Home | History | Annotate | Download | only in r200
      1 /*
      2 Copyright (C) The Weather Channel, Inc.  2002.  All Rights Reserved.
      3 
      4 The Weather Channel (TM) funded Tungsten Graphics to develop the
      5 initial release of the Radeon 8500 driver under the XFree86 license.
      6 This notice must be preserved.
      7 
      8 Permission is hereby granted, free of charge, to any person obtaining
      9 a copy of this software and associated documentation files (the
     10 "Software"), to deal in the Software without restriction, including
     11 without limitation the rights to use, copy, modify, merge, publish,
     12 distribute, sublicense, and/or sell copies of the Software, and to
     13 permit persons to whom the Software is furnished to do so, subject to
     14 the following conditions:
     15 
     16 The above copyright notice and this permission notice (including the
     17 next paragraph) shall be included in all copies or substantial
     18 portions of the Software.
     19 
     20 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
     21 EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
     22 MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
     23 IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
     24 LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
     25 OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
     26 WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
     27 */
     28 
     29 /*
     30  * Authors:
     31  *   Keith Whitwell <keithw (at) vmware.com>
     32  */
     33 
     34 #include "main/glheader.h"
     35 #include "main/imports.h"
     36 #include "main/enums.h"
     37 #include "main/api_arrayelt.h"
     38 
     39 #include "swrast/swrast.h"
     40 #include "vbo/vbo.h"
     41 #include "tnl/t_pipeline.h"
     42 #include "swrast_setup/swrast_setup.h"
     43 
     44 #include "radeon_common.h"
     45 #include "radeon_mipmap_tree.h"
     46 #include "r200_context.h"
     47 #include "r200_ioctl.h"
     48 #include "r200_state.h"
     49 #include "radeon_queryobj.h"
     50 
     51 #include "xmlpool.h"
     52 
     53 /* New (1.3) state mechanism.  3 commands (packet, scalar, vector) in
     54  * 1.3 cmdbuffers allow all previous state to be updated as well as
     55  * the tcl scalar and vector areas.
     56  */
     57 static struct {
     58 	int start;
     59 	int len;
     60 	const char *name;
     61 } packet[RADEON_MAX_STATE_PACKETS] = {
     62 	{RADEON_PP_MISC, 7, "RADEON_PP_MISC"},
     63 	{RADEON_PP_CNTL, 3, "RADEON_PP_CNTL"},
     64 	{RADEON_RB3D_COLORPITCH, 1, "RADEON_RB3D_COLORPITCH"},
     65 	{RADEON_RE_LINE_PATTERN, 2, "RADEON_RE_LINE_PATTERN"},
     66 	{RADEON_SE_LINE_WIDTH, 1, "RADEON_SE_LINE_WIDTH"},
     67 	{RADEON_PP_LUM_MATRIX, 1, "RADEON_PP_LUM_MATRIX"},
     68 	{RADEON_PP_ROT_MATRIX_0, 2, "RADEON_PP_ROT_MATRIX_0"},
     69 	{RADEON_RB3D_STENCILREFMASK, 3, "RADEON_RB3D_STENCILREFMASK"},
     70 	{RADEON_SE_VPORT_XSCALE, 6, "RADEON_SE_VPORT_XSCALE"},
     71 	{RADEON_SE_CNTL, 2, "RADEON_SE_CNTL"},
     72 	{RADEON_SE_CNTL_STATUS, 1, "RADEON_SE_CNTL_STATUS"},
     73 	{RADEON_RE_MISC, 1, "RADEON_RE_MISC"},
     74 	{RADEON_PP_TXFILTER_0, 6, "RADEON_PP_TXFILTER_0"},
     75 	{RADEON_PP_BORDER_COLOR_0, 1, "RADEON_PP_BORDER_COLOR_0"},
     76 	{RADEON_PP_TXFILTER_1, 6, "RADEON_PP_TXFILTER_1"},
     77 	{RADEON_PP_BORDER_COLOR_1, 1, "RADEON_PP_BORDER_COLOR_1"},
     78 	{RADEON_PP_TXFILTER_2, 6, "RADEON_PP_TXFILTER_2"},
     79 	{RADEON_PP_BORDER_COLOR_2, 1, "RADEON_PP_BORDER_COLOR_2"},
     80 	{RADEON_SE_ZBIAS_FACTOR, 2, "RADEON_SE_ZBIAS_FACTOR"},
     81 	{RADEON_SE_TCL_OUTPUT_VTX_FMT, 11, "RADEON_SE_TCL_OUTPUT_VTX_FMT"},
     82 	{RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED, 17,
     83 		    "RADEON_SE_TCL_MATERIAL_EMMISSIVE_RED"},
     84 	{R200_PP_TXCBLEND_0, 4, "R200_PP_TXCBLEND_0"},
     85 	{R200_PP_TXCBLEND_1, 4, "R200_PP_TXCBLEND_1"},
     86 	{R200_PP_TXCBLEND_2, 4, "R200_PP_TXCBLEND_2"},
     87 	{R200_PP_TXCBLEND_3, 4, "R200_PP_TXCBLEND_3"},
     88 	{R200_PP_TXCBLEND_4, 4, "R200_PP_TXCBLEND_4"},
     89 	{R200_PP_TXCBLEND_5, 4, "R200_PP_TXCBLEND_5"},
     90 	{R200_PP_TXCBLEND_6, 4, "R200_PP_TXCBLEND_6"},
     91 	{R200_PP_TXCBLEND_7, 4, "R200_PP_TXCBLEND_7"},
     92 	{R200_SE_TCL_LIGHT_MODEL_CTL_0, 6, "R200_SE_TCL_LIGHT_MODEL_CTL_0"},
     93 	{R200_PP_TFACTOR_0, 6, "R200_PP_TFACTOR_0"},
     94 	{R200_SE_VTX_FMT_0, 4, "R200_SE_VTX_FMT_0"},
     95 	{R200_SE_VAP_CNTL, 1, "R200_SE_VAP_CNTL"},
     96 	{R200_SE_TCL_MATRIX_SEL_0, 5, "R200_SE_TCL_MATRIX_SEL_0"},
     97 	{R200_SE_TCL_TEX_PROC_CTL_2, 5, "R200_SE_TCL_TEX_PROC_CTL_2"},
     98 	{R200_SE_TCL_UCP_VERT_BLEND_CTL, 1, "R200_SE_TCL_UCP_VERT_BLEND_CTL"},
     99 	{R200_PP_TXFILTER_0, 6, "R200_PP_TXFILTER_0"},
    100 	{R200_PP_TXFILTER_1, 6, "R200_PP_TXFILTER_1"},
    101 	{R200_PP_TXFILTER_2, 6, "R200_PP_TXFILTER_2"},
    102 	{R200_PP_TXFILTER_3, 6, "R200_PP_TXFILTER_3"},
    103 	{R200_PP_TXFILTER_4, 6, "R200_PP_TXFILTER_4"},
    104 	{R200_PP_TXFILTER_5, 6, "R200_PP_TXFILTER_5"},
    105 	{R200_PP_TXOFFSET_0, 1, "R200_PP_TXOFFSET_0"},
    106 	{R200_PP_TXOFFSET_1, 1, "R200_PP_TXOFFSET_1"},
    107 	{R200_PP_TXOFFSET_2, 1, "R200_PP_TXOFFSET_2"},
    108 	{R200_PP_TXOFFSET_3, 1, "R200_PP_TXOFFSET_3"},
    109 	{R200_PP_TXOFFSET_4, 1, "R200_PP_TXOFFSET_4"},
    110 	{R200_PP_TXOFFSET_5, 1, "R200_PP_TXOFFSET_5"},
    111 	{R200_SE_VTE_CNTL, 1, "R200_SE_VTE_CNTL"},
    112 	{R200_SE_TCL_OUTPUT_VTX_COMP_SEL, 1,
    113 	 "R200_SE_TCL_OUTPUT_VTX_COMP_SEL"},
    114 	{R200_PP_TAM_DEBUG3, 1, "R200_PP_TAM_DEBUG3"},
    115 	{R200_PP_CNTL_X, 1, "R200_PP_CNTL_X"},
    116 	{R200_RB3D_DEPTHXY_OFFSET, 1, "R200_RB3D_DEPTHXY_OFFSET"},
    117 	{R200_RE_AUX_SCISSOR_CNTL, 1, "R200_RE_AUX_SCISSOR_CNTL"},
    118 	{R200_RE_SCISSOR_TL_0, 2, "R200_RE_SCISSOR_TL_0"},
    119 	{R200_RE_SCISSOR_TL_1, 2, "R200_RE_SCISSOR_TL_1"},
    120 	{R200_RE_SCISSOR_TL_2, 2, "R200_RE_SCISSOR_TL_2"},
    121 	{R200_SE_VAP_CNTL_STATUS, 1, "R200_SE_VAP_CNTL_STATUS"},
    122 	{R200_SE_VTX_STATE_CNTL, 1, "R200_SE_VTX_STATE_CNTL"},
    123 	{R200_RE_POINTSIZE, 1, "R200_RE_POINTSIZE"},
    124 	{R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0, 4,
    125 		    "R200_SE_TCL_INPUT_VTX_VECTOR_ADDR_0"},
    126 	{R200_PP_CUBIC_FACES_0, 1, "R200_PP_CUBIC_FACES_0"},	/* 61 */
    127 	{R200_PP_CUBIC_OFFSET_F1_0, 5, "R200_PP_CUBIC_OFFSET_F1_0"}, /* 62 */
    128 	{R200_PP_CUBIC_FACES_1, 1, "R200_PP_CUBIC_FACES_1"},
    129 	{R200_PP_CUBIC_OFFSET_F1_1, 5, "R200_PP_CUBIC_OFFSET_F1_1"},
    130 	{R200_PP_CUBIC_FACES_2, 1, "R200_PP_CUBIC_FACES_2"},
    131 	{R200_PP_CUBIC_OFFSET_F1_2, 5, "R200_PP_CUBIC_OFFSET_F1_2"},
    132 	{R200_PP_CUBIC_FACES_3, 1, "R200_PP_CUBIC_FACES_3"},
    133 	{R200_PP_CUBIC_OFFSET_F1_3, 5, "R200_PP_CUBIC_OFFSET_F1_3"},
    134 	{R200_PP_CUBIC_FACES_4, 1, "R200_PP_CUBIC_FACES_4"},
    135 	{R200_PP_CUBIC_OFFSET_F1_4, 5, "R200_PP_CUBIC_OFFSET_F1_4"},
