drm/xe/pmu: Remove PMU from Xe till uapi is finalized
authorAshutosh Dixit <ashutosh.dixit@intel.com>
Fri, 15 Dec 2023 15:45:37 +0000 (15:45 +0000)
committerRodrigo Vivi <rodrigo.vivi@intel.com>
Thu, 21 Dec 2023 16:46:59 +0000 (11:46 -0500)
PMU uapi is likely to change in the future. Till the uapi is finalized,
remove PMU from Xe. PMU can be re-added after uapi is finalized.

v2: Include xe_drm.h in xe/tests/xe_dma_buf.c (Francois)

Signed-off-by: Ashutosh Dixit <ashutosh.dixit@intel.com>
Acked-by: Aravind Iddamsetty <aravind.iddamsetty@linux.intel.com>
Acked-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
Acked-by: José Roberto de Souza <jose.souza@intel.com>
Acked-by: Mateusz Naklicki <mateusz.naklicki@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
drivers/gpu/drm/xe/Makefile
drivers/gpu/drm/xe/regs/xe_gt_regs.h
drivers/gpu/drm/xe/tests/xe_dma_buf.c
drivers/gpu/drm/xe/xe_device.c
drivers/gpu/drm/xe/xe_device_types.h
drivers/gpu/drm/xe/xe_gt.c
drivers/gpu/drm/xe/xe_module.c
drivers/gpu/drm/xe/xe_pmu.c [deleted file]
drivers/gpu/drm/xe/xe_pmu.h [deleted file]
drivers/gpu/drm/xe/xe_pmu_types.h [deleted file]
include/uapi/drm/xe_drm.h

index 6790c049d89eaa0d222c2c48d68ff8578169b4d0..53bd2a8ba1ae5cea2535c59047d028f18bec8e65 100644 (file)
@@ -276,8 +276,6 @@ xe-$(CONFIG_DRM_XE_DISPLAY) += \
        i915-display/skl_universal_plane.o \
        i915-display/skl_watermark.o
 
-xe-$(CONFIG_PERF_EVENTS) += xe_pmu.o
-
 ifeq ($(CONFIG_ACPI),y)
        xe-$(CONFIG_DRM_XE_DISPLAY) += \
                i915-display/intel_acpi.o \
index d7f52a634c11613626b8315c3a02b5b3275ded67..1dd361046b5dc028c8f68b0fd4c52a71c482a68d 100644 (file)
 #define   INVALIDATION_BROADCAST_MODE_DIS      REG_BIT(12)
 #define   GLOBAL_INVALIDATION_MODE             REG_BIT(2)
 
-#define XE_OAG_RC0_ANY_ENGINE_BUSY_FREE                XE_REG(0xdb80)
-#define XE_OAG_ANY_MEDIA_FF_BUSY_FREE          XE_REG(0xdba0)
-#define XE_OAG_BLT_BUSY_FREE                   XE_REG(0xdbbc)
-#define XE_OAG_RENDER_BUSY_FREE                        XE_REG(0xdbdc)
-
 #define HALF_SLICE_CHICKEN5                    XE_REG_MCR(0xe188, XE_REG_OPTION_MASKED)
 #define   DISABLE_SAMPLE_G_PERFORMANCE         REG_BIT(0)
 
index bb6f6424e06f623a94b0d94a7c4afd8d03485a56..9f6d571d7fa9cb794fb86fec1f326ed3adfb85b9 100644 (file)
@@ -3,6 +3,8 @@
  * Copyright © 2022 Intel Corporation
  */
 
+#include <drm/xe_drm.h>
+
 #include <kunit/test.h>
 #include <kunit/visibility.h>
 
index 221e875843521e96e3ff4ca5a2594bc729775208..d9ae77fe7382ddf9997858995fe255108f7c5944 100644 (file)
@@ -529,8 +529,6 @@ int xe_device_probe(struct xe_device *xe)
 
        xe_debugfs_register(xe);
 
-       xe_pmu_register(&xe->pmu);
-
        xe_hwmon_register(xe);
 
        err = drmm_add_action_or_reset(&xe->drm, xe_device_sanitize, xe);
index d1a48456e9a312ab682d5cdc54fdf750936b68ba..c45ef17b347323801d397a964e53aa3fc5b060f4 100644 (file)
@@ -18,7 +18,6 @@
 #include "xe_lmtt_types.h"
 #include "xe_platform_types.h"
 #include "xe_pt_types.h"
-#include "xe_pmu.h"
 #include "xe_sriov_types.h"
 #include "xe_step_types.h"
 
