[PATCH 2/2] IOMMU/AMD: IOMMU PC PERF uncore PMU implementation
From: steven.kinney
Date: Mon May 13 2013 - 17:44:43 EST
From: Steven L Kinney <Steven.Kinney@xxxxxxx>
Implement a perf PMU to handle IOMMU PC perf events. This PMU will handle
static counter perf events relative to the AMD IOMMU Performance Counters.
To invoke the AMD IOMMU PMU issue a perf tool command such as:
./perf stat -e amd_iommu/config=<config-data>,config1=<config1-data>/u <command>
For example:
./perf stat -e amd_iommu/config=0x8000000000000005,config1=0/u <command>
The resulting count will be how many IOMMU total peripheral memory
operations were performed during the command execution window.
Signed-off-by: Steven Kinney <steven.kinney@xxxxxxx>
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@xxxxxxx>
---
arch/x86/kernel/cpu/Makefile | 2 +-
arch/x86/kernel/cpu/perf_event_amd_iommu.c | 409 ++++++++++++++++++++++++++
arch/x86/kernel/cpu/perf_event_amd_iommu.h | 55 ++++
3 files changed, 465 insertions(+), 1 deletion(-)
create mode 100644 arch/x86/kernel/cpu/perf_event_amd_iommu.c
create mode 100644 arch/x86/kernel/cpu/perf_event_amd_iommu.h
diff --git a/arch/x86/kernel/cpu/Makefile b/arch/x86/kernel/cpu/Makefile
index b0684e4..2a3b676 100644
--- a/arch/x86/kernel/cpu/Makefile
+++ b/arch/x86/kernel/cpu/Makefile
@@ -30,7 +30,7 @@ obj-$(CONFIG_CPU_SUP_UMC_32) += umc.o
obj-$(CONFIG_PERF_EVENTS) += perf_event.o
ifdef CONFIG_PERF_EVENTS
-obj-$(CONFIG_CPU_SUP_AMD) += perf_event_amd.o perf_event_amd_uncore.o
+obj-$(CONFIG_CPU_SUP_AMD) += perf_event_amd.o perf_event_amd_uncore.o perf_event_amd_iommu.o
obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_p6.o perf_event_knc.o perf_event_p4.o
obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_intel_lbr.o perf_event_intel_ds.o perf_event_intel.o
obj-$(CONFIG_CPU_SUP_INTEL) += perf_event_intel_uncore.o
diff --git a/arch/x86/kernel/cpu/perf_event_amd_iommu.c b/arch/x86/kernel/cpu/perf_event_amd_iommu.c
new file mode 100644
index 0000000..117a9aa
--- /dev/null
+++ b/arch/x86/kernel/cpu/perf_event_amd_iommu.c
@@ -0,0 +1,409 @@
+/*
+ * Copyright (C) 2013 Advanced Micro Devices, Inc.
