#define GEN9_RPSTAT0_PREV_GT_FREQ_SHIFT 0
#define GEN9_RPSTAT0_PREV_GT_FREQ_MASK INTEL_MASK(8, 0)
+#define GEN6_SO_PRIM_STORAGE_NEEDED 0x2280
+#define GEN7_SO_PRIM_STORAGE_NEEDED(n) (0x5240 + (n) * 8)
+#define GEN6_SO_NUM_PRIMS_WRITTEN 0x2288
+#define GEN7_SO_NUM_PRIMS_WRITTEN(n) (0x5200 + (n) * 8)
+
#define MAP_READ (1 << 0)
#define MAP_WRITE (1 << 1)
+/**
+ * Periodic OA samples are read() into these buffer structures via the
+ * i915 perf kernel interface and appended to the
+ * perf_ctx->sample_buffers linked list. When we process the
+ * results of an OA metrics query we need to consider all the periodic
+ * samples between the Begin and End MI_REPORT_PERF_COUNT command
+ * markers.
+ *
+ * 'Periodic' is a simplification as there are other automatic reports
+ * written by the hardware also buffered here.
+ *
+ * Considering three queries, A, B and C:
+ *
+ * Time ---->
+ * ________________A_________________
+ * | |
+ * | ________B_________ _____C___________
+ * | | | | | |
+ *
+ * And an illustration of sample buffers read over this time frame:
+ * [HEAD ][ ][ ][ ][ ][ ][ ][ ][TAIL ]
+ *
+ * These nodes may hold samples for query A:
+ * [ ][ ][ A ][ A ][ A ][ A ][ A ][ ][ ]
+ *
+ * These nodes may hold samples for query B:
+ * [ ][ ][ B ][ B ][ B ][ ][ ][ ][ ]
+ *
+ * These nodes may hold samples for query C:
+ * [ ][ ][ ][ ][ ][ C ][ C ][ C ][ ]
+ *
+ * The illustration assumes we have an even distribution of periodic
+ * samples so all nodes have the same size plotted against time:
+ *
+ * Note, to simplify code, the list is never empty.
+ *
+ * With overlapping queries we can see that periodic OA reports may
+ * relate to multiple queries and care needs to be take to keep
+ * track of sample buffers until there are no queries that might
+ * depend on their contents.
+ *
+ * We use a node ref counting system where a reference ensures that a
+ * node and all following nodes can't be freed/recycled until the
+ * reference drops to zero.
+ *
+ * E.g. with a ref of one here:
+ * [ 0 ][ 0 ][ 1 ][ 0 ][ 0 ][ 0 ][ 0 ][ 0 ][ 0 ]
+ *
+ * These nodes could be freed or recycled ("reaped"):
+ * [ 0 ][ 0 ]
+ *
+ * These must be preserved until the leading ref drops to zero:
+ * [ 1 ][ 0 ][ 0 ][ 0 ][ 0 ][ 0 ][ 0 ]
+ *
+ * When a query starts we take a reference on the current tail of
+ * the list, knowing that no already-buffered samples can possibly
+ * relate to the newly-started query. A pointer to this node is
+ * also saved in the query object's ->oa.samples_head.
+ *
+ * E.g. starting query A while there are two nodes in .sample_buffers:
+ * ________________A________
+ * |
+ *
+ * [ 0 ][ 1 ]
+ * ^_______ Add a reference and store pointer to node in
+ * A->oa.samples_head
+ *
+ * Moving forward to when the B query starts with no new buffer nodes:
+ * (for reference, i915 perf reads() are only done when queries finish)
+ * ________________A_______
+ * | ________B___
+ * | |
+ *
+ * [ 0 ][ 2 ]
+ * ^_______ Add a reference and store pointer to
+ * node in B->oa.samples_head
+ *
+ * Once a query is finished, after an OA query has become 'Ready',
+ * once the End OA report has landed and after we we have processed
+ * all the intermediate periodic samples then we drop the
+ * ->oa.samples_head reference we took at the start.
