *regs; + struct resource *res; + struct device_node *np = pdev->dev.of_node; + + if (np) { + ret = of_platform_populate(np, NULL, NULL, &pdev->dev); + if (ret) { + dev_err(&pdev->dev, + "failed to add resources for ocp2scp child\n"); + goto err0; + } + } + + pm_runtime_enable(&pdev->dev); + /* + * As per AM572x TRM: http://www.ti.com/lit/ug/spruhz6/spruhz6.pdf + * under section 26.3.2.2, table 26-26 OCP2SCP TIMING Caution; + * As per OMAP4430 TRM: http://www.ti.com/lit/ug/swpu231ap/swpu231ap.pdf + * under section 23.12.6.2.2 , Table 23-1213 OCP2SCP TIMING Caution; + * As per OMAP4460 TRM: http://www.ti.com/lit/ug/swpu235ab/swpu235ab.pdf + * under section 23.12.6.2.2, Table 23-1213 OCP2SCP TIMING Caution; + * As per OMAP543x TRM http://www.ti.com/lit/pdf/swpu249 + * under section 27.3.2.2, Table 27-27 OCP2SCP TIMING Caution; + * + * Read path of OCP2SCP is not working properly due to low reset value + * of SYNC2 parameter in OCP2SCP. Suggested reset value is 0x6 or more. + */ + if (!of_device_is_compatible(np, "ti,am437x-ocp2scp")) { + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + regs = devm_ioremap_resource(&pdev->dev, res); + if (IS_ERR(regs)) + goto err0; + + pm_runtime_get_sync(&pdev->dev); + reg = readl_relaxed(regs + OCP2SCP_TIMING); + reg &= ~(SYNC2_MASK); + reg |= 0x6; + writel_relaxed(reg, regs + OCP2SCP_TIMING); + pm_runtime_put_sync(&pdev->dev); + } + + return 0; + +err0: + device_for_each_child(&pdev->dev, NULL, ocp2scp_remove_devices); + + return ret; +} + +static int omap_ocp2scp_remove(struct platform_device *pdev) +{ + pm_runtime_disable(&pdev->dev); + device_for_each_child(&pdev->dev, NULL, ocp2scp_remove_devices); + + return 0; +} + +#ifdef CONFIG_OF +static const struct of_device_id omap_ocp2scp_id_table[] = { + { .compatible = "ti,omap-ocp2scp" }, + { .compatible = "ti,am437x-ocp2scp" }, + {} +}; +MODULE_DEVICE_TABLE(of, omap_ocp2scp_id_table); +#endif + +static struct platform_driver omap_ocp2scp_driver = { + .probe = omap_ocp2scp_probe, + .remove = omap_ocp2scp_remove, + .driver = { + .name = "omap-ocp2scp", + .of_match_table = of_match_ptr(omap_ocp2scp_id_table), + }, +}; + +module_platform_driver(omap_ocp2scp_driver); + +MODULE_ALIAS("platform: omap-ocp2scp"); +MODULE_AUTHOR("Texas Instruments Inc."); +MODULE_DESCRIPTION("OMAP OCP2SCP driver"); +MODULE_LICENSE("GPL v2"); diff --git a/drivers/bus/omap_l3_noc.c b/drivers/bus/omap_l3_noc.c new file mode 100644 index 000000000..ebee57d71 --- /dev/null +++ b/drivers/bus/omap_l3_noc.c @@ -0,0 +1,377 @@ +/* + * OMAP L3 Interconnect error handling driver + * + * Copyright (C) 2011-2015 Texas Instruments Incorporated - http://www.ti.com/ + * Santosh Shilimkar + * Sricharan + * + * 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. + * + * This program is distributed "as is" WITHOUT ANY WARRANTY of any + * kind, whether express or implied; without even the implied warranty + * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "omap_l3_noc.h" + +/** + * l3_handle_target() - Handle Target specific parse and reporting + * @l3: pointer to l3 struct + * @base: base address of clkdm + * @flag_mux: flagmux corresponding to the event + * @err_src: error source index of the slave (target) + * + * This does the second part of the error interrupt handling: + * 3) Parse in the slave information + * 4) Print the logged information. + * 5) Add dump stack to provide kernel trace. + * 6) Clear the source if known. + * + * This handles two types of errors: + * 1) Custom errors in L3 : + * Target like DMM/FW/EMIF generates SRESP=ERR error + * 2) Standard L3 error: + * - Unsupported CMD. + * L3 tries to access target while it is idle + * - OCP disconnect. + * - Address hole error: + * If DSS/ISS/FDIF/USBHOSTFS access a target where they + * do not have connectivity, the error is logged in + * their default target which is DMM2. + * + * On High Secure devices, firewall errors are possible and those + * can be trapped as well. But the trapping is implemented as part + * secure software and hence need not be implemented here. + */ +static int l3_handle_target(struct omap_l3 *l3, void __iomem *base, + struct l3_flagmux_data *flag_mux, int err_src) +{ + int k; + u32 std_err_main, clear, masterid; + u8 op_code, m_req_info; + void __iomem *l3_targ_base; + void __iomem *l3_targ_stderr, *l3_targ_slvofslsb, *l3_targ_mstaddr; + void __iomem *l3_targ_hdr, *l3_targ_info; + struct l3_target_data *l3_targ_inst; + struct l3_masters_data *master; + char *target_name, *master_name = "UN IDENTIFIED"; + char *err_description; + char err_string[30] = { 0 }; + char info_string[60] = { 0 }; + + /* We DONOT expect err_src to go out of bounds */ + BUG_ON(err_src > MAX_CLKDM_TARGETS); + + if (err_src < flag_mux->num_targ_data) { + l3_targ_inst = &flag_mux->l3_targ[err_src]; + target_name = l3_targ_inst->name; + l3_targ_base = base + l3_targ_inst->offset; + } else { + target_name = L3_TARGET_NOT_SUPPORTED; + } + + if (target_name == L3_TARGET_NOT_SUPPORTED) + return -ENODEV; + + /* Read the stderrlog_main_source from clk domain */ + l3_targ_stderr = l3_targ_base + L3_TARG_STDERRLOG_MAIN; + l3_targ_slvofslsb = l3_targ_base + L3_TARG_STDERRLOG_SLVOFSLSB; + + std_err_main = readl_relaxed(l3_targ_stderr); + + switch (std_err_main & CUSTOM_ERROR) { + case STANDARD_ERROR: + err_description = "Standard"; + snprintf(err_string, sizeof(err_string), + ": At Address: 0x%08X ", + readl_relaxed(l3_targ_slvofslsb)); + + l3_targ_mstaddr = l3_targ_base + L3_TARG_STDERRLOG_MSTADDR; + l3_targ_hdr = l3_targ_base + L3_TARG_STDERRLOG_HDR; + l3_targ_info = l3_targ_base + L3_TARG_STDERRLOG_INFO; + break; + + case CUSTOM_ERROR: + err_description = "Custom"; + + l3_targ_mstaddr = l3_targ_base + + L3_TARG_STDERRLOG_CINFO_MSTADDR; + l3_targ_hdr = l3_targ_base + L3_TARG_STDERRLOG_CINFO_OPCODE; + l3_targ_info = l3_targ_base + L3_TARG_STDERRLOG_CINFO_INFO; + break; + + default: + /* Nothing to be handled here as of now */ + return 0; + } + + /* STDERRLOG_MSTADDR Stores the NTTP master address. */ + masterid = (readl_relaxed(l3_targ_mstaddr) & + l3->mst_addr_mask) >> __ffs(l3->mst_addr_mask); + + for (k = 0, master = l3->l3_masters; k < l3->num_masters; + k++, master++) { + if (masterid == master->id) { + master_name = master->name; + break; + } + } + + op_code = readl_relaxed(l3_targ_hdr) & 0x7; + + m_req_info = readl_relaxed(l3_targ_info) & 0xF; + snprintf(info_string, sizeof(info_string), + ": %s in %s mode during %s access", + (m_req_info & BIT(0)) ? "Opcode Fetch" : "Data Access", + (m_req_info & BIT(1)) ? "Supervisor" : "User", + (m_req_info & BIT(3)) ? "Debug" : "Functional"); + + WARN(true, + "%s:L3 %s Error: MASTER %s TARGET %s (%s)%s%s\n", + dev_name(l3->dev), + err_description, + master_name, target_name, + l3_transaction_type[op_code], + err_string, info_string); + + /* clear the std error log*/ + clear = std_err_main | CLEAR_STDERR_LOG; + writel_relaxed(clear, l3_targ_stderr); + + return 0; +} + +/** + * l3_interrupt_handler() - interrupt handler for l3 events + * @irq: irq number + * @_l3: pointer to l3 structure + * + * Interrupt Handler for L3 error detection. + * 1) Identify the L3 clockdomain partition to which the error belongs to. + * 2) Identify the slave where the error information is logged + * ... handle the slave event.. + * 7) if the slave is unknown, mask out the slave. + */ +static irqreturn_t l3_interrupt_handler(int irq, void *_l3) +{ + struct omap_l3 *l3 = _l3; + int inttype, i, ret; + int err_src = 0; + u32 err_reg, mask_val; + void __iomem *base, *mask_reg; + struct l3_flagmux_data *flag_mux; + + /* Get the Type of interrupt */ + inttype = irq == l3->app_irq ? L3_APPLICATION_ERROR : L3_DEBUG_ERROR; + + for (i = 0; i < l3->num_modules; i++) { + /* + * Read the regerr register of the clock domain + * to determine the source + */ + base = l3->l3_base[i]; + flag_mux = l3->l3_flagmux[i]; + err_reg = readl_relaxed(base + flag_mux->offset + + L3_FLAGMUX_REGERR0 + (inttype << 3)); + + err_reg &= ~(inttype ? flag_mux->mask_app_bits : + flag_mux->mask_dbg_bits); + + /* Get the corresponding error and analyse */ + if (err_reg) { + /* Identify the source from control status register */ + err_src = __ffs(err_reg); + + ret = l3_handle_target(l3, base, flag_mux, err_src); + + /* + * Certain plaforms may have "undocumented" status + * pending on boot. So dont generate a severe warning + * here. Just mask it off to prevent the error from + * reoccuring and locking up the system. + */ + if (ret) { + dev_err(l3->dev, + "L3 %s error: target %d mod:%d %s\n", + inttype ? "debug" : "application", + err_src, i, "(unclearable)"); + + mask_reg = base + flag_mux->offset + + L3_FLAGMUX_MASK0 + (inttype << 3); + mask_val = readl_relaxed(mask_reg); + mask_val &= ~(1 << err_src); + writel_relaxed(mask_val, mask_reg); + + /* Mark these bits as to be ignored */ + if (inttype) + flag_mux->mask_app_bits |= 1 << err_src; + else + flag_mux->mask_dbg_bits |= 1 << err_src; + } + + /* Error found so break the for loop */ + return IRQ_HANDLED; + } + } + + dev_err(l3->dev, "L3 %s IRQ not handled!!\n", + inttype ? "debug" : "application"); + + return IRQ_NONE; +} + +static const struct of_device_id l3_noc_match[] = { + {.compatible = "ti,omap4-l3-noc", .data = &omap4_l3_data}, + {.compatible = "ti,omap5-l3-noc", .data = &omap5_l3_data}, + {.compatible = "ti,dra7-l3-noc", .data = &dra_l3_data}, + {.compatible = "ti,am4372-l3-noc", .data = &am4372_l3_data}, + {}, +}; +MODULE_DEVICE_TABLE(of, l3_noc_match); + +static int omap_l3_probe(struct platform_device *pdev) +{ + const struct of_device_id *of_id; + static struct omap_l3 *l3; + int ret, i, res_idx; + + of_id = of_match_device(l3_noc_match, &pdev->dev); + if (!of_id) { + dev_err(&pdev->dev, "OF data missing\n"); + return -EINVAL; + } + + l3 = devm_kzalloc(&pdev->dev, sizeof(*l3), GFP_KERNEL); + if (!l3) + return -ENOMEM; + + memcpy(l3, of_id->data, sizeof(*l3)); + l3->dev = &pdev->dev; + platform_set_drvdata(pdev, l3); + + /* Get mem resources */ + for (i = 0, res_idx = 0; i < l3->num_modules; i++) { + struct resource *res; + + if (l3->l3_base[i] == L3_BASE_IS_SUBMODULE) { + /* First entry cannot be submodule */ + BUG_ON(i == 0); + l3->l3_base[i] = l3->l3_base[i - 1]; + continue; + } + res = platform_get_resource(pdev, IORESOURCE_MEM, res_idx); + l3->l3_base[i] = devm_ioremap_resource(&pdev->dev, res); + if (IS_ERR(l3->l3_base[i])) { + dev_err(l3->dev, "ioremap %d failed\n", i); + return PTR_ERR(l3->l3_base[i]); + } + res_idx++; + } + + /* + * Setup interrupt Handlers + */ + l3->debug_irq = platform_get_irq(pdev, 0); + ret = devm_request_irq(l3->dev, l3->debug_irq, l3_interrupt_handler, + 0x0, "l3-dbg-irq", l3); + if (ret) { + dev_err(l3->dev, "request_irq failed for %d\n", + l3->debug_irq); + return ret; + } + + l3->app_irq = platform_get_irq(pdev, 1); + ret = devm_request_irq(l3->dev, l3->app_irq, l3_interrupt_handler, + 0x0, "l3-app-irq", l3); + if (ret) + dev_err(l3->dev, "request_irq failed for %d\n", l3->app_irq); + + return ret; +} + +#ifdef CONFIG_PM + +/** + * l3_resume_noirq() - resume function for l3_noc + * @dev: pointer to l3_noc device structure + * + * We only have the resume handler only since we + * have already maintained the delta register + * configuration as part of configuring the system + */ +static int l3_resume_noirq(struct device *dev) +{ + struct omap_l3 *l3 = dev_get_drvdata(dev); + int i; + struct l3_flagmux_data *flag_mux; + void __iomem *base, *mask_regx = NULL; + u32 mask_val; + + for (i = 0; i < l3->num_modules; i++) { + base = l3->l3_base[i]; + flag_mux = l3->l3_flagmux[i]; + if (!flag_mux->mask_app_bits && !flag_mux->mask_dbg_bits) + continue; + + mask_regx = base + flag_mux->offset + L3_FLAGMUX_MASK0 + + (L3_APPLICATION_ERROR << 3); + mask_val = readl_relaxed(mask_regx); + mask_val &= ~(flag_mux->mask_app_bits); + + writel_relaxed(mask_val, mask_regx); + mask_regx = base + flag_mux->offset + L3_FLAGMUX_MASK0 + + (L3_DEBUG_ERROR << 3); + mask_val = readl_relaxed(mask_regx); + mask_val &= ~(flag_mux->mask_dbg_bits); + + writel_relaxed(mask_val, mask_regx); + } + + /* Dummy read to force OCP barrier */ + if (mask_regx) + (void)readl(mask_regx); + + return 0; +} + +static const struct dev_pm_ops l3_dev_pm_ops = { + .resume_noirq = l3_resume_noirq, +}; + +#define L3_DEV_PM_OPS (&l3_dev_pm_ops) +#else +#define L3_DEV_PM_OPS NULL +#endif + +static struct platform_driver omap_l3_driver = { + .probe = omap_l3_probe, + .driver = { + .name = "omap_l3_noc", + .pm = L3_DEV_PM_OPS, + .of_match_table = of_match_ptr(l3_noc_match), + }, +}; + +static int __init omap_l3_init(void) +{ + return platform_driver_register(&omap_l3_driver); +} +postcore_initcall_sync(omap_l3_init); + +static void __exit omap_l3_exit(void) +{ + platform_driver_unregister(&omap_l3_driver); +} +module_exit(omap_l3_exit); diff --git a/drivers/bus/omap_l3_noc.h b/drivers/bus/omap_l3_noc.h new file mode 100644 index 000000000..73431f81d --- /dev/null +++ b/drivers/bus/omap_l3_noc.h @@ -0,0 +1,501 @@ +/* + * OMAP L3 Interconnect error handling driver header + * + * Copyright (C) 2011-2015 Texas Instruments Incorporated - http://www.ti.com/ + * Santosh Shilimkar + * sricharan + * + * 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. + * + * This program is distributed "as is" WITHOUT ANY WARRANTY of any + * kind, whether express or implied; without even the implied warranty + * of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + */ +#ifndef __OMAP_L3_NOC_H +#define __OMAP_L3_NOC_H + +#define MAX_L3_MODULES 3 +#define MAX_CLKDM_TARGETS 31 + +#define CLEAR_STDERR_LOG (1 << 31) +#define CUSTOM_ERROR 0x2 +#define STANDARD_ERROR 0x0 +#define INBAND_ERROR 0x0 +#define L3_APPLICATION_ERROR 0x0 +#define L3_DEBUG_ERROR 0x1 + +/* L3 TARG register offsets */ +#define L3_TARG_STDERRLOG_MAIN 0x48 +#define L3_TARG_STDERRLOG_HDR 0x4c +#define L3_TARG_STDERRLOG_MSTADDR 0x50 +#define L3_TARG_STDERRLOG_INFO 0x58 +#define L3_TARG_STDERRLOG_SLVOFSLSB 0x5c +#define L3_TARG_STDERRLOG_CINFO_INFO 0x64 +#define L3_TARG_STDERRLOG_CINFO_MSTADDR 0x68 +#define L3_TARG_STDERRLOG_CINFO_OPCODE 0x6c +#define L3_FLAGMUX_REGERR0 0xc +#define L3_FLAGMUX_MASK0 0x8 + +#define L3_TARGET_NOT_SUPPORTED NULL + +#define L3_BASE_IS_SUBMODULE ((void __iomem *)(1 << 0)) + +static const char * const l3_transaction_type[] = { + /* 0 0 0 */ "Idle", + /* 0 0 1 */ "Write", + /* 0 1 0 */ "Read", + /* 0 1 1 */ "ReadEx", + /* 1 0 0 */ "Read Link", + /* 1 0 1 */ "Write Non-Posted", + /* 1 1 0 */ "Write Conditional", + /* 1 1 1 */ "Write Broadcast", +}; + +/** + * struct l3_masters_data - L3 Master information + * @id: ID of the L3 Master + * @name: master name + */ +struct l3_masters_data { + u32 id; + char *name; +}; + +/** + * struct l3_target_data - L3 Target information + * @offset: Offset from base for L3 Target + * @name: Target name + * + * Target information is organized indexed by bit field definitions. + */ +struct l3_target_data { + u32 offset; + char *name; +}; + +/** + * struct l3_flagmux_data - Flag Mux information + * @offset: offset from base for flagmux register + * @l3_targ: array indexed by flagmux index (bit offset) pointing to the + * target data. unsupported ones are marked with + * L3_TARGET_NOT_SUPPORTED + * @num_targ_data: number of entries in target data + * @mask_app_bits: ignore these from raw application irq status + * @mask_dbg_bits: ignore these from raw debug irq status + */ +struct l3_flagmux_data { + u32 offset; + struct l3_target_data *l3_targ; + u8 num_targ_data; + u32 mask_app_bits; + u32 mask_dbg_bits; +}; + + +/** + * struct omap_l3 - Description of data relevant for L3 bus. + * @dev: device representing the bus (populated runtime) + * @l3_base: base addresses of modules (populated runtime if 0) + * if set to L3_BASE_IS_SUBMODULE, then uses previous + * module index as the base address + * @l3_flag_mux: array containing flag mux data per module + * offset from corresponding module base indexed per + * module. + * @num_modules: number of clock domains / modules. + * @l3_masters: array pointing to master data containing name and register + * offset for the master. + * @num_master: number of masters + * @mst_addr_mask: Mask representing MSTADDR information of NTTP packet + * @debug_irq: irq number of the debug interrupt (populated runtime) + * @app_irq: irq number of the application interrupt (populated runtime) + */ +struct omap_l3 { + struct device *dev; + + void __iomem *l3_base[MAX_L3_MODULES]; + struct l3_flagmux_data **l3_flagmux; + int num_modules; + + struct l3_masters_data *l3_masters; + int num_masters; + u32 mst_addr_mask; + + int debug_irq; + int app_irq; +}; + +static struct l3_target_data omap_l3_target_data_clk1[] = { + {0x100, "DMM1",}, + {0x200, "DMM2",}, + {0x300, "ABE",}, + {0x400, "L4CFG",}, + {0x600, "CLK2PWRDISC",}, + {0x0, "HOSTCLK1",}, + {0x900, "L4WAKEUP",}, +}; + +static struct l3_flagmux_data omap_l3_flagmux_clk1 = { + .offset = 0x500, + .l3_targ = omap_l3_target_data_clk1, + .num_targ_data = ARRAY_SIZE(omap_l3_target_data_clk1), +}; + + +static struct l3_target_data omap_l3_target_data_clk2[] = { + {0x500, "CORTEXM3",}, + {0x300, "DSS",}, + {0x100, "GPMC",}, + {0x400, "ISS",}, + {0x700, "IVAHD",}, + {0xD00, "AES1",}, + {0x900, "L4PER0",}, + {0x200, "OCMRAM",}, + {0x100, "GPMCsERROR",}, + {0x600, "SGX",}, + {0x800, "SL2",}, + {0x1600, "C2C",}, + {0x1100, "PWRDISCCLK1",}, + {0xF00, "SHA1",}, + {0xE00, "AES2",}, + {0xC00, "L4PER3",}, + {0xA00, "L4PER1",}, + {0xB00, "L4PER2",}, + {0x0, "HOSTCLK2",}, + {0x1800, "CAL",}, + {0x1700, "LLI",}, +}; + +static struct l3_flagmux_data omap_l3_flagmux_clk2 = { + .offset = 0x1000, + .l3_targ = omap_l3_target_data_clk2, + .num_targ_data = ARRAY_SIZE(omap_l3_target_data_clk2), +}; + + +static struct l3_target_data omap4_l3_target_data_clk3[] = { + {0x0100, "DEBUGSS",}, +}; + +static struct l3_flagmux_data omap4_l3_flagmux_clk3 = { + .offset = 0x0200, + .l3_targ = omap4_l3_target_data_clk3, + .num_targ_data = ARRAY_SIZE(omap4_l3_target_data_clk3), +}; + +static struct l3_masters_data omap_l3_masters[] = { + { 0x00, "MPU"}, + { 0x04, "CS_ADP"}, + { 0x05, "xxx"}, + { 0x08, "DSP"}, + { 0x0C, "IVAHD"}, + { 0x10, "ISS"}, + { 0x11, "DucatiM3"}, + { 0x12, "FaceDetect"}, + { 0x14, "SDMA_Rd"}, + { 0x15, "SDMA_Wr"}, + { 0x16, "xxx"}, + { 0x17, "xxx"}, + { 0x18, "SGX"}, + { 0x1C, "DSS"}, + { 0x20, "C2C"}, + { 0x22, "xxx"}, + { 0x23, "xxx"}, + { 0x24, "HSI"}, + { 0x28, "MMC1"}, + { 0x29, "MMC2"}, + { 0x2A, "MMC6"}, + { 0x2C, "UNIPRO1"}, + { 0x30, "USBHOSTHS"}, + { 0x31, "USBOTGHS"}, + { 0x32, "USBHOSTFS"} +}; + +static struct l3_flagmux_data *omap4_l3_flagmux[] = { + &omap_l3_flagmux_clk1, + &omap_l3_flagmux_clk2, + &omap4_l3_flagmux_clk3, +}; + +static const struct omap_l3 omap4_l3_data = { + .l3_flagmux = omap4_l3_flagmux, + .num_modules = ARRAY_SIZE(omap4_l3_flagmux), + .l3_masters = omap_l3_masters, + .num_masters = ARRAY_SIZE(omap_l3_masters), + /* The 6 MSBs of register field used to distinguish initiator */ + .mst_addr_mask = 0xFC, +}; + +/* OMAP5 data */ +static struct l3_target_data omap5_l3_target_data_clk3[] = { + {0x0100, "L3INSTR",}, + {0x0300, "DEBUGSS",}, + {0x0, "HOSTCLK3",}, +}; + +static struct l3_flagmux_data omap5_l3_flagmux_clk3 = { + .offset = 0x0200, + .l3_targ = omap5_l3_target_data_clk3, + .num_targ_data = ARRAY_SIZE(omap5_l3_target_data_clk3), +}; + +static struct l3_flagmux_data *omap5_l3_flagmux[] = { + &omap_l3_flagmux_clk1, + &omap_l3_flagmux_clk2, + &omap5_l3_flagmux_clk3, +}; + +static const struct omap_l3 omap5_l3_data = { + .l3_flagmux = omap5_l3_flagmux, + .num_modules = ARRAY_SIZE(omap5_l3_flagmux), + .l3_masters = omap_l3_masters, + .num_masters = ARRAY_SIZE(omap_l3_masters), + /* The 6 MSBs of register field used to distinguish initiator */ + .mst_addr_mask = 0x7E0, +}; + +/* DRA7 data */ +static struct l3_target_data dra_l3_target_data_clk1[] = { + {0x2a00, "AES1",}, + {0x0200, "DMM_P1",}, + {0x0600, "DSP2_SDMA",}, + {0x0b00, "EVE2",}, + {0x1300, "DMM_P2",}, + {0x2c00, "AES2",}, + {0x0300, "DSP1_SDMA",}, + {0x0a00, "EVE1",}, + {0x0c00, "EVE3",}, + {0x0d00, "EVE4",}, + {0x2900, "DSS",}, + {0x0100, "GPMC",}, + {0x3700, "PCIE1",}, + {0x1600, "IVA_CONFIG",}, + {0x1800, "IVA_SL2IF",}, + {0x0500, "L4_CFG",}, + {0x1d00, "L4_WKUP",}, + {0x3800, "PCIE2",}, + {0x3300, "SHA2_1",}, + {0x1200, "GPU",}, + {0x1000, "IPU1",}, + {0x1100, "IPU2",}, + {0x2000, "TPCC_EDMA",}, + {0x2e00, "TPTC1_EDMA",}, + {0x2b00, "TPTC2_EDMA",}, + {0x0700, "VCP1",}, + {0x2500, "L4_PER2_P3",}, + {0x0e00, "L4_PER3_P3",}, + {0x2200, "MMU1",}, + {0x1400, "PRUSS1",}, + {0x1500, "PRUSS2"}, + {0x0800, "VCP1",}, +}; + +static struct l3_flagmux_data dra_l3_flagmux_clk1 = { + .offset = 0x803500, + .l3_targ = dra_l3_target_data_clk1, + .num_targ_data = ARRAY_SIZE(dra_l3_target_data_clk1), +}; + +static struct l3_target_data dra_l3_target_data_clk2[] = { + {0x0, "HOST CLK1",}, + {0x800000, "HOST CLK2",}, + {0xdead, L3_TARGET_NOT_SUPPORTED,}, + {0x3400, "SHA2_2",}, + {0x0900, "BB2D",}, + {0xdead, L3_TARGET_NOT_SUPPORTED,}, + {0x2100, "L4_PER1_P3",}, + {0x1c00, "L4_PER1_P1",}, + {0x1f00, "L4_PER1_P2",}, + {0x2300, "L4_PER2_P1",}, + {0x2400, "L4_PER2_P2",}, + {0x2600, "L4_PER3_P1",}, + {0x2700, "L4_PER3_P2",}, + {0x2f00, "MCASP1",}, + {0x3000, "MCASP2",}, + {0x3100, "MCASP3",}, + {0x2800, "MMU2",}, + {0x0f00, "OCMC_RAM1",}, + {0x1700, "OCMC_RAM2",}, + {0x1900, "OCMC_RAM3",}, + {0x1e00, "OCMC_ROM",}, + {0x3900, "QSPI",}, +}; + +static struct l3_flagmux_data dra_l3_flagmux_clk2 = { + .offset = 0x803600, + .l3_targ = dra_l3_target_data_clk2, + .num_targ_data = ARRAY_SIZE(dra_l3_target_data_clk2), +}; + +static struct l3_target_data dra_l3_target_data_clk3[] = { + {0x0100, "L3_INSTR"}, + {0x0300, "DEBUGSS_CT_TBR"}, + {0x0, "HOST CLK3"}, +}; + +static struct l3_flagmux_data dra_l3_flagmux_clk3 = { + .offset = 0x200, + .l3_targ = dra_l3_target_data_clk3, + .num_targ_data = ARRAY_SIZE(dra_l3_target_data_clk3), +}; + +static struct l3_masters_data dra_l3_masters[] = { + { 0x0, "MPU" }, + { 0x4, "CS_DAP" }, + { 0x5, "IEEE1500_2_OCP" }, + { 0x8, "DSP1_MDMA" }, + { 0x9, "DSP1_CFG" }, + { 0xA, "DSP1_DMA" }, + { 0xB, "DSP2_MDMA" }, + { 0xC, "DSP2_CFG" }, + { 0xD, "DSP2_DMA" }, + { 0xE, "IVA" }, + { 0x10, "EVE1_P1" }, + { 0x11, "EVE2_P1" }, + { 0x12, "EVE3_P1" }, + { 0x13, "EVE4_P1" }, + { 0x14, "PRUSS1 PRU1" }, + { 0x15, "PRUSS1 PRU2" }, + { 0x16, "PRUSS2 PRU1" }, + { 0x17, "PRUSS2 PRU2" }, + { 0x18, "IPU1" }, + { 0x19, "IPU2" }, + { 0x1A, "SDMA" }, + { 0x1B, "CDMA" }, + { 0x1C, "TC1_EDMA" }, + { 0x1D, "TC2_EDMA" }, + { 0x20, "DSS" }, + { 0x21, "MMU1" }, + { 0x22, "PCIE1" }, + { 0x23, "MMU2" }, + { 0x24, "VIP1" }, + { 0x25, "VIP2" }, + { 0x26, "VIP3" }, + { 0x27, "VPE" }, + { 0x28, "GPU_P1" }, + { 0x29, "BB2D" }, + { 0x29, "GPU_P2" }, + { 0x2B, "GMAC_SW" }, + { 0x2C, "USB3" }, + { 0x2D, "USB2_SS" }, + { 0x2E, "USB2_ULPI_SS1" }, + { 0x2F, "USB2_ULPI_SS2" }, + { 0x30, "CSI2_1" }, + { 0x31, "CSI2_2" }, + { 0x33, "SATA" }, + { 0x34, "EVE1_P2" }, + { 0x35, "EVE2_P2" }, + { 0x36, "EVE3_P2" }, + { 0x37, "EVE4_P2" } +}; + +static struct l3_flagmux_data *dra_l3_flagmux[] = { + &dra_l3_flagmux_clk1, + &dra_l3_flagmux_clk2, + &dra_l3_flagmux_clk3, +}; + +static const struct omap_l3 dra_l3_data = { + .l3_base = { [1] = L3_BASE_IS_SUBMODULE }, + .l3_flagmux = dra_l3_flagmux, + .num_modules = ARRAY_SIZE(dra_l3_flagmux), + .l3_masters = dra_l3_masters, + .num_masters = ARRAY_SIZE(dra_l3_masters), + /* The 6 MSBs of register field used to distinguish initiator */ + .mst_addr_mask = 0xFC, +}; + +/* AM4372 data */ +static