    136 	{R200_PP_CUBIC_FACES_5, 1, "R200_PP_CUBIC_FACES_5"},
    137 	{R200_PP_CUBIC_OFFSET_F1_5, 5, "R200_PP_CUBIC_OFFSET_F1_5"},
    138 	{RADEON_PP_TEX_SIZE_0, 2, "RADEON_PP_TEX_SIZE_0"},
    139 	{RADEON_PP_TEX_SIZE_1, 2, "RADEON_PP_TEX_SIZE_1"},
    140 	{RADEON_PP_TEX_SIZE_2, 2, "RADEON_PP_TEX_SIZE_2"},
    141 	{R200_RB3D_BLENDCOLOR, 3, "R200_RB3D_BLENDCOLOR"},
    142 	{R200_SE_TCL_POINT_SPRITE_CNTL, 1, "R200_SE_TCL_POINT_SPRITE_CNTL"},
    143 	{RADEON_PP_CUBIC_FACES_0, 1, "RADEON_PP_CUBIC_FACES_0"},
    144 	{RADEON_PP_CUBIC_OFFSET_T0_0, 5, "RADEON_PP_CUBIC_OFFSET_T0_0"},
    145 	{RADEON_PP_CUBIC_FACES_1, 1, "RADEON_PP_CUBIC_FACES_1"},
    146 	{RADEON_PP_CUBIC_OFFSET_T1_0, 5, "RADEON_PP_CUBIC_OFFSET_T1_0"},
    147 	{RADEON_PP_CUBIC_FACES_2, 1, "RADEON_PP_CUBIC_FACES_2"},
    148 	{RADEON_PP_CUBIC_OFFSET_T2_0, 5, "RADEON_PP_CUBIC_OFFSET_T2_0"},
    149 	{R200_PP_TRI_PERF, 2, "R200_PP_TRI_PERF"},
    150 	{R200_PP_TXCBLEND_8, 32, "R200_PP_AFS_0"},     /* 85 */
    151 	{R200_PP_TXCBLEND_0, 32, "R200_PP_AFS_1"},
    152 	{R200_PP_TFACTOR_0, 8, "R200_ATF_TFACTOR"},
    153 	{R200_PP_TXFILTER_0, 8, "R200_PP_TXCTLALL_0"},
    154 	{R200_PP_TXFILTER_1, 8, "R200_PP_TXCTLALL_1"},
    155 	{R200_PP_TXFILTER_2, 8, "R200_PP_TXCTLALL_2"},
    156 	{R200_PP_TXFILTER_3, 8, "R200_PP_TXCTLALL_3"},
    157 	{R200_PP_TXFILTER_4, 8, "R200_PP_TXCTLALL_4"},
    158 	{R200_PP_TXFILTER_5, 8, "R200_PP_TXCTLALL_5"},
    159 	{R200_VAP_PVS_CNTL_1, 2, "R200_VAP_PVS_CNTL"},
    160 };
    161 
    162 /* =============================================================
    163  * State initialization
    164  */
    165 static int cmdpkt( r200ContextPtr rmesa, int id )
    166 {
    167    return CP_PACKET0(packet[id].start, packet[id].len - 1);
    168 }
    169 
    170 static int cmdvec( int offset, int stride, int count )
    171 {
    172    drm_radeon_cmd_header_t h;
    173    h.i = 0;
    174    h.vectors.cmd_type = RADEON_CMD_VECTORS;
    175    h.vectors.offset = offset;
    176    h.vectors.stride = stride;
    177    h.vectors.count = count;
    178    return h.i;
    179 }
    180 
    181 /* warning: the count here is divided by 4 compared to other cmds
    182    (so it doesn't exceed the char size)! */
    183 static int cmdveclinear( int offset, int count )
    184 {
    185    drm_radeon_cmd_header_t h;
    186    h.i = 0;
    187    h.veclinear.cmd_type = RADEON_CMD_VECLINEAR;
    188    h.veclinear.addr_lo = offset & 0xff;
    189    h.veclinear.addr_hi = (offset & 0xff00) >> 8;
    190    h.veclinear.count = count;
    191    return h.i;
    192 }
    193 
    194 static int cmdscl( int offset, int stride, int count )
    195 {
    196    drm_radeon_cmd_header_t h;
    197    h.i = 0;
    198    h.scalars.cmd_type = RADEON_CMD_SCALARS;
    199    h.scalars.offset = offset;
    200    h.scalars.stride = stride;
    201    h.scalars.count = count;
    202    return h.i;
    203 }
    204 
    205 static int cmdscl2( int offset, int stride, int count )
    206 {
    207    drm_radeon_cmd_header_t h;
    208    h.i = 0;
    209    h.scalars.cmd_type = RADEON_CMD_SCALARS2;
    210    h.scalars.offset = offset - 0x100;
    211    h.scalars.stride = stride;
    212    h.scalars.count = count;
    213    return h.i;
    214 }
    215 
    216 /**
    217  * Check functions are used to check if state is active.
    218  * If it is active check function returns maximum emit size.
    219  */
    220 #define CHECK( NM, FLAG, ADD )				\
    221 static int check_##NM( struct gl_context *ctx, struct radeon_state_atom *atom) \
    222 {							\
    223    r200ContextPtr rmesa = R200_CONTEXT(ctx);		\
    224    (void) rmesa;					\
    225    return (FLAG) ? atom->cmd_size + (ADD) : 0;			\
    226 }
    227 
    228 #define TCL_CHECK( NM, FLAG, ADD )				\
    229 static int check_##NM( struct gl_context *ctx, struct radeon_state_atom *atom) \
    230 {									\
    231    r200ContextPtr rmesa = R200_CONTEXT(ctx);				\
    232    return (!rmesa->radeon.TclFallback && !ctx->VertexProgram._Enabled && (FLAG)) ? atom->cmd_size + (ADD) : 0; \
    233 }
    234 
    235 #define TCL_OR_VP_CHECK( NM, FLAG, ADD )			\
    236 static int check_##NM( struct gl_context *ctx, struct radeon_state_atom *atom ) \
    237 {							\
    238    r200ContextPtr rmesa = R200_CONTEXT(ctx);		\
    239    return (!rmesa->radeon.TclFallback && (FLAG)) ? atom->cmd_size + (ADD) : 0;	\
    240 }
    241 
    242 #define VP_CHECK( NM, FLAG, ADD )				\
    243 static int check_##NM( struct gl_context *ctx, struct radeon_state_atom *atom ) \
    244 {									\
    245    r200ContextPtr rmesa = R200_CONTEXT(ctx);				\
    246    (void) atom;								\
    247    return (!rmesa->radeon.TclFallback && ctx->VertexProgram._Enabled && (FLAG)) ? atom->cmd_size + (ADD) : 0; \
    248 }
    249 
    250 CHECK( always, GL_TRUE, 0 )
    251 CHECK( always_add4, GL_TRUE, 4 )
    252 CHECK( never, GL_FALSE, 0 )
    253 CHECK( tex_any, ctx->Texture._MaxEnabledTexImageUnit != -1, 0 )
    254 CHECK( tf, (ctx->Texture._MaxEnabledTexImageUnit != -1 && !ctx->ATIFragmentShader._Enabled), 0 );
    255 CHECK( pix_zero, !ctx->ATIFragmentShader._Enabled, 0 )
    256 CHECK( texenv, (rmesa->state.envneeded & (1 << (atom->idx)) && !ctx->ATIFragmentShader._Enabled), 0 )
    257 CHECK( afs_pass1, (ctx->ATIFragmentShader._Enabled && (ctx->ATIFragmentShader.Current->NumPasses > 1)), 0 )
    258 CHECK( afs, ctx->ATIFragmentShader._Enabled, 0 )
    259 CHECK( tex_cube, rmesa->state.texture.unit[atom->idx].unitneeded & TEXTURE_CUBE_BIT, 3 + 3*5 - CUBE_STATE_SIZE )
    260 CHECK( tex_cube_cs, rmesa->state.texture.unit[atom->idx].unitneeded & TEXTURE_CUBE_BIT, 2 + 4*5 - CUBE_STATE_SIZE )
    261 TCL_CHECK( tcl_fog_add4, ctx->Fog.Enabled, 4 )
    262 TCL_CHECK( tcl, GL_TRUE, 0 )
    263 TCL_CHECK( tcl_add8, GL_TRUE, 8 )
    264 TCL_CHECK( tcl_add4, GL_TRUE, 4 )
    265 TCL_CHECK( tcl_tex_add4, rmesa->state.texture.unit[atom->idx].unitneeded, 4 )
    266 TCL_CHECK( tcl_lighting_add4, ctx->Light.Enabled, 4 )
    267 TCL_CHECK( tcl_lighting_add6, ctx->Light.Enabled, 6 )
    268 TCL_CHECK( tcl_light_add6, ctx->Light.Enabled && ctx->Light.Light[atom->idx].Enabled, 6 )
    269 TCL_OR_VP_CHECK( tcl_ucp_add4, (ctx->Transform.ClipPlanesEnabled & (1 << (atom->idx))), 4 )
    270 TCL_OR_VP_CHECK( tcl_or_vp, GL_TRUE, 0 )
    271 TCL_OR_VP_CHECK( tcl_or_vp_add2, GL_TRUE, 2 )
    272 VP_CHECK( tcl_vp, GL_TRUE, 0 )
    273 VP_CHECK( tcl_vp_add4, GL_TRUE, 4 )
    274 VP_CHECK( tcl_vp_size_add4, ctx->VertexProgram.Current->arb.NumNativeInstructions > 64, 4 )
    275 VP_CHECK( tcl_vpp_size_add4, ctx->VertexProgram.Current->arb.NumNativeParameters > 96, 4 )
    276 
    277 #define OUT_VEC(hdr, data) do {			\
    278     drm_radeon_cmd_header_t h;					\
    279     h.i = hdr;								\
    280     OUT_BATCH(CP_PACKET0(RADEON_SE_TCL_STATE_FLUSH, 0));		\
    281     OUT_BATCH(0);							\
    282     OUT_BATCH(CP_PACKET0(R200_SE_TCL_VECTOR_INDX_REG, 0));		\
