linux/drivers/perf/arm-cci.c

1731 lines
48 KiB
C
Raw Normal View History

// SPDX-License-Identifier: GPL-2.0
// CCI Cache Coherent Interconnect PMU driver
// Copyright (C) 2013-2018 Arm Ltd.
// Author: Punit Agrawal <punit.agrawal@arm.com>, Suzuki Poulose <suzuki.poulose@arm.com>
#include <linux/arm-cci.h>
#include <linux/io.h>
#include <linux/interrupt.h>
#include <linux/module.h>
#include <linux/of_address.h>
#include <linux/of_device.h>
#include <linux/of_irq.h>
#include <linux/of_platform.h>
#include <linux/perf_event.h>
#include <linux/platform_device.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#define DRIVER_NAME "ARM-CCI 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_SIZE(model) ((model)->cntr_size)
#define CCI_PMU_CNTR_BASE(model, idx) ((idx) * CCI_PMU_CNTR_SIZE(model))
#define CCI_PMU_CNTR_MASK ((1ULL << 32) -1)
#define CCI_PMU_CNTR_LAST(cci_pmu) (cci_pmu->num_cntrs - 1)
#define CCI_PMU_MAX_HW_CNTRS(model) \
((model)->num_hw_cntrs + (model)->fixed_hw_cntrs)
/* Types of interfaces that can generate events */
enum {
CCI_IF_SLAVE,
CCI_IF_MASTER,
#ifdef CONFIG_ARM_CCI5xx_PMU
CCI_IF_GLOBAL,
#endif
CCI_IF_MAX,
};
#define NUM_HW_CNTRS_CII_4XX 4
#define NUM_HW_CNTRS_CII_5XX 8
#define NUM_HW_CNTRS_MAX NUM_HW_CNTRS_CII_5XX
#define FIXED_HW_CNTRS_CII_4XX 1
#define FIXED_HW_CNTRS_CII_5XX 0
#define FIXED_HW_CNTRS_MAX FIXED_HW_CNTRS_CII_4XX
#define HW_CNTRS_MAX (NUM_HW_CNTRS_MAX + FIXED_HW_CNTRS_MAX)
struct event_range {
u32 min;
u32 max;
};
struct cci_pmu_hw_events {
struct perf_event **events;
unsigned long *used_mask;
raw_spinlock_t pmu_lock;
};
struct cci_pmu;
/*
* struct cci_pmu_model:
* @fixed_hw_cntrs - Number of fixed event counters
* @num_hw_cntrs - Maximum number of programmable event counters
* @cntr_size - Size of an event counter mapping
*/
struct cci_pmu_model {
char *name;
u32 fixed_hw_cntrs;
u32 num_hw_cntrs;
u32 cntr_size;
struct attribute **format_attrs;
struct attribute **event_attrs;
struct event_range event_ranges[CCI_IF_MAX];
int (*validate_hw_event)(struct cci_pmu *, unsigned long);
int (*get_event_idx)(struct cci_pmu *, struct cci_pmu_hw_events *, unsigned long);
void (*write_counters)(struct cci_pmu *, unsigned long *);
};
static struct cci_pmu_model cci_pmu_models[];
struct cci_pmu {
void __iomem *base;
void __iomem *ctrl_base;
struct pmu pmu;
int cpu;
int nr_irqs;
int *irqs;
unsigned long active_irqs;
const struct cci_pmu_model *model;
struct cci_pmu_hw_events hw_events;
struct platform_device *plat_device;
int num_cntrs;
atomic_t active_events;
struct mutex reserve_mutex;
};
#define to_cci_pmu(c) (container_of(c, struct cci_pmu, pmu))
static struct cci_pmu *g_cci_pmu;
enum cci_models {
#ifdef CONFIG_ARM_CCI400_PMU
CCI400_R0,
CCI400_R1,
#endif
#ifdef CONFIG_ARM_CCI5xx_PMU
CCI500_R0,
CCI550_R0,
#endif
CCI_MODEL_MAX
};
static void pmu_write_counters(struct cci_pmu *cci_pmu,
unsigned long *mask);
static ssize_t __maybe_unused cci_pmu_format_show(struct device *dev,
struct device_attribute *attr, char *buf);
static ssize_t __maybe_unused cci_pmu_event_show(struct device *dev,
struct device_attribute *attr, char *buf);
#define CCI_EXT_ATTR_ENTRY(_name, _func, _config) \
&((struct dev_ext_attribute[]) { \
{ __ATTR(_name, S_IRUGO, _func, NULL), (void *)_config } \
})[0].attr.attr
#define CCI_FORMAT_EXT_ATTR_ENTRY(_name, _config) \
CCI_EXT_ATTR_ENTRY(_name, cci_pmu_format_show, (char *)_config)
#define CCI_EVENT_EXT_ATTR_ENTRY(_name, _config) \
CCI_EXT_ATTR_ENTRY(_name, cci_pmu_event_show, (unsigned long)_config)
/* CCI400 PMU Specific definitions */
#ifdef CONFIG_ARM_CCI400_PMU
/* Port ids */
#define CCI400_PORT_S0 0
#define CCI400_PORT_S1 1
#define CCI400_PORT_S2 2
#define CCI400_PORT_S3 3
#define CCI400_PORT_S4 4
#define CCI400_PORT_M0 5
#define CCI400_PORT_M1 6
#define CCI400_PORT_M2 7
#define CCI400_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 {
CCI400_PMU_CYCLES = 0xff
};
#define CCI400_PMU_CYCLE_CNTR_IDX 0
#define CCI400_PMU_CNTR0_IDX 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 CCI400_PMU_EVENT_MASK 0xffUL
#define CCI400_PMU_EVENT_SOURCE_SHIFT 5
#define CCI400_PMU_EVENT_SOURCE_MASK 0x7
#define CCI400_PMU_EVENT_CODE_SHIFT 0
#define CCI400_PMU_EVENT_CODE_MASK 0x1f
#define CCI400_PMU_EVENT_SOURCE(event) \
((event >> CCI400_PMU_EVENT_SOURCE_SHIFT) & \
CCI400_PMU_EVENT_SOURCE_MASK)
#define CCI400_PMU_EVENT_CODE(event) \
((event >> CCI400_PMU_EVENT_CODE_SHIFT) & CCI400_PMU_EVENT_CODE_MASK)
#define CCI400_R0_SLAVE_PORT_MIN_EV 0x00
#define CCI400_R0_SLAVE_PORT_MAX_EV 0x13
#define CCI400_R0_MASTER_PORT_MIN_EV 0x14
#define CCI400_R0_MASTER_PORT_MAX_EV 0x1a
#define CCI400_R1_SLAVE_PORT_MIN_EV 0x00
#define CCI400_R1_SLAVE_PORT_MAX_EV 0x14
#define CCI400_R1_MASTER_PORT_MIN_EV 0x00
#define CCI400_R1_MASTER_PORT_MAX_EV 0x11
#define CCI400_CYCLE_EVENT_EXT_ATTR_ENTRY(_name, _config) \
CCI_EXT_ATTR_ENTRY(_name, cci400_pmu_cycle_event_show, \
(unsigned long)_config)
static ssize_t cci400_pmu_cycle_event_show(struct device *dev,
struct device_attribute *attr, char *buf);
static struct attribute *cci400_pmu_format_attrs[] = {
CCI_FORMAT_EXT_ATTR_ENTRY(event, "config:0-4"),
CCI_FORMAT_EXT_ATTR_ENTRY(source, "config:5-7"),
NULL
};
static struct attribute *cci400_r0_pmu_event_attrs[] = {
/* Slave events */
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_any, 0x0),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_device, 0x01),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_normal_or_nonshareable, 0x2),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_inner_or_outershareable, 0x3),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_cache_maintenance, 0x4),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_mem_barrier, 0x5),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_sync_barrier, 0x6),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_dvm_msg, 0x7),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_dvm_msg_sync, 0x8),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_stall_tt_full, 0x9),
CCI_EVENT_EXT_ATTR_ENTRY(si_r_data_last_hs_snoop, 0xA),
