drm/i915/pmu: Expose a PMU interface for perf queries

From: Chris Wilson <chris@chris-wilson.co.uk>
From: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
From: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>

The first goal is to be able to measure GPU (and invidual ring) busyness
without having to poll registers from userspace. (Which not only incurs
holding the forcewake lock indefinitely, perturbing the system, but also
runs the risk of hanging the machine.) As an alternative we can use the
perf event counter interface to sample the ring registers periodically
and send those results to userspace.

Functionality we are exporting to userspace is via the existing perf PMU
API and can be exercised via the existing tools. For example:

  perf stat -a -e i915/rcs0-busy/ -I 1000

Will print the render engine busynnes once per second. All the performance
counters can be enumerated (perf list) and have their unit of measure
correctly reported in sysfs.

v1-v2 (Chris Wilson):

v2: Use a common timer for the ring sampling.

v3: (Tvrtko Ursulin)
 * Decouple uAPI from i915 engine ids.
 * Complete uAPI defines.
 * Refactor some code to helpers for clarity.
 * Skip sampling disabled engines.
 * Expose counters in sysfs.
 * Pass in fake regs to avoid null ptr deref in perf core.
 * Convert to class/instance uAPI.
 * Use shared driver code for rc6 residency, power and frequency.

v4: (Dmitry Rogozhkin)
 * Register PMU with .task_ctx_nr=perf_invalid_context
 * Expose cpumask for the PMU with the single CPU in the mask
 * Properly support pmu->stop(): it should call pmu->read()
 * Properly support pmu->del(): it should call stop(event, PERF_EF_UPDATE)
 * Introduce refcounting of event subscriptions.
 * Make pmu.busy_stats a refcounter to avoid busy stats going away
   with some deleted event.
 * Expose cpumask for i915 PMU to avoid multiple events creation of
   the same type followed by counter aggregation by perf-stat.
 * Track CPUs getting online/offline to migrate perf context. If (likely)
   cpumask will initially set CPU0, CONFIG_BOOTPARAM_HOTPLUG_CPU0 will be
   needed to see effect of CPU status tracking.
 * End result is that only global events are supported and perf stat
   works correctly.
 * Deny perf driver level sampling - it is prohibited for uncore PMU.

v5: (Tvrtko Ursulin)

 * Don't hardcode number of engine samplers.
 * Rewrite event ref-counting for correctness and simplicity.
 * Store initial counter value when starting already enabled events
   to correctly report values to all listeners.
 * Fix RC6 residency readout.
 * Comments, GPL header.

v6:
 * Add missing entry to v4 changelog.
 * Fix accounting in CPU hotplug case by copying the approach from
   arch/x86/events/intel/cstate.c. (Dmitry Rogozhkin)

v7:
 * Log failure message only on failure.
 * Remove CPU hotplug notification state on unregister.

v8:
 * Fix error unwind on failed registration.
 * Checkpatch cleanup.

v9:
 * Drop the energy metric, it is available via intel_rapl_perf.
   (Ville Syrjälä)
 * Use HAS_RC6(p). (Chris Wilson)
 * Handle unsupported non-engine events. (Dmitry Rogozhkin)
 * Rebase for intel_rc6_residency_ns needing caller managed
   runtime pm.
 * Drop HAS_RC6 checks from the read callback since creating those
   events will be rejected at init time already.
 * Add counter units to sysfs so perf stat output is nicer.
 * Cleanup the attribute tables for brevity and readability.

v10:
 * Fixed queued accounting.

v11:
 * Move intel_engine_lookup_user to intel_engine_cs.c
 * Commit update. (Joonas Lahtinen)

v12:
 * More accurate sampling. (Chris Wilson)
 * Store and report frequency in MHz for better usability from
   perf stat.
 * Removed metrics: queued, interrupts, rc6 counters.
 * Sample engine busyness based on seqno difference only
   for less MMIO (and forcewake) on all platforms. (Chris Wilson)

v13:
 * Comment spelling, use mul_u32_u32 to work around potential GCC
   issue and somne code alignment changes. (Chris Wilson)

v14:
 * Rebase.

v15:
 * Rebase for RPS refactoring.

v16:
 * Use the dynamic slot in the CPU hotplug state machine so that we are
   free to setup our state as multi-instance. Previously we were re-using
   the CPUHP_AP_PERF_X86_UNCORE_ONLINE slot which is neither used as
   multi-instance, nor owned by our driver to start with.
 * Register the CPU hotplug handlers after the PMU, otherwise the callback
   will get called before the PMU is initialized which can end up in
   perf_pmu_migrate_context with an un-initialized base.
 * Added workaround for a probable bug in cpuhp core.