+ *
+ * Author: Steven Kinney <Steven.Kinney@xxxxxxx>
+ * Author: Suravee Suthikulpanit <Suraveee.Suthikulpanit@xxxxxxx>
+ *
+ * Performance event - AMD IOMMU Performance Counter
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/perf_event.h>
+#include <linux/module.h>
+#include <linux/pci.h>
+#include <linux/ptrace.h>
+
+#include "perf_event.h"
+#include "perf_event_amd_iommu.h"
+
+#define _GET_DEVID(ev) ((u16)((ev->hw.config \
+ & IOMMU_PC_DEVICEID_MATCH) >> 8))
+#define _GET_BANK(ev) ((u8)(ev->hw.extra_reg.reg >> 8))
+#define _GET_CNTR(ev) ((u8)(ev->hw.extra_reg.reg))
+
+/* define iommu pc states based on ibs, add states if needed */
+enum iommu_pc_states {
+ IOMMU_PC_ENABLED = 0,
+ IOMMU_PC_STARTED = 1,
+ IOMMU_PC_STOPPING = 2,
+ IOMMU_PC_MAX_STATES,
+};
+
+
+struct cpu_perf_iommu {
+ struct perf_event *event;
+ unsigned long state[BITS_TO_LONGS(IOMMU_PC_MAX_STATES)];
+};
+
+
+struct perf_iommu {
+ struct pmu pmu;
+ struct attribute **format_attrs;
+ struct attribute_group format_group;
+ const struct attribute_group *attr_groups[11];
+ u16 iommu_devid;
+ u8 max_banks;
+ u8 max_counters;
+
+ struct cpu_perf_iommu __percpu *pcpu;
+};
+
+
+PMU_FORMAT_ATTR(iommu_raw, "config:63");
+PMU_FORMAT_ATTR(csource, "config:7-0");
+PMU_FORMAT_ATTR(deviceid, "config:23-8");
+PMU_FORMAT_ATTR(pasid, "config:39-24");
+PMU_FORMAT_ATTR(domain, "config:55-40");
+PMU_FORMAT_ATTR(en_deviceid_filter, "config:56");
+PMU_FORMAT_ATTR(en_pasid_filter, "config:57");
+PMU_FORMAT_ATTR(en_domain_filter, "config:58");
+PMU_FORMAT_ATTR(deviceid_mask, "config1:15-0");
+PMU_FORMAT_ATTR(pasid_mask, "config1:31-16");
+PMU_FORMAT_ATTR(domain_mask, "config1:47-32");
+
+
+static struct attribute *iommu_fetch_format_attrs[] = {
+ &format_attr_iommu_raw.attr,
+ &format_attr_csource.attr,
+ &format_attr_deviceid.attr,
+ &format_attr_pasid.attr,
+ &format_attr_domain.attr,
+ &format_attr_en_deviceid_filter.attr,
+ &format_attr_en_pasid_filter.attr,
+ &format_attr_en_domain_filter.attr,
+ &format_attr_deviceid_mask.attr,
+ &format_attr_pasid_mask.attr,
+ &format_attr_domain_mask.attr,
+ NULL,
+};
+
+
+static u64 cntr_assign_mask;
+
+static struct perf_iommu perf_iommu_fetch;
+
+
+static u16 get_next_avail_iommu_bnk_cntr(struct perf_iommu *perf_iommu)
+{
+ int bank_index, cntr_index;
+ int max_banks, max_cntrs;
+ int shift = 0;
+ int bank_offset = 0;
+ u16 retval;
+
+ max_banks = perf_iommu->max_banks;
+ max_cntrs = perf_iommu->max_counters;
+
+ for (bank_index = 0; bank_index < max_banks; bank_index++) {
+ for (cntr_index = 0; cntr_index < max_cntrs; cntr_index++) {
+ shift = bank_index + cntr_index + bank_offset;
+ if (cntr_assign_mask & (1ULL<<shift))
+ continue;
+ else {
+ cntr_assign_mask |= (1ULL<<shift);
+ retval = ((u16)((u16)bank_index<<8) |
+ (u8)(cntr_index));
+ goto out;
+ }
+ }
+ bank_offset += 3;
+ }
+ retval = -EINVAL;
+out:
+ return retval;
+}
+
+
+static int clear_avail_iommu_bnk_cntr(struct perf_iommu *perf_iommu,