CCI_EVENT_EXT_ATTR_ENTRY(si_r_data_stall_rvalids_h_rready_l, 0xB),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_any, 0xC),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_device, 0xD),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_normal_or_nonshareable, 0xE),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_inner_or_outershare_wback_wclean, 0xF),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_write_unique, 0x10),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_write_line_unique, 0x11),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_evict, 0x12),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_stall_tt_full, 0x13),
/* Master events */
CCI_EVENT_EXT_ATTR_ENTRY(mi_retry_speculative_fetch, 0x14),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_addr_hazard, 0x15),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_id_hazard, 0x16),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_tt_full, 0x17),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_barrier_hazard, 0x18),
CCI_EVENT_EXT_ATTR_ENTRY(mi_wrq_stall_barrier_hazard, 0x19),
CCI_EVENT_EXT_ATTR_ENTRY(mi_wrq_stall_tt_full, 0x1A),
/* Special event for cycles counter */
CCI400_CYCLE_EVENT_EXT_ATTR_ENTRY(cycles, 0xff),
NULL
};
static struct attribute *cci400_r1_pmu_event_attrs[] = {
/* Slave events */
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_any, 0x0),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_device, 0x01),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_normal_or_nonshareable, 0x2),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_inner_or_outershareable, 0x3),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_cache_maintenance, 0x4),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_mem_barrier, 0x5),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_sync_barrier, 0x6),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_dvm_msg, 0x7),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_dvm_msg_sync, 0x8),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_stall_tt_full, 0x9),
CCI_EVENT_EXT_ATTR_ENTRY(si_r_data_last_hs_snoop, 0xA),
CCI_EVENT_EXT_ATTR_ENTRY(si_r_data_stall_rvalids_h_rready_l, 0xB),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_any, 0xC),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_device, 0xD),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_normal_or_nonshareable, 0xE),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_inner_or_outershare_wback_wclean, 0xF),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_write_unique, 0x10),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_write_line_unique, 0x11),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_evict, 0x12),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_stall_tt_full, 0x13),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_stall_slave_id_hazard, 0x14),
/* Master events */
CCI_EVENT_EXT_ATTR_ENTRY(mi_retry_speculative_fetch, 0x0),
CCI_EVENT_EXT_ATTR_ENTRY(mi_stall_cycle_addr_hazard, 0x1),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_master_id_hazard, 0x2),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_hi_prio_rtq_full, 0x3),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_barrier_hazard, 0x4),
CCI_EVENT_EXT_ATTR_ENTRY(mi_wrq_stall_barrier_hazard, 0x5),
CCI_EVENT_EXT_ATTR_ENTRY(mi_wrq_stall_wtq_full, 0x6),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_low_prio_rtq_full, 0x7),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_mid_prio_rtq_full, 0x8),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_qvn_vn0, 0x9),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_qvn_vn1, 0xA),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_qvn_vn2, 0xB),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall_qvn_vn3, 0xC),
CCI_EVENT_EXT_ATTR_ENTRY(mi_wrq_stall_qvn_vn0, 0xD),
CCI_EVENT_EXT_ATTR_ENTRY(mi_wrq_stall_qvn_vn1, 0xE),
CCI_EVENT_EXT_ATTR_ENTRY(mi_wrq_stall_qvn_vn2, 0xF),
CCI_EVENT_EXT_ATTR_ENTRY(mi_wrq_stall_qvn_vn3, 0x10),
CCI_EVENT_EXT_ATTR_ENTRY(mi_wrq_unique_or_line_unique_addr_hazard, 0x11),
/* Special event for cycles counter */
CCI400_CYCLE_EVENT_EXT_ATTR_ENTRY(cycles, 0xff),
NULL
};
static ssize_t cci400_pmu_cycle_event_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct dev_ext_attribute *eattr = container_of(attr,
struct dev_ext_attribute, attr);
return snprintf(buf, PAGE_SIZE, "config=0x%lx\n", (unsigned long)eattr->var);
}
static int cci400_get_event_idx(struct cci_pmu *cci_pmu,
struct cci_pmu_hw_events *hw,
unsigned long cci_event)
{
int idx;
/* cycles event idx is fixed */
if (cci_event == CCI400_PMU_CYCLES) {
if (test_and_set_bit(CCI400_PMU_CYCLE_CNTR_IDX, hw->used_mask))
return -EAGAIN;
return CCI400_PMU_CYCLE_CNTR_IDX;
}
for (idx = CCI400_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 cci400_validate_hw_event(struct cci_pmu *cci_pmu, unsigned long hw_event)
{
u8 ev_source = CCI400_PMU_EVENT_SOURCE(hw_event);
u8 ev_code = CCI400_PMU_EVENT_CODE(hw_event);
int if_type;
if (hw_event & ~CCI400_PMU_EVENT_MASK)
return -ENOENT;
if (hw_event == CCI400_PMU_CYCLES)
return hw_event;
switch (ev_source) {
case CCI400_PORT_S0:
case CCI400_PORT_S1:
case CCI400_PORT_S2:
case CCI400_PORT_S3:
case CCI400_PORT_S4:
/* Slave Interface */
if_type = CCI_IF_SLAVE;
break;
case CCI400_PORT_M0:
case CCI400_PORT_M1:
case CCI400_PORT_M2:
/* Master Interface */
if_type = CCI_IF_MASTER;
break;
default:
return -ENOENT;
}
if (ev_code >= cci_pmu->model->event_ranges[if_type].min &&
ev_code <= cci_pmu->model->event_ranges[if_type].max)
return hw_event;
return -ENOENT;
}
static int probe_cci400_revision(struct cci_pmu *cci_pmu)
{
int rev;
rev = readl_relaxed(cci_pmu->ctrl_base + CCI_PID2) & CCI_PID2_REV_MASK;
rev >>= CCI_PID2_REV_SHIFT;
if (rev < CCI400_R1_PX)
return CCI400_R0;
else
return CCI400_R1;
}
static const struct cci_pmu_model *probe_cci_model(struct cci_pmu *cci_pmu)
{
if (platform_has_secure_cci_access())
return &cci_pmu_models[probe_cci400_revision(cci_pmu)];
return NULL;
}
#else /* !CONFIG_ARM_CCI400_PMU */
static inline struct cci_pmu_model *probe_cci_model(struct cci_pmu *cci_pmu)
{
return NULL;
}
#endif /* CONFIG_ARM_CCI400_PMU */
#ifdef CONFIG_ARM_CCI5xx_PMU
/*
* CCI5xx PMU event id is an 9-bit value made of two parts.