    283     OUT_BATCH(h.vectors.offset | (h.vectors.stride << RADEON_VEC_INDX_OCTWORD_STRIDE_SHIFT)); \
    284     OUT_BATCH(CP_PACKET0_ONE(R200_SE_TCL_VECTOR_DATA_REG, h.vectors.count - 1));	\
    285     OUT_BATCH_TABLE((data), h.vectors.count);				\
    286   } while(0)
    287 
    288 #define OUT_VECLINEAR(hdr, data) do {					\
    289     drm_radeon_cmd_header_t h;						\
    290     uint32_t _start, _sz;						\
    291     h.i = hdr;								\
    292     _start = h.veclinear.addr_lo | (h.veclinear.addr_hi << 8);		\
    293     _sz = h.veclinear.count * 4;					\
    294     if (_sz) {								\
    295     BEGIN_BATCH(dwords); \
    296     OUT_BATCH(CP_PACKET0(RADEON_SE_TCL_STATE_FLUSH, 0));		\
    297     OUT_BATCH(0);							\
    298     OUT_BATCH(CP_PACKET0(R200_SE_TCL_VECTOR_INDX_REG, 0));		\
    299     OUT_BATCH(_start | (1 << RADEON_VEC_INDX_OCTWORD_STRIDE_SHIFT));	\
    300     OUT_BATCH(CP_PACKET0_ONE(R200_SE_TCL_VECTOR_DATA_REG, _sz - 1));	\
    301     OUT_BATCH_TABLE((data), _sz);					\
    302     END_BATCH(); \
    303     } \
    304   } while(0)
    305 
    306 #define OUT_SCL(hdr, data) do {					\
    307     drm_radeon_cmd_header_t h;						\
    308     h.i = hdr;								\
    309     OUT_BATCH(CP_PACKET0(R200_SE_TCL_SCALAR_INDX_REG, 0));		\
    310     OUT_BATCH((h.scalars.offset) | (h.scalars.stride << RADEON_SCAL_INDX_DWORD_STRIDE_SHIFT)); \
    311     OUT_BATCH(CP_PACKET0_ONE(R200_SE_TCL_SCALAR_DATA_REG, h.scalars.count - 1));	\
    312     OUT_BATCH_TABLE((data), h.scalars.count);				\
    313   } while(0)
    314 
    315 #define OUT_SCL2(hdr, data) do {					\
    316     drm_radeon_cmd_header_t h;						\
    317     h.i = hdr;								\
    318     OUT_BATCH(CP_PACKET0(R200_SE_TCL_SCALAR_INDX_REG, 0));		\
    319     OUT_BATCH((h.scalars.offset + 0x100) | (h.scalars.stride << RADEON_SCAL_INDX_DWORD_STRIDE_SHIFT)); \
    320     OUT_BATCH(CP_PACKET0_ONE(R200_SE_TCL_SCALAR_DATA_REG, h.scalars.count - 1));	\
    321     OUT_BATCH_TABLE((data), h.scalars.count);				\
    322   } while(0)
    323 static int check_rrb(struct gl_context *ctx, struct radeon_state_atom *atom)
    324 {
    325    r200ContextPtr r200 = R200_CONTEXT(ctx);
    326    struct radeon_renderbuffer *rrb;
    327    rrb = radeon_get_colorbuffer(&r200->radeon);
    328    if (!rrb || !rrb->bo)
    329       return 0;
    330    return atom->cmd_size;
    331 }
    332 
    333 static int check_polygon_stipple(struct gl_context *ctx,
    334 		struct radeon_state_atom *atom)
    335 {
    336    r200ContextPtr r200 = R200_CONTEXT(ctx);
    337    if (r200->hw.set.cmd[SET_RE_CNTL] & R200_STIPPLE_ENABLE)
    338 	   return atom->cmd_size;
    339    return 0;
    340 }
    341 
    342 static void mtl_emit(struct gl_context *ctx, struct radeon_state_atom *atom)
    343 {
    344    r200ContextPtr r200 = R200_CONTEXT(ctx);
    345    BATCH_LOCALS(&r200->radeon);
    346    uint32_t dwords = atom->check(ctx, atom);
    347 
    348    BEGIN_BATCH(dwords);
    349    OUT_VEC(atom->cmd[MTL_CMD_0], (atom->cmd+1));
    350    OUT_SCL2(atom->cmd[MTL_CMD_1], (atom->cmd + 18));
    351    END_BATCH();
    352 }
    353 
    354 static void lit_emit(struct gl_context *ctx, struct radeon_state_atom *atom)
    355 {
    356    r200ContextPtr r200 = R200_CONTEXT(ctx);
    357    BATCH_LOCALS(&r200->radeon);
    358    uint32_t dwords = atom->check(ctx, atom);
    359 
    360    BEGIN_BATCH(dwords);
    361    OUT_VEC(atom->cmd[LIT_CMD_0], atom->cmd+1);
    362    OUT_SCL(atom->cmd[LIT_CMD_1], atom->cmd+LIT_CMD_1+1);
    363    END_BATCH();
    364 }
    365 
    366 static void ptp_emit(struct gl_context *ctx, struct radeon_state_atom *atom)
    367 {
    368    r200ContextPtr r200 = R200_CONTEXT(ctx);
    369    BATCH_LOCALS(&r200->radeon);
    370    uint32_t dwords = atom->check(ctx, atom);
    371 
    372    BEGIN_BATCH(dwords);
    373    OUT_VEC(atom->cmd[PTP_CMD_0], atom->cmd+1);
    374    OUT_VEC(atom->cmd[PTP_CMD_1], atom->cmd+PTP_CMD_1+1);
    375    END_BATCH();
    376 }
    377 
    378 static void veclinear_emit(struct gl_context *ctx, struct radeon_state_atom *atom)
    379 {
    380    r200ContextPtr r200 = R200_CONTEXT(ctx);
    381    BATCH_LOCALS(&r200->radeon);
    382    uint32_t dwords = atom->check(ctx, atom);
    383 
    384    OUT_VECLINEAR(atom->cmd[0], atom->cmd+1);
    385 }
    386 
    387 static void scl_emit(struct gl_context *ctx, struct radeon_state_atom *atom)
    388 {
    389    r200ContextPtr r200 = R200_CONTEXT(ctx);
    390    BATCH_LOCALS(&r200->radeon);
    391    uint32_t dwords = atom->check(ctx, atom);
    392 
    393    BEGIN_BATCH(dwords);
    394    OUT_SCL(atom->cmd[0], atom->cmd+1);
    395    END_BATCH();
    396 }
    397 
    398 
    399 static void vec_emit(struct gl_context *ctx, struct radeon_state_atom *atom)
    400 {
    401    r200ContextPtr r200 = R200_CONTEXT(ctx);
    402    BATCH_LOCALS(&r200->radeon);
    403    uint32_t dwords = atom->check(ctx, atom);
    404 
    405    BEGIN_BATCH(dwords);
    406    OUT_VEC(atom->cmd[0], atom->cmd+1);
    407    END_BATCH();
    408 }
    409 
    410 static int check_always_ctx( struct gl_context *ctx, struct radeon_state_atom *atom)
    411 {
    412    r200ContextPtr r200 = R200_CONTEXT(ctx);
    413    struct radeon_renderbuffer *rrb, *drb;
    414    uint32_t dwords;
    415 
    416    rrb = radeon_get_colorbuffer(&r200->radeon);
    417    if (!rrb || !rrb->bo) {
    418       return 0;
    419    }
    420 
    421    drb = radeon_get_depthbuffer(&r200->radeon);
    422 
    423    dwords = 10;
    424    if (drb)
    425      dwords += 6;
    426    if (rrb)
    427      dwords += 8;
    428    if (atom->cmd_size == CTX_STATE_SIZE_NEWDRM)
    429      dwords += 4;
    430 
    431 
    432    return dwords;
    433 }
    434 
    435 static void ctx_emit_cs(struct gl_context *ctx, struct radeon_state_atom *atom)
    436 {
    437    r200ContextPtr r200 = R200_CONTEXT(ctx);
    438    BATCH_LOCALS(&r200->radeon);
    439    struct radeon_renderbuffer *rrb, *drb;
    440    uint32_t cbpitch = 0;
    441    uint32_t zbpitch = 0;
    442    uint32_t dwords = atom->check(ctx, atom);
    443    uint32_t depth_fmt;
    444 
    445    rrb = radeon_get_colorbuffer(&r200->radeon);
    446    if (!rrb || !rrb->bo) {
    447       return;
    448    }
    449 
    450    atom->cmd[CTX_RB3D_CNTL] &= ~(0xf << 10);
    451    if (rrb->cpp == 4)
    452 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_ARGB8888;
    453    else switch (rrb->base.Base.Format) {
    454    case MESA_FORMAT_B5G6R5_UNORM:
    455    case MESA_FORMAT_R5G6B5_UNORM:
    456 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_RGB565;
    457 	break;
    458    case MESA_FORMAT_B4G4R4A4_UNORM:
    459    case MESA_FORMAT_A4R4G4B4_UNORM:
    460 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_ARGB4444;
    461 	break;
    462    case MESA_FORMAT_B5G5R5A1_UNORM:
    463    case MESA_FORMAT_A1R5G5B5_UNORM:
    464 	atom->cmd[CTX_RB3D_CNTL] |= RADEON_COLOR_FORMAT_ARGB1555;