v17:
 * Remove workaround for the cpuhp bug.

v18:
 * Rebase for drm_i915_gem_engine_class getting upstream before us.

v19:
 * Rebase. (trivial)
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Dmitry Rogozhkin <dmitry.v.rogozhkin@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20171121181852.16128-2-tvrtko.ursulin@linux.intel.com

drivers/gpu/drm/i915/Makefile

@@ -45,6 +45,7 @@ i915-y := i915_drv.o \
 
 i915-$(CONFIG_COMPAT)   += i915_ioc32.o
 i915-$(CONFIG_DEBUG_FS) += i915_debugfs.o intel_pipe_crc.o
+i915-$(CONFIG_PERF_EVENTS) += i915_pmu.o
 
 # GEM code
 i915-y += i915_cmd_parser.o \

drivers/gpu/drm/i915/i915_drv.c

@@ -48,6 +48,7 @@
 
 #include "i915_drv.h"
 #include "i915_trace.h"
+#include "i915_pmu.h"
 #include "i915_vgpu.h"
 #include "intel_drv.h"
 #include "intel_uc.h"
@@ -1215,6 +1216,7 @@ static void i915_driver_register(struct drm_i915_private *dev_priv)
 	struct drm_device *dev = &dev_priv->drm;
 
 	i915_gem_shrinker_init(dev_priv);
+	i915_pmu_register(dev_priv);
 
 	/*
 	 * Notify a valid surface after modesetting,
@@ -1269,6 +1271,7 @@ static void i915_driver_unregister(struct drm_i915_private *dev_priv)
 	intel_opregion_unregister(dev_priv);
 
 	i915_perf_unregister(dev_priv);
+	i915_pmu_unregister(dev_priv);
 
 	i915_teardown_sysfs(dev_priv);
 	i915_guc_log_unregister(dev_priv);

drivers/gpu/drm/i915/i915_drv.h

@@ -40,6 +40,7 @@
 #include <linux/hash.h>
 #include <linux/intel-iommu.h>
 #include <linux/kref.h>
+#include <linux/perf_event.h>
 #include <linux/pm_qos.h>
 #include <linux/reservation.h>
 #include <linux/shmem_fs.h>
@@ -2290,6 +2291,8 @@ struct drm_i915_private {
 	struct i915_gem_context *kernel_context;
 	/* Context only to be used for injecting preemption commands */
 	struct i915_gem_context *preempt_context;
+	struct intel_engine_cs *engine_class[MAX_ENGINE_CLASS + 1]
+					    [MAX_ENGINE_INSTANCE + 1];
 
 	struct drm_dma_handle *status_page_dmah;
 	struct resource mch_res;
@@ -2761,6 +2764,8 @@ struct drm_i915_private {
 		int	irq;
 	} lpe_audio;
 
+	struct i915_pmu pmu;
+
 	/*
 	 * NOTE: This is the dri1/ums dungeon, don't add stuff here. Your patch
 	 * will be rejected. Instead look for a better place.