* bits [8:5] - Source for the event
* bits [4:0] - Event code (specific to type of interface)
*
*
*/
/* Port ids */
#define CCI5xx_PORT_S0 0x0
#define CCI5xx_PORT_S1 0x1
#define CCI5xx_PORT_S2 0x2
#define CCI5xx_PORT_S3 0x3
#define CCI5xx_PORT_S4 0x4
#define CCI5xx_PORT_S5 0x5
#define CCI5xx_PORT_S6 0x6
#define CCI5xx_PORT_M0 0x8
#define CCI5xx_PORT_M1 0x9
#define CCI5xx_PORT_M2 0xa
#define CCI5xx_PORT_M3 0xb
#define CCI5xx_PORT_M4 0xc
#define CCI5xx_PORT_M5 0xd
#define CCI5xx_PORT_M6 0xe
#define CCI5xx_PORT_GLOBAL 0xf
#define CCI5xx_PMU_EVENT_MASK 0x1ffUL
#define CCI5xx_PMU_EVENT_SOURCE_SHIFT 0x5
#define CCI5xx_PMU_EVENT_SOURCE_MASK 0xf
#define CCI5xx_PMU_EVENT_CODE_SHIFT 0x0
#define CCI5xx_PMU_EVENT_CODE_MASK 0x1f
#define CCI5xx_PMU_EVENT_SOURCE(event) \
((event >> CCI5xx_PMU_EVENT_SOURCE_SHIFT) & CCI5xx_PMU_EVENT_SOURCE_MASK)
#define CCI5xx_PMU_EVENT_CODE(event) \
((event >> CCI5xx_PMU_EVENT_CODE_SHIFT) & CCI5xx_PMU_EVENT_CODE_MASK)
#define CCI5xx_SLAVE_PORT_MIN_EV 0x00
#define CCI5xx_SLAVE_PORT_MAX_EV 0x1f
#define CCI5xx_MASTER_PORT_MIN_EV 0x00
#define CCI5xx_MASTER_PORT_MAX_EV 0x06
#define CCI5xx_GLOBAL_PORT_MIN_EV 0x00
#define CCI5xx_GLOBAL_PORT_MAX_EV 0x0f
#define CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(_name, _config) \
CCI_EXT_ATTR_ENTRY(_name, cci5xx_pmu_global_event_show, \
(unsigned long) _config)
static ssize_t cci5xx_pmu_global_event_show(struct device *dev,
struct device_attribute *attr, char *buf);
static struct attribute *cci5xx_pmu_format_attrs[] = {
CCI_FORMAT_EXT_ATTR_ENTRY(event, "config:0-4"),
CCI_FORMAT_EXT_ATTR_ENTRY(source, "config:5-8"),
NULL,
};
static struct attribute *cci5xx_pmu_event_attrs[] = {
/* Slave events */
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_arvalid, 0x0),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_dev, 0x1),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_nonshareable, 0x2),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_shareable_non_alloc, 0x3),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_shareable_alloc, 0x4),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_invalidate, 0x5),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_cache_maint, 0x6),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_dvm_msg, 0x7),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_rval, 0x8),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_hs_rlast_snoop, 0x9),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_hs_awalid, 0xA),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_dev, 0xB),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_non_shareable, 0xC),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_share_wb, 0xD),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_share_wlu, 0xE),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_share_wunique, 0xF),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_evict, 0x10),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_wrevict, 0x11),
CCI_EVENT_EXT_ATTR_ENTRY(si_w_data_beat, 0x12),
CCI_EVENT_EXT_ATTR_ENTRY(si_srq_acvalid, 0x13),
CCI_EVENT_EXT_ATTR_ENTRY(si_srq_read, 0x14),
CCI_EVENT_EXT_ATTR_ENTRY(si_srq_clean, 0x15),
CCI_EVENT_EXT_ATTR_ENTRY(si_srq_data_transfer_low, 0x16),
CCI_EVENT_EXT_ATTR_ENTRY(si_rrq_stall_arvalid, 0x17),
CCI_EVENT_EXT_ATTR_ENTRY(si_r_data_stall, 0x18),
CCI_EVENT_EXT_ATTR_ENTRY(si_wrq_stall, 0x19),
CCI_EVENT_EXT_ATTR_ENTRY(si_w_data_stall, 0x1A),
CCI_EVENT_EXT_ATTR_ENTRY(si_w_resp_stall, 0x1B),
CCI_EVENT_EXT_ATTR_ENTRY(si_srq_stall, 0x1C),
CCI_EVENT_EXT_ATTR_ENTRY(si_s_data_stall, 0x1D),
CCI_EVENT_EXT_ATTR_ENTRY(si_rq_stall_ot_limit, 0x1E),
CCI_EVENT_EXT_ATTR_ENTRY(si_r_stall_arbit, 0x1F),
/* Master events */
CCI_EVENT_EXT_ATTR_ENTRY(mi_r_data_beat_any, 0x0),
CCI_EVENT_EXT_ATTR_ENTRY(mi_w_data_beat_any, 0x1),
CCI_EVENT_EXT_ATTR_ENTRY(mi_rrq_stall, 0x2),
CCI_EVENT_EXT_ATTR_ENTRY(mi_r_data_stall, 0x3),
CCI_EVENT_EXT_ATTR_ENTRY(mi_wrq_stall, 0x4),
CCI_EVENT_EXT_ATTR_ENTRY(mi_w_data_stall, 0x5),
CCI_EVENT_EXT_ATTR_ENTRY(mi_w_resp_stall, 0x6),
/* Global events */
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_access_filter_bank_0_1, 0x0),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_access_filter_bank_2_3, 0x1),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_access_filter_bank_4_5, 0x2),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_access_filter_bank_6_7, 0x3),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_access_miss_filter_bank_0_1, 0x4),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_access_miss_filter_bank_2_3, 0x5),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_access_miss_filter_bank_4_5, 0x6),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_access_miss_filter_bank_6_7, 0x7),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_back_invalidation, 0x8),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_stall_alloc_busy, 0x9),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_stall_tt_full, 0xA),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_wrq, 0xB),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_cd_hs, 0xC),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_rq_stall_addr_hazard, 0xD),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_rq_stall_tt_full, 0xE),
CCI5xx_GLOBAL_EVENT_EXT_ATTR_ENTRY(cci_snoop_rq_tzmp1_prot, 0xF),
NULL
};
static ssize_t cci5xx_pmu_global_event_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct dev_ext_attribute *eattr = container_of(attr,
struct dev_ext_attribute, attr);
/* Global events have single fixed source code */
return snprintf(buf, PAGE_SIZE, "event=0x%lx,source=0x%x\n",
(unsigned long)eattr->var, CCI5xx_PORT_GLOBAL);
}
/*
* CCI500 provides 8 independent event counters that can count
* any of the events available.
* CCI500 PMU event source ids
* 0x0-0x6 - Slave interfaces
* 0x8-0xD - Master interfaces
* 0xf - Global Events
* 0x7,0xe - Reserved
*/
static int cci500_validate_hw_event(struct cci_pmu *cci_pmu,
unsigned long hw_event)
{
u32 ev_source = CCI5xx_PMU_EVENT_SOURCE(hw_event);
u32 ev_code = CCI5xx_PMU_EVENT_CODE(hw_event);
int if_type;
if (hw_event & ~CCI5xx_PMU_EVENT_MASK)
return -ENOENT;
switch (ev_source) {
case CCI5xx_PORT_S0:
case CCI5xx_PORT_S1:
case CCI5xx_PORT_S2:
case CCI5xx_PORT_S3:
case CCI5xx_PORT_S4:
case CCI5xx_PORT_S5:
case CCI5xx_PORT_S6:
if_type = CCI_IF_SLAVE;
break;
case CCI5xx_PORT_M0:
case CCI5xx_PORT_M1:
case CCI5xx_PORT_M2:
case CCI5xx_PORT_M3:
case CCI5xx_PORT_M4:
case CCI5xx_PORT_M5:
if_type = CCI_IF_MASTER;
break;
case CCI5xx_PORT_GLOBAL:
if_type = CCI_IF_GLOBAL;
break;
default:
return -ENOENT;
}
if (ev_code >= cci_pmu->model->event_ranges[if_type].min &&
ev_code <= cci_pmu->model->event_ranges[if_type].max)
return hw_event;
return -ENOENT;
}
/*
* CCI550 provides 8 independent event counters that can count
* any of the events available.