@@ -427,9 +426,6 @@ struct xe_device {
         */
        struct task_struct *pm_callback_task;
 
-       /** @pmu: performance monitoring unit */
-       struct xe_pmu pmu;
-
        /** @hwmon: hwmon subsystem integration */
        struct xe_hwmon *hwmon;
 
index dfd9cf01a5d581b8e930a367a255cdf7de9d4203..f5d18e98f8b6c7322a4a1c1b909e66b87e43b4af 100644 (file)
@@ -709,8 +709,6 @@ int xe_gt_suspend(struct xe_gt *gt)
        if (err)
                goto err_msg;
 
-       xe_pmu_suspend(gt);
-
        err = xe_uc_suspend(&gt->uc);
        if (err)
                goto err_force_wake;
index 51bf69b7ab222dda4b368f9a83e54eaf1bfb9c55..110b698646565ec80adbb46e83ba1dfb6bfa3c46 100644 (file)
@@ -11,7 +11,6 @@
 #include "xe_drv.h"
 #include "xe_hw_fence.h"
 #include "xe_pci.h"
-#include "xe_pmu.h"
 #include "xe_sched_job.h"
 
 struct xe_modparam xe_modparam = {
@@ -63,10 +62,6 @@ static const struct init_funcs init_funcs[] = {
                .init = xe_sched_job_module_init,
                .exit = xe_sched_job_module_exit,
        },
-       {
-               .init = xe_pmu_init,
-               .exit = xe_pmu_exit,
-       },
        {
                .init = xe_register_pci_driver,
                .exit = xe_unregister_pci_driver,
diff --git a/drivers/gpu/drm/xe/xe_pmu.c b/drivers/gpu/drm/xe/xe_pmu.c
deleted file mode 100644 (file)
index 9d0b788..0000000
+++ /dev/null
@@ -1,645 +0,0 @@
-// SPDX-License-Identifier: MIT
-/*
- * Copyright © 2023 Intel Corporation
- */
-
-#include <drm/drm_drv.h>
-#include <drm/drm_managed.h>
-#include <drm/xe_drm.h>
-
-#include "regs/xe_gt_regs.h"
-#include "xe_device.h"
-#include "xe_gt_clock.h"
-#include "xe_mmio.h"
-
-static cpumask_t xe_pmu_cpumask;
-static unsigned int xe_pmu_target_cpu = -1;
-
-static unsigned int config_gt_id(const u64 config)
-{
-       return config >> __DRM_XE_PMU_GT_SHIFT;
-}
-
-static u64 config_counter(const u64 config)
-{
-       return config & ~(~0ULL << __DRM_XE_PMU_GT_SHIFT);
-}
-
-static void xe_pmu_event_destroy(struct perf_event *event)
-{
-       struct xe_device *xe =
-               container_of(event->pmu, typeof(*xe), pmu.base);
-
-       drm_WARN_ON(&xe->drm, event->parent);
-
-       drm_dev_put(&xe->drm);
-}
-
-static u64 __engine_group_busyness_read(struct xe_gt *gt, int sample_type)
-{
-       u64 val;
-
-       switch (sample_type) {
-       case __XE_SAMPLE_RENDER_GROUP_BUSY:
-               val = xe_mmio_read32(gt, XE_OAG_RENDER_BUSY_FREE);
-               break;
-       case __XE_SAMPLE_COPY_GROUP_BUSY:
-               val = xe_mmio_read32(gt, XE_OAG_BLT_BUSY_FREE);
-               break;
-       case __XE_SAMPLE_MEDIA_GROUP_BUSY:
-               val = xe_mmio_read32(gt, XE_OAG_ANY_MEDIA_FF_BUSY_FREE);
-               break;
-       case __XE_SAMPLE_ANY_ENGINE_GROUP_BUSY:
-               val = xe_mmio_read32(gt, XE_OAG_RC0_ANY_ENGINE_BUSY_FREE);
-               break;
-       default:
-               drm_warn(&gt->tile->xe->drm, "unknown pmu event\n");
-       }
-
-       return xe_gt_clock_cycles_to_ns(gt, val * 16);
-}
-
-static u64 engine_group_busyness_read(struct xe_gt *gt, u64 config)
-{
-       int sample_type = config_counter(config);
-       const unsigned int gt_id = gt->info.id;
-       struct xe_device *xe = gt->tile->xe;
-       struct xe_pmu *pmu = &xe->pmu;
-       unsigned long flags;
-       bool device_awake;
-       u64 val;
-
-       device_awake = xe_device_mem_access_get_if_ongoing(xe);
-       if (device_awake) {
-               XE_WARN_ON(xe_force_wake_get(gt_to_fw(gt), XE_FW_GT));