+ *
+ * So when the B query has finished we have:
+ * ________________A________
+ * | ______B___________
+ * | | |
+ * [ 0 ][ 1 ][ 0 ][ 0 ][ 0 ]
+ * ^_______ Drop B->oa.samples_head reference
+ *
+ * We still can't free these due to the A->oa.samples_head ref:
+ * [ 1 ][ 0 ][ 0 ][ 0 ]
+ *
+ * When the A query finishes: (note there's a new ref for C's samples_head)
+ * ________________A_________________
+ * | |
+ * | _____C_________
+ * | | |
+ * [ 0 ][ 0 ][ 0 ][ 0 ][ 1 ][ 0 ][ 0 ]
+ * ^_______ Drop A->oa.samples_head reference
+ *
+ * And we can now reap these nodes up to the C->oa.samples_head:
+ * [ X ][ X ][ X ][ X ]
+ * keeping -> [ 1 ][ 0 ][ 0 ]
+ *
+ * We reap old sample buffers each time we finish processing an OA
+ * query by iterating the sample_buffers list from the head until we
+ * find a referenced node and stop.
+ *
+ * Reaped buffers move to a perfquery.free_sample_buffers list and
+ * when we come to read() we first look to recycle a buffer from the
+ * free_sample_buffers list before allocating a new buffer.
+ */
+struct oa_sample_buf {
+ struct exec_node link;
+ int refcount;
+ int len;
+ uint8_t buf[I915_PERF_OA_SAMPLE_SIZE * 10];
+ uint32_t last_timestamp;
+};
+
+struct gen_perf_query_object *
+gen_perf_new_query(struct gen_perf_context *perf_ctx, unsigned query_index)
+{
+ const struct gen_perf_query_info *query =
+ &perf_ctx->perf->queries[query_index];
+ struct gen_perf_query_object *obj =
+ calloc(1, sizeof(struct gen_perf_query_object));
+
+ if (!obj)
+ return NULL;
+
+ obj->queryinfo = query;
+
+ perf_ctx->n_query_instances++;
+ return obj;
+}
+
static bool
get_sysfs_dev_dir(struct gen_perf_config *perf, int fd)
{
return read_file_uint64(buf, value);
}
+static inline struct gen_perf_query_info *
+append_query_info(struct gen_perf_config *perf, int max_counters)
+{
+ struct gen_perf_query_info *query;
+
+ perf->queries = reralloc(perf, perf->queries,
+ struct gen_perf_query_info,
+ ++perf->n_queries);
+ query = &perf->queries[perf->n_queries - 1];
+ memset(query, 0, sizeof(*query));
+
+ if (max_counters > 0) {
+ query->max_counters = max_counters;
+ query->counters =
+ rzalloc_array(perf, struct gen_perf_query_counter, max_counters);
+ }
+
+ return query;
+}
+
static void
register_oa_config(struct gen_perf_config *perf,
const struct gen_perf_query_info *query,
uint64_t config_id)
{
- struct gen_perf_query_info *registred_query =
- gen_perf_query_append_query_info(perf, 0);
+ struct gen_perf_query_info *registred_query = append_query_info(perf, 0);
*registred_query = *query;
registred_query->oa_metrics_set_id = config_id;
&invalid_config_id) < 0 && errno == ENOENT;
}
-bool
-gen_perf_load_metric_id(struct gen_perf_config *perf, const char *guid,
- uint64_t *metric_id)
+static bool
+load_metric_id(struct gen_perf_config *perf, const char *guid,
+ uint64_t *metric_id)
{
char config_path[280];
uint64_t config_id;
int ret;
- if (gen_perf_load_metric_id(perf, query->guid, &config_id)) {
+ if (load_metric_id(perf, query->guid, &config_id)) {
DBG("metric set: %s (already loaded)\n", query->guid);
register_oa_config(perf, query, config_id);
continue;
return NULL;
}
-bool
-gen_perf_load_oa_metrics(struct gen_perf_config *perf, int fd,
+static inline void
+add_stat_reg(struct gen_perf_query_info *query, uint32_t reg,
+ uint32_t numerator, uint32_t denominator,
+ const char *name, const char *description)
+{
