aboutsummaryrefslogtreecommitdiffstats
path: root/src/gallium/drivers/radeonsi/si_hw_context.c
blob: 6c2734eac29548ba2f08ca5567f0a0c2e7e00b0c (plain)
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
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
/*
 * Copyright 2010 Jerome Glisse <glisse@freedesktop.org>
 *
 * 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
 * on the rights to use, copy, modify, merge, publish, distribute, sub
 * license, 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 NON-INFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHOR(S) AND/OR THEIR SUPPLIERS 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.
 *
 * Authors:
 *      Jerome Glisse
 */

#include "si_pipe.h"

/* initialize */
void si_need_cs_space(struct si_context *ctx, unsigned num_dw,
			boolean count_draw_in)
{
	int i;

	/* The number of dwords we already used in the CS so far. */
	num_dw += ctx->b.rings.gfx.cs->cdw;

	if (count_draw_in) {
		for (i = 0; i < SI_NUM_ATOMS(ctx); i++) {
			if (ctx->atoms.array[i]->dirty) {
				num_dw += ctx->atoms.array[i]->num_dw;
			}
		}

		/* The number of dwords all the dirty states would take. */
		num_dw += si_pm4_dirty_dw(ctx);

		/* The upper-bound of how much a draw command would take. */
		num_dw += SI_MAX_DRAW_CS_DWORDS;
	}

	/* Count in queries_suspend. */
	num_dw += ctx->b.num_cs_dw_nontimer_queries_suspend;

	/* Count in streamout_end at the end of CS. */
	if (ctx->b.streamout.begin_emitted) {
		num_dw += ctx->b.streamout.num_dw_for_end;
	}

	/* Count in render_condition(NULL) at the end of CS. */
	if (ctx->b.predicate_drawing) {
		num_dw += 3;
	}

	/* Count in framebuffer cache flushes at the end of CS. */
	num_dw += ctx->atoms.s.cache_flush->num_dw;

#if SI_TRACE_CS
	if (ctx->screen->b.trace_bo) {
		num_dw += SI_TRACE_CS_DWORDS;
	}
#endif

	/* Flush if there's not enough space. */
	if (num_dw > RADEON_MAX_CMDBUF_DWORDS) {
		ctx->b.rings.gfx.flush(ctx, RADEON_FLUSH_ASYNC, NULL);
	}
}

void si_context_gfx_flush(void *context, unsigned flags,
			  struct pipe_fence_handle **fence)
{
	struct si_context *ctx = context;
	struct radeon_winsys_cs *cs = ctx->b.rings.gfx.cs;

	if (cs->cdw == ctx->b.initial_gfx_cs_size && !fence)
		return;

	ctx->b.rings.gfx.flushing = true;

	r600_preflush_suspend_features(&ctx->b);

	ctx->b.flags |= R600_CONTEXT_FLUSH_AND_INV_CB |
			R600_CONTEXT_FLUSH_AND_INV_CB_META |
			R600_CONTEXT_FLUSH_AND_INV_DB |
			R600_CONTEXT_FLUSH_AND_INV_DB_META |
			R600_CONTEXT_INV_TEX_CACHE |
			/* this is probably not needed anymore */
			R600_CONTEXT_PS_PARTIAL_FLUSH;
	si_emit_cache_flush(&ctx->b, NULL);

	/* force to keep tiling flags */
	flags |= RADEON_FLUSH_KEEP_TILING_FLAGS;

	/* Flush the CS. */
	ctx->b.ws->cs_flush(cs, flags, fence, ctx->screen->b.cs_count++);
	ctx->b.rings.gfx.flushing = false;

#if SI_TRACE_CS
	if (ctx->screen->b.trace_bo) {
		struct si_screen *sscreen = ctx->screen;
		unsigned i;

		for (i = 0; i < 10; i++) {
			usleep(5);
			if (!ctx->b.ws->buffer_is_busy(sscreen->b.trace_bo->buf, RADEON_USAGE_READWRITE)) {
				break;
			}
		}
		if (i == 10) {
			fprintf(stderr, "timeout on cs lockup likely happen at cs %d dw %d\n",
				sscreen->b.trace_ptr[1], sscreen->b.trace_ptr[0]);
		} else {
			fprintf(stderr, "cs %d executed in %dms\n", sscreen->b.trace_ptr[1], i * 5);
		}
	}
#endif

	si_begin_new_cs(ctx);
}

void si_begin_new_cs(struct si_context *ctx)
{
	/* Flush read caches at the beginning of CS. */
	ctx->b.flags |= R600_CONTEXT_INV_TEX_CACHE |
			R600_CONTEXT_INV_CONST_CACHE |
			R600_CONTEXT_INV_SHADER_CACHE;

	/* set all valid group as dirty so they get reemited on
	 * next draw command
	 */
	si_pm4_reset_emitted(ctx);

	/* The CS initialization should be emitted before everything else. */
	si_pm4_emit(ctx, ctx->queued.named.init);
	ctx->emitted.named.init = ctx->queued.named.init;

	ctx->clip_regs.dirty = true;
	ctx->framebuffer.atom.dirty = true;
	ctx->msaa_config.dirty = true;
	ctx->db_render_state.dirty = true;
	ctx->b.streamout.enable_atom.dirty = true;
	si_all_descriptors_begin_new_cs(ctx);

	r600_postflush_resume_features(&ctx->b);

	ctx->b.initial_gfx_cs_size = ctx->b.rings.gfx.cs->cdw;
	si_invalidate_draw_sh_constants(ctx);
}