    465 	break;
    466    default:
    467 	_mesa_problem(ctx, "Unexpected format in ctx_emit_cs");
    468    }
    469 
    470    cbpitch = (rrb->pitch / rrb->cpp);
    471    if (rrb->bo->flags & RADEON_BO_FLAGS_MACRO_TILE)
    472        cbpitch |= R200_COLOR_TILE_ENABLE;
    473    if (rrb->bo->flags & RADEON_BO_FLAGS_MICRO_TILE)
    474        cbpitch |= R200_COLOR_MICROTILE_ENABLE;
    475 
    476 
    477    drb = radeon_get_depthbuffer(&r200->radeon);
    478    if (drb) {
    479      zbpitch = (drb->pitch / drb->cpp);
    480      if (drb->cpp == 4)
    481         depth_fmt = RADEON_DEPTH_FORMAT_24BIT_INT_Z;
    482      else
    483         depth_fmt = RADEON_DEPTH_FORMAT_16BIT_INT_Z;
    484      atom->cmd[CTX_RB3D_ZSTENCILCNTL] &= ~RADEON_DEPTH_FORMAT_MASK;
    485      atom->cmd[CTX_RB3D_ZSTENCILCNTL] |= depth_fmt;
    486    }
    487 
    488    /* output the first 7 bytes of context */
    489    BEGIN_BATCH(dwords);
    490 
    491    /* In the CS case we need to split this up */
    492    OUT_BATCH(CP_PACKET0(packet[0].start, 3));
    493    OUT_BATCH_TABLE((atom->cmd + 1), 4);
    494 
    495    if (drb) {
    496      OUT_BATCH(CP_PACKET0(RADEON_RB3D_DEPTHOFFSET, 0));
    497      OUT_BATCH_RELOC(0, drb->bo, 0, 0, RADEON_GEM_DOMAIN_VRAM, 0);
    498 
    499      OUT_BATCH(CP_PACKET0(RADEON_RB3D_DEPTHPITCH, 0));
    500      OUT_BATCH(zbpitch);
    501    }
    502 
    503    OUT_BATCH(CP_PACKET0(RADEON_RB3D_ZSTENCILCNTL, 0));
    504    OUT_BATCH(atom->cmd[CTX_RB3D_ZSTENCILCNTL]);
    505    OUT_BATCH(CP_PACKET0(RADEON_PP_CNTL, 1));
    506    OUT_BATCH(atom->cmd[CTX_PP_CNTL]);
    507    OUT_BATCH(atom->cmd[CTX_RB3D_CNTL]);
    508 
    509 
    510    if (rrb) {
    511      OUT_BATCH(CP_PACKET0(RADEON_RB3D_COLOROFFSET, 0));
    512      OUT_BATCH_RELOC(rrb->draw_offset, rrb->bo, rrb->draw_offset, 0, RADEON_GEM_DOMAIN_VRAM, 0);
    513 
    514      OUT_BATCH(CP_PACKET0(RADEON_RB3D_COLORPITCH, 0));
    515      OUT_BATCH_RELOC(cbpitch, rrb->bo, cbpitch, 0, RADEON_GEM_DOMAIN_VRAM, 0);
    516    }
    517 
    518    if (atom->cmd_size == CTX_STATE_SIZE_NEWDRM) {
    519      OUT_BATCH_TABLE((atom->cmd + 14), 4);
    520    }
    521 
    522    END_BATCH();
    523 }
    524 
    525 static int get_tex_mm_size(struct gl_context* ctx, struct radeon_state_atom *atom)
    526 {
    527    r200ContextPtr r200 = R200_CONTEXT(ctx);
    528    uint32_t dwords = atom->cmd_size + 2;
    529    int hastexture = 1;
    530    int i = atom->idx;
    531    radeonTexObj *t = r200->state.texture.unit[i].texobj;
    532    if (!t)
    533 	hastexture = 0;
    534    else {
    535 	if (!t->mt && !t->bo)
    536 		hastexture = 0;
    537    }
    538 
    539    if (!hastexture)
    540      dwords -= 4;
    541    return dwords;
    542 }
    543 
    544 static int check_tex_pair_mm(struct gl_context* ctx, struct radeon_state_atom *atom)
    545 {
    546    r200ContextPtr r200 = R200_CONTEXT(ctx);
    547    /** XOR is bit flip operation so use it for finding pair */
    548    if (!(r200->state.texture.unit[atom->idx].unitneeded | r200->state.texture.unit[atom->idx ^ 1].unitneeded))
    549      return 0;
    550 
    551    return get_tex_mm_size(ctx, atom);
    552 }
    553 
    554 static int check_tex_mm(struct gl_context* ctx, struct radeon_state_atom *atom)
    555 {
    556    r200ContextPtr r200 = R200_CONTEXT(ctx);
    557    if (!(r200->state.texture.unit[atom->idx].unitneeded))
    558      return 0;
    559 
    560    return get_tex_mm_size(ctx, atom);
    561 }
    562 
    563 
    564 static void tex_emit_mm(struct gl_context *ctx, struct radeon_state_atom *atom)
    565 {
    566    r200ContextPtr r200 = R200_CONTEXT(ctx);
    567    BATCH_LOCALS(&r200->radeon);
    568    uint32_t dwords = atom->check(ctx, atom);
    569    int i = atom->idx;
    570    radeonTexObj *t = r200->state.texture.unit[i].texobj;
    571 
    572    if (!r200->state.texture.unit[i].unitneeded && !(dwords <= atom->cmd_size))
    573         dwords -= 4;
    574    BEGIN_BATCH(dwords);
    575 
    576    OUT_BATCH(CP_PACKET0(R200_PP_TXFILTER_0 + (32 * i), 7));
    577    OUT_BATCH_TABLE((atom->cmd + 1), 8);
    578 
    579    if (dwords > atom->cmd_size) {
    580      OUT_BATCH(CP_PACKET0(R200_PP_TXOFFSET_0 + (24 * i), 0));
    581      if (t->mt && !t->image_override) {
    582         OUT_BATCH_RELOC(t->tile_bits, t->mt->bo, 0,
    583 		  RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
    584       } else {
    585 	if (t->bo)
    586             OUT_BATCH_RELOC(t->tile_bits, t->bo, 0,
    587                             RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
    588       }
    589    }
    590    END_BATCH();
    591 }
    592 
    593 static void cube_emit_cs(struct gl_context *ctx, struct radeon_state_atom *atom)
    594 {
    595    r200ContextPtr r200 = R200_CONTEXT(ctx);
    596    BATCH_LOCALS(&r200->radeon);
    597    uint32_t dwords = atom->check(ctx, atom);
    598    int i = atom->idx, j;
    599    radeonTexObj *t = r200->state.texture.unit[i].texobj;
    600    radeon_mipmap_level *lvl;
    601    if (!(t && !t->image_override))
    602      dwords = 2;
    603 
    604    BEGIN_BATCH(dwords);
    605    OUT_BATCH_TABLE(atom->cmd, 2);
    606 
    607    if (t && !t->image_override) {
    608      lvl = &t->mt->levels[0];
    609      for (j = 1; j <= 5; j++) {
    610        OUT_BATCH(CP_PACKET0(R200_PP_CUBIC_OFFSET_F1_0 + (24*i) + (4 * (j-1)), 0));
    611        OUT_BATCH_RELOC(lvl->faces[j].offset, t->mt->bo, lvl->faces[j].offset,
    612 			RADEON_GEM_DOMAIN_GTT|RADEON_GEM_DOMAIN_VRAM, 0, 0);
    613      }
    614    }
    615    END_BATCH();
    616 }
    617 
    618 /* Initialize the context's hardware state.
    619  */
    620 void r200InitState( r200ContextPtr rmesa )
    621 {
    622    struct gl_context *ctx = &rmesa->radeon.glCtx;
    623    GLuint i;
    624 
    625    rmesa->radeon.Fallback = 0;
    626 
    627    rmesa->radeon.hw.max_state_size = 0;
    628 
    629 #define ALLOC_STATE( ATOM, CHK, SZ, NM, IDX )				\
    630    do {								\
    631       rmesa->hw.ATOM.cmd_size = SZ;				\
    632       rmesa->hw.ATOM.cmd = (GLuint *) calloc(SZ, sizeof(int));          \
    633       rmesa->hw.ATOM.lastcmd = (GLuint *) calloc(SZ, sizeof(int));	\
    634       rmesa->hw.ATOM.name = NM;					\
    635       rmesa->hw.ATOM.idx = IDX;					\
    636       if (check_##CHK != check_never) {				\
    637          rmesa->hw.ATOM.check = check_##CHK;			\
    638          rmesa->radeon.hw.max_state_size += SZ * sizeof(int);	\
    639       } else {							\
    640          rmesa->hw.ATOM.check = NULL;				\
    641       }								\
    642       rmesa->hw.ATOM.dirty = GL_FALSE;				\
    643    } while (0)
    644 
    645 
    646    /* Allocate state buffers:
    647     */
    648    ALLOC_STATE( ctx, always_add4, CTX_STATE_SIZE_NEWDRM, "CTX/context", 0 );
    649 
    650    rmesa->hw.ctx.emit = ctx_emit_cs;
