1 /*
2  * Copyright © 2011 Intel Corporation
3  *
4  * Permission is hereby granted, free of charge, to any person obtaining a
5  * copy of this software and associated documentation files (the "Software"),
6  * to deal in the Software without restriction, including without limitation
7  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8  * and/or sell copies of the Software, and to permit persons to whom the
9  * Software is furnished to do so, subject to the following conditions:
10  *
11  * The above copyright notice and this permission notice (including the next
12  * paragraph) shall be included in all copies or substantial portions of the
13  * Software.
14  *
15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
18  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
20  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
21  * IN THE SOFTWARE.
22  */
23 
24 #include "main/macros.h"
25 #include "intel_batchbuffer.h"
26 #include "brw_context.h"
27 #include "brw_state.h"
28 #include "brw_defines.h"
29 
30 /**
31  * The following diagram shows how we partition the URB:
32  *
33  *      8kB         8kB              Rest of the URB space
34  *   ____-____   ____-____   _________________-_________________
35  *  /         \ /         \ /                                   \
36  * +-------------------------------------------------------------+
37  * | VS Push   | FS Push   | VS                                  |
38  * | Constants | Constants | Handles                             |
39  * +-------------------------------------------------------------+
40  *
41  * Notably, push constants must be stored at the beginning of the URB
42  * space, while entries can be stored anywhere.  Ivybridge has a maximum
43  * constant buffer size of 16kB.
44  *
45  * Currently we split the constant buffer space evenly between VS and FS.
46  * This is probably not ideal, but simple.
47  *
48  * Ivybridge GT1 has 128kB of URB space.
49  * Ivybridge GT2 has 256kB of URB space.
50  *
51  * See "Volume 2a: 3D Pipeline," section 1.8.
52  */
53 void
gen7_allocate_push_constants(struct brw_context * brw)54 gen7_allocate_push_constants(struct brw_context *brw)
55 {
56    struct intel_context *intel = &brw->intel;
57    BEGIN_BATCH(2);
58    OUT_BATCH(_3DSTATE_PUSH_CONSTANT_ALLOC_VS << 16 | (2 - 2));
59    OUT_BATCH(8);
60    ADVANCE_BATCH();
61 
62    BEGIN_BATCH(2);
63    OUT_BATCH(_3DSTATE_PUSH_CONSTANT_ALLOC_PS << 16 | (2 - 2));
64    OUT_BATCH(8 | 8 << GEN7_PUSH_CONSTANT_BUFFER_OFFSET_SHIFT);
65    ADVANCE_BATCH();
66 }
67 
68 const struct brw_tracked_state gen7_push_constant_alloc = {
69    .dirty = {
70       .mesa = 0,
71       .brw = BRW_NEW_CONTEXT,
72       .cache = 0,
73    },
74    .emit = gen7_allocate_push_constants,
75 };
76 
77 static void
gen7_upload_urb(struct brw_context * brw)78 gen7_upload_urb(struct brw_context *brw)
79 {
80    struct intel_context *intel = &brw->intel;
81    /* Total space for entries is URB size - 16kB for push constants */
82    int handle_region_size = (brw->urb.size - 16) * 1024; /* bytes */
83 
84    /* CACHE_NEW_VS_PROG */
85    brw->urb.vs_size = MAX2(brw->vs.prog_data->urb_entry_size, 1);
86 
87    int nr_vs_entries = handle_region_size / (brw->urb.vs_size * 64);
88    if (nr_vs_entries > brw->urb.max_vs_entries)
89       nr_vs_entries = brw->urb.max_vs_entries;
90 
91    /* According to volume 2a, nr_vs_entries must be a multiple of 8. */
92    brw->urb.nr_vs_entries = ROUND_DOWN_TO(nr_vs_entries, 8);
93 
94    /* URB Starting Addresses are specified in multiples of 8kB. */
95    brw->urb.vs_start = 2; /* skip over push constants */
96 
97    assert(brw->urb.nr_vs_entries % 8 == 0);
98    assert(brw->urb.nr_gs_entries % 8 == 0);
99    /* GS requirement */
100    assert(!brw->gs.prog_active);
101 
102    gen7_emit_vs_workaround_flush(intel);
103    gen7_emit_urb_state(brw, brw->urb.nr_vs_entries, brw->urb.vs_size,
104                        brw->urb.vs_start);
105 }
106 
107 void
gen7_emit_urb_state(struct brw_context * brw,GLuint nr_vs_entries,GLuint vs_size,GLuint vs_start)108 gen7_emit_urb_state(struct brw_context *brw, GLuint nr_vs_entries,
109                     GLuint vs_size, GLuint vs_start)
110 {
111    struct intel_context *intel = &brw->intel;
112 
113    BEGIN_BATCH(2);
114    OUT_BATCH(_3DSTATE_URB_VS << 16 | (2 - 2));
115    OUT_BATCH(nr_vs_entries |
116              ((vs_size - 1) << GEN7_URB_ENTRY_SIZE_SHIFT) |
117              (vs_start << GEN7_URB_STARTING_ADDRESS_SHIFT));
118    ADVANCE_BATCH();
119 
120    /* Allocate the GS, HS, and DS zero space - we don't use them. */
121    BEGIN_BATCH(2);
122    OUT_BATCH(_3DSTATE_URB_GS << 16 | (2 - 2));
123    OUT_BATCH((0 << GEN7_URB_ENTRY_SIZE_SHIFT) |
124              (vs_start << GEN7_URB_STARTING_ADDRESS_SHIFT));
125    ADVANCE_BATCH();
126 
127    BEGIN_BATCH(2);
128    OUT_BATCH(_3DSTATE_URB_HS << 16 | (2 - 2));
129    OUT_BATCH((0 << GEN7_URB_ENTRY_SIZE_SHIFT) |
130              (vs_start << GEN7_URB_STARTING_ADDRESS_SHIFT));
131    ADVANCE_BATCH();
132 
133    BEGIN_BATCH(2);
134    OUT_BATCH(_3DSTATE_URB_DS << 16 | (2 - 2));
135    OUT_BATCH((0 << GEN7_URB_ENTRY_SIZE_SHIFT) |
136              (vs_start << GEN7_URB_STARTING_ADDRESS_SHIFT));
137    ADVANCE_BATCH();
138 }
139 
140 const struct brw_tracked_state gen7_urb = {
141    .dirty = {
142       .mesa = 0,
143       .brw = BRW_NEW_CONTEXT,
144       .cache = (CACHE_NEW_VS_PROG | CACHE_NEW_GS_PROG),
145    },
146    .emit = gen7_upload_urb,
147 };
148