* CCI550 PMU event source ids
* 0x0-0x6 - Slave interfaces
* 0x8-0xe - Master interfaces
* 0xf - Global Events
* 0x7 - Reserved
*/
static int cci550_validate_hw_event(struct cci_pmu *cci_pmu,
unsigned long hw_event)
{
u32 ev_source = CCI5xx_PMU_EVENT_SOURCE(hw_event);
u32 ev_code = CCI5xx_PMU_EVENT_CODE(hw_event);
int if_type;
if (hw_event & ~CCI5xx_PMU_EVENT_MASK)
return -ENOENT;
switch (ev_source) {
case CCI5xx_PORT_S0:
case CCI5xx_PORT_S1:
case CCI5xx_PORT_S2:
case CCI5xx_PORT_S3:
case CCI5xx_PORT_S4:
case CCI5xx_PORT_S5:
case CCI5xx_PORT_S6:
if_type = CCI_IF_SLAVE;
break;
case CCI5xx_PORT_M0:
case CCI5xx_PORT_M1:
case CCI5xx_PORT_M2:
case CCI5xx_PORT_M3:
case CCI5xx_PORT_M4:
case CCI5xx_PORT_M5:
case CCI5xx_PORT_M6:
if_type = CCI_IF_MASTER;
break;
case CCI5xx_PORT_GLOBAL:
if_type = CCI_IF_GLOBAL;
break;
default:
return -ENOENT;
}
if (ev_code >= cci_pmu->model->event_ranges[if_type].min &&
ev_code <= cci_pmu->model->event_ranges[if_type].max)
return hw_event;
return -ENOENT;
}
#endif /* CONFIG_ARM_CCI5xx_PMU */
/*
* Program the CCI PMU counters which have PERF_HES_ARCH set
* with the event period and mark them ready before we enable
* PMU.
*/
static void cci_pmu_sync_counters(struct cci_pmu *cci_pmu)
{
int i;
struct cci_pmu_hw_events *cci_hw = &cci_pmu->hw_events;
DECLARE_BITMAP(mask, HW_CNTRS_MAX);
bitmap_zero(mask, cci_pmu->num_cntrs);
for_each_set_bit(i, cci_pmu->hw_events.used_mask, cci_pmu->num_cntrs) {
struct perf_event *event = cci_hw->events[i];
if (WARN_ON(!event))
continue;
/* Leave the events which are not counting */
if (event->hw.state & PERF_HES_STOPPED)
continue;
if (event->hw.state & PERF_HES_ARCH) {
set_bit(i, mask);
event->hw.state &= ~PERF_HES_ARCH;
}
}
pmu_write_counters(cci_pmu, mask);
}
/* Should be called with cci_pmu->hw_events->pmu_lock held */
static void __cci_pmu_enable_nosync(struct cci_pmu *cci_pmu)
{
u32 val;
/* Enable all the PMU counters. */
val = readl_relaxed(cci_pmu->ctrl_base + CCI_PMCR) | CCI_PMCR_CEN;
writel(val, cci_pmu->ctrl_base + CCI_PMCR);
}
/* Should be called with cci_pmu->hw_events->pmu_lock held */
static void __cci_pmu_enable_sync(struct cci_pmu *cci_pmu)
{
cci_pmu_sync_counters(cci_pmu);
__cci_pmu_enable_nosync(cci_pmu);
}
/* Should be called with cci_pmu->hw_events->pmu_lock held */
static void __cci_pmu_disable(struct cci_pmu *cci_pmu)
{
u32 val;
/* Disable all the PMU counters. */
val = readl_relaxed(cci_pmu->ctrl_base + CCI_PMCR) & ~CCI_PMCR_CEN;
writel(val, cci_pmu->ctrl_base + CCI_PMCR);
}
static ssize_t cci_pmu_format_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct dev_ext_attribute *eattr = container_of(attr,
struct dev_ext_attribute, attr);
return snprintf(buf, PAGE_SIZE, "%s\n", (char *)eattr->var);
}
static ssize_t cci_pmu_event_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct dev_ext_attribute *eattr = container_of(attr,
struct dev_ext_attribute, attr);
/* source parameter is mandatory for normal PMU events */
return snprintf(buf, PAGE_SIZE, "source=?,event=0x%lx\n",
(unsigned long)eattr->var);
}
static int pmu_is_valid_counter(struct cci_pmu *cci_pmu, int idx)
{
return 0 <= idx && idx <= CCI_PMU_CNTR_LAST(cci_pmu);
}
static u32 pmu_read_register(struct cci_pmu *cci_pmu, int idx, unsigned int offset)
{
return readl_relaxed(cci_pmu->base +
CCI_PMU_CNTR_BASE(cci_pmu->model, idx) + offset);
}
static void pmu_write_register(struct cci_pmu *cci_pmu, u32 value,
int idx, unsigned int offset)
{
writel_relaxed(value, cci_pmu->base +
CCI_PMU_CNTR_BASE(cci_pmu->model, idx) + offset);
}
static void pmu_disable_counter(struct cci_pmu *cci_pmu, int idx)
{
pmu_write_register(cci_pmu, 0, idx, CCI_PMU_CNTR_CTRL);
}
static void pmu_enable_counter(struct cci_pmu *cci_pmu, int idx)
{
pmu_write_register(cci_pmu, 1, idx, CCI_PMU_CNTR_CTRL);
}
static bool __maybe_unused
pmu_counter_is_enabled(struct cci_pmu *cci_pmu, int idx)
{
return (pmu_read_register(cci_pmu, idx, CCI_PMU_CNTR_CTRL) & 0x1) != 0;
}
static void pmu_set_event(struct cci_pmu *cci_pmu, int idx, unsigned long event)
{
pmu_write_register(cci_pmu, event, idx, CCI_PMU_EVT_SEL);
}
/*
* For all counters on the CCI-PMU, disable any 'enabled' counters,
* saving the changed counters in the mask, so that we can restore
* it later using pmu_restore_counters. The mask is private to the
* caller. We cannot rely on the used_mask maintained by the CCI_PMU
* as it only tells us if the counter is assigned to perf_event or not.
* The state of the perf_event cannot be locked by the PMU layer, hence
* we check the individual counter status (which can be locked by
* cci_pm->hw_events->pmu_lock).
*
* @mask should be initialised to empty by the caller.
*/
static void __maybe_unused
pmu_save_counters(struct cci_pmu *cci_pmu, unsigned long *mask)
{
int i;
for (i = 0; i < cci_pmu->num_cntrs; i++) {
if (pmu_counter_is_enabled(cci_pmu, i)) {
set_bit(i, mask);
pmu_disable_counter(cci_pmu, i);
}
}
}
/*
* Restore the status of the counters. Reversal of the pmu_save_counters().
* For each counter set in the mask, enable the counter back.
*/
static void __maybe_unused
pmu_restore_counters(struct cci_pmu *cci_pmu, unsigned long *mask)
{
int i;
for_each_set_bit(i, mask, cci_pmu->num_cntrs)
pmu_enable_counter(cci_pmu, i);
}
/*
* Returns the number of programmable counters actually implemented
* by the cci
*/
static u32 pmu_get_max_counters(struct cci_pmu *cci_pmu)
{
return (readl_relaxed(cci_pmu->ctrl_base + CCI_PMCR) &
CCI_PMCR_NCNT_MASK) >> CCI_PMCR_NCNT_SHIFT;
}
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);
unsigned long cci_event = event->hw.config_base;
int idx;
if (cci_pmu->model->get_event_idx)
return cci_pmu->model->get_event_idx(cci_pmu, hw, cci_event);
/* Generic code to find an unused idx from the mask */
for(idx = 0; 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)
{
struct cci_pmu *cci_pmu = to_cci_pmu(event->pmu);
if (event->attr.type < PERF_TYPE_MAX ||
!cci_pmu->model->validate_hw_event)
return -ENOENT;
return cci_pmu->model->validate_hw_event(cci_pmu, event->attr.config);
}
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 (cci_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 < cci_pmu->nr_irqs; i++) {
int err = request_irq(cci_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",
cci_pmu->irqs[i]);
return err;
}
set_bit(i, &cci_pmu->active_irqs);
}
return 0;
}
static void pmu_free_irq(struct cci_pmu *cci_pmu)
{
int i;
for (i = 0; i < cci_pmu->nr_irqs; i++) {
if (!test_and_clear_bit(i, &cci_pmu->active_irqs))
continue;
free_irq(cci_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(cci_pmu, idx, CCI_PMU_CNTR);
return value;
}
static void pmu_write_counter(struct cci_pmu *cci_pmu, u32 value, int idx)
{
pmu_write_register(cci_pmu, value, idx, CCI_PMU_CNTR);
}
static void __pmu_write_counters(struct cci_pmu *cci_pmu, unsigned long *mask)
{
int i;
struct cci_pmu_hw_events *cci_hw = &cci_pmu->hw_events;
for_each_set_bit(i, mask, cci_pmu->num_cntrs) {
struct perf_event *event = cci_hw->events[i];
if (WARN_ON(!event))
continue;
pmu_write_counter(cci_pmu, local64_read(&event->hw.prev_count), i);
}
}
static void pmu_write_counters(struct cci_pmu *cci_pmu, unsigned long *mask)
{
if (cci_pmu->model->write_counters)
cci_pmu->model->write_counters(cci_pmu, mask);
else
__pmu_write_counters(cci_pmu, mask);
}
#ifdef CONFIG_ARM_CCI5xx_PMU
/*
* CCI-500/CCI-550 has advanced power saving policies, which could gate the
* clocks to the PMU counters, which makes the writes to them ineffective.