+ u8 bank, u8 cntr)
+{
+ int max_banks, max_cntrs;
+ int shift = 0;
+
+ max_banks = perf_iommu->max_banks;
+ max_cntrs = perf_iommu->max_counters;
+
+ if ((bank > max_banks) || (cntr > max_cntrs))
+ return -EINVAL;
+
+ shift = bank + cntr + (bank*3);
+ cntr_assign_mask &= ~(1ULL<<shift);
+
+ return 0;
+}
+
+
+static int perf_iommu_init(struct perf_event *event)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct perf_iommu *perf_iommu;
+ u64 config, config1;
+
+ /* initialize the cntr_assign_mask */
+ cntr_assign_mask = 0;
+
+ /* test the event attr type check for PMU enumeration */
+ if (event->attr.type != event->pmu->type)
+ return -ENOENT;
+
+ perf_iommu = &perf_iommu_fetch;
+
+ if (perf_iommu) {
+ config = event->attr.config;
+ config1 = event->attr.config1;
+ } else
+ return -EINVAL;
+
+ if (event->pmu != &perf_iommu->pmu)
+ return -ENOENT;
+
+ /* make sure the event->attr.config is raw (vendor specific) */
+ if (!(config & IOMMU_PC_RAW_CONFIG_MASK))
+ return -EINVAL;
+
+ /* integrate with iommu base devid (0000), assume one iommu */
+ perf_iommu->max_banks =
+ amd_iommu_pc_get_max_banks(IOMMU_BASE_DEVID);
+ perf_iommu->max_counters =
+ amd_iommu_pc_get_max_counters(IOMMU_BASE_DEVID);
+
+ /* update the hw_perf_event struct with the iommu config data */
+ hwc->config = config;
+ hwc->extra_reg.config = config1;
+
+ return 0;
+}
+
+
+static void perf_iommu_enable_event(struct perf_event *event)
+{
+ u64 csource = (u64)(event->hw.config & IOMMU_PC_CSOURCE);
+
+ amd_iommu_pc_get_set_reg_val(_GET_DEVID(event),
+ _GET_BANK(event),
+ _GET_CNTR(event) ,
+ IOMMU_PC_COUNTER_SRC_REG,
+ &csource,
+ true);
+}
+
+
+static void perf_iommu_start(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct perf_iommu *perf_iommu =
+ container_of(event->pmu, struct perf_iommu, pmu);
+ struct cpu_perf_iommu *pcpu = this_cpu_ptr(perf_iommu->pcpu);
+
+ if (WARN_ON_ONCE(!(hwc->state & PERF_HES_STOPPED)))
+ return;
+
+ WARN_ON_ONCE(!(hwc->state & PERF_HES_UPTODATE));
+ hwc->state = 0;
+
+ if (flags & PERF_EF_RELOAD) {
+ u64 prev_raw_count = local64_read(&hwc->prev_count);
+
+ amd_iommu_pc_get_set_reg_val(_GET_DEVID(event),
+ _GET_BANK(event),
+ _GET_CNTR(event),
+ IOMMU_PC_COUNTER_REG,
+ &prev_raw_count,
+ true);
+ }
+
+ set_bit(IOMMU_PC_STARTED, pcpu->state);
+ perf_iommu_enable_event(event);
+ perf_event_update_userpage(event);
+
+}
+
+
+static void perf_iommu_event_update(struct perf_event *event)
+{
+ u64 count;
+ u64 prev_raw_count;
+ u64 delta;
+ struct hw_perf_event *hwc = &event->hw;
+
+ amd_iommu_pc_get_set_reg_val(_GET_DEVID(event),
+ _GET_BANK(event),
+ _GET_CNTR(event),
+ IOMMU_PC_COUNTER_REG,
+ &count,
+ false);
+
+ prev_raw_count = local64_read(&hwc->prev_count);
+ if (local64_cmpxchg(&hwc->prev_count, prev_raw_count,
+ count) != prev_raw_count)
+ return;
+
+ delta = count - prev_raw_count;
+ local64_add(delta, &event->count);
+
+}
+
+
+static void perf_iommu_disable_event(struct perf_event *event)
+{