    651    rmesa->hw.ctx.check = check_always_ctx;
    652    ALLOC_STATE( set, always, SET_STATE_SIZE, "SET/setup", 0 );
    653    ALLOC_STATE( lin, always, LIN_STATE_SIZE, "LIN/line", 0 );
    654    ALLOC_STATE( msk, always, MSK_STATE_SIZE, "MSK/mask", 0 );
    655    ALLOC_STATE( vpt, always, VPT_STATE_SIZE, "VPT/viewport", 0 );
    656    ALLOC_STATE( vtx, always, VTX_STATE_SIZE, "VTX/vertex", 0 );
    657    ALLOC_STATE( vap, always, VAP_STATE_SIZE, "VAP/vap", 0 );
    658    ALLOC_STATE( vte, always, VTE_STATE_SIZE, "VTE/vte", 0 );
    659    ALLOC_STATE( msc, always, MSC_STATE_SIZE, "MSC/misc", 0 );
    660    ALLOC_STATE( cst, always, CST_STATE_SIZE, "CST/constant", 0 );
    661    ALLOC_STATE( zbs, always, ZBS_STATE_SIZE, "ZBS/zbias", 0 );
    662    ALLOC_STATE( tf, tf, TF_STATE_SIZE, "TF/tfactor", 0 );
    663    {
    664       int state_size = TEX_STATE_SIZE_NEWDRM;
    665       if (rmesa->radeon.radeonScreen->chip_family == CHIP_FAMILY_R200) {
    666          /* make sure texture units 0/1 are emitted pair-wise for r200 t0 hang workaround */
    667          ALLOC_STATE( tex[0], tex_pair_mm, state_size, "TEX/tex-0", 0 );
    668          ALLOC_STATE( tex[1], tex_pair_mm, state_size, "TEX/tex-1", 1 );
    669          ALLOC_STATE( tam, tex_any, TAM_STATE_SIZE, "TAM/tam", 0 );
    670       }
    671       else {
    672          ALLOC_STATE( tex[0], tex_mm, state_size, "TEX/tex-0", 0 );
    673          ALLOC_STATE( tex[1], tex_mm, state_size, "TEX/tex-1", 1 );
    674          ALLOC_STATE( tam, never, TAM_STATE_SIZE, "TAM/tam", 0 );
    675       }
    676       ALLOC_STATE( tex[2], tex_mm, state_size, "TEX/tex-2", 2 );
    677       ALLOC_STATE( tex[3], tex_mm, state_size, "TEX/tex-3", 3 );
    678       ALLOC_STATE( tex[4], tex_mm, state_size, "TEX/tex-4", 4 );
    679       ALLOC_STATE( tex[5], tex_mm, state_size, "TEX/tex-5", 5 );
    680       ALLOC_STATE( atf, afs, ATF_STATE_SIZE, "ATF/tfactor", 0 );
    681       ALLOC_STATE( afs[0], afs_pass1, AFS_STATE_SIZE, "AFS/afsinst-0", 0 );
    682       ALLOC_STATE( afs[1], afs, AFS_STATE_SIZE, "AFS/afsinst-1", 1 );
    683    }
    684 
    685    ALLOC_STATE( stp, polygon_stipple, STP_STATE_SIZE, "STP/stp", 0 );
    686 
    687    for (i = 0; i < 6; i++)
    688       rmesa->hw.tex[i].emit = tex_emit_mm;
    689    ALLOC_STATE( cube[0], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-0", 0 );
    690    ALLOC_STATE( cube[1], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-1", 1 );
    691    ALLOC_STATE( cube[2], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-2", 2 );
    692    ALLOC_STATE( cube[3], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-3", 3 );
    693    ALLOC_STATE( cube[4], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-4", 4 );
    694    ALLOC_STATE( cube[5], tex_cube, CUBE_STATE_SIZE, "CUBE/tex-5", 5 );
    695    for (i = 0; i < 6; i++) {
    696       rmesa->hw.cube[i].emit = cube_emit_cs;
    697       rmesa->hw.cube[i].check = check_tex_cube_cs;
    698    }
    699 
    700    ALLOC_STATE( pvs, tcl_vp, PVS_STATE_SIZE, "PVS/pvscntl", 0 );
    701    ALLOC_STATE( vpi[0], tcl_vp_add4, VPI_STATE_SIZE, "VP/vertexprog-0", 0 );
    702    ALLOC_STATE( vpi[1], tcl_vp_size_add4, VPI_STATE_SIZE, "VP/vertexprog-1", 1 );
    703    ALLOC_STATE( vpp[0], tcl_vp_add4, VPP_STATE_SIZE, "VPP/vertexparam-0", 0 );
    704    ALLOC_STATE( vpp[1], tcl_vpp_size_add4, VPP_STATE_SIZE, "VPP/vertexparam-1", 1 );
    705 
    706    /* FIXME: this atom has two commands, we need only one (ucp_vert_blend) for vp */
    707    ALLOC_STATE( tcl, tcl_or_vp, TCL_STATE_SIZE, "TCL/tcl", 0 );
    708    ALLOC_STATE( msl, tcl, MSL_STATE_SIZE, "MSL/matrix-select", 0 );
    709    ALLOC_STATE( tcg, tcl, TCG_STATE_SIZE, "TCG/texcoordgen", 0 );
    710    ALLOC_STATE( mtl[0], tcl_lighting_add6, MTL_STATE_SIZE, "MTL0/material0", 0 );
    711    ALLOC_STATE( mtl[1], tcl_lighting_add6, MTL_STATE_SIZE, "MTL1/material1", 1 );
    712    ALLOC_STATE( grd, tcl_or_vp_add2, GRD_STATE_SIZE, "GRD/guard-band", 0 );
    713    ALLOC_STATE( fog, tcl_fog_add4, FOG_STATE_SIZE, "FOG/fog", 0 );
    714    ALLOC_STATE( glt, tcl_lighting_add4, GLT_STATE_SIZE, "GLT/light-global", 0 );
    715    ALLOC_STATE( eye, tcl_lighting_add4, EYE_STATE_SIZE, "EYE/eye-vector", 0 );
    716    ALLOC_STATE( mat[R200_MTX_MV], tcl_add4, MAT_STATE_SIZE, "MAT/modelview", 0 );
    717    ALLOC_STATE( mat[R200_MTX_IMV], tcl_add4, MAT_STATE_SIZE, "MAT/it-modelview", 0 );
    718    ALLOC_STATE( mat[R200_MTX_MVP], tcl_add4, MAT_STATE_SIZE, "MAT/modelproject", 0 );
    719    ALLOC_STATE( mat[R200_MTX_TEX0], tcl_tex_add4, MAT_STATE_SIZE, "MAT/texmat0", 0 );
    720    ALLOC_STATE( mat[R200_MTX_TEX1], tcl_tex_add4, MAT_STATE_SIZE, "MAT/texmat1", 1 );
    721    ALLOC_STATE( mat[R200_MTX_TEX2], tcl_tex_add4, MAT_STATE_SIZE, "MAT/texmat2", 2 );
    722    ALLOC_STATE( mat[R200_MTX_TEX3], tcl_tex_add4, MAT_STATE_SIZE, "MAT/texmat3", 3 );
    723    ALLOC_STATE( mat[R200_MTX_TEX4], tcl_tex_add4, MAT_STATE_SIZE, "MAT/texmat4", 4 );
    724    ALLOC_STATE( mat[R200_MTX_TEX5], tcl_tex_add4, MAT_STATE_SIZE, "MAT/texmat5", 5 );
    725    ALLOC_STATE( ucp[0], tcl_ucp_add4, UCP_STATE_SIZE, "UCP/userclip-0", 0 );
    726    ALLOC_STATE( ucp[1], tcl_ucp_add4, UCP_STATE_SIZE, "UCP/userclip-1", 1 );
    727    ALLOC_STATE( ucp[2], tcl_ucp_add4, UCP_STATE_SIZE, "UCP/userclip-2", 2 );
    728    ALLOC_STATE( ucp[3], tcl_ucp_add4, UCP_STATE_SIZE, "UCP/userclip-3", 3 );
    729    ALLOC_STATE( ucp[4], tcl_ucp_add4, UCP_STATE_SIZE, "UCP/userclip-4", 4 );
    730    ALLOC_STATE( ucp[5], tcl_ucp_add4, UCP_STATE_SIZE, "UCP/userclip-5", 5 );
    731    ALLOC_STATE( lit[0], tcl_light_add6, LIT_STATE_SIZE, "LIT/light-0", 0 );
    732    ALLOC_STATE( lit[1], tcl_light_add6, LIT_STATE_SIZE, "LIT/light-1", 1 );
    733    ALLOC_STATE( lit[2], tcl_light_add6, LIT_STATE_SIZE, "LIT/light-2", 2 );
    734    ALLOC_STATE( lit[3], tcl_light_add6, LIT_STATE_SIZE, "LIT/light-3", 3 );
    735    ALLOC_STATE( lit[4], tcl_light_add6, LIT_STATE_SIZE, "LIT/light-4", 4 );
    736    ALLOC_STATE( lit[5], tcl_light_add6, LIT_STATE_SIZE, "LIT/light-5", 5 );
    737    ALLOC_STATE( lit[6], tcl_light_add6, LIT_STATE_SIZE, "LIT/light-6", 6 );
    738    ALLOC_STATE( lit[7], tcl_light_add6, LIT_STATE_SIZE, "LIT/light-7", 7 );
    739    ALLOC_STATE( sci, rrb, SCI_STATE_SIZE, "SCI/scissor", 0 );
    740    ALLOC_STATE( pix[0], pix_zero, PIX_STATE_SIZE, "PIX/pixstage-0", 0 );
    741    ALLOC_STATE( pix[1], texenv, PIX_STATE_SIZE, "PIX/pixstage-1", 1 );
    742    ALLOC_STATE( pix[2], texenv, PIX_STATE_SIZE, "PIX/pixstage-2", 2 );
    743    ALLOC_STATE( pix[3], texenv, PIX_STATE_SIZE, "PIX/pixstage-3", 3 );
    744    ALLOC_STATE( pix[4], texenv, PIX_STATE_SIZE, "PIX/pixstage-4", 4 );
    745    ALLOC_STATE( pix[5], texenv, PIX_STATE_SIZE, "PIX/pixstage-5", 5 );
    746    ALLOC_STATE( prf, always, PRF_STATE_SIZE, "PRF/performance-tri", 0 );