drivers/gpu/drm/i915/i915_pmu.h

+/*
+ * Copyright © 2017 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ *
+ */
+#ifndef __I915_PMU_H__
+#define __I915_PMU_H__
+
+enum {
+	__I915_SAMPLE_FREQ_ACT = 0,
+	__I915_SAMPLE_FREQ_REQ,
+	__I915_NUM_PMU_SAMPLERS
+};
+
+/**
+ * How many different events we track in the global PMU mask.
+ *
+ * It is also used to know to needed number of event reference counters.
+ */
+#define I915_PMU_MASK_BITS \
+	((1 << I915_PMU_SAMPLE_BITS) + \
+	 (I915_PMU_LAST + 1 - __I915_PMU_OTHER(0)))
+
+struct i915_pmu_sample {
+	u64 cur;
+	u32 prev;
+};
+
+struct i915_pmu {
+	/**
+	 * @node: List node for CPU hotplug handling.
+	 */
+	struct hlist_node node;
+	/**
+	 * @base: PMU base.
+	 */
+	struct pmu base;
+	/**
+	 * @lock: Lock protecting enable mask and ref count handling.
+	 */
+	spinlock_t lock;
+	/**
+	 * @timer: Timer for internal i915 PMU sampling.
+	 */
+	struct hrtimer timer;
+	/**
+	 * @enable: Bitmask of all currently enabled events.
+	 *
+	 * Bits are derived from uAPI event numbers in a way that low 16 bits
+	 * correspond to engine event _sample_ _type_ (I915_SAMPLE_QUEUED is
+	 * bit 0), and higher bits correspond to other events (for instance
+	 * I915_PMU_ACTUAL_FREQUENCY is bit 16 etc).
+	 *
+	 * In other words, low 16 bits are not per engine but per engine
+	 * sampler type, while the upper bits are directly mapped to other
+	 * event types.
+	 */
+	u64 enable;
+	/**
+	 * @enable_count: Reference counts for the enabled events.
+	 *
+	 * Array indices are mapped in the same way as bits in the @enable field
+	 * and they are used to control sampling on/off when multiple clients
+	 * are using the PMU API.
+	 */
+	unsigned int enable_count[I915_PMU_MASK_BITS];
+	/**
+	 * @sample: Current and previous (raw) counters for sampling events.
+	 *
+	 * These counters are updated from the i915 PMU sampling timer.
+	 *
+	 * Only global counters are held here, while the per-engine ones are in
+	 * struct intel_engine_cs.
+	 */
+	struct i915_pmu_sample sample[__I915_NUM_PMU_SAMPLERS];
+};
+
+#ifdef CONFIG_PERF_EVENTS
+void i915_pmu_register(struct drm_i915_private *i915);
+void i915_pmu_unregister(struct drm_i915_private *i915);
+#else
+static inline void i915_pmu_register(struct drm_i915_private *i915) {}
+static inline void i915_pmu_unregister(struct drm_i915_private *i915) {}
+#endif
+
+#endif

drivers/gpu/drm/i915/i915_reg.h

@@ -186,6 +186,9 @@ static inline bool i915_mmio_reg_valid(i915_reg_t reg)
 #define VIDEO_ENHANCEMENT_CLASS	2
 #define COPY_ENGINE_CLASS	3
 #define OTHER_CLASS		4
+#define MAX_ENGINE_CLASS	4
+
+#define MAX_ENGINE_INSTANCE    1
 
 /* PCI config space */
 

drivers/gpu/drm/i915/intel_engine_cs.c

@@ -205,6 +205,15 @@ intel_engine_setup(struct drm_i915_private *dev_priv,
 	GEM_BUG_ON(info->class >= ARRAY_SIZE(intel_engine_classes));
 	class_info = &intel_engine_classes[info->class];
 
+	if (GEM_WARN_ON(info->class > MAX_ENGINE_CLASS))
+		return -EINVAL;
+
+	if (GEM_WARN_ON(info->instance > MAX_ENGINE_INSTANCE))
+		return -EINVAL;
+
+	if (GEM_WARN_ON(dev_priv->engine_class[info->class][info->instance]))
+		return -EINVAL;
+
 	GEM_BUG_ON(dev_priv->engine[id]);
 	engine = kzalloc(sizeof(*engine), GFP_KERNEL);
 	if (!engine)
@@ -234,6 +243,7 @@ intel_engine_setup(struct drm_i915_private *dev_priv,
 
 	ATOMIC_INIT_NOTIFIER_HEAD(&engine->context_status_notifier);
 
+	dev_priv->engine_class[info->class][info->instance] = engine;
 	dev_priv->engine[id] = engine;
 	return 0;
 }
@@ -1816,6 +1826,29 @@ void intel_engine_dump(struct intel_engine_cs *engine, struct drm_printer *m)
 	drm_printf(m, "\n");
 }
 
+static u8 user_class_map[] = {
+	[I915_ENGINE_CLASS_RENDER] = RENDER_CLASS,
+	[I915_ENGINE_CLASS_COPY] = COPY_ENGINE_CLASS,
+	[I915_ENGINE_CLASS_VIDEO] = VIDEO_DECODE_CLASS,
+	[I915_ENGINE_CLASS_VIDEO_ENHANCE] = VIDEO_ENHANCEMENT_CLASS,
+};
+
+struct intel_engine_cs *
+intel_engine_lookup_user(struct drm_i915_private *i915, u8 class, u8 instance)
+{
+	if (class >= ARRAY_SIZE(user_class_map))
+		return NULL;
+
+	class = user_class_map[class];
+
+	GEM_BUG_ON(class > MAX_ENGINE_CLASS);
+
+	if (instance > MAX_ENGINE_INSTANCE)
+		return NULL;
+
+	return i915->engine_class[class][instance];
+}
+
 #if IS_ENABLED(CONFIG_DRM_I915_SELFTEST)
 #include "selftests/mock_engine.c"
 #endif

drivers/gpu/drm/i915/intel_ringbuffer.h

@@ -5,6 +5,7 @@
 #include "i915_gem_batch_pool.h"
 #include "i915_gem_request.h"
 #include "i915_gem_timeline.h"
+#include "i915_pmu.h"
 #include "i915_selftest.h"
 
 struct drm_printer;
@@ -338,6 +339,28 @@ struct intel_engine_cs {
 		I915_SELFTEST_DECLARE(bool mock : 1);
 	} breadcrumbs;
 
+	struct {
+		/**
+		 * @enable: Bitmask of enable sample events on this engine.
+		 *
+		 * Bits correspond to sample event types, for instance
+		 * I915_SAMPLE_QUEUED is bit 0 etc.
+		 */
+		u32 enable;
+		/**
+		 * @enable_count: Reference count for the enabled samplers.
+		 *
+		 * Index number corresponds to the bit number from @enable.
+		 */
+		unsigned int enable_count[I915_PMU_SAMPLE_BITS];
+		/**
+		 * @sample: Counter values for sampling events.
+		 *
+		 * Our internal timer stores the current counters in this field.
+		 */
+		struct i915_pmu_sample sample[I915_ENGINE_SAMPLE_MAX];
+	} pmu;
+
 	/*
 	 * A pool of objects to use as shadow copies of client batch buffers
 	 * when the command parser is enabled. Prevents the client from
@@ -926,4 +949,7 @@ bool intel_engine_can_store_dword(struct intel_engine_cs *engine);
 
 void intel_engine_dump(struct intel_engine_cs *engine, struct drm_printer *p);
 
+struct intel_engine_cs *
+intel_engine_lookup_user(struct drm_i915_private *i915, u8 class, u8 instance);
+
 #endif /* _INTEL_RINGBUFFER_H_ */

include/uapi/drm/i915_drm.h

@@ -102,6 +102,45 @@ enum drm_i915_gem_engine_class {
 	I915_ENGINE_CLASS_INVALID	= -1
 };
 
+/**
+ * DOC: perf_events exposed by i915 through /sys/bus/event_sources/drivers/i915
+ *
+ */
+
+enum drm_i915_pmu_engine_sample {
+	I915_SAMPLE_BUSY = 0,
+	I915_SAMPLE_WAIT = 1,
+	I915_SAMPLE_SEMA = 2,
+	I915_ENGINE_SAMPLE_MAX /* non-ABI */
+};
+
+#define I915_PMU_SAMPLE_BITS (4)
+#define I915_PMU_SAMPLE_MASK (0xf)
+#define I915_PMU_SAMPLE_INSTANCE_BITS (8)
+#define I915_PMU_CLASS_SHIFT \
+	(I915_PMU_SAMPLE_BITS + I915_PMU_SAMPLE_INSTANCE_BITS)
+
+#define __I915_PMU_ENGINE(class, instance, sample) \
+	((class) << I915_PMU_CLASS_SHIFT | \
+	(instance) << I915_PMU_SAMPLE_BITS | \
+	(sample))
+
+#define I915_PMU_ENGINE_BUSY(class, instance) \
+	__I915_PMU_ENGINE(class, instance, I915_SAMPLE_BUSY)
+
+#define I915_PMU_ENGINE_WAIT(class, instance) \
+	__I915_PMU_ENGINE(class, instance, I915_SAMPLE_WAIT)
+
+#define I915_PMU_ENGINE_SEMA(class, instance) \
+	__I915_PMU_ENGINE(class, instance, I915_SAMPLE_SEMA)
+
+#define __I915_PMU_OTHER(x) (__I915_PMU_ENGINE(0xff, 0xff, 0xf) + 1 + (x))
+
+#define I915_PMU_ACTUAL_FREQUENCY	__I915_PMU_OTHER(0)
+#define I915_PMU_REQUESTED_FREQUENCY	__I915_PMU_OTHER(1)
+
+#define I915_PMU_LAST I915_PMU_REQUESTED_FREQUENCY
+
 /* Each region is a minimum of 16k, and there are at most 255 of them.
  */
 #define I915_NR_TEX_REGIONS 255	/* table size 2k - maximum due to use