+ u64 val = PC_CSOURCE_DISABLE_CNT;
+
+ amd_iommu_pc_get_set_reg_val(_GET_DEVID(event),
+ _GET_BANK(event),
+ _GET_CNTR(event),
+ IOMMU_PC_COUNTER_SRC_REG,
+ &val,
+ true);
+}
+
+
+static void perf_iommu_stop(struct perf_event *event, int flags)
+{
+ struct hw_perf_event *hwc = &event->hw;
+ struct perf_iommu *perf_iommu =
+ container_of(event->pmu, struct perf_iommu, pmu);
+ struct cpu_perf_iommu *pcpu = this_cpu_ptr(perf_iommu->pcpu);
+ int stopping;
+ u64 config;
+
+ stopping = test_and_clear_bit(IOMMU_PC_STARTED, pcpu->state);
+
+ if (!stopping && (hwc->state & PERF_HES_UPTODATE))
+ return;
+
+ if (stopping) {
+ set_bit(IOMMU_PC_STOPPING, pcpu->state);
+ perf_iommu_disable_event(event);
+ WARN_ON_ONCE(hwc->state & PERF_HES_STOPPED);
+ hwc->state |= PERF_HES_STOPPED;
+ }
+
+ if (hwc->state & PERF_HES_UPTODATE)
+ return;
+
+ config = hwc->config;
+ perf_iommu_event_update(event);
+ hwc->state |= PERF_HES_UPTODATE;
+}
+
+
+static int perf_iommu_add(struct perf_event *event, int flags)
+{
+ struct perf_iommu *perf_iommu =
+ container_of(event->pmu, struct perf_iommu, pmu);
+ struct cpu_perf_iommu *pcpu = this_cpu_ptr(perf_iommu->pcpu);
+
+ if (test_and_set_bit(IOMMU_PC_ENABLED, pcpu->state))
+ return -ENOSPC;
+
+ event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED;
+
+ pcpu->event = event;
+
+ /* request an iommu bank/counter */
+ event->hw.extra_reg.reg = get_next_avail_iommu_bnk_cntr(perf_iommu);
+
+ if (flags & PERF_EF_START)
+ perf_iommu_start(event, PERF_EF_RELOAD);
+
+ return 0;
+}
+
+
+static void perf_iommu_del(struct perf_event *event, int flags)
+{
+ struct perf_iommu *perf_iommu =
+ container_of(event->pmu, struct perf_iommu, pmu);
+ struct cpu_perf_iommu *pcpu = this_cpu_ptr(perf_iommu->pcpu);
+
+ if (!test_and_clear_bit(IOMMU_PC_ENABLED, pcpu->state))
+ return;
+
+ perf_iommu_stop(event, PERF_EF_UPDATE);
+
+ /* clear the assigned iommu bank/counter */
+ clear_avail_iommu_bnk_cntr(perf_iommu,
+ _GET_BANK(event),
+ _GET_CNTR(event));
+ pcpu->event = NULL;
+
+ perf_event_update_userpage(event);
+}
+
+
+static void perf_iommu_read(struct perf_event *event)
+{
+ pr_debug("perf: AMD IOMMU: perf_iommu_read called\n");
+}
+
+
+static struct perf_iommu perf_iommu_fetch = {
+ .pmu = {
+ .event_init = perf_iommu_init,
+ .add = perf_iommu_add,
+ .del = perf_iommu_del,
+ .start = perf_iommu_start,
+ .stop = perf_iommu_stop,
+ .read = perf_iommu_read,
+ },
+ .format_attrs = iommu_fetch_format_attrs,
+ .max_banks = 0x00,
+ .max_counters = 0x00,
+};
+
+
+static __init int perf_iommu_pmu_init(struct perf_iommu *perf_iommu,
+ char *name)
+{
+ struct cpu_perf_iommu __percpu *pcpu;
+ int ret;
+
+ pcpu = alloc_percpu(struct cpu_perf_iommu);
+ if (!pcpu)
+ return -ENOMEM;
+
+ perf_iommu->pcpu = pcpu;
+
+ /* allocate and register atttributes */
+ if (perf_iommu->format_attrs[0]) {
+ memset(&perf_iommu->format_group, 0,
+ sizeof(perf_iommu->format_group));
+ perf_iommu->format_group.name = "format";