    747    ALLOC_STATE( spr, always, SPR_STATE_SIZE, "SPR/pointsprite", 0 );
    748    ALLOC_STATE( ptp, tcl_add8, PTP_STATE_SIZE, "PTP/pointparams", 0 );
    749 
    750    r200SetUpAtomList( rmesa );
    751 
    752    /* Fill in the packet headers:
    753     */
    754    rmesa->hw.ctx.cmd[CTX_CMD_0] = cmdpkt(rmesa, RADEON_EMIT_PP_MISC);
    755    rmesa->hw.ctx.cmd[CTX_CMD_1] = cmdpkt(rmesa, RADEON_EMIT_PP_CNTL);
    756    rmesa->hw.ctx.cmd[CTX_CMD_2] = cmdpkt(rmesa, RADEON_EMIT_RB3D_COLORPITCH);
    757    rmesa->hw.ctx.cmd[CTX_CMD_3] = cmdpkt(rmesa, R200_EMIT_RB3D_BLENDCOLOR);
    758    rmesa->hw.lin.cmd[LIN_CMD_0] = cmdpkt(rmesa, RADEON_EMIT_RE_LINE_PATTERN);
    759    rmesa->hw.lin.cmd[LIN_CMD_1] = cmdpkt(rmesa, RADEON_EMIT_SE_LINE_WIDTH);
    760    rmesa->hw.msk.cmd[MSK_CMD_0] = cmdpkt(rmesa, RADEON_EMIT_RB3D_STENCILREFMASK);
    761    rmesa->hw.vpt.cmd[VPT_CMD_0] = cmdpkt(rmesa, RADEON_EMIT_SE_VPORT_XSCALE);
    762    rmesa->hw.set.cmd[SET_CMD_0] = cmdpkt(rmesa, RADEON_EMIT_SE_CNTL);
    763    rmesa->hw.msc.cmd[MSC_CMD_0] = cmdpkt(rmesa, RADEON_EMIT_RE_MISC);
    764    rmesa->hw.cst.cmd[CST_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_CNTL_X);
    765    rmesa->hw.cst.cmd[CST_CMD_1] = cmdpkt(rmesa, R200_EMIT_RB3D_DEPTHXY_OFFSET);
    766    rmesa->hw.cst.cmd[CST_CMD_2] = cmdpkt(rmesa, R200_EMIT_RE_AUX_SCISSOR_CNTL);
    767    rmesa->hw.cst.cmd[CST_CMD_4] = cmdpkt(rmesa, R200_EMIT_SE_VAP_CNTL_STATUS);
    768    rmesa->hw.cst.cmd[CST_CMD_5] = cmdpkt(rmesa, R200_EMIT_RE_POINTSIZE);
    769    rmesa->hw.cst.cmd[CST_CMD_6] = cmdpkt(rmesa, R200_EMIT_TCL_INPUT_VTX_VECTOR_ADDR_0);
    770    rmesa->hw.tam.cmd[TAM_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TAM_DEBUG3);
    771    rmesa->hw.tf.cmd[TF_CMD_0] = cmdpkt(rmesa, R200_EMIT_TFACTOR_0);
    772    rmesa->hw.atf.cmd[ATF_CMD_0] = cmdpkt(rmesa, R200_EMIT_ATF_TFACTOR);
    773    rmesa->hw.tex[0].cmd[TEX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCTLALL_0);
    774    rmesa->hw.tex[0].cmd[TEX_CMD_1_NEWDRM] = cmdpkt(rmesa, R200_EMIT_PP_TXOFFSET_0);
    775    rmesa->hw.tex[1].cmd[TEX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCTLALL_1);
    776    rmesa->hw.tex[1].cmd[TEX_CMD_1_NEWDRM] = cmdpkt(rmesa, R200_EMIT_PP_TXOFFSET_1);
    777    rmesa->hw.tex[2].cmd[TEX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCTLALL_2);
    778    rmesa->hw.tex[2].cmd[TEX_CMD_1_NEWDRM] = cmdpkt(rmesa, R200_EMIT_PP_TXOFFSET_2);
    779    rmesa->hw.tex[3].cmd[TEX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCTLALL_3);
    780    rmesa->hw.tex[3].cmd[TEX_CMD_1_NEWDRM] = cmdpkt(rmesa, R200_EMIT_PP_TXOFFSET_3);
    781    rmesa->hw.tex[4].cmd[TEX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCTLALL_4);
    782    rmesa->hw.tex[4].cmd[TEX_CMD_1_NEWDRM] = cmdpkt(rmesa, R200_EMIT_PP_TXOFFSET_4);
    783    rmesa->hw.tex[5].cmd[TEX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCTLALL_5);
    784    rmesa->hw.tex[5].cmd[TEX_CMD_1_NEWDRM] = cmdpkt(rmesa, R200_EMIT_PP_TXOFFSET_5);
    785    rmesa->hw.afs[0].cmd[AFS_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_AFS_0);
    786    rmesa->hw.afs[1].cmd[AFS_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_AFS_1);
    787    rmesa->hw.pvs.cmd[PVS_CMD_0] = cmdpkt(rmesa, R200_EMIT_VAP_PVS_CNTL);
    788    rmesa->hw.cube[0].cmd[CUBE_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_FACES_0);
    789    rmesa->hw.cube[0].cmd[CUBE_CMD_1] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_OFFSETS_0);
    790    rmesa->hw.cube[1].cmd[CUBE_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_FACES_1);
    791    rmesa->hw.cube[1].cmd[CUBE_CMD_1] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_OFFSETS_1);
    792    rmesa->hw.cube[2].cmd[CUBE_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_FACES_2);
    793    rmesa->hw.cube[2].cmd[CUBE_CMD_1] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_OFFSETS_2);
    794    rmesa->hw.cube[3].cmd[CUBE_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_FACES_3);
    795    rmesa->hw.cube[3].cmd[CUBE_CMD_1] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_OFFSETS_3);
    796    rmesa->hw.cube[4].cmd[CUBE_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_FACES_4);
    797    rmesa->hw.cube[4].cmd[CUBE_CMD_1] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_OFFSETS_4);
    798    rmesa->hw.cube[5].cmd[CUBE_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_FACES_5);
    799    rmesa->hw.cube[5].cmd[CUBE_CMD_1] = cmdpkt(rmesa, R200_EMIT_PP_CUBIC_OFFSETS_5);
    800    rmesa->hw.pix[0].cmd[PIX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCBLEND_0);
    801    rmesa->hw.pix[1].cmd[PIX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCBLEND_1);
    802    rmesa->hw.pix[2].cmd[PIX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCBLEND_2);
    803    rmesa->hw.pix[3].cmd[PIX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCBLEND_3);
    804    rmesa->hw.pix[4].cmd[PIX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCBLEND_4);
    805    rmesa->hw.pix[5].cmd[PIX_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TXCBLEND_5);
    806    rmesa->hw.zbs.cmd[ZBS_CMD_0] = cmdpkt(rmesa, RADEON_EMIT_SE_ZBIAS_FACTOR);
    807    rmesa->hw.tcl.cmd[TCL_CMD_0] = cmdpkt(rmesa, R200_EMIT_TCL_LIGHT_MODEL_CTL_0);
    808    rmesa->hw.tcl.cmd[TCL_CMD_1] = cmdpkt(rmesa, R200_EMIT_TCL_UCP_VERT_BLEND_CTL);
    809    rmesa->hw.tcg.cmd[TCG_CMD_0] = cmdpkt(rmesa, R200_EMIT_TEX_PROC_CTL_2);
    810    rmesa->hw.msl.cmd[MSL_CMD_0] = cmdpkt(rmesa, R200_EMIT_MATRIX_SELECT_0);
    811    rmesa->hw.vap.cmd[VAP_CMD_0] = cmdpkt(rmesa, R200_EMIT_VAP_CTL);
    812    rmesa->hw.vtx.cmd[VTX_CMD_0] = cmdpkt(rmesa, R200_EMIT_VTX_FMT_0);
    813    rmesa->hw.vtx.cmd[VTX_CMD_1] = cmdpkt(rmesa, R200_EMIT_OUTPUT_VTX_COMP_SEL);
    814    rmesa->hw.vtx.cmd[VTX_CMD_2] = cmdpkt(rmesa, R200_EMIT_SE_VTX_STATE_CNTL);
    815    rmesa->hw.vte.cmd[VTE_CMD_0] = cmdpkt(rmesa, R200_EMIT_VTE_CNTL);
    816    rmesa->hw.prf.cmd[PRF_CMD_0] = cmdpkt(rmesa, R200_EMIT_PP_TRI_PERF_CNTL);
    817    rmesa->hw.spr.cmd[SPR_CMD_0] = cmdpkt(rmesa, R200_EMIT_TCL_POINT_SPRITE_CNTL);
    818 
    819    rmesa->hw.sci.cmd[SCI_CMD_1] = CP_PACKET0(R200_RE_TOP_LEFT, 0);
    820    rmesa->hw.sci.cmd[SCI_CMD_2] = CP_PACKET0(R200_RE_WIDTH_HEIGHT, 0);
    821 
    822    rmesa->hw.stp.cmd[STP_CMD_0] = CP_PACKET0(RADEON_RE_STIPPLE_ADDR, 0);
    823    rmesa->hw.stp.cmd[STP_DATA_0] = 0;
    824    rmesa->hw.stp.cmd[STP_CMD_1] = CP_PACKET0_ONE(RADEON_RE_STIPPLE_DATA, 31);
    825 
    826    rmesa->hw.mtl[0].emit = mtl_emit;
    827    rmesa->hw.mtl[1].emit = mtl_emit;
    828 
    829    rmesa->hw.vpi[0].emit = veclinear_emit;
    830    rmesa->hw.vpi[1].emit = veclinear_emit;
    831    rmesa->hw.vpp[0].emit = veclinear_emit;
    832    rmesa->hw.vpp[1].emit = veclinear_emit;
    833 
    834    rmesa->hw.grd.emit = scl_emit;
    835    rmesa->hw.fog.emit = vec_emit;
    836    rmesa->hw.glt.emit = vec_emit;
    837    rmesa->hw.eye.emit = vec_emit;