+ struct gen_perf_query_counter *counter;
+
+ assert(query->n_counters < query->max_counters);
+
+ counter = &query->counters[query->n_counters];
+ counter->name = name;
+ counter->desc = description;
+ counter->type = GEN_PERF_COUNTER_TYPE_RAW;
+ counter->data_type = GEN_PERF_COUNTER_DATA_TYPE_UINT64;
+ counter->offset = sizeof(uint64_t) * query->n_counters;
+ counter->pipeline_stat.reg = reg;
+ counter->pipeline_stat.numerator = numerator;
+ counter->pipeline_stat.denominator = denominator;
+
+ query->n_counters++;
+}
+
+static inline void
+add_basic_stat_reg(struct gen_perf_query_info *query,
+ uint32_t reg, const char *name)
+{
+ add_stat_reg(query, reg, 1, 1, name, name);
+}
+
+static void
+load_pipeline_statistic_metrics(struct gen_perf_config *perf_cfg,
+ const struct gen_device_info *devinfo)
+{
+ struct gen_perf_query_info *query =
+ append_query_info(perf_cfg, MAX_STAT_COUNTERS);
+
+ query->kind = GEN_PERF_QUERY_TYPE_PIPELINE;
+ query->name = "Pipeline Statistics Registers";
+
+ add_basic_stat_reg(query, IA_VERTICES_COUNT,
+ "N vertices submitted");
+ add_basic_stat_reg(query, IA_PRIMITIVES_COUNT,
+ "N primitives submitted");
+ add_basic_stat_reg(query, VS_INVOCATION_COUNT,
+ "N vertex shader invocations");
+
+ if (devinfo->gen == 6) {
+ add_stat_reg(query, GEN6_SO_PRIM_STORAGE_NEEDED, 1, 1,
+ "SO_PRIM_STORAGE_NEEDED",
+ "N geometry shader stream-out primitives (total)");
+ add_stat_reg(query, GEN6_SO_NUM_PRIMS_WRITTEN, 1, 1,
+ "SO_NUM_PRIMS_WRITTEN",
+ "N geometry shader stream-out primitives (written)");
+ } else {
+ add_stat_reg(query, GEN7_SO_PRIM_STORAGE_NEEDED(0), 1, 1,
+ "SO_PRIM_STORAGE_NEEDED (Stream 0)",
+ "N stream-out (stream 0) primitives (total)");
+ add_stat_reg(query, GEN7_SO_PRIM_STORAGE_NEEDED(1), 1, 1,
+ "SO_PRIM_STORAGE_NEEDED (Stream 1)",
+ "N stream-out (stream 1) primitives (total)");
+ add_stat_reg(query, GEN7_SO_PRIM_STORAGE_NEEDED(2), 1, 1,
+ "SO_PRIM_STORAGE_NEEDED (Stream 2)",
+ "N stream-out (stream 2) primitives (total)");
+ add_stat_reg(query, GEN7_SO_PRIM_STORAGE_NEEDED(3), 1, 1,
+ "SO_PRIM_STORAGE_NEEDED (Stream 3)",
+ "N stream-out (stream 3) primitives (total)");
+ add_stat_reg(query, GEN7_SO_NUM_PRIMS_WRITTEN(0), 1, 1,
+ "SO_NUM_PRIMS_WRITTEN (Stream 0)",
+ "N stream-out (stream 0) primitives (written)");
+ add_stat_reg(query, GEN7_SO_NUM_PRIMS_WRITTEN(1), 1, 1,
+ "SO_NUM_PRIMS_WRITTEN (Stream 1)",
+ "N stream-out (stream 1) primitives (written)");
+ add_stat_reg(query, GEN7_SO_NUM_PRIMS_WRITTEN(2), 1, 1,
+ "SO_NUM_PRIMS_WRITTEN (Stream 2)",
+ "N stream-out (stream 2) primitives (written)");
+ add_stat_reg(query, GEN7_SO_NUM_PRIMS_WRITTEN(3), 1, 1,
+ "SO_NUM_PRIMS_WRITTEN (Stream 3)",
+ "N stream-out (stream 3) primitives (written)");
+ }
+
+ add_basic_stat_reg(query, HS_INVOCATION_COUNT,
+ "N TCS shader invocations");
+ add_basic_stat_reg(query, DS_INVOCATION_COUNT,
+ "N TES shader invocations");
+
+ add_basic_stat_reg(query, GS_INVOCATION_COUNT,
+ "N geometry shader invocations");
+ add_basic_stat_reg(query, GS_PRIMITIVES_COUNT,
+ "N geometry shader primitives emitted");
+
+ add_basic_stat_reg(query, CL_INVOCATION_COUNT,
+ "N primitives entering clipping");
+ add_basic_stat_reg(query, CL_PRIMITIVES_COUNT,
+ "N primitives leaving clipping");
+
+ if (devinfo->is_haswell || devinfo->gen == 8) {
+ add_stat_reg(query, PS_INVOCATION_COUNT, 1, 4,
+ "N fragment shader invocations",