+ perf_iommu->format_group.attrs = perf_iommu->format_attrs;
+ memset(&perf_iommu->attr_groups, 0,
+ sizeof(perf_iommu->attr_groups));
+ perf_iommu->attr_groups[0] = &perf_iommu->format_group;
+ perf_iommu->pmu.attr_groups = perf_iommu->attr_groups;
+ }
+
+ ret = perf_pmu_register(&perf_iommu->pmu, name, -1);
+ if (ret) {
+ pr_err("perf: AMD IOMMU PMU failed to initialized.\n");
+ perf_iommu->pcpu = NULL;
+ free_percpu(pcpu);
+ }
+
+ pr_info("perf: AMD IOMMU PMU detected. (%d banks, %d counters/bank)\n",
+ amd_iommu_pc_get_max_banks(IOMMU_BASE_DEVID),
+ amd_iommu_pc_get_max_counters(IOMMU_BASE_DEVID));
+
+ return ret;
+}
+
+static __init int amd_iommu_pmc_init(void)
+{
+ /* Make sure the IOMMU PC resource is available */
+ if (!amd_iommu_pc_supported()) {
+ pr_err("perf: AMD IOMMU PMU not installed. No support!\n");
+ return -ENODEV;
+ }
+
+ perf_iommu_pmu_init(&perf_iommu_fetch, "amd_iommu");
+
+ return 0;
+}
+
+device_initcall(amd_iommu_pmc_init);
diff --git a/arch/x86/kernel/cpu/perf_event_amd_iommu.h b/arch/x86/kernel/cpu/perf_event_amd_iommu.h
new file mode 100644
index 0000000..d832033
--- /dev/null
+++ b/arch/x86/kernel/cpu/perf_event_amd_iommu.h
@@ -0,0 +1,55 @@
+/*
+ * Copyright (C) 2013 Advanced Micro Devices, Inc.
+ *
+ * Author: Steven Kinney <Steven.Kinney@xxxxxxx>
+ * Author: Suravee Suthikulpanit <Suraveee.Suthikulpanit@xxxxxxx>
+ *
+ * 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.
+ */
+
+#ifndef _PERF_EVENT_AMD_IOMMU_H_
+#define _PERF_EVENT_AMD_IOMMU_H_
+
+/* iommu pc mmio region register indexes */
+#define IOMMU_PC_COUNTER_REG 0x00
+#define IOMMU_PC_COUNTER_SRC_REG 0x08
+#define IOMMU_PC_PASID_MATCH_REG 0x10
+#define IOMMU_PC_DOMAIN_MATCH_REG 0x18
+#define IOMMU_PC_DEVICEID_MATCH_REG 0x20
+#define IOMMU_PC_COUNTER_REPORT_REG 0x28
+
+/* iommu pc csource register values */
+#define PC_CSOURCE_DISABLE_CNT 0ULL
+
+/* maximun specified bank/counters */
+#define PC_MAX_SPEC_BNKS 64
+#define PC_MAX_SPEC_CNTRS 16
+
+/* iommu pmu config masks */
+#define IOMMU_PC_RAW_CONFIG_MASK (1ULL << 63)
+#define IOMMU_PC_CSOURCE (0xFFULL)
+#define IOMMU_PC_DEVICEID_MATCH (0xFFFFULL << 8)
+#define IOMMU_PC_PASID_MATCH (0xFFFFULL << 24)
+#define IOMMU_PC_DOMAIN_MATCH (0xFFFFULL << 40)
+#define IOMMU_PC_EN_DEVICEID (1ULL << 56)
+#define IOMMU_PC_EN_PASID (1ULL << 57)
+#define IOMMU_PC_EN_DOMAIN (1ULL << 58)
+#define IOMMU_PC_DEVICEID_MASK (0xFFFFULL)
+#define IOMMU_PC_PASID_MASK (0xFFFFULL << 16)
+#define IOMMU_PC_DOMAIN_MASK (0xFFFFULL << 32)
+
+#define IOMMU_BASE_DEVID 0x0000
+
+/* amd_iommu_init.c external support functions */
+extern bool amd_iommu_pc_supported(void);
+
+extern u8 amd_iommu_pc_get_max_banks(u16 devid);
+
+extern u8 amd_iommu_pc_get_max_counters(u16 devid);
+
+extern int amd_iommu_pc_get_set_reg_val(u16 devid, u8 bank, u8 cntr,
+ u8 fxn, long long *value, bool is_write);
+
+#endif /*_PERF_EVENT_AMD_IOMMU_H_*/
--
1.7.10.4
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/