* The only way to write to those counters is when the global counters
* are enabled and the particular counter is enabled.
*
* So we do the following :
*
* 1) Disable all the PMU counters, saving their current state
* 2) Enable the global PMU profiling, now that all counters are
* disabled.
*
* For each counter to be programmed, repeat steps 3-7:
*
* 3) Write an invalid event code to the event control register for the
counter, so that the counters are not modified.
* 4) Enable the counter control for the counter.
* 5) Set the counter value
* 6) Disable the counter
* 7) Restore the event in the target counter
*
* 8) Disable the global PMU.
* 9) Restore the status of the rest of the counters.
*
* We choose an event which for CCI-5xx is guaranteed not to count.
* We use the highest possible event code (0x1f) for the master interface 0.
*/
#define CCI5xx_INVALID_EVENT ((CCI5xx_PORT_M0 << CCI5xx_PMU_EVENT_SOURCE_SHIFT) | \
(CCI5xx_PMU_EVENT_CODE_MASK << CCI5xx_PMU_EVENT_CODE_SHIFT))
static void cci5xx_pmu_write_counters(struct cci_pmu *cci_pmu, unsigned long *mask)
{
int i;
DECLARE_BITMAP(saved_mask, HW_CNTRS_MAX);
bitmap_zero(saved_mask, cci_pmu->num_cntrs);
pmu_save_counters(cci_pmu, saved_mask);
/*
* Now that all the counters are disabled, we can safely turn the PMU on,
* without syncing the status of the counters
*/
__cci_pmu_enable_nosync(cci_pmu);
for_each_set_bit(i, mask, cci_pmu->num_cntrs) {
struct perf_event *event = cci_pmu->hw_events.events[i];
if (WARN_ON(!event))
continue;
pmu_set_event(cci_pmu, i, CCI5xx_INVALID_EVENT);
pmu_enable_counter(cci_pmu, i);
pmu_write_counter(cci_pmu, local64_read(&event->hw.prev_count), i);
pmu_disable_counter(cci_pmu, i);
pmu_set_event(cci_pmu, i, event->hw.config_base);
}
__cci_pmu_disable(cci_pmu);
pmu_restore_counters(cci_pmu, saved_mask);
}
#endif /* CONFIG_ARM_CCI5xx_PMU */
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);
}
static 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);
/*
* CCI PMU uses PERF_HES_ARCH to keep track of the counters, whose
* values needs to be sync-ed with the s/w state before the PMU is
* enabled.
* Mark this counter for sync.
*/
hwc->state |= PERF_HES_ARCH;
}
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 = &cci_pmu->hw_events;
int idx, handled = IRQ_NONE;
raw_spin_lock_irqsave(&events->pmu_lock, flags);
/* Disable the PMU while we walk through the counters */
__cci_pmu_disable(cci_pmu);
/*
* 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 = 0; idx <= CCI_PMU_CNTR_LAST(cci_pmu); idx++) {
struct perf_event *event = events->events[idx];
if (!event)
continue;
/* Did this counter overflow? */
if (!(pmu_read_register(cci_pmu, idx, CCI_PMU_OVRFLW) &
CCI_PMU_OVRFLW_FLAG))
continue;
pmu_write_register(cci_pmu, CCI_PMU_OVRFLW_FLAG, idx,
CCI_PMU_OVRFLW);
pmu_event_update(event);
pmu_event_set_period(event);
handled = IRQ_HANDLED;
}
/* Enable the PMU and sync possibly overflowed counters */
__cci_pmu_enable_sync(cci_pmu);
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_cntrs);
unsigned long flags;
if (!enabled)
return;
raw_spin_lock_irqsave(&hw_events->pmu_lock, flags);
__cci_pmu_enable_sync(cci_pmu);
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;
raw_spin_lock_irqsave(&hw_events->pmu_lock, flags);
__cci_pmu_disable(cci_pmu);
raw_spin_unlock_irqrestore(&hw_events->pmu_lock, flags);
}
/*
* Check if the idx represents a non-programmable counter.
* All the fixed event counters are mapped before the programmable
* counters.
*/
static bool pmu_fixed_hw_idx(struct cci_pmu *cci_pmu, int idx)
{
return (idx >= 0) && (idx < cci_pmu->model->fixed_hw_cntrs);
}
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 counter unless you are counting a fixed event */
if (!pmu_fixed_hw_idx(cci_pmu, idx))
pmu_set_event(cci_pmu, idx, hwc->config_base);
pmu_event_set_period(event);
pmu_enable_counter(cci_pmu, 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(cci_pmu, 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;
/* If we don't have a space for the counter then finish early. */
idx = pmu_get_event_idx(hw_events, event);
if (idx < 0)
return idx;
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);
return 0;
}
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 *cci_pmu = to_cci_pmu(event->pmu);
unsigned long mask[BITS_TO_LONGS(HW_CNTRS_MAX)];
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 = mask,
};
memset(mask, 0, BITS_TO_LONGS(cci_pmu->num_cntrs) * sizeof(unsigned long));
if (!validate_event(event->pmu, &fake_pmu, leader))
return -EINVAL;
ARM: SoC driver updates for 4.17 The main addition this time around is the new ARM "SCMI" framework, which is the latest in a series of standards coming from ARM to do power management in a platform independent way. This has been through many review cycles, and it relies on a rather interesting way of using the mailbox subsystem, but in the end I agreed that Sudeep's version was the best we could do after all. Other changes include: - the ARM CCN driver is moved out of drivers/bus into drivers/perf, which makes more sense. Similarly, the performance monitoring portion of the CCI driver are moved the same way and cleaned up a little more. - a series of updates to the SCPI framework - support for the Mediatek mt7623a SoC in drivers/soc - support for additional NVIDIA Tegra hardware in drivers/soc - a new reset driver for Socionext Uniphier - lesser bug