+ "N fragment shader invocations");
+ } else {
+ add_basic_stat_reg(query, PS_INVOCATION_COUNT,
+ "N fragment shader invocations");
+ }
+
+ add_basic_stat_reg(query, PS_DEPTH_COUNT,
+ "N z-pass fragments");
+
+ if (devinfo->gen >= 7) {
+ add_basic_stat_reg(query, CS_INVOCATION_COUNT,
+ "N compute shader invocations");
+ }
+
+ query->data_size = sizeof(uint64_t) * query->n_counters;
+}
+
+static bool
+load_oa_metrics(struct gen_perf_config *perf, int fd,
const struct gen_device_info *devinfo)
{
perf_register_oa_queries_t oa_register = get_register_queries_function(devinfo);
*unslice_freq_hz = unslice_freq * 16666667ULL;
}
-void
-gen_perf_query_result_read_frequencies(struct gen_perf_query_result *result,
- const struct gen_device_info *devinfo,
- const uint32_t *start,
- const uint32_t *end)
+static void
+query_result_read_frequencies(struct gen_perf_query_result *result,
+ const struct gen_device_info *devinfo,
+ const uint32_t *start,
+ const uint32_t *end)
{
/* Slice/Unslice frequency is only available in the OA reports when the
* "Disable OA reports due to clock ratio change" field in
&result->unslice_frequency[1]);
}
-void
-gen_perf_query_result_accumulate(struct gen_perf_query_result *result,
- const struct gen_perf_query_info *query,
- const uint32_t *start,
- const uint32_t *end)
+static void
+query_result_accumulate(struct gen_perf_query_result *result,
+ const struct gen_perf_query_info *query,
+ const uint32_t *start,
+ const uint32_t *end)
{
int i, idx = 0;
}
-void
-gen_perf_query_result_clear(struct gen_perf_query_result *result)
+static void
+query_result_clear(struct gen_perf_query_result *result)
{
memset(result, 0, sizeof(*result));
result->hw_id = 0xffffffff; /* invalid */
}
+static void
+register_mdapi_statistic_query(struct gen_perf_config *perf_cfg,
+ const struct gen_device_info *devinfo)
+{
+ if (!(devinfo->gen >= 7 && devinfo->gen <= 11))
+ return;
+
+ struct gen_perf_query_info *query =
+ append_query_info(perf_cfg, MAX_STAT_COUNTERS);
+
+ query->kind = GEN_PERF_QUERY_TYPE_PIPELINE;
+ query->name = "Intel_Raw_Pipeline_Statistics_Query";
+
+ /* The order has to match mdapi_pipeline_metrics. */
+ add_basic_stat_reg(query, IA_VERTICES_COUNT,
+ "N vertices submitted");
+ add_basic_stat_reg(query, IA_PRIMITIVES_COUNT,
+ "N primitives submitted");
+ add_basic_stat_reg(query, VS_INVOCATION_COUNT,
+ "N vertex shader invocations");
+ add_basic_stat_reg(query, GS_INVOCATION_COUNT,
+ "N geometry shader invocations");
+ add_basic_stat_reg(query, GS_PRIMITIVES_COUNT,
+ "N geometry shader primitives emitted");
+ add_basic_stat_reg(query, CL_INVOCATION_COUNT,
+ "N primitives entering clipping");
+ add_basic_stat_reg(query, CL_PRIMITIVES_COUNT,
+ "N primitives leaving clipping");
+ if (devinfo->is_haswell || devinfo->gen == 8) {
+ add_stat_reg(query, PS_INVOCATION_COUNT, 1, 4,
+ "N fragment shader invocations",
+ "N fragment shader invocations");
+ } else {
+ add_basic_stat_reg(query, PS_INVOCATION_COUNT,
+ "N fragment shader invocations");
+ }
+ add_basic_stat_reg(query, HS_INVOCATION_COUNT,
+ "N TCS shader invocations");
+ add_basic_stat_reg(query, DS_INVOCATION_COUNT,
+ "N TES shader invocations");
+ if (devinfo->gen >= 7) {
+ add_basic_stat_reg(query, CS_INVOCATION_COUNT,
+ "N compute shader invocations");
+ }
+
+ if (devinfo->gen >= 10) {
+ /* Reuse existing CS invocation register until we can expose this new
+ * one.