-               val = __engine_group_busyness_read(gt, sample_type);
-               XE_WARN_ON(xe_force_wake_put(gt_to_fw(gt), XE_FW_GT));
-               xe_device_mem_access_put(xe);
-       }
-
-       spin_lock_irqsave(&pmu->lock, flags);
-
-       if (device_awake)
-               pmu->sample[gt_id][sample_type] = val;
-       else
-               val = pmu->sample[gt_id][sample_type];
-
-       spin_unlock_irqrestore(&pmu->lock, flags);
-
-       return val;
-}
-
-static void engine_group_busyness_store(struct xe_gt *gt)
-{
-       struct xe_pmu *pmu = &gt->tile->xe->pmu;
-       unsigned int gt_id = gt->info.id;
-       unsigned long flags;
-       int i;
-
-       spin_lock_irqsave(&pmu->lock, flags);
-
-       for (i = __XE_SAMPLE_RENDER_GROUP_BUSY; i <= __XE_SAMPLE_ANY_ENGINE_GROUP_BUSY; i++)
-               pmu->sample[gt_id][i] = __engine_group_busyness_read(gt, i);
-
-       spin_unlock_irqrestore(&pmu->lock, flags);
-}
-
-static int
-config_status(struct xe_device *xe, u64 config)
-{
-       unsigned int gt_id = config_gt_id(config);
-       struct xe_gt *gt = xe_device_get_gt(xe, gt_id);
-
-       if (gt_id >= XE_PMU_MAX_GT)
-               return -ENOENT;
-
-       switch (config_counter(config)) {
-       case DRM_XE_PMU_RENDER_GROUP_BUSY(0):
-       case DRM_XE_PMU_COPY_GROUP_BUSY(0):
-       case DRM_XE_PMU_ANY_ENGINE_GROUP_BUSY(0):
-               if (gt->info.type == XE_GT_TYPE_MEDIA)
-                       return -ENOENT;
-               break;
-       case DRM_XE_PMU_MEDIA_GROUP_BUSY(0):
-               if (!(gt->info.engine_mask & (BIT(XE_HW_ENGINE_VCS0) | BIT(XE_HW_ENGINE_VECS0))))
-                       return -ENOENT;
-               break;
-       default:
-               return -ENOENT;
-       }
-
-       return 0;
-}
-
-static int xe_pmu_event_init(struct perf_event *event)
-{
-       struct xe_device *xe =
-               container_of(event->pmu, typeof(*xe), pmu.base);
-       struct xe_pmu *pmu = &xe->pmu;
-       int ret;
-
-       if (pmu->closed)
-               return -ENODEV;
-
-       if (event->attr.type != event->pmu->type)
-               return -ENOENT;
-
-       /* unsupported modes and filters */
-       if (event->attr.sample_period) /* no sampling */
-               return -EINVAL;
-
-       if (has_branch_stack(event))
-               return -EOPNOTSUPP;
-
-       if (event->cpu < 0)
-               return -EINVAL;
-
-       /* only allow running on one cpu at a time */
-       if (!cpumask_test_cpu(event->cpu, &xe_pmu_cpumask))
-               return -EINVAL;
-
-       ret = config_status(xe, event->attr.config);
-       if (ret)
-               return ret;
-
-       if (!event->parent) {
-               drm_dev_get(&xe->drm);
-               event->destroy = xe_pmu_event_destroy;
-       }
-
-       return 0;
-}
-
-static u64 __xe_pmu_event_read(struct perf_event *event)
-{
-       struct xe_device *xe =
-               container_of(event->pmu, typeof(*xe), pmu.base);
-       const unsigned int gt_id = config_gt_id(event->attr.config);
-       const u64 config = event->attr.config;
-       struct xe_gt *gt = xe_device_get_gt(xe, gt_id);
-       u64 val;
-
-       switch (config_counter(config)) {
-       case DRM_XE_PMU_RENDER_GROUP_BUSY(0):
-       case DRM_XE_PMU_COPY_GROUP_BUSY(0):
-       case DRM_XE_PMU_ANY_ENGINE_GROUP_BUSY(0):
-       case DRM_XE_PMU_MEDIA_GROUP_BUSY(0):
-               val = engine_group_busyness_read(gt, config);
-               break;
-       default:
-               drm_warn(&gt->tile->xe->drm, "unknown pmu event\n");