fixes in drivers/soc, drivers/tee, drivers/memory, and drivers/firmware and drivers/reset across platforms -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iQIcBAABAgAGBQJaxiNzAAoJEGCrR//JCVInYhYP/2kPhc5t/kszA1bcklcbO9dY eX37Ra/RR4yQ5yeQZVIZ4UkUovxk9PmG2tM4K5oJaTDsz5pPEgavVOOr3sbfj6vb 4O9auTeysEQlHcbVdNFum0YS2gUY2YD7D12DTRorotLxCqod184ccWXq0XGfIWaY l3YRrcL/lPlqmyS3z/GNx9oNygOMUzEfXfIQYICyzHuYiLBUGnkKC1vIb+Hx1TDq Cxk++AUqH13Mss24O2A2QQh+oBHj2BybDLLqwcC5PSpsUbFrVCfzG54l43mig32T NOxV0Qnml2wAtU4H0QcgtSgwRimHD0YOiX8ssquvDDiqTqM5G+llSTGkEbYe+AUW 4GIZYoBOwGkfEXS+tyymHe9yfc5h1OLYAeFU1jRm723c7phanuu67rPn35YC8UMK zSql10JpkAGNzMikrxxb6wnis951w2UFlzhgZQ6ItA/nRq3l+oEQA0Qiljv965nz DVLsD5+gdhK6GBctkzlsD5HFn6GjM8JilnsOVPHD765nKnVBSxKiXRLV228XVug2 rChF1FhQqLnM54jCMqHZX5fS9SbSgtYswHqIXpVw6GmJkqq/Ly10yGR0vuWD+uyn BV7q5AKpGrwm6wZkMM2uZ1VdUtWzn856AbkqrvX/QhmJcX4McuqaLUrC8bSOj1ty KeVil0akq3nU+xHl5Ojs =Pmsx -----END PGP SIGNATURE----- Merge tag 'armsoc-drivers' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc Pull ARM SoC driver updates from Arnd Bergmann: "The main addition this time around is the new ARM "SCMI" framework, which is the latest in a series of standards coming from ARM to do power management in a platform independent way. This has been through many review cycles, and it relies on a rather interesting way of using the mailbox subsystem, but in the end I agreed that Sudeep's version was the best we could do after all. Other changes include: - the ARM CCN driver is moved out of drivers/bus into drivers/perf, which makes more sense. Similarly, the performance monitoring portion of the CCI driver are moved the same way and cleaned up a little more. - a series of updates to the SCPI framework - support for the Mediatek mt7623a SoC in drivers/soc - support for additional NVIDIA Tegra hardware in drivers/soc - a new reset driver for Socionext Uniphier - lesser bug fixes in drivers/soc, drivers/tee, drivers/memory, and drivers/firmware and drivers/reset across platforms" * tag 'armsoc-drivers' of git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc: (87 commits) reset: uniphier: add ethernet reset control support for PXs3 reset: stm32mp1: Enable stm32mp1 reset driver dt-bindings: reset: add STM32MP1 resets reset: uniphier: add Pro4/Pro5/PXs2 audio systems reset control reset: imx7: add 'depends on HAS_IOMEM' to fix unmet dependency reset: modify the way reset lookup works for board files reset: add support for non-DT systems clk: scmi: use devm_of_clk_add_hw_provider() API and drop scmi_clocks_remove firmware: arm_scmi: prevent accessing rate_discrete uninitialized hwmon: (scmi) return -EINVAL when sensor information is unavailable amlogic: meson-gx-socinfo: Update soc ids soc/tegra: pmc: Use the new reset APIs to manage reset controllers soc: mediatek: update power domain data of MT2712 dt-bindings: soc: update MT2712 power dt-bindings cpufreq: scmi: add thermal dependency soc: mediatek: fix the mistaken pointer accessed when subdomains are added soc: mediatek: add SCPSYS power domain driver for MediaTek MT7623A SoC soc: mediatek: avoid hardcoded value with bus_prot_mask dt-bindings: soc: add header files required for MT7623A SCPSYS dt-binding dt-bindings: soc: add SCPSYS binding for MT7623 and MT7623A SoC ...
2018-04-06 12:29:35 +08:00
for_each_sibling_event(sibling, leader) {
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;
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;
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;
/*
* 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.
*/
if (event->cpu < 0)
return -EINVAL;
event->cpu = cci_pmu->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_cpumask_attr_show(struct device *dev,
struct device_attribute *attr, char *buf)
{
struct pmu *pmu = dev_get_drvdata(dev);
struct cci_pmu *cci_pmu = to_cci_pmu(pmu);
return cpumap_print_to_pagebuf(true, buf, cpumask_of(cci_pmu->cpu));
}
static struct device_attribute pmu_cpumask_attr =
__ATTR(cpumask, S_IRUGO, pmu_cpumask_attr_show, NULL);
static struct attribute *pmu_attrs[] = {
&pmu_cpumask_attr.attr,
NULL,
};
static struct attribute_group pmu_attr_group = {
.attrs = pmu_attrs,
};
static struct attribute_group pmu_format_attr_group = {
.name = "format",
.attrs = NULL, /* Filled in cci_pmu_init_attrs */
};
static struct attribute_group pmu_event_attr_group = {
.name = "events",
.attrs = NULL, /* Filled in cci_pmu_init_attrs */
};
static const struct attribute_group *pmu_attr_groups[] = {
&pmu_attr_group,
&pmu_format_attr_group,
&pmu_event_attr_group,
NULL
};
static int cci_pmu_init(struct cci_pmu *cci_pmu, struct platform_device *pdev)
{
const struct cci_pmu_model *model = cci_pmu->model;
char *name = model->name;
u32 num_cntrs;
if (WARN_ON(model->num_hw_cntrs > NUM_HW_CNTRS_MAX))
return -EINVAL;
if (WARN_ON(model->fixed_hw_cntrs > FIXED_HW_CNTRS_MAX))
return -EINVAL;
pmu_event_attr_group.attrs = model->event_attrs;
pmu_format_attr_group.attrs = model->format_attrs;
cci_pmu->pmu = (struct pmu) {
.module = THIS_MODULE,
.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,
.capabilities = PERF_PMU_CAP_NO_EXCLUDE,
};
cci_pmu->plat_device = pdev;
num_cntrs = pmu_get_max_counters(cci_pmu);
if (num_cntrs > cci_pmu->model->num_hw_cntrs) {
dev_warn(&pdev->dev,
"PMU implements more counters(%d) than supported by"
" the model(%d), truncated.",
num_cntrs, cci_pmu->model->num_hw_cntrs);
