drm/i915: Enable i915 perf stream for Haswell OA unit

Gen graphics hardware can be set up to periodically write snapshots of
performance counters into a circular buffer via its Observation
Architecture and this patch exposes that capability to userspace via the
i915 perf interface.

v2:
   Make sure to initialize ->specific_ctx_id when opening, without
   relying on _pin_notify hook, in case ctx already pinned.
v3:
   Revert back to pinning ctx upfront when opening stream, removing
   need to hook in to pinning and to update OACONTROL on the fly.
Signed-off-by: Robert Bragg <robert@sixbynine.org>
Signed-off-by: Zhenyu Wang <zhenyuw@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Sourab Gupta <sourab.gupta@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: http://patchwork.freedesktop.org/patch/msgid/20161107194957.3385-7-robert@sixbynine.org

drivers/gpu/drm/i915/i915_drv.h

@@ -1797,6 +1797,11 @@ struct intel_wm_config {
 	bool sprites_scaled;
 };
 
+struct i915_oa_format {
+	u32 format;
+	int size;
+};
+
 struct i915_oa_reg {
 	i915_reg_t addr;
 	u32 value;
@@ -1817,11 +1822,6 @@ struct i915_perf_stream_ops {
 	 */
 	void (*disable)(struct i915_perf_stream *stream);
 
-	/* Return: true if any i915 perf records are ready to read()
-	 * for this stream.
-	 */
-	bool (*can_read)(struct i915_perf_stream *stream);
-
 	/* Call poll_wait, passing a wait queue that will be woken
 	 * once there is something ready to read() for the stream
 	 */
@@ -1831,9 +1831,7 @@ struct i915_perf_stream_ops {
 
 	/* For handling a blocking read, wait until there is something
 	 * to ready to read() for the stream. E.g. wait on the same
-	 * wait queue that would be passed to poll_wait() until
-	 * ->can_read() returns true (if its safe to call ->can_read()
-	 * without the i915 perf lock held).
+	 * wait queue that would be passed to poll_wait().
 	 */
 	int (*wait_unlocked)(struct i915_perf_stream *stream);
 
@@ -1873,11 +1871,28 @@ struct i915_perf_stream {
 	struct list_head link;
 
 	u32 sample_flags;
+	int sample_size;
 
 	struct i915_gem_context *ctx;
 	bool enabled;
 
-	struct i915_perf_stream_ops *ops;
+	const struct i915_perf_stream_ops *ops;
+};
+
+struct i915_oa_ops {
+	void (*init_oa_buffer)(struct drm_i915_private *dev_priv);
+	int (*enable_metric_set)(struct drm_i915_private *dev_priv);
+	void (*disable_metric_set)(struct drm_i915_private *dev_priv);
+	void (*oa_enable)(struct drm_i915_private *dev_priv);
+	void (*oa_disable)(struct drm_i915_private *dev_priv);
+	void (*update_oacontrol)(struct drm_i915_private *dev_priv);
+	void (*update_hw_ctx_id_locked)(struct drm_i915_private *dev_priv,
+					u32 ctx_id);
+	int (*read)(struct i915_perf_stream *stream,
+		    char __user *buf,
+		    size_t count,
+		    size_t *offset);
+	bool (*oa_buffer_is_empty)(struct drm_i915_private *dev_priv);
 };
 
 struct drm_i915_private {
@@ -2177,16 +2192,47 @@ struct drm_i915_private {
 
 	struct {
 		bool initialized;
+
 		struct mutex lock;
 		struct list_head streams;
 
+		spinlock_t hook_lock;
+
 		struct {
-			u32 metrics_set;
+			struct i915_perf_stream *exclusive_stream;
+
+			u32 specific_ctx_id;
+			struct i915_vma *pinned_rcs_vma;
+
+			struct hrtimer poll_check_timer;
+			wait_queue_head_t poll_wq;
+			bool pollin;
+
+			bool periodic;
+			int period_exponent;
+			int timestamp_frequency;
+
+			int tail_margin;
+
+			int metrics_set;
 
 			const struct i915_oa_reg *mux_regs;
 			int mux_regs_len;
 			const struct i915_oa_reg *b_counter_regs;
 			int b_counter_regs_len;
+
+			struct {
+				struct i915_vma *vma;
+				u8 *vaddr;
+				int format;
+				int format_size;
+			} oa_buffer;
+
+			u32 gen7_latched_oastatus1;
+
+			struct i915_oa_ops ops;
+			const struct i915_oa_format *oa_formats;
+			int n_builtin_sets;
 		} oa;
 	} perf;
 

include/uapi/drm/i915_drm.h

@@ -1230,6 +1230,18 @@ struct drm_i915_gem_context_param {
 	__u64 value;
 };
 