-       }
-
-       return val;
-}
-
-static void xe_pmu_event_read(struct perf_event *event)
-{
-       struct xe_device *xe =
-               container_of(event->pmu, typeof(*xe), pmu.base);
-       struct hw_perf_event *hwc = &event->hw;
-       struct xe_pmu *pmu = &xe->pmu;
-       u64 prev, new;
-
-       if (pmu->closed) {
-               event->hw.state = PERF_HES_STOPPED;
-               return;
-       }
-again:
-       prev = local64_read(&hwc->prev_count);
-       new = __xe_pmu_event_read(event);
-
-       if (local64_cmpxchg(&hwc->prev_count, prev, new) != prev)
-               goto again;
-
-       local64_add(new - prev, &event->count);
-}
-
-static void xe_pmu_enable(struct perf_event *event)
-{
-       /*
-        * Store the current counter value so we can report the correct delta
-        * for all listeners. Even when the event was already enabled and has
-        * an existing non-zero value.
-        */
-       local64_set(&event->hw.prev_count, __xe_pmu_event_read(event));
-}
-
-static void xe_pmu_event_start(struct perf_event *event, int flags)
-{
-       struct xe_device *xe =
-               container_of(event->pmu, typeof(*xe), pmu.base);
-       struct xe_pmu *pmu = &xe->pmu;
-
-       if (pmu->closed)
-               return;
-
-       xe_pmu_enable(event);
-       event->hw.state = 0;
-}
-
-static void xe_pmu_event_stop(struct perf_event *event, int flags)
-{
-       if (flags & PERF_EF_UPDATE)
-               xe_pmu_event_read(event);
-
-       event->hw.state = PERF_HES_STOPPED;
-}
-
-static int xe_pmu_event_add(struct perf_event *event, int flags)
-{
-       struct xe_device *xe =
-               container_of(event->pmu, typeof(*xe), pmu.base);
-       struct xe_pmu *pmu = &xe->pmu;
-
-       if (pmu->closed)
-               return -ENODEV;
-
-       if (flags & PERF_EF_START)
-               xe_pmu_event_start(event, flags);
-
-       return 0;
-}
-
-static void xe_pmu_event_del(struct perf_event *event, int flags)
-{
-       xe_pmu_event_stop(event, PERF_EF_UPDATE);
-}
-
-static int xe_pmu_event_event_idx(struct perf_event *event)
-{
-       return 0;
-}
-
-struct xe_ext_attribute {
-       struct device_attribute attr;
-       unsigned long val;
-};
-
-static ssize_t xe_pmu_event_show(struct device *dev,
-                                struct device_attribute *attr, char *buf)
-{
-       struct xe_ext_attribute *eattr;
-
-       eattr = container_of(attr, struct xe_ext_attribute, attr);
-       return sprintf(buf, "config=0x%lx\n", eattr->val);
-}
-
-static ssize_t cpumask_show(struct device *dev,
-                           struct device_attribute *attr, char *buf)
-{
-       return cpumap_print_to_pagebuf(true, buf, &xe_pmu_cpumask);
-}
-
-static DEVICE_ATTR_RO(cpumask);
-
-static struct attribute *xe_cpumask_attrs[] = {
-       &dev_attr_cpumask.attr,
-       NULL,
-};
-
-static const struct attribute_group xe_pmu_cpumask_attr_group = {
-       .attrs = xe_cpumask_attrs,
-};
-
-#define __event(__counter, __name, __unit) \
-{ \
-       .counter = (__counter), \
-       .name = (__name), \
-       .unit = (__unit), \
-       .global = false, \
-}
-
-#define __global_event(__counter, __name, __unit) \
-{ \
-       .counter = (__counter), \
-       .name = (__name), \
-       .unit = (__unit), \
-       .global = true, \
-}
-
-static struct xe_ext_attribute *
-add_xe_attr(struct xe_ext_attribute *attr, const char *name, u64 config)
-{
-       sysfs_attr_init(&attr->attr.attr);
-       attr->attr.attr.name = name;
-       attr->attr.attr.mode = 0444;