    838 
    839    for (i = R200_MTX_MV; i <= R200_MTX_TEX5; i++)
    840       rmesa->hw.mat[i].emit = vec_emit;
    841 
    842    for (i = 0; i < 8; i++)
    843       rmesa->hw.lit[i].emit = lit_emit;
    844 
    845    for (i = 0; i < 6; i++)
    846       rmesa->hw.ucp[i].emit = vec_emit;
    847 
    848    rmesa->hw.ptp.emit = ptp_emit;
    849 
    850    rmesa->hw.mtl[0].cmd[MTL_CMD_0] =
    851       cmdvec( R200_VS_MAT_0_EMISS, 1, 16 );
    852    rmesa->hw.mtl[0].cmd[MTL_CMD_1] =
    853       cmdscl2( R200_SS_MAT_0_SHININESS, 1, 1 );
    854    rmesa->hw.mtl[1].cmd[MTL_CMD_0] =
    855       cmdvec( R200_VS_MAT_1_EMISS, 1, 16 );
    856    rmesa->hw.mtl[1].cmd[MTL_CMD_1] =
    857       cmdscl2( R200_SS_MAT_1_SHININESS, 1, 1 );
    858 
    859    rmesa->hw.vpi[0].cmd[VPI_CMD_0] =
    860       cmdveclinear( R200_PVS_PROG0, 64 );
    861    rmesa->hw.vpi[1].cmd[VPI_CMD_0] =
    862       cmdveclinear( R200_PVS_PROG1, 64 );
    863    rmesa->hw.vpp[0].cmd[VPP_CMD_0] =
    864       cmdveclinear( R200_PVS_PARAM0, 96 );
    865    rmesa->hw.vpp[1].cmd[VPP_CMD_0] =
    866       cmdveclinear( R200_PVS_PARAM1, 96 );
    867 
    868    rmesa->hw.grd.cmd[GRD_CMD_0] =
    869       cmdscl( R200_SS_VERT_GUARD_CLIP_ADJ_ADDR, 1, 4 );
    870    rmesa->hw.fog.cmd[FOG_CMD_0] =
    871       cmdvec( R200_VS_FOG_PARAM_ADDR, 1, 4 );
    872    rmesa->hw.glt.cmd[GLT_CMD_0] =
    873       cmdvec( R200_VS_GLOBAL_AMBIENT_ADDR, 1, 4 );
    874    rmesa->hw.eye.cmd[EYE_CMD_0] =
    875       cmdvec( R200_VS_EYE_VECTOR_ADDR, 1, 4 );
    876 
    877    rmesa->hw.mat[R200_MTX_MV].cmd[MAT_CMD_0] =
    878       cmdvec( R200_VS_MATRIX_0_MV, 1, 16);
    879    rmesa->hw.mat[R200_MTX_IMV].cmd[MAT_CMD_0] =
    880       cmdvec( R200_VS_MATRIX_1_INV_MV, 1, 16);
    881    rmesa->hw.mat[R200_MTX_MVP].cmd[MAT_CMD_0] =
    882       cmdvec( R200_VS_MATRIX_2_MVP, 1, 16);
    883    rmesa->hw.mat[R200_MTX_TEX0].cmd[MAT_CMD_0] =
    884       cmdvec( R200_VS_MATRIX_3_TEX0, 1, 16);
    885    rmesa->hw.mat[R200_MTX_TEX1].cmd[MAT_CMD_0] =
    886       cmdvec( R200_VS_MATRIX_4_TEX1, 1, 16);
    887    rmesa->hw.mat[R200_MTX_TEX2].cmd[MAT_CMD_0] =
    888       cmdvec( R200_VS_MATRIX_5_TEX2, 1, 16);
    889    rmesa->hw.mat[R200_MTX_TEX3].cmd[MAT_CMD_0] =
    890       cmdvec( R200_VS_MATRIX_6_TEX3, 1, 16);
    891    rmesa->hw.mat[R200_MTX_TEX4].cmd[MAT_CMD_0] =
    892       cmdvec( R200_VS_MATRIX_7_TEX4, 1, 16);
    893    rmesa->hw.mat[R200_MTX_TEX5].cmd[MAT_CMD_0] =
    894       cmdvec( R200_VS_MATRIX_8_TEX5, 1, 16);
    895 
    896    for (i = 0 ; i < 8; i++) {
    897       rmesa->hw.lit[i].cmd[LIT_CMD_0] =
    898 	 cmdvec( R200_VS_LIGHT_AMBIENT_ADDR + i, 8, 24 );
    899       rmesa->hw.lit[i].cmd[LIT_CMD_1] =
    900 	 cmdscl( R200_SS_LIGHT_DCD_ADDR + i, 8, 7 );
    901    }
    902 
    903    for (i = 0 ; i < 6; i++) {
    904       rmesa->hw.ucp[i].cmd[UCP_CMD_0] =
    905 	 cmdvec( R200_VS_UCP_ADDR + i, 1, 4 );
    906    }
    907 
    908    rmesa->hw.ptp.cmd[PTP_CMD_0] =
    909       cmdvec( R200_VS_PNT_SPRITE_VPORT_SCALE, 1, 4 );
    910    rmesa->hw.ptp.cmd[PTP_CMD_1] =
    911       cmdvec( R200_VS_PNT_SPRITE_ATT_CONST, 1, 12 );
    912 
    913    /* Initial Harware state:
    914     */
    915    rmesa->hw.ctx.cmd[CTX_PP_MISC] = (R200_ALPHA_TEST_PASS
    916 				     /* | R200_RIGHT_HAND_CUBE_OGL*/);
    917 
    918    rmesa->hw.ctx.cmd[CTX_PP_FOG_COLOR] = (R200_FOG_VERTEX |
    919 					  R200_FOG_USE_SPEC_ALPHA);
    920 
    921    rmesa->hw.ctx.cmd[CTX_RE_SOLID_COLOR] = 0x00000000;
    922 
    923    rmesa->hw.ctx.cmd[CTX_RB3D_BLENDCNTL] = (R200_COMB_FCN_ADD_CLAMP |
    924 				(R200_BLEND_GL_ONE << R200_SRC_BLEND_SHIFT) |
    925 				(R200_BLEND_GL_ZERO << R200_DST_BLEND_SHIFT));
    926 
    927    rmesa->hw.ctx.cmd[CTX_RB3D_BLENDCOLOR] = 0x00000000;
    928    rmesa->hw.ctx.cmd[CTX_RB3D_ABLENDCNTL] = (R200_COMB_FCN_ADD_CLAMP |
    929 				(R200_BLEND_GL_ONE << R200_SRC_BLEND_SHIFT) |
    930 				(R200_BLEND_GL_ZERO << R200_DST_BLEND_SHIFT));
    931    rmesa->hw.ctx.cmd[CTX_RB3D_CBLENDCNTL] = (R200_COMB_FCN_ADD_CLAMP |
    932 				(R200_BLEND_GL_ONE << R200_SRC_BLEND_SHIFT) |
    933 				(R200_BLEND_GL_ZERO << R200_DST_BLEND_SHIFT));
    934 
    935    rmesa->hw.ctx.cmd[CTX_RB3D_DEPTHOFFSET] =
    936       rmesa->radeon.radeonScreen->depthOffset + rmesa->radeon.radeonScreen->fbLocation;
    937 
    938    rmesa->hw.ctx.cmd[CTX_RB3D_DEPTHPITCH] =
    939       ((rmesa->radeon.radeonScreen->depthPitch &
    940 	R200_DEPTHPITCH_MASK) |
    941        R200_DEPTH_ENDIAN_NO_SWAP);
    942 
    943    if (rmesa->using_hyperz)
    944       rmesa->hw.ctx.cmd[CTX_RB3D_DEPTHPITCH] |= R200_DEPTH_HYPERZ;
    945 
    946    rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] = (R200_Z_TEST_LESS |
    947 					       R200_STENCIL_TEST_ALWAYS |
    948 					       R200_STENCIL_FAIL_KEEP |
    949 					       R200_STENCIL_ZPASS_KEEP |
    950 					       R200_STENCIL_ZFAIL_KEEP |
    951 					       R200_Z_WRITE_ENABLE);
    952 
    953    if (rmesa->using_hyperz) {
    954       rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= R200_Z_COMPRESSION_ENABLE |
    955 						  R200_Z_DECOMPRESSION_ENABLE;
    956 /*      if (rmesa->radeon.radeonScreen->chip_family == CHIP_FAMILY_R200)
    957 	 rmesa->hw.ctx.cmd[CTX_RB3D_ZSTENCILCNTL] |= RADEON_Z_HIERARCHY_ENABLE;*/
    958    }
    959 
    960    rmesa->hw.ctx.cmd[CTX_PP_CNTL] = (R200_ANTI_ALIAS_NONE
    961  				     | R200_TEX_BLEND_0_ENABLE);
    962 
    963    switch ( driQueryOptioni( &rmesa->radeon.optionCache, "dither_mode" ) ) {
    964    case DRI_CONF_DITHER_XERRORDIFFRESET:
    965       rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= R200_DITHER_INIT;
    966       break;
    967    case DRI_CONF_DITHER_ORDERED:
    968       rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= R200_SCALE_DITHER_ENABLE;
    969       break;
    970    }
    971    if ( driQueryOptioni( &rmesa->radeon.optionCache, "round_mode" ) ==
    972 	DRI_CONF_ROUND_ROUND )
    973       rmesa->radeon.state.color.roundEnable = R200_ROUND_ENABLE;
    974    else
    975       rmesa->radeon.state.color.roundEnable = 0;
    976    if ( driQueryOptioni (&rmesa->radeon.optionCache, "color_reduction" ) ==
    977 	DRI_CONF_COLOR_REDUCTION_DITHER )
    978       rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= R200_DITHER_ENABLE;
    979    else
    980       rmesa->hw.ctx.cmd[CTX_RB3D_CNTL] |= rmesa->radeon.state.color.roundEnable;
    981 
    982    rmesa->hw.prf.cmd[PRF_PP_TRI_PERF] = R200_TRI_CUTOFF_MASK - R200_TRI_CUTOFF_MASK *
    983 			driQueryOptionf (&rmesa->radeon.optionCache,"texture_blend_quality");
    984    rmesa->hw.prf.cmd[PRF_PP_PERF_CNTL] = 0;
    985 
    986    rmesa->hw.set.cmd[SET_SE_CNTL] = (R200_FFACE_CULL_CCW |
    987 				     R200_BFACE_SOLID |
    988 				     R200_FFACE_SOLID |
    989 				     R200_FLAT_SHADE_VTX_LAST |
    990 				     R200_DIFFUSE_SHADE_GOURAUD |