+ */
+ add_basic_stat_reg(query, CS_INVOCATION_COUNT,
+ "Reserved1");
+ }
+
+ query->data_size = sizeof(uint64_t) * query->n_counters;
+}
+
static void
fill_mdapi_perf_query_counter(struct gen_perf_query_info *query,
const char *name,
sizeof(struct_name.field_name[0]), \
GEN_PERF_COUNTER_DATA_TYPE_##type_name)
-void
-gen_perf_query_register_mdapi_oa_query(const struct gen_device_info *devinfo,
- struct gen_perf_config *perf)
+static void
+register_mdapi_oa_query(const struct gen_device_info *devinfo,
+ struct gen_perf_config *perf)
{
struct gen_perf_query_info *query = NULL;
switch (devinfo->gen) {
case 7: {
- query = gen_perf_query_append_query_info(perf, 1 + 45 + 16 + 7);
+ query = append_query_info(perf, 1 + 45 + 16 + 7);
query->oa_format = I915_OA_FORMAT_A45_B8_C8;
struct gen7_mdapi_metrics metric_data;
break;
}
case 8: {
- query = gen_perf_query_append_query_info(perf, 2 + 36 + 16 + 16);
+ query = append_query_info(perf, 2 + 36 + 16 + 16);
query->oa_format = I915_OA_FORMAT_A32u40_A4u32_B8_C8;
struct gen8_mdapi_metrics metric_data;
case 9:
case 10:
case 11: {
- query = gen_perf_query_append_query_info(perf, 2 + 36 + 16 + 16 + 16 + 2);
+ query = append_query_info(perf, 2 + 36 + 16 + 16 + 16 + 2);
query->oa_format = I915_OA_FORMAT_A32u40_A4u32_B8_C8;
struct gen9_mdapi_metrics metric_data;
}
}
-void
-gen_perf_query_register_mdapi_statistic_query(const struct gen_device_info *devinfo,
- struct gen_perf_config *perf)
-{
- if (!(devinfo->gen >= 7 && devinfo->gen <= 11))
- return;
-
- struct gen_perf_query_info *query =
- gen_perf_query_append_query_info(perf, MAX_STAT_COUNTERS);
-
- query->kind = GEN_PERF_QUERY_TYPE_PIPELINE;
- query->name = "Intel_Raw_Pipeline_Statistics_Query";
-
- /* The order has to match mdapi_pipeline_metrics. */
- gen_perf_query_info_add_basic_stat_reg(query, IA_VERTICES_COUNT,
- "N vertices submitted");
- gen_perf_query_info_add_basic_stat_reg(query, IA_PRIMITIVES_COUNT,
- "N primitives submitted");
- gen_perf_query_info_add_basic_stat_reg(query, VS_INVOCATION_COUNT,
- "N vertex shader invocations");
- gen_perf_query_info_add_basic_stat_reg(query, GS_INVOCATION_COUNT,
- "N geometry shader invocations");
- gen_perf_query_info_add_basic_stat_reg(query, GS_PRIMITIVES_COUNT,
- "N geometry shader primitives emitted");
- gen_perf_query_info_add_basic_stat_reg(query, CL_INVOCATION_COUNT,
- "N primitives entering clipping");
- gen_perf_query_info_add_basic_stat_reg(query, CL_PRIMITIVES_COUNT,
- "N primitives leaving clipping");
- if (devinfo->is_haswell || devinfo->gen == 8) {
- gen_perf_query_info_add_stat_reg(query, PS_INVOCATION_COUNT, 1, 4,
- "N fragment shader invocations",
- "N fragment shader invocations");
- } else {
- gen_perf_query_info_add_basic_stat_reg(query, PS_INVOCATION_COUNT,
- "N fragment shader invocations");
- }
- gen_perf_query_info_add_basic_stat_reg(query, HS_INVOCATION_COUNT,
- "N TCS shader invocations");
- gen_perf_query_info_add_basic_stat_reg(query, DS_INVOCATION_COUNT,
- "N TES shader invocations");
- if (devinfo->gen >= 7) {
- gen_perf_query_info_add_basic_stat_reg(query, CS_INVOCATION_COUNT,
- "N compute shader invocations");
- }
-
- if (devinfo->gen >= 10) {
- /* Reuse existing CS invocation register until we can expose this new
- * one.