-       attr->attr.show = xe_pmu_event_show;
-       attr->val = config;
-
-       return ++attr;
-}
-
-static struct perf_pmu_events_attr *
-add_pmu_attr(struct perf_pmu_events_attr *attr, const char *name,
-            const char *str)
-{
-       sysfs_attr_init(&attr->attr.attr);
-       attr->attr.attr.name = name;
-       attr->attr.attr.mode = 0444;
-       attr->attr.show = perf_event_sysfs_show;
-       attr->event_str = str;
-
-       return ++attr;
-}
-
-static struct attribute **
-create_event_attributes(struct xe_pmu *pmu)
-{
-       struct xe_device *xe = container_of(pmu, typeof(*xe), pmu);
-       static const struct {
-               unsigned int counter;
-               const char *name;
-               const char *unit;
-               bool global;
-       } events[] = {
-               __event(0, "render-group-busy", "ns"),
-               __event(1, "copy-group-busy", "ns"),
-               __event(2, "media-group-busy", "ns"),
-               __event(3, "any-engine-group-busy", "ns"),
-       };
-
-       struct perf_pmu_events_attr *pmu_attr = NULL, *pmu_iter;
-       struct xe_ext_attribute *xe_attr = NULL, *xe_iter;
-       struct attribute **attr = NULL, **attr_iter;
-       unsigned int count = 0;
-       unsigned int i, j;
-       struct xe_gt *gt;
-
-       /* Count how many counters we will be exposing. */
-       for_each_gt(gt, xe, j) {
-               for (i = 0; i < ARRAY_SIZE(events); i++) {
-                       u64 config = ___DRM_XE_PMU_OTHER(j, events[i].counter);
-
-                       if (!config_status(xe, config))
-                               count++;
-               }
-       }
-
-       /* Allocate attribute objects and table. */
-       xe_attr = kcalloc(count, sizeof(*xe_attr), GFP_KERNEL);
-       if (!xe_attr)
-               goto err_alloc;
-
-       pmu_attr = kcalloc(count, sizeof(*pmu_attr), GFP_KERNEL);
-       if (!pmu_attr)
-               goto err_alloc;
-
-       /* Max one pointer of each attribute type plus a termination entry. */
-       attr = kcalloc(count * 2 + 1, sizeof(*attr), GFP_KERNEL);
-       if (!attr)
-               goto err_alloc;
-
-       xe_iter = xe_attr;
-       pmu_iter = pmu_attr;
-       attr_iter = attr;
-
-       for_each_gt(gt, xe, j) {
-               for (i = 0; i < ARRAY_SIZE(events); i++) {
-                       u64 config = ___DRM_XE_PMU_OTHER(j, events[i].counter);
-                       char *str;
-
-                       if (config_status(xe, config))
-                               continue;
-
-                       if (events[i].global)
-                               str = kstrdup(events[i].name, GFP_KERNEL);
-                       else
-                               str = kasprintf(GFP_KERNEL, "%s-gt%u",
-                                               events[i].name, j);
-                       if (!str)
-                               goto err;
-
-                       *attr_iter++ = &xe_iter->attr.attr;
-                       xe_iter = add_xe_attr(xe_iter, str, config);
-
-                       if (events[i].unit) {
-                               if (events[i].global)
-                                       str = kasprintf(GFP_KERNEL, "%s.unit",
-                                                       events[i].name);
-                               else
-                                       str = kasprintf(GFP_KERNEL, "%s-gt%u.unit",
-                                                       events[i].name, j);
-                               if (!str)
