Merge branch 'master' into autoconf2
[mesa.git] / src / mesa / drivers / dri / i965 / brw_sf_state.c
1 /*
2 Copyright (C) Intel Corp. 2006. All Rights Reserved.
3 Intel funded Tungsten Graphics (http://www.tungstengraphics.com) to
4 develop this 3D driver.
5
6 Permission is hereby granted, free of charge, to any person obtaining
7 a copy of this software and associated documentation files (the
8 "Software"), to deal in the Software without restriction, including
9 without limitation the rights to use, copy, modify, merge, publish,
10 distribute, sublicense, and/or sell copies of the Software, and to
11 permit persons to whom the Software is furnished to do so, subject to
12 the following conditions:
13
14 The above copyright notice and this permission notice (including the
15 next paragraph) shall be included in all copies or substantial
16 portions of the Software.
17
18 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
19 EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
20 MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
21 IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
22 LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
23 OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
24 WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
25
26 **********************************************************************/
27 /*
28 * Authors:
29 * Keith Whitwell <keith@tungstengraphics.com>
30 */
31
32
33
34 #include "brw_context.h"
35 #include "brw_state.h"
36 #include "brw_defines.h"
37 #include "macros.h"
38 #include "intel_fbo.h"
39
40 static void upload_sf_vp(struct brw_context *brw)
41 {
42 GLcontext *ctx = &brw->intel.ctx;
43 const GLfloat depth_scale = 1.0F / ctx->DrawBuffer->_DepthMaxF;
44 struct brw_sf_viewport sfv;
45 struct intel_renderbuffer *irb =
46 intel_renderbuffer(ctx->DrawBuffer->_ColorDrawBuffers[0][0]);
47 GLfloat y_scale, y_bias;
48
49 memset(&sfv, 0, sizeof(sfv));
50
51 if (ctx->DrawBuffer->Name) {
52 /* User-created FBO */
53 if (irb && !irb->RenderToTexture) {
54 y_scale = -1.0;
55 y_bias = ctx->DrawBuffer->Height;
56 } else {
57 y_scale = 1.0;
58 y_bias = 0;
59 }
60 } else {
61 y_scale = -1.0;
62 y_bias = ctx->DrawBuffer->Height;
63 }
64
65 /* _NEW_VIEWPORT, BRW_NEW_METAOPS */
66
67 if (!brw->metaops.active) {
68 const GLfloat *v = ctx->Viewport._WindowMap.m;
69
70 sfv.viewport.m00 = v[MAT_SX];
71 sfv.viewport.m11 = v[MAT_SY] * y_scale;
72 sfv.viewport.m22 = v[MAT_SZ] * depth_scale;
73 sfv.viewport.m30 = v[MAT_TX];
74 sfv.viewport.m31 = v[MAT_TY] * y_scale + y_bias;
75 sfv.viewport.m32 = v[MAT_TZ] * depth_scale;
76 } else {
77 sfv.viewport.m00 = 1;
78 sfv.viewport.m11 = - 1;
79 sfv.viewport.m22 = 1;
80 sfv.viewport.m30 = 0;
81 sfv.viewport.m31 = ctx->DrawBuffer->Height;
82 sfv.viewport.m32 = 0;
83 }
84
85 /* _NEW_SCISSOR */
86
87 /* The scissor only needs to handle the intersection of drawable and
88 * scissor rect. Clipping to the boundaries of static shared buffers
89 * for front/back/depth is covered by looping over cliprects in brw_draw.c.
90 *
91 * Note that the hardware's coordinates are inclusive, while Mesa's min is
92 * inclusive but max is exclusive.
93 */
94 sfv.scissor.xmin = ctx->DrawBuffer->_Xmin;
95 sfv.scissor.xmax = ctx->DrawBuffer->_Xmax - 1;
96 sfv.scissor.ymin = ctx->DrawBuffer->Height - ctx->DrawBuffer->_Ymax;
97 sfv.scissor.ymax = ctx->DrawBuffer->Height - ctx->DrawBuffer->_Ymin - 1;
98
99 dri_bo_unreference(brw->sf.vp_bo);
100 brw->sf.vp_bo = brw_cache_data( &brw->cache, BRW_SF_VP, &sfv, NULL, 0 );
101 }
102
103 const struct brw_tracked_state brw_sf_vp = {
104 .dirty = {
105 .mesa = (_NEW_VIEWPORT |
106 _NEW_SCISSOR),
107 .brw = BRW_NEW_METAOPS,
108 .cache = 0
109 },
110 .update = upload_sf_vp
111 };
112
113
114
115 static void upload_sf_unit( struct brw_context *brw )
116 {
117 struct brw_sf_unit_state sf;
118 memset(&sf, 0, sizeof(sf));
