1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
|
/*
* Copyright © 2015 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 BRW_PERFORMANCE_QUERY_H
#define BRW_PERFORMANCE_QUERY_H
#include <stdint.h>
#include "brw_context.h"
#include "perf/gen_perf.h"
struct gen_perf_query_info;
/*
* When currently allocate only one page for pipeline statistics queries. Here
* we derived the maximum number of counters for that amount.
*/
#define STATS_BO_SIZE 4096
#define STATS_BO_END_OFFSET_BYTES (STATS_BO_SIZE / 2)
#define MAX_STAT_COUNTERS (STATS_BO_END_OFFSET_BYTES / 8)
/**
* i965 representation of a performance query object.
*
* NB: We want to keep this structure relatively lean considering that
* applications may expect to allocate enough objects to be able to
* query around all draw calls in a frame.
*/
struct brw_perf_query_object
{
struct gl_perf_query_object base;
const struct gen_perf_query_info *query;
/* See query->kind to know which state below is in use... */
union {
struct {
/**
* BO containing OA counter snapshots at query Begin/End time.
*/
struct brw_bo *bo;
/**
* Address of mapped of @bo
*/
void *map;
/**
* The MI_REPORT_PERF_COUNT command lets us specify a unique
* ID that will be reflected in the resulting OA report
* that's written by the GPU. This is the ID we're expecting
* in the begin report and the the end report should be
* @begin_report_id + 1.
*/
int begin_report_id;
/**
* Reference the head of the brw->perfquery.sample_buffers
* list at the time that the query started (so we only need
* to look at nodes after this point when looking for samples
* related to this query)
*
* (See struct brw_oa_sample_buf description for more details)
*/
struct exec_node *samples_head;
/**
* false while in the unaccumulated_elements list, and set to
* true when the final, end MI_RPC snapshot has been
* accumulated.
*/
bool results_accumulated;
/**
* Frequency of the GT at begin and end of the query.
*/
uint64_t gt_frequency[2];
/**
* Accumulated OA results between begin and end of the query.
*/
struct gen_perf_query_result result;
} oa;
struct {
/**
* BO containing starting and ending snapshots for the
* statistics counters.
*/
struct brw_bo *bo;
} pipeline_stats;
};
};
int brw_perf_query_get_mdapi_oa_data(struct brw_context *brw,
struct brw_perf_query_object *obj,
size_t data_size,
uint8_t *data);
void brw_perf_query_register_mdapi_oa_query(struct brw_context *brw);
void brw_perf_query_register_mdapi_statistic_query(struct brw_context *brw);
#endif /* BRW_PERFORMANCE_QUERY_H */
|