-                                       goto err;
-
-                               *attr_iter++ = &pmu_iter->attr.attr;
-                               pmu_iter = add_pmu_attr(pmu_iter, str,
-                                                       events[i].unit);
-                       }
-               }
-       }
-
-       pmu->xe_attr = xe_attr;
-       pmu->pmu_attr = pmu_attr;
-
-       return attr;
-
-err:
-       for (attr_iter = attr; *attr_iter; attr_iter++)
-               kfree((*attr_iter)->name);
-
-err_alloc:
-       kfree(attr);
-       kfree(xe_attr);
-       kfree(pmu_attr);
-
-       return NULL;
-}
-
-static void free_event_attributes(struct xe_pmu *pmu)
-{
-       struct attribute **attr_iter = pmu->events_attr_group.attrs;
-
-       for (; *attr_iter; attr_iter++)
-               kfree((*attr_iter)->name);
-
-       kfree(pmu->events_attr_group.attrs);
-       kfree(pmu->xe_attr);
-       kfree(pmu->pmu_attr);
-
-       pmu->events_attr_group.attrs = NULL;
-       pmu->xe_attr = NULL;
-       pmu->pmu_attr = NULL;
-}
-
-static int xe_pmu_cpu_online(unsigned int cpu, struct hlist_node *node)
-{
-       struct xe_pmu *pmu = hlist_entry_safe(node, typeof(*pmu), cpuhp.node);
-
-       /* Select the first online CPU as a designated reader. */
-       if (cpumask_empty(&xe_pmu_cpumask))
-               cpumask_set_cpu(cpu, &xe_pmu_cpumask);
-
-       return 0;
-}
-
-static int xe_pmu_cpu_offline(unsigned int cpu, struct hlist_node *node)
-{
-       struct xe_pmu *pmu = hlist_entry_safe(node, typeof(*pmu), cpuhp.node);
-       unsigned int target = xe_pmu_target_cpu;
-
-       /*
-        * Unregistering an instance generates a CPU offline event which we must
-        * ignore to avoid incorrectly modifying the shared xe_pmu_cpumask.
-        */
-       if (pmu->closed)
-               return 0;
-
-       if (cpumask_test_and_clear_cpu(cpu, &xe_pmu_cpumask)) {
-               target = cpumask_any_but(topology_sibling_cpumask(cpu), cpu);
-
-               /* Migrate events if there is a valid target */
-               if (target < nr_cpu_ids) {
-                       cpumask_set_cpu(target, &xe_pmu_cpumask);
-                       xe_pmu_target_cpu = target;
-               }
-       }
-
-       if (target < nr_cpu_ids && target != pmu->cpuhp.cpu) {
-               perf_pmu_migrate_context(&pmu->base, cpu, target);
-               pmu->cpuhp.cpu = target;
-       }
-
-       return 0;
-}
-
-static enum cpuhp_state cpuhp_slot = CPUHP_INVALID;
-
-int xe_pmu_init(void)
-{
-       int ret;
-
-       ret = cpuhp_setup_state_multi(CPUHP_AP_ONLINE_DYN,
-                                     "perf/x86/intel/xe:online",
-                                     xe_pmu_cpu_online,
-                                     xe_pmu_cpu_offline);
-       if (ret < 0)
-               pr_notice("Failed to setup cpuhp state for xe PMU! (%d)\n",
-                         ret);
-       else
-               cpuhp_slot = ret;
-
-       return 0;
-}
-
-void xe_pmu_exit(void)
-{
-       if (cpuhp_slot != CPUHP_INVALID)
-               cpuhp_remove_multi_state(cpuhp_slot);
-}
-
-static int xe_pmu_register_cpuhp_state(struct xe_pmu *pmu)
-{
-       if (cpuhp_slot == CPUHP_INVALID)
-               return -EINVAL;
-
-       return cpuhp_state_add_instance(cpuhp_slot, &pmu->cpuhp.node);
-}
-
-static void xe_pmu_unregister_cpuhp_state(struct xe_pmu *pmu)
-{
-       cpuhp_state_remove_instance(cpuhp_slot, &pmu->cpuhp.node);
-}
-
-void xe_pmu_suspend(struct xe_gt *gt)
-{
-       engine_group_busyness_store(gt);
-}
-