num_cntrs = cci_pmu->model->num_hw_cntrs;
}
cci_pmu->num_cntrs = num_cntrs + cci_pmu->model->fixed_hw_cntrs;
return perf_pmu_register(&cci_pmu->pmu, name, -1);
}
static int cci_pmu_offline_cpu(unsigned int cpu)
{
int target;
if (!g_cci_pmu || cpu != g_cci_pmu->cpu)
return 0;
target = cpumask_any_but(cpu_online_mask, cpu);
if (target >= nr_cpu_ids)
return 0;
perf_pmu_migrate_context(&g_cci_pmu->pmu, cpu, target);
g_cci_pmu->cpu = target;
return 0;
}
static __maybe_unused struct cci_pmu_model cci_pmu_models[] = {
#ifdef CONFIG_ARM_CCI400_PMU
[CCI400_R0] = {
.name = "CCI_400",
.fixed_hw_cntrs = FIXED_HW_CNTRS_CII_4XX, /* Cycle counter */
.num_hw_cntrs = NUM_HW_CNTRS_CII_4XX,
.cntr_size = SZ_4K,
.format_attrs = cci400_pmu_format_attrs,
.event_attrs = cci400_r0_pmu_event_attrs,
.event_ranges = {
[CCI_IF_SLAVE] = {
CCI400_R0_SLAVE_PORT_MIN_EV,
CCI400_R0_SLAVE_PORT_MAX_EV,
},
[CCI_IF_MASTER] = {
CCI400_R0_MASTER_PORT_MIN_EV,
CCI400_R0_MASTER_PORT_MAX_EV,
},
},
.validate_hw_event = cci400_validate_hw_event,
.get_event_idx = cci400_get_event_idx,
},
[CCI400_R1] = {
.name = "CCI_400_r1",
.fixed_hw_cntrs = FIXED_HW_CNTRS_CII_4XX, /* Cycle counter */
.num_hw_cntrs = NUM_HW_CNTRS_CII_4XX,
.cntr_size = SZ_4K,
.format_attrs = cci400_pmu_format_attrs,
.event_attrs = cci400_r1_pmu_event_attrs,
.event_ranges = {
[CCI_IF_SLAVE] = {
CCI400_R1_SLAVE_PORT_MIN_EV,
CCI400_R1_SLAVE_PORT_MAX_EV,
},
[CCI_IF_MASTER] = {
CCI400_R1_MASTER_PORT_MIN_EV,
CCI400_R1_MASTER_PORT_MAX_EV,
},
},
.validate_hw_event = cci400_validate_hw_event,
.get_event_idx = cci400_get_event_idx,
},
#endif
#ifdef CONFIG_ARM_CCI5xx_PMU
[CCI500_R0] = {
.name = "CCI_500",
.fixed_hw_cntrs = FIXED_HW_CNTRS_CII_5XX,
.num_hw_cntrs = NUM_HW_CNTRS_CII_5XX,
.cntr_size = SZ_64K,
.format_attrs = cci5xx_pmu_format_attrs,
.event_attrs = cci5xx_pmu_event_attrs,
.event_ranges = {
[CCI_IF_SLAVE] = {
CCI5xx_SLAVE_PORT_MIN_EV,
CCI5xx_SLAVE_PORT_MAX_EV,
},
[CCI_IF_MASTER] = {
CCI5xx_MASTER_PORT_MIN_EV,
CCI5xx_MASTER_PORT_MAX_EV,
},
[CCI_IF_GLOBAL] = {
CCI5xx_GLOBAL_PORT_MIN_EV,
CCI5xx_GLOBAL_PORT_MAX_EV,
},
},
.validate_hw_event = cci500_validate_hw_event,
.write_counters = cci5xx_pmu_write_counters,
},
[CCI550_R0] = {
.name = "CCI_550",
.fixed_hw_cntrs = FIXED_HW_CNTRS_CII_5XX,
.num_hw_cntrs = NUM_HW_CNTRS_CII_5XX,
.cntr_size = SZ_64K,
.format_attrs = cci5xx_pmu_format_attrs,
.event_attrs = cci5xx_pmu_event_attrs,
.event_ranges = {
[CCI_IF_SLAVE] = {
CCI5xx_SLAVE_PORT_MIN_EV,
CCI5xx_SLAVE_PORT_MAX_EV,
},
[CCI_IF_MASTER] = {
CCI5xx_MASTER_PORT_MIN_EV,
CCI5xx_MASTER_PORT_MAX_EV,
},
[CCI_IF_GLOBAL] = {
CCI5xx_GLOBAL_PORT_MIN_EV,
CCI5xx_GLOBAL_PORT_MAX_EV,
},
},
.validate_hw_event = cci550_validate_hw_event,
.write_counters = cci5xx_pmu_write_counters,
},
#endif
};
static const struct of_device_id arm_cci_pmu_matches[] = {
#ifdef CONFIG_ARM_CCI400_PMU
{
.compatible = "arm,cci-400-pmu",
.data = NULL,
},
{
.compatible = "arm,cci-400-pmu,r0",
.data = &cci_pmu_models[CCI400_R0],
},
{
.compatible = "arm,cci-400-pmu,r1",
.data = &cci_pmu_models[CCI400_R1],
},
#endif
#ifdef CONFIG_ARM_CCI5xx_PMU
{
.compatible = "arm,cci-500-pmu,r0",
.data = &cci_pmu_models[CCI500_R0],
},
{
.compatible = "arm,cci-550-pmu,r0",
.data = &cci_pmu_models[CCI550_R0],
},
#endif
{},
};
MODULE_DEVICE_TABLE(of, arm_cci_pmu_matches);
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 struct cci_pmu *cci_pmu_alloc(struct device *dev)
{
struct cci_pmu *cci_pmu;
const struct cci_pmu_model *model;
/*
* All allocations are devm_* hence we don't have to free
* them explicitly on an error, as it would end up in driver
* detach.
*/
cci_pmu = devm_kzalloc(dev, sizeof(*cci_pmu), GFP_KERNEL);
if (!cci_pmu)
return ERR_PTR(-ENOMEM);
cci_pmu->ctrl_base = *(void __iomem **)dev->platform_data;
model = of_device_get_match_data(dev);
if (!model) {
dev_warn(dev,
"DEPRECATED compatible property, requires secure access to CCI registers");
model = probe_cci_model(cci_pmu);
}
if (!model) {
dev_warn(dev, "CCI PMU version not supported\n");
return ERR_PTR(-ENODEV);
}
cci_pmu->model = model;
cci_pmu->irqs = devm_kcalloc(dev, CCI_PMU_MAX_HW_CNTRS(model),
sizeof(*cci_pmu->irqs), GFP_KERNEL);
if (!cci_pmu->irqs)
return ERR_PTR(-ENOMEM);
cci_pmu->hw_events.events = devm_kcalloc(dev,
CCI_PMU_MAX_HW_CNTRS(model),
sizeof(*cci_pmu->hw_events.events),
GFP_KERNEL);
if (!cci_pmu->hw_events.events)
return ERR_PTR(-ENOMEM);
cci_pmu->hw_events.used_mask = devm_kcalloc(dev,
BITS_TO_LONGS(CCI_PMU_MAX_HW_CNTRS(model)),
sizeof(*cci_pmu->hw_events.used_mask),
GFP_KERNEL);
if (!cci_pmu->hw_events.used_mask)
return ERR_PTR(-ENOMEM);
return cci_pmu;
}
static int cci_pmu_probe(struct platform_device *pdev)
{
struct resource *res;
struct cci_pmu *cci_pmu;
int i, ret, irq;
cci_pmu = cci_pmu_alloc(&pdev->dev);
if (IS_ERR(cci_pmu))
return PTR_ERR(cci_pmu);
res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
cci_pmu->base = devm_ioremap_resource(&pdev->dev, res);
if (IS_ERR(cci_pmu->base))
return -ENOMEM;
/*
* CCI PMU has one overflow interrupt per counter; but some may be tied
* together to a common interrupt.
*/
cci_pmu->nr_irqs = 0;
for (i = 0; i < CCI_PMU_MAX_HW_CNTRS(cci_pmu->model); i++) {
irq = platform_get_irq(pdev, i);
if (irq < 0)
break;
if (is_duplicate_irq(irq, cci_pmu->irqs, cci_pmu->nr_irqs))
continue;
cci_pmu->irqs[cci_pmu->nr_irqs++] = irq;
}
/*
* Ensure that the device tree has as many interrupts as the number
* of counters.