119 dri_bo *reloc_bufs[2];
120
121 /* CACHE_NEW_SF_PROG */
122 sf.thread0.grf_reg_count = ALIGN(brw->sf.prog_data->total_grf, 16) / 16 - 1;
123 sf.thread0.kernel_start_pointer = brw->sf.prog_bo->offset >> 6; /* reloc */
124 sf.thread3.urb_entry_read_length = brw->sf.prog_data->urb_read_length;
125
126 sf.thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
127 sf.thread3.dispatch_grf_start_reg = 3;
128 sf.thread3.urb_entry_read_offset = 1;
129
130 /* BRW_NEW_URB_FENCE */
131 sf.thread4.nr_urb_entries = brw->urb.nr_sf_entries;
132 sf.thread4.urb_entry_allocation_size = brw->urb.sfsize - 1;
133 sf.thread4.max_threads = MIN2(12, brw->urb.nr_sf_entries / 2) - 1;
134
135 if (INTEL_DEBUG & DEBUG_SINGLE_THREAD)
136 sf.thread4.max_threads = 0;
137
138 if (INTEL_DEBUG & DEBUG_STATS)
139 sf.thread4.stats_enable = 1;
140
141 /* CACHE_NEW_SF_VP */
142 sf.sf5.sf_viewport_state_offset = brw->sf.vp_bo->offset >> 5; /* reloc */
143
144 sf.sf5.viewport_transform = 1;
145
146 /* _NEW_SCISSOR */
147 if (brw->attribs.Scissor->Enabled)
148 sf.sf6.scissor = 1;
149
150 /* _NEW_POLYGON */
151 if (brw->attribs.Polygon->FrontFace == GL_CCW)
152 sf.sf5.front_winding = BRW_FRONTWINDING_CCW;
153 else
154 sf.sf5.front_winding = BRW_FRONTWINDING_CW;
155
156 if (brw->attribs.Polygon->CullFlag) {
157 switch (brw->attribs.Polygon->CullFaceMode) {
158 case GL_FRONT:
159 sf.sf6.cull_mode = BRW_CULLMODE_FRONT;
160 break;
161 case GL_BACK:
162 sf.sf6.cull_mode = BRW_CULLMODE_BACK;
163 break;
164 case GL_FRONT_AND_BACK:
165 sf.sf6.cull_mode = BRW_CULLMODE_BOTH;
166 break;
167 default:
168 assert(0);
169 break;
170 }
171 }
172 else
173 sf.sf6.cull_mode = BRW_CULLMODE_NONE;
174
175
176 /* _NEW_LINE */
177 /* XXX use ctx->Const.Min/MaxLineWidth here */
178 sf.sf6.line_width = CLAMP(brw->attribs.Line->Width, 1.0, 5.0) * (1<<1);
179
180 sf.sf6.line_endcap_aa_region_width = 1;
181 if (brw->attribs.Line->SmoothFlag)
182 sf.sf6.aa_enable = 1;
183 else if (sf.sf6.line_width <= 0x2)
184 sf.sf6.line_width = 0;
185
186 /* _NEW_POINT */
187 sf.sf6.point_rast_rule = 1; /* opengl conventions */
188 /* XXX clamp max depends on AA vs. non-AA */
189
190 sf.sf7.sprite_point = brw->attribs.Point->PointSprite;
191 sf.sf7.point_size = CLAMP(brw->attribs.Point->Size, 1.0, 255.0) * (1<<3);
192 sf.sf7.use_point_size_state = !brw->attribs.Point->_Attenuated;
193
194 /* might be BRW_NEW_PRIMITIVE if we have to adjust pv for polygons:
195 */
196 sf.sf7.trifan_pv = 2;
197 sf.sf7.linestrip_pv = 1;
198 sf.sf7.tristrip_pv = 2;
199 sf.sf7.line_last_pixel_enable = 0;
200
201 /* Set bias for OpenGL rasterization rules:
202 */
203 sf.sf6.dest_org_vbias = 0x8;
204 sf.sf6.dest_org_hbias = 0x8;
205
206 reloc_bufs[0] = brw->sf.prog_bo;
207 reloc_bufs[1] = brw->sf.vp_bo;
208
209 brw->sf.thread0_delta = sf.thread0.grf_reg_count << 1;
210 brw->sf.sf5_delta = sf.sf5.front_winding | (sf.sf5.viewport_transform << 1);
211
212 dri_bo_unreference(brw->sf.state_bo);
213 brw->sf.state_bo = brw_cache_data( &brw->cache, BRW_SF_UNIT, &sf,
214 reloc_bufs, 2 );
215 }
216
217 static void emit_reloc_sf_unit(struct brw_context *brw)
218 {
219 /* Emit SF program relocation */
220 dri_emit_reloc(brw->sf.state_bo,
221 DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
222 brw->sf.thread0_delta,
223 offsetof(struct brw_sf_unit_state, thread0),
224 brw->sf.prog_bo);
225
226 /* Emit SF viewport relocation */
227 dri_emit_reloc(brw->sf.state_bo,
228 DRM_BO_FLAG_MEM_TT | DRM_BO_FLAG_READ,
229 brw->sf.sf5_delta,
230 offsetof(struct brw_sf_unit_state, sf5),
231 brw->sf.vp_bo);
232 }
233
234 const struct brw_tracked_state brw_sf_unit = {
235 .dirty = {
236 .mesa = (_NEW_POLYGON |
237 _NEW_LINE |
238 _NEW_POINT |
239 _NEW_SCISSOR),
240 .brw = (BRW_NEW_URB_FENCE |
241 BRW_NEW_METAOPS),
242 .cache = (CACHE_NEW_SF_VP |
243 CACHE_NEW_SF_PROG)
244 },
245 .update = upload_sf_unit,
246 .emit_reloc = emit_reloc_sf_unit,
247 };