-static void xe_pmu_unregister(struct drm_device *device, void *arg)
-{
-       struct xe_pmu *pmu = arg;
-
-       if (!pmu->base.event_init)
-               return;
-
-       /*
-        * "Disconnect" the PMU callbacks - since all are atomic synchronize_rcu
-        * ensures all currently executing ones will have exited before we
-        * proceed with unregistration.
-        */
-       pmu->closed = true;
-       synchronize_rcu();
-
-       xe_pmu_unregister_cpuhp_state(pmu);
-
-       perf_pmu_unregister(&pmu->base);
-       pmu->base.event_init = NULL;
-       kfree(pmu->base.attr_groups);
-       kfree(pmu->name);
-       free_event_attributes(pmu);
-}
-
-void xe_pmu_register(struct xe_pmu *pmu)
-{
-       struct xe_device *xe = container_of(pmu, typeof(*xe), pmu);
-       const struct attribute_group *attr_groups[] = {
-               &pmu->events_attr_group,
-               &xe_pmu_cpumask_attr_group,
-               NULL
-       };
-
-       int ret = -ENOMEM;
-
-       spin_lock_init(&pmu->lock);
-       pmu->cpuhp.cpu = -1;
-
-       pmu->name = kasprintf(GFP_KERNEL,
-                             "xe_%s",
-                             dev_name(xe->drm.dev));
-       if (pmu->name)
-               /* tools/perf reserves colons as special. */
-               strreplace((char *)pmu->name, ':', '_');
-
-       if (!pmu->name)
-               goto err;
-
-       pmu->events_attr_group.name = "events";
-       pmu->events_attr_group.attrs = create_event_attributes(pmu);
-       if (!pmu->events_attr_group.attrs)
-               goto err_name;
-
-       pmu->base.attr_groups = kmemdup(attr_groups, sizeof(attr_groups),
-                                       GFP_KERNEL);
-       if (!pmu->base.attr_groups)
-               goto err_attr;
-
-       pmu->base.module        = THIS_MODULE;
-       pmu->base.task_ctx_nr   = perf_invalid_context;
-       pmu->base.event_init    = xe_pmu_event_init;
-       pmu->base.add           = xe_pmu_event_add;
-       pmu->base.del           = xe_pmu_event_del;
-       pmu->base.start         = xe_pmu_event_start;
-       pmu->base.stop          = xe_pmu_event_stop;
-       pmu->base.read          = xe_pmu_event_read;
-       pmu->base.event_idx     = xe_pmu_event_event_idx;
-
-       ret = perf_pmu_register(&pmu->base, pmu->name, -1);
-       if (ret)
-               goto err_groups;
-
-       ret = xe_pmu_register_cpuhp_state(pmu);
-       if (ret)
-               goto err_unreg;
-
-       ret = drmm_add_action_or_reset(&xe->drm, xe_pmu_unregister, pmu);
-       if (ret)
-               goto err_cpuhp;
-
-       return;
-
-err_cpuhp:
-       xe_pmu_unregister_cpuhp_state(pmu);
-err_unreg:
-       perf_pmu_unregister(&pmu->base);
-err_groups:
-       kfree(pmu->base.attr_groups);
-err_attr:
-       pmu->base.event_init = NULL;
-       free_event_attributes(pmu);
-err_name:
-       kfree(pmu->name);
-err:
-       drm_notice(&xe->drm, "Failed to register PMU!\n");
-}
diff --git a/drivers/gpu/drm/xe/xe_pmu.h b/drivers/gpu/drm/xe/xe_pmu.h
deleted file mode 100644 (file)
index a99d4dd..0000000
+++ /dev/null
@@ -1,25 +0,0 @@
-/* SPDX-License-Identifier: MIT */
-/*
- * Copyright © 2023 Intel Corporation
- */
-
-#ifndef _XE_PMU_H_
-#define _XE_PMU_H_
-
-#include "xe_gt_types.h"
-#include "xe_pmu_types.h"
-
-#if IS_ENABLED(CONFIG_PERF_EVENTS)
-int xe_pmu_init(void);
-void xe_pmu_exit(void);
-void xe_pmu_register(struct xe_pmu *pmu);
-void xe_pmu_suspend(struct xe_gt *gt);
-#else
-static inline int xe_pmu_init(void) { return 0; }
-static inline void xe_pmu_exit(void) {}