*/
if (i < CCI_PMU_MAX_HW_CNTRS(cci_pmu->model)) {
dev_warn(&pdev->dev, "In-correct number of interrupts: %d, should be %d\n",
i, CCI_PMU_MAX_HW_CNTRS(cci_pmu->model));
return -EINVAL;
}
raw_spin_lock_init(&cci_pmu->hw_events.pmu_lock);
mutex_init(&cci_pmu->reserve_mutex);
atomic_set(&cci_pmu->active_events, 0);
perf/arm-cci: Remove broken race mitigation Uncore PMU drivers face an awkward cyclic dependency wherein: - They have to pick a valid online CPU to associate with before registering the PMU device, since it will get exposed to userspace immediately. - The PMU registration has to be be at least partly complete before hotplug events can be handled, since trying to migrate an uninitialised context would be bad. - The hotplug handler has to be ready as soon as a CPU is chosen, lest it go offline without the user-visible cpumask value getting updated. The arm-cci driver has tried to solve this by using get_cpu() to pick the current CPU and prevent it from disappearing while both registrations are performed, but that results in taking mutexes with preemption disabled, which makes certain configurations very unhappy: [ 1.983337] BUG: sleeping function called from invalid context at kernel/locking/rtmutex.c:2004 [ 1.983340] in_atomic(): 1, irqs_disabled(): 0, pid: 1, name: swapper/0 [ 1.983342] Preemption disabled at: [ 1.983353] [<ffffff80089801f4>] cci_pmu_probe+0x1dc/0x488 [ 1.983360] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.18.20-rt8-yocto-preempt-rt #1 [ 1.983362] Hardware name: ZynqMP ZCU102 Rev1.0 (DT) [ 1.983364] Call trace: [ 1.983369] dump_backtrace+0x0/0x158 [ 1.983372] show_stack+0x24/0x30 [ 1.983378] dump_stack+0x80/0xa4 [ 1.983383] ___might_sleep+0x138/0x160 [ 1.983386] __might_sleep+0x58/0x90 [ 1.983391] __rt_mutex_lock_state+0x30/0xc0 [ 1.983395] _mutex_lock+0x24/0x30 [ 1.983400] perf_pmu_register+0x2c/0x388 [ 1.983404] cci_pmu_probe+0x2bc/0x488 [ 1.983409] platform_drv_probe+0x58/0xa8 It is not feasible to resolve all the possible races outside of the perf core itself, so address the immediate bug by following the example of nearly every other PMU driver and not even trying to do so. Registering the hotplug notifier first should minimise the window in which things can go wrong, so that's about as much as we can reasonably do here. This also revealed an additional race in assigning the global pointer too late relative to the hotplug notifier, which gets fixed in the process. Reported-by: Li, Meng <Meng.Li@windriver.com> Tested-by: Corentin Labbe <clabbe.montjoie@gmail.com> Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Robin Murphy <robin.murphy@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2019-04-16 23:24:24 +08:00
cci_pmu->cpu = raw_smp_processor_id();
g_cci_pmu = cci_pmu;
cpuhp_setup_state_nocalls(CPUHP_AP_PERF_ARM_CCI_ONLINE,
"perf/arm/cci:online", NULL,
cci_pmu_offline_cpu);
perf/arm-cci: Remove broken race mitigation Uncore PMU drivers face an awkward cyclic dependency wherein: - They have to pick a valid online CPU to associate with before registering the PMU device, since it will get exposed to userspace immediately. - The PMU registration has to be be at least partly complete before hotplug events can be handled, since trying to migrate an uninitialised context would be bad. - The hotplug handler has to be ready as soon as a CPU is chosen, lest it go offline without the user-visible cpumask value getting updated. The arm-cci driver has tried to solve this by using get_cpu() to pick the current CPU and prevent it from disappearing while both registrations are performed, but that results in taking mutexes with preemption disabled, which makes certain configurations very unhappy: [ 1.983337] BUG: sleeping function called from invalid context at kernel/locking/rtmutex.c:2004 [ 1.983340] in_atomic(): 1, irqs_disabled(): 0, pid: 1, name: swapper/0 [ 1.983342] Preemption disabled at: [ 1.983353] [<ffffff80089801f4>] cci_pmu_probe+0x1dc/0x488 [ 1.983360] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.18.20-rt8-yocto-preempt-rt #1 [ 1.983362] Hardware name: ZynqMP ZCU102 Rev1.0 (DT) [ 1.983364] Call trace: [ 1.983369] dump_backtrace+0x0/0x158 [ 1.983372] show_stack+0x24/0x30 [ 1.983378] dump_stack+0x80/0xa4 [ 1.983383] ___might_sleep+0x138/0x160 [ 1.983386] __might_sleep+0x58/0x90 [ 1.983391] __rt_mutex_lock_state+0x30/0xc0 [ 1.983395] _mutex_lock+0x24/0x30 [ 1.983400] perf_pmu_register+0x2c/0x388 [ 1.983404] cci_pmu_probe+0x2bc/0x488 [ 1.983409] platform_drv_probe+0x58/0xa8 It is not feasible to resolve all the possible races outside of the perf core itself, so address the immediate bug by following the example of nearly every other PMU driver and not even trying to do so. Registering the hotplug notifier first should minimise the window in which things can go wrong, so that's about as much as we can reasonably do here. This also revealed an additional race in assigning the global pointer too late relative to the hotplug notifier, which gets fixed in the process. Reported-by: Li, Meng <Meng.Li@windriver.com> Tested-by: Corentin Labbe <clabbe.montjoie@gmail.com> Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Robin Murphy <robin.murphy@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2019-04-16 23:24:24 +08:00
ret = cci_pmu_init(cci_pmu, pdev);
if (ret)
goto error_pmu_init;
pr_info("ARM %s PMU driver probed", cci_pmu->model->name);
return 0;
perf/arm-cci: Remove broken race mitigation Uncore PMU drivers face an awkward cyclic dependency wherein: - They have to pick a valid online CPU to associate with before registering the PMU device, since it will get exposed to userspace immediately. - The PMU registration has to be be at least partly complete before hotplug events can be handled, since trying to migrate an uninitialised context would be bad. - The hotplug handler has to be ready as soon as a CPU is chosen, lest it go offline without the user-visible cpumask value getting updated. The arm-cci driver has tried to solve this by using get_cpu() to pick the current CPU and prevent it from disappearing while both registrations are performed, but that results in taking mutexes with preemption disabled, which makes certain configurations very unhappy: [ 1.983337] BUG: sleeping function called from invalid context at kernel/locking/rtmutex.c:2004 [ 1.983340] in_atomic(): 1, irqs_disabled(): 0, pid: 1, name: swapper/0 [ 1.983342] Preemption disabled at: [ 1.983353] [<ffffff80089801f4>] cci_pmu_probe+0x1dc/0x488 [ 1.983360] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.18.20-rt8-yocto-preempt-rt #1 [ 1.983362] Hardware name: ZynqMP ZCU102 Rev1.0 (DT) [ 1.983364] Call trace: [ 1.983369] dump_backtrace+0x0/0x158 [ 1.983372] show_stack+0x24/0x30 [ 1.983378] dump_stack+0x80/0xa4 [ 1.983383] ___might_sleep+0x138/0x160 [ 1.983386] __might_sleep+0x58/0x90 [ 1.983391] __rt_mutex_lock_state+0x30/0xc0 [ 1.983395] _mutex_lock+0x24/0x30 [ 1.983400] perf_pmu_register+0x2c/0x388 [ 1.983404] cci_pmu_probe+0x2bc/0x488 [ 1.983409] platform_drv_probe+0x58/0xa8 It is not feasible to resolve all the possible races outside of the perf core itself, so address the immediate bug by following the example of nearly every other PMU driver and not even trying to do so. Registering the hotplug notifier first should minimise the window in which things can go wrong, so that's about as much as we can reasonably do here. This also revealed an additional race in assigning the global pointer too late relative to the hotplug notifier, which gets fixed in the process. Reported-by: Li, Meng <Meng.Li@windriver.com> Tested-by: Corentin Labbe <clabbe.montjoie@gmail.com> Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com> Signed-off-by: Robin Murphy <robin.murphy@arm.com> Signed-off-by: Will Deacon <will.deacon@arm.com>
2019-04-16 23:24:24 +08:00
error_pmu_init:
cpuhp_remove_state(CPUHP_AP_PERF_ARM_CCI_ONLINE);
g_cci_pmu = NULL;
return ret;
}
static int cci_pmu_remove(struct platform_device *pdev)
{
if (!g_cci_pmu)
return 0;
cpuhp_remove_state(CPUHP_AP_PERF_ARM_CCI_ONLINE);
perf_pmu_unregister(&g_cci_pmu->pmu);
g_cci_pmu = NULL;
return 0;
}
static struct platform_driver cci_pmu_driver = {
.driver = {
.name = DRIVER_NAME,
.of_match_table = arm_cci_pmu_matches,
},
.probe = cci_pmu_probe,
.remove = cci_pmu_remove,
};
module_platform_driver(cci_pmu_driver);
MODULE_LICENSE("GPL v2");
MODULE_DESCRIPTION("ARM CCI PMU support");