- */
- gen_perf_query_info_add_basic_stat_reg(query, CS_INVOCATION_COUNT,
- "Reserved1");
- }
-
- query->data_size = sizeof(uint64_t) * query->n_counters;
-}
-
-uint64_t
-gen_perf_query_get_metric_id(struct gen_perf_config *perf,
- const struct gen_perf_query_info *query)
+static uint64_t
+get_metric_id(struct gen_perf_config *perf,
+ const struct gen_perf_query_info *query)
{
/* These queries are know not to ever change, their config ID has been
* loaded upon the first query creation. No need to look them up again.
}
struct gen_perf_query_info *raw_query = (struct gen_perf_query_info *)query;
- if (!gen_perf_load_metric_id(perf, query->guid,
- &raw_query->oa_metrics_set_id)) {
+ if (!load_metric_id(perf, query->guid,
+ &raw_query->oa_metrics_set_id)) {
DBG("Unable to read query guid=%s ID, falling back to test config\n", query->guid);
raw_query->oa_metrics_set_id = 1ULL;
} else {
return query->oa_metrics_set_id;
}
-struct oa_sample_buf *
-gen_perf_get_free_sample_buf(struct gen_perf_context *perf_ctx)
+static struct oa_sample_buf *
+get_free_sample_buf(struct gen_perf_context *perf_ctx)
{
struct exec_node *node = exec_list_pop_head(&perf_ctx->free_sample_buffers);
struct oa_sample_buf *buf;
return buf;
}
-void
-gen_perf_reap_old_sample_buffers(struct gen_perf_context *perf_ctx)
+static void
+reap_old_sample_buffers(struct gen_perf_context *perf_ctx)
{
struct exec_node *tail_node =
exec_list_get_tail(&perf_ctx->sample_buffers);
}
}
-void
-gen_perf_free_sample_bufs(struct gen_perf_context *perf_ctx)
+static void
+free_sample_bufs(struct gen_perf_context *perf_ctx)
{
foreach_list_typed_safe(struct oa_sample_buf, buf, link,
&perf_ctx->free_sample_buffers)
* Emit MI_STORE_REGISTER_MEM commands to capture all of the
* pipeline statistics for the performance query object.
*/
-void
-gen_perf_snapshot_statistics_registers(void *context,
- struct gen_perf_config *perf,
- struct gen_perf_query_object *obj,
- uint32_t offset_in_bytes)
+static void
+snapshot_statistics_registers(void *context,
+ struct gen_perf_config *perf,
+ struct gen_perf_query_object *obj,
+ uint32_t offset_in_bytes)
{
const struct gen_perf_query_info *query = obj->queryinfo;
const int n_counters = query->n_counters;
}
}
-void
+static void
gen_perf_close(struct gen_perf_context *perfquery,
const struct gen_perf_query_info *query)
{
}
}
-bool
+static bool
gen_perf_open(struct gen_perf_context *perf_ctx,
int metrics_set_id,
int report_format,
return true;
}
-bool
-gen_perf_inc_n_users(struct gen_perf_context *perf_ctx)
+static bool
+inc_n_users(struct gen_perf_context *perf_ctx)
{
if (perf_ctx->n_oa_users == 0 &&
gen_ioctl(perf_ctx->oa_stream_fd, I915_PERF_IOCTL_ENABLE, 0) < 0)
return true;
}
-void
-gen_perf_dec_n_users(struct gen_perf_context *perf_ctx)
+static void
+dec_n_users(struct gen_perf_context *perf_ctx)
{
/* Disabling the i915 perf stream will effectively disable the OA
* counters. Note it's important to be sure there are no outstanding
}
}
+void
+gen_perf_init_metrics(struct gen_perf_config *perf_cfg,
+ const struct gen_device_info *devinfo,
+ int drm_fd)
+{
+ load_pipeline_statistic_metrics(perf_cfg, devinfo);
+ register_mdapi_statistic_query(perf_cfg, devinfo);
+ if (load_oa_metrics(perf_cfg, drm_fd, devinfo))
+ register_mdapi_oa_query(devinfo, perf_cfg);
+}
+
void
gen_perf_init_context(struct gen_perf_context *perf_ctx,
struct gen_perf_config *perf_cfg,
* Begin an OA query we can always take a reference on a buffer
* in this list.