-static inline void xe_pmu_register(struct xe_pmu *pmu) {}
-static inline void xe_pmu_suspend(struct xe_gt *gt) {}
-#endif
-
-#endif
-
diff --git a/drivers/gpu/drm/xe/xe_pmu_types.h b/drivers/gpu/drm/xe/xe_pmu_types.h
deleted file mode 100644 (file)
index 9cadbd2..0000000
+++ /dev/null
@@ -1,68 +0,0 @@
-/* SPDX-License-Identifier: MIT */
-/*
- * Copyright © 2023 Intel Corporation
- */
-
-#ifndef _XE_PMU_TYPES_H_
-#define _XE_PMU_TYPES_H_
-
-#include <linux/perf_event.h>
-#include <linux/spinlock_types.h>
-#include <uapi/drm/xe_drm.h>
-
-enum {
-       __XE_SAMPLE_RENDER_GROUP_BUSY,
-       __XE_SAMPLE_COPY_GROUP_BUSY,
-       __XE_SAMPLE_MEDIA_GROUP_BUSY,
-       __XE_SAMPLE_ANY_ENGINE_GROUP_BUSY,
-       __XE_NUM_PMU_SAMPLERS
-};
-
-#define XE_PMU_MAX_GT 2
-
-struct xe_pmu {
-       /**
-        * @cpuhp: Struct used for CPU hotplug handling.
-        */
-       struct {
-               struct hlist_node node;
-               unsigned int cpu;
-       } cpuhp;
-       /**
-        * @base: PMU base.
-        */
-       struct pmu base;
-       /**
-        * @closed: xe is unregistering.
-        */
-       bool closed;
-       /**
-        * @name: Name as registered with perf core.
-        */
-       const char *name;
-       /**
-        * @lock: Lock protecting enable mask and ref count handling.
-        */
-       spinlock_t lock;
-       /**
-        * @sample: Current and previous (raw) counters.
-        *
-        * These counters are updated when the device is awake.
-        *
-        */
-       u64 sample[XE_PMU_MAX_GT][__XE_NUM_PMU_SAMPLERS];
-       /**
-        * @events_attr_group: Device events attribute group.
-        */
-       struct attribute_group events_attr_group;
-       /**
-        * @xe_attr: Memory block holding device attributes.
-        */
-       void *xe_attr;
-       /**
-        * @pmu_attr: Memory block holding device attributes.
-        */
-       void *pmu_attr;
-};
-
-#endif
index d122f985435a00444d757556532c31246782e723..e1e8fb1846eac7176897345c4b4f2fba5971b56c 100644 (file)
@@ -1074,46 +1074,6 @@ struct drm_xe_wait_user_fence {
        /** @reserved: Reserved */
        __u64 reserved[2];
 };
-
-/**
- * DOC: XE PMU event config IDs
- *
- * Check 'man perf_event_open' to use the ID's DRM_XE_PMU_XXXX listed in xe_drm.h
- * in 'struct perf_event_attr' as part of perf_event_open syscall to read a
- * particular event.
- *
- * For example to open the DRMXE_PMU_RENDER_GROUP_BUSY(0):
- *
- * .. code-block:: C
- *
- *     struct perf_event_attr attr;
- *     long long count;
- *     int cpu = 0;
- *     int fd;
- *
- *     memset(&attr, 0, sizeof(struct perf_event_attr));
- *     attr.type = type; // eg: /sys/bus/event_source/devices/xe_0000_56_00.0/type
- *     attr.read_format = PERF_FORMAT_TOTAL_TIME_ENABLED;
- *     attr.use_clockid = 1;
- *     attr.clockid = CLOCK_MONOTONIC;
- *     attr.config = DRM_XE_PMU_RENDER_GROUP_BUSY(0);
- *
- *     fd = syscall(__NR_perf_event_open, &attr, -1, cpu, -1, 0);
- */
-
-/*
- * Top bits of every counter are GT id.
- */
-#define __DRM_XE_PMU_GT_SHIFT (56)
-
-#define ___DRM_XE_PMU_OTHER(gt, x) \
-       (((__u64)(x)) | ((__u64)(gt) << __DRM_XE_PMU_GT_SHIFT))
-
-#define DRM_XE_PMU_RENDER_GROUP_BUSY(gt)       ___DRM_XE_PMU_OTHER(gt, 0)
-#define DRM_XE_PMU_COPY_GROUP_BUSY(gt)         ___DRM_XE_PMU_OTHER(gt, 1)
-#define DRM_XE_PMU_MEDIA_GROUP_BUSY(gt)                ___DRM_XE_PMU_OTHER(gt, 2)
-#define DRM_XE_PMU_ANY_ENGINE_GROUP_BUSY(gt)   ___DRM_XE_PMU_OTHER(gt, 3)
-
 #if defined(__cplusplus)
 }
 #endif