summaryrefslogtreecommitdiff
path: root/src/gallium/drivers/freedreno/a6xx/fd6_emit.h
blob: a603017dd847abba23aaf4d274f5432bc7bc35da (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
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
/*
 * Copyright (C) 2016 Rob Clark <robclark@freedesktop.org>
 * Copyright © 2018 Google, Inc.
 *
 * 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.
 *
 * Authors:
 *    Rob Clark <robclark@freedesktop.org>
 */

#ifndef FD6_EMIT_H
#define FD6_EMIT_H

#include "pipe/p_context.h"

#include "freedreno_context.h"
#include "fd6_context.h"
#include "fd6_format.h"
#include "fd6_program.h"
#include "ir3_shader.h"

struct fd_ringbuffer;

/* grouped together emit-state for prog/vertex/state emit: */
struct fd6_emit {
	struct pipe_debug_callback *debug;
	const struct fd_vertex_state *vtx;
	const struct fd_program_stateobj *prog;
	const struct pipe_draw_info *info;
	struct ir3_shader_key key;
	enum fd_dirty_3d_state dirty;

	uint32_t sprite_coord_enable;  /* bitmask */
	bool sprite_coord_mode;
	bool rasterflat;
	bool no_decode_srgb;

	/* in binning pass, we don't have real frag shader, so we
	 * don't know if real draw disqualifies lrz write.  So just
	 * figure that out up-front and stash it in the emit.
	 */
	bool no_lrz_write;

	/* cached to avoid repeated lookups of same variants: */
	const struct ir3_shader_variant *vp, *fp;
	/* TODO: other shader stages.. */

	unsigned streamout_mask;
};

static inline enum a6xx_color_fmt fd6_emit_format(struct pipe_surface *surf)
{
	if (!surf)
		return 0;
	return fd6_pipe2color(surf->format);
}

static inline const struct ir3_shader_variant *
fd6_emit_get_vp(struct fd6_emit *emit)
{
	if (!emit->vp) {
		struct fd6_shader_stateobj *so = emit->prog->vp;
		emit->vp = ir3_shader_variant(so->shader, emit->key, emit->debug);
	}
	return emit->vp;
}

static inline const struct ir3_shader_variant *
fd6_emit_get_fp(struct fd6_emit *emit)
{
	if (!emit->fp) {
		if (emit->key.binning_pass) {
			/* use dummy stateobj to simplify binning vs non-binning: */
			static const struct ir3_shader_variant binning_fp = {};
			emit->fp = &binning_fp;
		} else {
			struct fd6_shader_stateobj *so = emit->prog->fp;
			emit->fp = ir3_shader_variant(so->shader, emit->key, emit->debug);
		}
	}
	return emit->fp;
}

static inline void
fd6_cache_flush(struct fd_batch *batch, struct fd_ringbuffer *ring)
{
	fd_reset_wfi(batch);
#if 0
	OUT_PKT4(ring, REG_A6XX_UCHE_CACHE_INVALIDATE_MIN_LO, 5);
	OUT_RING(ring, 0x00000000);   /* UCHE_CACHE_INVALIDATE_MIN_LO */
	OUT_RING(ring, 0x00000000);   /* UCHE_CACHE_INVALIDATE_MIN_HI */
	OUT_RING(ring, 0x00000000);   /* UCHE_CACHE_INVALIDATE_MAX_LO */
	OUT_RING(ring, 0x00000000);   /* UCHE_CACHE_INVALIDATE_MAX_HI */
	OUT_RING(ring, 0x00000012);   /* UCHE_CACHE_INVALIDATE */
	fd_wfi(batch, ring);
#else
	DBG("fd6_cache_flush stub");
#endif
}

static inline void
fd6_emit_blit(struct fd_context *ctx, struct fd_ringbuffer *ring)
{
	emit_marker6(ring, 7);

	OUT_PKT7(ring, CP_EVENT_WRITE, 1);
	OUT_RING(ring, CP_EVENT_WRITE_0_EVENT(BLIT));

	emit_marker6(ring, 7);
}

static inline void
fd6_emit_render_cntl(struct fd_context *ctx, bool blit, bool binning)
{
#if 0
	struct fd_ringbuffer *ring = binning ? ctx->batch->binning : ctx->batch->draw;

	/* TODO eventually this partially depends on the pfb state, ie.
	 * which of the cbuf(s)/zsbuf has an UBWC flag buffer.. that part
	 * we could probably cache and just regenerate if framebuffer
	 * state is dirty (or something like that)..
	 *
	 * Other bits seem to depend on query state, like if samples-passed
	 * query is active.
	 */
	bool samples_passed = (fd6_context(ctx)->samples_passed_queries > 0);
	OUT_PKT4(ring, REG_A6XX_RB_RENDER_CNTL, 1);
	OUT_RING(ring, 0x00000000 |   /* RB_RENDER_CNTL */
			COND(binning, A6XX_RB_RENDER_CNTL_BINNING_PASS) |
			COND(binning, A6XX_RB_RENDER_CNTL_DISABLE_COLOR_PIPE) |
			COND(samples_passed, A6XX_RB_RENDER_CNTL_SAMPLES_PASSED) |
			COND(!blit, 0x8));
	OUT_PKT4(ring, REG_A6XX_GRAS_SC_CNTL, 1);
	OUT_RING(ring, 0x00000008 |   /* GRAS_SC_CNTL */
			COND(binning, A6XX_GRAS_SC_CNTL_BINNING_PASS) |
			COND(samples_passed, A6XX_GRAS_SC_CNTL_SAMPLES_PASSED));
#else
	DBG("render ctl stub");
#endif
}

static inline void
fd6_emit_lrz_flush(struct fd_ringbuffer *ring)
{
	/* TODO I think the extra writes to GRAS_LRZ_CNTL are probably
	 * a workaround and not needed on all a5xx.
	 */
	OUT_PKT4(ring, REG_A6XX_GRAS_LRZ_CNTL, 1);
	OUT_RING(ring, A6XX_GRAS_LRZ_CNTL_ENABLE);

	OUT_PKT7(ring, CP_EVENT_WRITE, 1);
	OUT_RING(ring, LRZ_FLUSH);

	OUT_PKT4(ring, REG_A6XX_GRAS_LRZ_CNTL, 1);
	OUT_RING(ring, 0x0);
}

static inline enum a6xx_state_block
fd6_stage2shadersb(enum shader_t type)
{
	switch (type) {
	case SHADER_VERTEX:
		return SB6_VS_SHADER;
	case SHADER_FRAGMENT:
		return SB6_FS_SHADER;
	case SHADER_COMPUTE:
		return SB6_CS_SHADER;
	default:
		unreachable("bad shader type");
		return ~0;
	}
}

void fd6_emit_vertex_bufs(struct fd_ringbuffer *ring, struct fd6_emit *emit);

void fd6_emit_state(struct fd_context *ctx, struct fd_ringbuffer *ring,
		struct fd6_emit *emit);

void fd6_emit_cs_state(struct fd_context *ctx, struct fd_ringbuffer *ring,
		struct ir3_shader_variant *cp);

void fd6_emit_restore(struct fd_batch *batch, struct fd_ringbuffer *ring);

void fd6_emit_init(struct pipe_context *pctx);

#define WRITE(reg, val) do {					\
		OUT_PKT4(ring, reg, 1);					\
		OUT_RING(ring, val);					\
	} while (0)


#endif /* FD6_EMIT_H */