*/
- struct oa_sample_buf *buf = gen_perf_get_free_sample_buf(perf_ctx);
+ struct oa_sample_buf *buf = get_free_sample_buf(perf_ctx);
exec_list_push_head(&perf_ctx->sample_buffers, &buf->link);
perf_ctx->oa_stream_fd = -1;
* require a different counter set or format unless we get an opportunity
* to close the stream and open a new one...
*/
- uint64_t metric_id = gen_perf_query_get_metric_id(perf_ctx->perf, queryinfo);
+ uint64_t metric_id = get_metric_id(perf_ctx->perf, queryinfo);
if (perf_ctx->oa_stream_fd != -1 &&
perf_ctx->current_oa_metrics_set_id != metric_id) {
perf_ctx->current_oa_format == queryinfo->oa_format);
}
- if (!gen_perf_inc_n_users(perf_ctx)) {
+ if (!inc_n_users(perf_ctx)) {
DBG("WARNING: Error enabling i915 perf stream: %m\n");
return false;
}
*/
buf->refcount++;
- gen_perf_query_result_clear(&query->oa.result);
+ query_result_clear(&query->oa.result);
query->oa.results_accumulated = false;
add_to_unaccumulated_query_list(perf_ctx, query);
STATS_BO_SIZE);
/* Take starting snapshots. */
- gen_perf_snapshot_statistics_registers(perf_ctx->ctx , perf_cfg, query, 0);
+ snapshot_statistics_registers(perf_ctx->ctx , perf_cfg, query, 0);
++perf_ctx->n_active_pipeline_stats_queries;
break;
break;
case GEN_PERF_QUERY_TYPE_PIPELINE:
- gen_perf_snapshot_statistics_registers(perf_ctx->ctx, perf_cfg, query,
- STATS_BO_END_OFFSET_BYTES);
+ snapshot_statistics_registers(perf_ctx->ctx, perf_cfg, query,
+ STATS_BO_END_OFFSET_BYTES);
--perf_ctx->n_active_pipeline_stats_queries;
break;
uint32_t last_timestamp = tail_buf->last_timestamp;
while (1) {
- struct oa_sample_buf *buf = gen_perf_get_free_sample_buf(perf_ctx);
+ struct oa_sample_buf *buf = get_free_sample_buf(perf_ctx);
uint32_t offset;
int len;
query->oa.samples_head = NULL;
- gen_perf_reap_old_sample_buffers(perf_ctx);
+ reap_old_sample_buffers(perf_ctx);
}
/* In general if we see anything spurious while accumulating results,
query->oa.results_accumulated = true;
drop_from_unaccumulated_query_list(perf_ctx, query);
- gen_perf_dec_n_users(perf_ctx);
+ dec_n_users(perf_ctx);
}
}
}
if (add) {
- gen_perf_query_result_accumulate(&query->oa.result, query->queryinfo,
- last, report);
+ query_result_accumulate(&query->oa.result, query->queryinfo,
+ last, report);
}
last = report;
end:
- gen_perf_query_result_accumulate(&query->oa.result, query->queryinfo,
- last, end);
+ query_result_accumulate(&query->oa.result, query->queryinfo,
+ last, end);
query->oa.results_accumulated = true;
drop_from_unaccumulated_query_list(perf_ctx, query);
- gen_perf_dec_n_users(perf_ctx);
+ dec_n_users(perf_ctx);
return;
if (query->oa.bo) {
if (!query->oa.results_accumulated) {
drop_from_unaccumulated_query_list(perf_ctx, query);
- gen_perf_dec_n_users(perf_ctx);
+ dec_n_users(perf_ctx);
}
perf_cfg->vtbl.bo_unreference(query->oa.bo);
* buffers and close any current i915-perf stream.
*/
if (--perf_ctx->n_query_instances == 0) {
- gen_perf_free_sample_bufs(perf_ctx);
+ free_sample_bufs(perf_ctx);
gen_perf_close(perf_ctx, query->queryinfo);
}
read_gt_frequency(perf_ctx, query);
uint32_t *begin_report = query->oa.map;
uint32_t *end_report = query->oa.map + MI_RPC_BO_END_OFFSET_BYTES;
- gen_perf_query_result_read_frequencies(&query->oa.result,
- perf_ctx->devinfo,
- begin_report,
- end_report);
+ query_result_read_frequencies(&query->oa.result,
+ perf_ctx->devinfo,
+ begin_report,
+ end_report);
accumulate_oa_reports(perf_ctx, query);
assert(query->oa.results_accumulated);