+enum drm_i915_oa_format {
+	I915_OA_FORMAT_A13 = 1,
+	I915_OA_FORMAT_A29,
+	I915_OA_FORMAT_A13_B8_C8,
+	I915_OA_FORMAT_B4_C8,
+	I915_OA_FORMAT_A45_B8_C8,
+	I915_OA_FORMAT_B4_C8_A16,
+	I915_OA_FORMAT_C4_B8,
+
+	I915_OA_FORMAT_MAX	    /* non-ABI */
+};
+
 enum drm_i915_perf_property_id {
 	/**
 	 * Open the stream for a specific context handle (as used with
@@ -1238,6 +1250,32 @@ enum drm_i915_perf_property_id {
 	 */
 	DRM_I915_PERF_PROP_CTX_HANDLE = 1,
 
+	/**
+	 * A value of 1 requests the inclusion of raw OA unit reports as
+	 * part of stream samples.
+	 */
+	DRM_I915_PERF_PROP_SAMPLE_OA,
+
+	/**
+	 * The value specifies which set of OA unit metrics should be
+	 * be configured, defining the contents of any OA unit reports.
+	 */
+	DRM_I915_PERF_PROP_OA_METRICS_SET,
+
+	/**
+	 * The value specifies the size and layout of OA unit reports.
+	 */
+	DRM_I915_PERF_PROP_OA_FORMAT,
+
+	/**
+	 * Specifying this property implicitly requests periodic OA unit
+	 * sampling and (at least on Haswell) the sampling frequency is derived
+	 * from this exponent as follows:
+	 *
+	 *   80ns * 2^(period_exponent + 1)
+	 */
+	DRM_I915_PERF_PROP_OA_EXPONENT,
+
 	DRM_I915_PERF_PROP_MAX /* non-ABI */
 };
 
@@ -1257,7 +1295,23 @@ struct drm_i915_perf_open_param {
 	__u64 __user properties_ptr;
 };
 
+/**
+ * Enable data capture for a stream that was either opened in a disabled state
+ * via I915_PERF_FLAG_DISABLED or was later disabled via
+ * I915_PERF_IOCTL_DISABLE.
+ *
+ * It is intended to be cheaper to disable and enable a stream than it may be
+ * to close and re-open a stream with the same configuration.
+ *
+ * It's undefined whether any pending data for the stream will be lost.
+ */
 #define I915_PERF_IOCTL_ENABLE	_IO('i', 0x0)
+
+/**
+ * Disable data capture for a stream.
+ *
+ * It is an error to try and read a stream that is disabled.
+ */
 #define I915_PERF_IOCTL_DISABLE	_IO('i', 0x1)
 
 /**
@@ -1281,17 +1335,30 @@ enum drm_i915_perf_record_type {
 	 * every sample.
 	 *
 	 * The order of these sample properties given by userspace has no
-	 * affect on the ordering of data within a sample. The order will be
+	 * affect on the ordering of data within a sample. The order is
 	 * documented here.
 	 *
 	 * struct {
 	 *     struct drm_i915_perf_record_header header;
 	 *
-	 *     TODO: itemize extensible sample data here
+	 *     { u32 oa_report[]; } && DRM_I915_PERF_PROP_SAMPLE_OA
 	 * };
 	 */
 	DRM_I915_PERF_RECORD_SAMPLE = 1,
 
+	/*
+	 * Indicates that one or more OA reports were not written by the
+	 * hardware. This can happen for example if an MI_REPORT_PERF_COUNT
+	 * command collides with periodic sampling - which would be more likely
+	 * at higher sampling frequencies.
+	 */
+	DRM_I915_PERF_RECORD_OA_REPORT_LOST = 2,
+
+	/**
+	 * An error occurred that resulted in all pending OA reports being lost.
+	 */
+	DRM_I915_PERF_RECORD_OA_BUFFER_LOST = 3,
+
 	DRM_I915_PERF_RECORD_MAX /* non-ABI */
 };