struct l3_target_data am4372_l3_target_data_200f[] = { + {0xf00, "EMIF",}, + {0x1200, "DES",}, + {0x400, "OCMCRAM",}, + {0x700, "TPTC0",}, + {0x800, "TPTC1",}, + {0x900, "TPTC2"}, + {0xb00, "TPCC",}, + {0xd00, "DEBUGSS",}, + {0xdead, L3_TARGET_NOT_SUPPORTED,}, + {0x200, "SHA",}, + {0xc00, "SGX530",}, + {0x500, "AES0",}, + {0xa00, "L4_FAST",}, + {0x300, "MPUSS_L2_RAM",}, + {0x100, "ICSS",}, +}; + +static struct l3_flagmux_data am4372_l3_flagmux_200f = { + .offset = 0x1000, + .l3_targ = am4372_l3_target_data_200f, + .num_targ_data = ARRAY_SIZE(am4372_l3_target_data_200f), +}; + +static struct l3_target_data am4372_l3_target_data_100s[] = { + {0x100, "L4_PER_0",}, + {0x200, "L4_PER_1",}, + {0x300, "L4_PER_2",}, + {0x400, "L4_PER_3",}, + {0x800, "McASP0",}, + {0x900, "McASP1",}, + {0xC00, "MMCHS2",}, + {0x700, "GPMC",}, + {0xD00, "L4_FW",}, + {0xdead, L3_TARGET_NOT_SUPPORTED,}, + {0x500, "ADCTSC",}, + {0xE00, "L4_WKUP",}, + {0xA00, "MAG_CARD",}, +}; + +static struct l3_flagmux_data am4372_l3_flagmux_100s = { + .offset = 0x600, + .l3_targ = am4372_l3_target_data_100s, + .num_targ_data = ARRAY_SIZE(am4372_l3_target_data_100s), +}; + +static struct l3_masters_data am4372_l3_masters[] = { + { 0x0, "M1 (128-bit)"}, + { 0x1, "M2 (64-bit)"}, + { 0x4, "DAP"}, + { 0x5, "P1500"}, + { 0xC, "ICSS0"}, + { 0xD, "ICSS1"}, + { 0x14, "Wakeup Processor"}, + { 0x18, "TPTC0 Read"}, + { 0x19, "TPTC0 Write"}, + { 0x1A, "TPTC1 Read"}, + { 0x1B, "TPTC1 Write"}, + { 0x1C, "TPTC2 Read"}, + { 0x1D, "TPTC2 Write"}, + { 0x20, "SGX530"}, + { 0x21, "OCP WP Traffic Probe"}, + { 0x22, "OCP WP DMA Profiling"}, + { 0x23, "OCP WP Event Trace"}, + { 0x25, "DSS"}, + { 0x28, "Crypto DMA RD"}, + { 0x29, "Crypto DMA WR"}, + { 0x2C, "VPFE0"}, + { 0x2D, "VPFE1"}, + { 0x30, "GEMAC"}, + { 0x34, "USB0 RD"}, + { 0x35, "USB0 WR"}, + { 0x36, "USB1 RD"}, + { 0x37, "USB1 WR"}, +}; + +static struct l3_flagmux_data *am4372_l3_flagmux[] = { + &am4372_l3_flagmux_200f, + &am4372_l3_flagmux_100s, +}; + +static const struct omap_l3 am4372_l3_data = { + .l3_flagmux = am4372_l3_flagmux, + .num_modules = ARRAY_SIZE(am4372_l3_flagmux), + .l3_masters = am4372_l3_masters, + .num_masters = ARRAY_SIZE(am4372_l3_masters), + /* All 6 bits of register field used to distinguish initiator */ + .mst_addr_mask = 0x3F, +}; + +#endif /* __OMAP_L3_NOC_H */ diff --git a/drivers/bus/omap_l3_smx.c b/drivers/bus/omap_l3_smx.c new file mode 100644 index 000000000..360a5c0a4 --- /dev/null +++ b/drivers/bus/omap_l3_smx.c @@ -0,0 +1,311 @@ +/* + * OMAP3XXX L3 Interconnect Driver + * + * Copyright (C) 2011 Texas Corporation + * Felipe Balbi + * Santosh Shilimkar + * Sricharan + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 + * USA + */ + +#include +#include +#include +#include +#include +#include +#include +#include + +#include "omap_l3_smx.h" + +static inline u64 omap3_l3_readll(void __iomem *base, u16 reg) +{ + return __raw_readll(base + reg); +} + +static inline void omap3_l3_writell(void __iomem *base, u16 reg, u64 value) +{ + __raw_writell(value, base + reg); +} + +static inline enum omap3_l3_code omap3_l3_decode_error_code(u64 error) +{ + return (error & 0x0f000000) >> L3_ERROR_LOG_CODE; +} + +static inline u32 omap3_l3_decode_addr(u64 error_addr) +{ + return error_addr & 0xffffffff; +} + +static inline unsigned omap3_l3_decode_cmd(u64 error) +{ + return (error & 0x07) >> L3_ERROR_LOG_CMD; +} + +static inline enum omap3_l3_initiator_id omap3_l3_decode_initid(u64 error) +{ + return (error & 0xff00) >> L3_ERROR_LOG_INITID; +} + +static inline unsigned omap3_l3_decode_req_info(u64 error) +{ + return (error >> 32) & 0xffff; +} + +static char *omap3_l3_code_string(u8 code) +{ + switch (code) { + case OMAP_L3_CODE_NOERROR: + return "No Error"; + case OMAP_L3_CODE_UNSUP_CMD: + return "Unsupported Command"; + case OMAP_L3_CODE_ADDR_HOLE: + return "Address Hole"; + case OMAP_L3_CODE_PROTECT_VIOLATION: + return "Protection Violation"; + case OMAP_L3_CODE_IN_BAND_ERR: + return "In-band Error"; + case OMAP_L3_CODE_REQ_TOUT_NOT_ACCEPT: + return "Request Timeout Not Accepted"; + case OMAP_L3_CODE_REQ_TOUT_NO_RESP: + return "Request Timeout, no response"; + default: + return "UNKNOWN error"; + } +} + +static char *omap3_l3_initiator_string(u8 initid) +{ + switch (initid) { + case OMAP_L3_LCD: + return "LCD"; + case OMAP_L3_SAD2D: + return "SAD2D"; + case OMAP_L3_IA_MPU_SS_1: + case OMAP_L3_IA_MPU_SS_2: + case OMAP_L3_IA_MPU_SS_3: + case OMAP_L3_IA_MPU_SS_4: + case OMAP_L3_IA_MPU_SS_5: + return "MPU"; + case OMAP_L3_IA_IVA_SS_1: + case OMAP_L3_IA_IVA_SS_2: + case OMAP_L3_IA_IVA_SS_3: + return "IVA_SS"; + case OMAP_L3_IA_IVA_SS_DMA_1: + case OMAP_L3_IA_IVA_SS_DMA_2: + case OMAP_L3_IA_IVA_SS_DMA_3: + case OMAP_L3_IA_IVA_SS_DMA_4: + case OMAP_L3_IA_IVA_SS_DMA_5: + case OMAP_L3_IA_IVA_SS_DMA_6: + return "IVA_SS_DMA"; + case OMAP_L3_IA_SGX: + return "SGX"; + case OMAP_L3_IA_CAM_1: + case OMAP_L3_IA_CAM_2: + case OMAP_L3_IA_CAM_3: + return "CAM"; + case OMAP_L3_IA_DAP: + return "DAP"; + case OMAP_L3_SDMA_WR_1: + case OMAP_L3_SDMA_WR_2: + return "SDMA_WR"; + case OMAP_L3_SDMA_RD_1: + case OMAP_L3_SDMA_RD_2: + case OMAP_L3_SDMA_RD_3: + case OMAP_L3_SDMA_RD_4: + return "SDMA_RD"; + case OMAP_L3_USBOTG: + return "USB_OTG"; + case OMAP_L3_USBHOST: + return "USB_HOST"; + default: + return "UNKNOWN Initiator"; + } +} + +/* + * omap3_l3_block_irq - handles a register block's irq + * @l3: struct omap3_l3 * + * @base: register block base address + * @error: L3_ERROR_LOG register of our block + * + * Called in hard-irq context. Caller should take care of locking + * + * OMAP36xx TRM gives, on page 2001, Figure 9-10, the Typical Error + * Analysis Sequence, we are following that sequence here, please + * refer to that Figure for more information on the subject. + */ +static irqreturn_t omap3_l3_block_irq(struct omap3_l3 *l3, + u64 error, int error_addr) +{ + u8 code = omap3_l3_decode_error_code(error); + u8 initid = omap3_l3_decode_initid(error); + u8 multi = error & L3_ERROR_LOG_MULTI; + u32 address = omap3_l3_decode_addr(error_addr); + + pr_err("%s seen by %s %s at address %x\n", + omap3_l3_code_string(code), + omap3_l3_initiator_string(initid), + multi ? "Multiple Errors" : "", address); + WARN_ON(1); + + return IRQ_HANDLED; +} + +static irqreturn_t omap3_l3_app_irq(int irq, void *_l3) +{ + struct omap3_l3 *l3 = _l3; + u64 status, clear; + u64 error; + u64 error_addr; + u64 err_source = 0; + void __iomem *base; + int int_type; + irqreturn_t ret = IRQ_NONE; + + int_type = irq == l3->app_irq ? L3_APPLICATION_ERROR : L3_DEBUG_ERROR; + if (!int_type) { + status = omap3_l3_readll(l3->rt, L3_SI_FLAG_STATUS_0); + /* + * if we have a timeout error, there's nothing we can + * do besides rebooting the board. So let's BUG on any + * of such errors and handle the others. timeout error + * is severe and not expected to occur. + */ + BUG_ON(status & L3_STATUS_0_TIMEOUT_MASK); + } else { + status = omap3_l3_readll(l3->rt, L3_SI_FLAG_STATUS_1); + /* No timeout error for debug sources */ + } + + /* identify the error source */ + err_source = __ffs(status); + + base = l3->rt + omap3_l3_bases[int_type][err_source]; + error = omap3_l3_readll(base, L3_ERROR_LOG); + if (error) { + error_addr = omap3_l3_readll(base, L3_ERROR_LOG_ADDR); + ret |= omap3_l3_block_irq(l3, error, error_addr); + } + + /* Clear the status register */ + clear = (L3_AGENT_STATUS_CLEAR_IA << int_type) | + L3_AGENT_STATUS_CLEAR_TA; + omap3_l3_writell(base, L3_AGENT_STATUS, clear); + + /* clear the error log register */ + omap3_l3_writell(base, L3_ERROR_LOG, error); + + return ret; +} + +#if IS_BUILTIN(CONFIG_OF) +static const struct of_device_id omap3_l3_match[] = { + { + .compatible = "ti,omap3-l3-smx", + }, + { }, +}; +MODULE_DEVICE_TABLE(of, omap3_l3_match); +#endif + +static int omap3_l3_probe(struct platform_device *pdev) +{ + struct omap3_l3 *l3; + struct resource *res; + int ret; + + l3 = kzalloc(sizeof(*l3), GFP_KERNEL); + if (!l3) + return -ENOMEM; + + platform_set_drvdata(pdev, l3); + + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!res) { + dev_err(&pdev->dev, "couldn't find resource\n"); + ret = -ENODEV; + goto err0; + } + l3->rt = ioremap(res->start, resource_size(res)); + if (!l3->rt) { + dev_err(&pdev->dev, "ioremap failed\n"); + ret = -ENOMEM; + goto err0; + } + + l3->debug_irq = platform_get_irq(pdev, 0); + ret = request_irq(l3->debug_irq, omap3_l3_app_irq, IRQF_TRIGGER_RISING, + "l3-debug-irq", l3); + if (ret) { + dev_err(&pdev->dev, "couldn't request debug irq\n"); + goto err1; + } + + l3->app_irq = platform_get_irq(pdev, 1); + ret = request_irq(l3->app_irq, omap3_l3_app_irq, IRQF_TRIGGER_RISING, + "l3-app-irq", l3); + if (ret) { + dev_err(&pdev->dev, "couldn't request app irq\n"); + goto err2; + } + + return 0; + +err2: + free_irq(l3->debug_irq, l3); +err1: + iounmap(l3->rt); +err0: + kfree(l3); + return ret; +} + +static int omap3_l3_remove(struct platform_device *pdev) +{ + struct omap3_l3 *l3 = platform_get_drvdata(pdev); + + free_irq(l3->app_irq, l3); + free_irq(l3->debug_irq, l3); + iounmap(l3->rt); + kfree(l3); + + return 0; +} + +static struct platform_driver omap3_l3_driver = { + .probe = omap3_l3_probe, + .remove = omap3_l3_remove, + .driver = { + .name = "omap_l3_smx", + .of_match_table = of_match_ptr(omap3_l3_match), + }, +}; + +static int __init omap3_l3_init(void) +{ + return platform_driver_register(&omap3_l3_driver); +} +postcore_initcall_sync(omap3_l3_init); + +static void __exit omap3_l3_exit(void) +{ + platform_driver_unregister(&omap3_l3_driver); +} +module_exit(omap3_l3_exit); diff --git a/drivers/bus/omap_l3_smx.h b/drivers/bus/omap_l3_smx.h new file mode 100644 index 000000000..4f3cebca4 --- /dev/null +++ b/drivers/bus/omap_l3_smx.h @@ -0,0 +1,338 @@ +/* + * OMAP3XXX L3 Interconnect Driver header + * + * Copyright (C) 2011 Texas Corporation + * Felipe Balbi + * Santosh Shilimkar + * sricharan + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 + * USA + */ +#ifndef __ARCH_ARM_MACH_OMAP2_L3_INTERCONNECT_3XXX_H +#define __ARCH_ARM_MACH_OMAP2_L3_INTERCONNECT_3XXX_H + +/* Register definitions. All 64-bit wide */ +#define L3_COMPONENT 0x000 +#define L3_CORE 0x018 +#define L3_AGENT_CONTROL 0x020 +#define L3_AGENT_STATUS 0x028 +#define L3_ERROR_LOG 0x058 + +#define L3_ERROR_LOG_MULTI (1 << 31) +#define L3_ERROR_LOG_SECONDARY (1 << 30) + +#define L3_ERROR_LOG_ADDR 0x060 + +/* Register definitions for Sideband Interconnect */ +#define L3_SI_CONTROL 0x020 +#define L3_SI_FLAG_STATUS_0 0x510 + +static const u64 shift = 1; + +#define L3_STATUS_0_MPUIA_BRST (shift << 0) +#define L3_STATUS_0_MPUIA_RSP (shift << 1) +#define L3_STATUS_0_MPUIA_INBAND (shift << 2) +#define L3_STATUS_0_IVAIA_BRST (shift << 6) +#define L3_STATUS_0_IVAIA_RSP (shift << 7) +#define L3_STATUS_0_IVAIA_INBAND (shift << 8) +#define L3_STATUS_0_SGXIA_BRST (shift << 9) +#define L3_STATUS_0_SGXIA_RSP (shift << 10) +#define L3_STATUS_0_SGXIA_MERROR (shift << 11) +#define L3_STATUS_0_CAMIA_BRST (shift << 12) +#define L3_STATUS_0_CAMIA_RSP (shift << 13) +#define L3_STATUS_0_CAMIA_INBAND (shift << 14) +#define L3_STATUS_0_DISPIA_BRST (shift << 15) +#define L3_STATUS_0_DISPIA_RSP (shift << 16) +#define L3_STATUS_0_DMARDIA_BRST (shift << 18) +#define L3_STATUS_0_DMARDIA_RSP (shift << 19) +#define L3_STATUS_0_DMAWRIA_BRST (shift << 21) +#define L3_STATUS_0_DMAWRIA_RSP (shift << 22) +#define L3_STATUS_0_USBOTGIA_BRST (shift << 24) +#define L3_STATUS_0_USBOTGIA_RSP (shift << 25) +#define L3_STATUS_0_USBOTGIA_INBAND (shift << 26) +#define L3_STATUS_0_USBHOSTIA_BRST (shift << 27) +#define L3_STATUS_0_USBHOSTIA_INBAND (shift << 28) +#define L3_STATUS_0_SMSTA_REQ (shift << 48) +#define L3_STATUS_0_GPMCTA_REQ (shift << 49) +#define L3_STATUS_0_OCMRAMTA_REQ (shift << 50) +#define L3_STATUS_0_OCMROMTA_REQ (shift << 51) +#define L3_STATUS_0_IVATA_REQ (shift << 54) +#define L3_STATUS_0_SGXTA_REQ (shift << 55) +#define L3_STATUS_0_SGXTA_SERROR (shift << 56) +#define L3_STATUS_0_GPMCTA_SERROR (shift << 57) +#define L3_STATUS_0_L4CORETA_REQ (shift << 58) +#define L3_STATUS_0_L4PERTA_REQ (shift << 59) +#define L3_STATUS_0_L4EMUTA_REQ (shift << 60) +#define L3_STATUS_0_MAD2DTA_REQ (shift << 61) + +#define L3_STATUS_0_TIMEOUT_MASK (L3_STATUS_0_MPUIA_BRST \ + | L3_STATUS_0_MPUIA_RSP \ + | L3_STATUS_0_IVAIA_BRST \ + | L3_STATUS_0_IVAIA_RSP \ + | L3_STATUS_0_SGXIA_BRST \ + | L3_STATUS_0_SGXIA_RSP \ + | L3_STATUS_0_CAMIA_BRST \ + | L3_STATUS_0_CAMIA_RSP \ + | L3_STATUS_0_DISPIA_BRST \ + | L3_STATUS_0_DISPIA_RSP \ + | L3_STATUS_0_DMARDIA_BRST \ + | L3_STATUS_0_DMARDIA_RSP \ + | L3_STATUS_0_DMAWRIA_BRST \ + | L3_STATUS_0_DMAWRIA_RSP \ + | L3_STATUS_0_USBOTGIA_BRST \ + | L3_STATUS_0_USBOTGIA_RSP \ + | L3_STATUS_0_USBHOSTIA_BRST \ + | L3_STATUS_0_SMSTA_REQ \ + | L3_STATUS_0_GPMCTA_REQ \ + | L3_STATUS_0_OCMRAMTA_REQ \ + | L3_STATUS_0_OCMROMTA_REQ \ + | L3_STATUS_0_IVATA_REQ \ + | L3_STATUS_0_SGXTA_REQ \ + | L3_STATUS_0_L4CORETA_REQ \ + | L3_STATUS_0_L4PERTA_REQ \ + | L3_STATUS_0_L4EMUTA_REQ \ + | L3_STATUS_0_MAD2DTA_REQ) + +#define L3_SI_FLAG_STATUS_1 0x530 + +#define L3_STATUS_1_MPU_DATAIA (1 << 0) +#define L3_STATUS_1_DAPIA0 (1 << 3) +#define L3_STATUS_1_DAPIA1 (1 << 4) +#define L3_STATUS_1_IVAIA (1 << 6) + +#define L3_PM_ERROR_LOG 0x020 +#define L3_PM_CONTROL 0x028 +#define L3_PM_ERROR_CLEAR_SINGLE 0x030 +#define L3_PM_ERROR_CLEAR_MULTI 0x038 +#define L3_PM_REQ_INFO_PERMISSION(n) (0x048 + (0x020 * n)) +#define L3_PM_READ_PERMISSION(n) (0x050 + (0x020 * n)) +#define L3_PM_WRITE_PERMISSION(n) (0x058 + (0x020 * n)) +#define L3_PM_ADDR_MATCH(n) (0x060 + (0x020 * n)) + +/* L3 error log bit fields. Common for IA and TA */ +#define L3_ERROR_LOG_CODE 24 +#define L3_ERROR_LOG_INITID 8 +#define L3_ERROR_LOG_CMD 0 + +/* L3 agent status bit fields. */ +#define L3_AGENT_STATUS_CLEAR_IA 0x10000000 +#define L3_AGENT_STATUS_CLEAR_TA 0x01000000 + +#define OMAP34xx_IRQ_L3_APP 10 +#define L3_APPLICATION_ERROR 0x0 +#define L3_DEBUG_ERROR 0x1 + +enum omap3_l3_initiator_id { + /* LCD has 1 ID */ + OMAP_L3_LCD = 29, + /* SAD2D has 1 ID */ + OMAP_L3_SAD2D = 28, + /* MPU has 5 IDs */ + OMAP_L3_IA_MPU_SS_1 = 27, + OMAP_L3_IA_MPU_SS_2 = 26, + OMAP_L3_IA_MPU_SS_3 = 25, + OMAP_L3_IA_MPU_SS_4 = 24, + OMAP_L3_IA_MPU_SS_5 = 23, + /* IVA2.2 SS has 3 IDs*/ + OMAP_L3_IA_IVA_SS_1 = 22, + OMAP_L3_IA_IVA_SS_2 = 21, + OMAP_L3_IA_IVA_SS_3 = 20, + /* IVA 2.2 SS DMA has 6 IDS */ + OMAP_L3_IA_IVA_SS_DMA_1 = 19, + OMAP_L3_IA_IVA_SS_DMA_2 = 18, + OMAP_L3_IA_IVA_SS_DMA_3 = 17, + OMAP_L3_IA_IVA_SS_DMA_4 = 16, + OMAP_L3_IA_IVA_SS_DMA_5 = 15, + OMAP_L3_IA_IVA_SS_DMA_6 = 14, + /* SGX has 1 ID */ + OMAP_L3_IA_SGX = 13, + /* CAM has 3 ID */ + OMAP_L3_IA_CAM_1 = 12, + OMAP_L3_IA_CAM_2 = 11, + OMAP_L3_IA_CAM_3 = 10, + /* DAP has 1 ID */ + OMAP_L3_IA_DAP = 9, + /* SDMA WR has 2 IDs */ + OMAP_L3_SDMA_WR_1 = 8, + OMAP_L3_SDMA_WR_2 = 7, + /* SDMA RD has 4 IDs */ + OMAP_L3_SDMA_RD_1 = 6, + OMAP_L3_SDMA_RD_2 = 5, + OMAP_L3_SDMA_RD_3 = 4, + OMAP_L3_SDMA_RD_4 = 3, + /* HSUSB OTG has 1 ID */ + OMAP_L3_USBOTG = 2, + /* HSUSB HOST has 1 ID */ + OMAP_L3_USBHOST = 1, +}; + +enum omap3_l3_code { + OMAP_L3_CODE_NOERROR = 0, + OMAP_L3_CODE_UNSUP_CMD = 1, + OMAP_L3_CODE_ADDR_HOLE = 2, + OMAP_L3_CODE_PROTECT_VIOLATION = 3, + OMAP_L3_CODE_IN_BAND_ERR = 4, + /* codes 5 and 6 are reserved */ + OMAP_L3_CODE_REQ_TOUT_NOT_ACCEPT = 7, + OMAP_L3_CODE_REQ_TOUT_NO_RESP = 8, + /* codes 9 - 15 are also reserved */ +}; + +struct omap3_l3 { + struct device *dev; + struct clk *ick; + + /* memory base*/ + void __iomem *rt; + + int debug_irq; + int app_irq; + + /* true when and inband functional error occurs */ + unsigned inband:1; +}; + +/* offsets for l3 agents in order with the Flag status register */ +static unsigned int omap3_l3_app_bases[] = { + /* MPU IA */ + 0x1400, + 0x1400, + 0x1400, + /* RESERVED */ + 0, + 0, + 0, + /* IVA 2.2 IA */ + 0x1800, + 0x1800, + 0x1800, + /* SGX IA */ + 0x1c00, + 0x1c00, + /* RESERVED */ + 0, + /* CAMERA IA */ + 0x5800, + 0x5800, + 0x5800, + /* DISPLAY IA */ + 0x5400, + 0x5400, + /* RESERVED */ + 0, + /*SDMA RD IA */ + 0x4c00, + 0x4c00, + /* RESERVED */ + 0, + /* SDMA WR IA */ + 0x5000, + 0x5000, + /* RESERVED */ + 0, + /* USB OTG IA */ + 0x4400, + 0x4400, + 0x4400, + /* USB HOST IA */ + 0x4000, + 0x4000, + /* RESERVED */ + 0, + 0, + 0, + 0, + /* SAD2D IA */ + 0x3000, + 0x3000, + 0x3000, + /* RESERVED */ + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + 0, + /* SMA TA */ + 0x2000, + /* GPMC TA */ + 0x2400, + /* OCM RAM TA */ + 0x2800, + /* OCM ROM TA */ + 0x2C00, + /* L4 CORE TA */ + 0x6800, + /* L4 PER TA */ + 0x6c00, + /* IVA 2.2 TA */ + 0x6000, + /* SGX TA */ + 0x6400, + /* L4 EMU TA */ + 0x7000, + /* GPMC TA */ + 0x2400, + /* L4 CORE TA */ + 0x6800, + /* L4 PER TA */ + 0x6c00, + /* L4 EMU TA */ + 0x7000, + /* MAD2D TA */ + 0x3400, + /* RESERVED */ + 0, + 0, +}; + +static unsigned int omap3_l3_debug_bases[] = { + /* MPU DATA IA */ + 0x1400, + /* RESERVED */ + 0, + 0, + /* DAP IA */ + 0x5c00, + 0x5c00, + /* RESERVED */ + 0, + /* IVA 2.2 IA */ + 0x1800, + /* REST RESERVED */ +}; + +static u32 *omap3_l3_bases[] = { + omap3_l3_app_bases, + omap3_l3_debug_bases, +}; + +/* + * REVISIT define __raw_readll/__raw_writell here, but move them to + * at some point + */ +#define __raw_writell(v, a) (__chk_io_ptr(a), \ + *(volatile u64 __force *)(a) = (v)) +#define __raw_readll(a) (__chk_io_ptr(a), \ + *(volatile u64 __force *)(a)) + +#endif diff --git a/drivers/bus/simple-pm-bus.c b/drivers/bus/simple-pm-bus.c new file mode 100644 index 000000000..c5eb46cbf --- /dev/null +++ b/drivers/bus/simple-pm-bus.c @@ -0,0 +1,58 @@ +/* + * Simple Power-Managed Bus Driver + * + * Copyright (C) 2014-2015 Glider bvba + * + * This file is subject to the terms and conditions of the GNU General Public + * License. See the file "COPYING" in the main directory of this archive + * for more details. + */ + +#include +#include +#include +#include + + +static int simple_pm_bus_probe(struct platform_device *pdev) +{ + struct device_node *np = pdev->dev.of_node; + + dev_dbg(&pdev->dev, "%s\n", __func__); + + pm_runtime_enable(&pdev->dev); + + if (np) + of_platform_populate(np, NULL, NULL, &pdev->dev); + + return 0; +} + +static int simple_pm_bus_remove(struct platform_device *pdev) +{ + dev_dbg(&pdev->dev, "%s\n", __func__); + + pm_runtime_disable(&pdev->dev); + return 0; +} + +static const struct of_device_id simple_pm_bus_of_match[] = { + { .compatible = "simple-pm-bus", }, + { /* sentinel */ } +}; +MODULE_DEVICE_TABLE(of, simple_pm_bus_of_match); + +static struct platform_driver simple_pm_bus_driver = { + .probe = simple_pm_bus_probe, + .remove = simple_pm_bus_remove, + .driver = { + .name = "simple-pm-bus", + .of_match_table = simple_pm_bus_of_match, + }, +}; + +module_platform_driver(simple_pm_bus_driver); + +MODULE_DESCRIPTION("Simple Power-Managed Bus Driver"); +MODULE_AUTHOR("Geert Uytterhoeven "); +MODULE_LICENSE("GPL v2"); diff --git a/drivers/bus/vexpress-config.c b/drivers/bus/vexpress-config.c new file mode 100644 index 000000000..a64763b6b --- /dev/null +++ b/drivers/bus/vexpress-config.c @@ -0,0 +1,202 @@ +/* + * 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. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * Copyright (C) 2014 ARM Limited + */ + +#include +#include +#include +#include +#include + + +struct vexpress_config_bridge { + struct vexpress_config_bridge_ops *ops; + void *context; +}; + + +static DEFINE_MUTEX(vexpress_config_mutex); +static struct class *vexpress_config_class; +static u32 vexpress_config_site_master = VEXPRESS_SITE_MASTER; + + +void vexpress_config_set_master(u32 site) +{ + vexpress_config_site_master = site; +} + +u32 vexpress_config_get_master(void) +{ + return vexpress_config_site_master; +} + +void vexpress_config_lock(void *arg) +{ + mutex_lock(&vexpress_config_mutex); +} + +void vexpress_config_unlock(void *arg) +{ + mutex_unlock(&vexpress_config_mutex); +} + + +static void vexpress_config_find_prop(struct device_node *node, + const char *name, u32 *val) +{ + /* Default value */ + *val = 0; + + of_node_get(node); + while (node) { + if (of_property_read_u32(node, name, val) == 0) { + of_node_put(node); + return; + } + node = of_get_next_parent(node); + } +} + +int vexpress_config_get_topo(struct device_node *node, u32 *site, + u32 *position, u32 *dcc) +{ + vexpress_config_find_prop(node, "arm,vexpress,site", site); + if (*site == VEXPRESS_SITE_MASTER) + *site = vexpress_config_site_master; + if (WARN_ON(vexpress_config_site_master == VEXPRESS_SITE_MASTER)) + return -EINVAL; + vexpress_config_find_prop(node, "arm,vexpress,position", position); + vexpress_config_find_prop(node, "arm,vexpress,dcc", dcc); + + return 0; +} + + +static void vexpress_config_devres_release(struct device *dev, void *res) +{ + struct vexpress_config_bridge *bridge = dev_get_drvdata(dev->parent); + struct regmap *regmap = res; + + bridge->ops->regmap_exit(regmap, bridge->context); +} + +struct regmap *devm_regmap_init_vexpress_config(struct device *dev) +{ + struct vexpress_config_bridge *bridge; + struct regmap *regmap; + struct regmap **res; + + if (WARN_ON(dev->parent->class != vexpress_config_class)) + return ERR_PTR(-ENODEV); + + bridge = dev_get_drvdata(dev->parent); + if (WARN_ON(!bridge)) + return ERR_PTR(-EINVAL); + + res = devres_alloc(vexpress_config_devres_release, sizeof(*res), + GFP_KERNEL); + if (!res) + return ERR_PTR(-ENOMEM); + + regmap = bridge->ops->regmap_init(dev, bridge->context); + if (IS_ERR(regmap)) { + devres_free(res); + return regmap; + } + + *res = regmap; + devres_add(dev, res); + + return regmap; +} +EXPORT_SYMBOL_GPL(devm_regmap_init_vexpress_config); + +struct device *vexpress_config_bridge_register(struct device *parent, + struct vexpress_config_bridge_ops *ops, void *context) +{ + struct device *dev; + struct vexpress_config_bridge *bridge; + + if (!vexpress_config_class) { + vexpress_config_class = class_create(THIS_MODULE, + "vexpress-config"); + if (IS_ERR(vexpress_config_class)) + return (void *)vexpress_config_class; + } + + dev = device_create(vexpress_config_class, parent, 0, + NULL, "%s.bridge", dev_name(parent)); + + if (IS_ERR(dev)) + return dev; + + bridge = devm_kmalloc(dev, sizeof(*bridge), GFP_KERNEL); + if (!bridge) { + put_device(dev); + device_unregister(dev); + return ERR_PTR(-ENOMEM); + } + bridge->ops = ops; + bridge->context = context; + + dev_set_drvdata(dev, bridge); + + dev_dbg(parent, "Registered bridge '%s', parent node %p\n", + dev_name(dev), parent->of_node); + + return dev; +} + + +static int vexpress_config_node_match(struct device *dev, const void *data) +{ + const struct device_node *node = data; + + dev_dbg(dev, "Parent node %p, looking for %p\n", + dev->parent->of_node, node); + + return dev->parent->of_node == node; +} + +static int vexpress_config_populate(struct device_node *node) +{ + struct device_node *bridge; + struct device *parent; + + bridge = of_parse_phandle(node, "arm,vexpress,config-bridge", 0); + if (!bridge) + return -EINVAL; + + parent = class_find_device(vexpress_config_class, NULL, bridge, + vexpress_config_node_match); + if (WARN_ON(!parent)) + return -ENODEV; + + return of_platform_populate(node, NULL, NULL, parent); +} + +static int __init vexpress_config_init(void) +{ + int err = 0; + struct device_node *node; + + /* Need the config devices early, before the "normal" devices... */ + for_each_compatible_node(node, NULL, "arm,vexpress,config-bus") { + err = vexpress_config_populate(node); + if (err) + break; + } + + return err; +} +postcore_initcall(vexpress_config_init); + -- cgit v1.2.3-54-g00ecf