    991 				     R200_ALPHA_SHADE_GOURAUD |
    992 				     R200_SPECULAR_SHADE_GOURAUD |
    993 				     R200_FOG_SHADE_GOURAUD |
    994 				     R200_DISC_FOG_SHADE_GOURAUD |
    995 				     R200_VTX_PIX_CENTER_OGL |
    996 				     R200_ROUND_MODE_TRUNC |
    997 				     R200_ROUND_PREC_8TH_PIX);
    998 
    999    rmesa->hw.set.cmd[SET_RE_CNTL] = (R200_PERSPECTIVE_ENABLE |
   1000 				     R200_SCISSOR_ENABLE);
   1001 
   1002    rmesa->hw.lin.cmd[LIN_RE_LINE_PATTERN] = ((1 << 16) | 0xffff);
   1003 
   1004    rmesa->hw.lin.cmd[LIN_RE_LINE_STATE] =
   1005       ((0 << R200_LINE_CURRENT_PTR_SHIFT) |
   1006        (1 << R200_LINE_CURRENT_COUNT_SHIFT));
   1007 
   1008    rmesa->hw.lin.cmd[LIN_SE_LINE_WIDTH] = (1 << 4);
   1009 
   1010    rmesa->hw.msk.cmd[MSK_RB3D_STENCILREFMASK] =
   1011       ((0x00 << R200_STENCIL_REF_SHIFT) |
   1012        (0xff << R200_STENCIL_MASK_SHIFT) |
   1013        (0xff << R200_STENCIL_WRITEMASK_SHIFT));
   1014 
   1015    rmesa->hw.msk.cmd[MSK_RB3D_ROPCNTL] = R200_ROP_COPY;
   1016    rmesa->hw.msk.cmd[MSK_RB3D_PLANEMASK] = 0xffffffff;
   1017 
   1018    rmesa->hw.tam.cmd[TAM_DEBUG3] = 0;
   1019 
   1020    rmesa->hw.msc.cmd[MSC_RE_MISC] =
   1021       ((0 << R200_STIPPLE_X_OFFSET_SHIFT) |
   1022        (0 << R200_STIPPLE_Y_OFFSET_SHIFT) |
   1023        R200_STIPPLE_BIG_BIT_ORDER);
   1024 
   1025 
   1026    rmesa->hw.cst.cmd[CST_PP_CNTL_X] = 0;
   1027    rmesa->hw.cst.cmd[CST_RB3D_DEPTHXY_OFFSET] = 0;
   1028    rmesa->hw.cst.cmd[CST_RE_AUX_SCISSOR_CNTL] = 0x0;
   1029    rmesa->hw.cst.cmd[CST_SE_VAP_CNTL_STATUS] =
   1030 #ifdef MESA_BIG_ENDIAN
   1031 						R200_VC_32BIT_SWAP;
   1032 #else
   1033 						R200_VC_NO_SWAP;
   1034 #endif
   1035 
   1036    if (!(rmesa->radeon.radeonScreen->chip_flags & RADEON_CHIPSET_TCL)) {
   1037       /* Bypass TCL */
   1038       rmesa->hw.cst.cmd[CST_SE_VAP_CNTL_STATUS] |= (1<<8);
   1039    }
   1040 
   1041    rmesa->hw.cst.cmd[CST_RE_POINTSIZE] =
   1042       (((GLuint)(ctx->Const.MaxPointSize * 16.0)) << R200_MAXPOINTSIZE_SHIFT) | 0x10;
   1043    rmesa->hw.cst.cmd[CST_SE_TCL_INPUT_VTX_0] =
   1044       (0x0 << R200_VERTEX_POSITION_ADDR__SHIFT);
   1045    rmesa->hw.cst.cmd[CST_SE_TCL_INPUT_VTX_1] =
   1046       (0x02 << R200_VTX_COLOR_0_ADDR__SHIFT) |
   1047       (0x03 << R200_VTX_COLOR_1_ADDR__SHIFT);
   1048    rmesa->hw.cst.cmd[CST_SE_TCL_INPUT_VTX_2] =
   1049       (0x06 << R200_VTX_TEX_0_ADDR__SHIFT) |
   1050       (0x07 << R200_VTX_TEX_1_ADDR__SHIFT) |
   1051       (0x08 << R200_VTX_TEX_2_ADDR__SHIFT) |
   1052       (0x09 << R200_VTX_TEX_3_ADDR__SHIFT);
   1053    rmesa->hw.cst.cmd[CST_SE_TCL_INPUT_VTX_3] =
   1054       (0x0A << R200_VTX_TEX_4_ADDR__SHIFT) |
   1055       (0x0B << R200_VTX_TEX_5_ADDR__SHIFT);
   1056 
   1057 
   1058    rmesa->hw.vpt.cmd[VPT_SE_VPORT_XSCALE]  = 0x00000000;
   1059    rmesa->hw.vpt.cmd[VPT_SE_VPORT_XOFFSET] = 0x00000000;
   1060    rmesa->hw.vpt.cmd[VPT_SE_VPORT_YSCALE]  = 0x00000000;
   1061    rmesa->hw.vpt.cmd[VPT_SE_VPORT_YOFFSET] = 0x00000000;
   1062    rmesa->hw.vpt.cmd[VPT_SE_VPORT_ZSCALE]  = 0x00000000;
   1063    rmesa->hw.vpt.cmd[VPT_SE_VPORT_ZOFFSET] = 0x00000000;
   1064 
   1065    for ( i = 0 ; i < ctx->Const.MaxTextureUnits ; i++ ) {
   1066       rmesa->hw.tex[i].cmd[TEX_PP_TXFILTER] = R200_BORDER_MODE_OGL;
   1067       rmesa->hw.tex[i].cmd[TEX_PP_TXFORMAT] =
   1068          ((i << R200_TXFORMAT_ST_ROUTE_SHIFT) |  /* <-- note i */
   1069           (2 << R200_TXFORMAT_WIDTH_SHIFT) |
   1070           (2 << R200_TXFORMAT_HEIGHT_SHIFT));
   1071       rmesa->hw.tex[i].cmd[TEX_PP_BORDER_COLOR] = 0;
   1072       rmesa->hw.tex[i].cmd[TEX_PP_TXFORMAT_X] =
   1073          (/* R200_TEXCOORD_PROJ | */
   1074           R200_LOD_BIAS_CORRECTION);	/* Small default bias */
   1075       rmesa->hw.tex[i].cmd[TEX_PP_TXOFFSET_NEWDRM] =
   1076 	     rmesa->radeon.radeonScreen->texOffset[RADEON_LOCAL_TEX_HEAP];
   1077       rmesa->hw.tex[i].cmd[TEX_PP_CUBIC_FACES] = 0;
   1078       rmesa->hw.tex[i].cmd[TEX_PP_TXMULTI_CTL] = 0;
   1079 
   1080       rmesa->hw.cube[i].cmd[CUBE_PP_CUBIC_FACES] = 0;
   1081       rmesa->hw.cube[i].cmd[CUBE_PP_CUBIC_OFFSET_F1] =
   1082          rmesa->radeon.radeonScreen->texOffset[RADEON_LOCAL_TEX_HEAP];
   1083       rmesa->hw.cube[i].cmd[CUBE_PP_CUBIC_OFFSET_F2] =
   1084          rmesa->radeon.radeonScreen->texOffset[RADEON_LOCAL_TEX_HEAP];
   1085       rmesa->hw.cube[i].cmd[CUBE_PP_CUBIC_OFFSET_F3] =
   1086          rmesa->radeon.radeonScreen->texOffset[RADEON_LOCAL_TEX_HEAP];
   1087       rmesa->hw.cube[i].cmd[CUBE_PP_CUBIC_OFFSET_F4] =
   1088          rmesa->radeon.radeonScreen->texOffset[RADEON_LOCAL_TEX_HEAP];
   1089       rmesa->hw.cube[i].cmd[CUBE_PP_CUBIC_OFFSET_F5] =
   1090          rmesa->radeon.radeonScreen->texOffset[RADEON_LOCAL_TEX_HEAP];
   1091 
   1092       rmesa->hw.pix[i].cmd[PIX_PP_TXCBLEND] =
   1093          (R200_TXC_ARG_A_ZERO |
   1094           R200_TXC_ARG_B_ZERO |
   1095           R200_TXC_ARG_C_DIFFUSE_COLOR |
   1096           R200_TXC_OP_MADD);
   1097 
   1098       rmesa->hw.pix[i].cmd[PIX_PP_TXCBLEND2] =
   1099          ((i << R200_TXC_TFACTOR_SEL_SHIFT) |
   1100           R200_TXC_SCALE_1X |
   1101           R200_TXC_CLAMP_0_1 |
   1102           R200_TXC_OUTPUT_REG_R0);
   1103 
   1104       rmesa->hw.pix[i].cmd[PIX_PP_TXABLEND] =
   1105          (R200_TXA_ARG_A_ZERO |
   1106           R200_TXA_ARG_B_ZERO |
   1107           R200_TXA_ARG_C_DIFFUSE_ALPHA |
   1108           R200_TXA_OP_MADD);
   1109 
   1110       rmesa->hw.pix[i].cmd[PIX_PP_TXABLEND2] =
   1111          ((i << R200_TXA_TFACTOR_SEL_SHIFT) |
   1112           R200_TXA_SCALE_1X |
   1113           R200_TXA_CLAMP_0_1 |
   1114           R200_TXA_OUTPUT_REG_R0);
   1115    }
   1116 
   1117    rmesa->hw.tf.cmd[TF_TFACTOR_0] = 0;
   1118    rmesa->hw.tf.cmd[TF_TFACTOR_1] = 0;
   1119    rmesa->hw.tf.cmd[TF_TFACTOR_2] = 0;
   1120    rmesa->hw.tf.cmd[TF_TFACTOR_3] = 0;
   1121    rmesa->hw.tf.cmd[TF_TFACTOR_4] = 0;
   1122    rmesa->hw.tf.cmd[TF_TFACTOR_5] = 0;
   1123 
   1124    rmesa->hw.vap.cmd[VAP_SE_VAP_CNTL] =
   1125       (R200_VAP_TCL_ENABLE |
   1126        (0x9 << R200_VAP_VF_MAX_VTX_NUM__SHIFT));
   1127 
   1128    rmesa->hw.vte.cmd[VTE_SE_VTE_CNTL] =
   1129       (R200_VPORT_X_SCALE_ENA |
   1130        R200_VPORT_Y_SCALE_ENA |
   1131        R200_VPORT_Z_SCALE_ENA |
   1132        R200_VPORT_X_OFFSET_ENA |
   1133        R200_VPORT_Y_OFFSET_ENA |
   1134        R200_VPORT_Z_OFFSET_ENA |
   1135 /* FIXME: Turn on for tex rect only */
   1136        R200_VTX_ST_DENORMALIZED |
   1137        R200_VTX_W0_FMT);
   1138 
   1139 
   1140    rmesa->hw.vtx.cmd[VTX_VTXFMT_0] = 0;
   1141    rmesa->hw.vtx.cmd[VTX_VTXFMT_1] = 0;
   1142    rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_0] =
   1143       ((R200_VTX_Z0 | R200_VTX_W0 |
   1144        (R200_VTX_FP_RGBA << R200_VTX_COLOR_0_SHIFT)));
   1145    rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_VTXFMT_1] = 0;
   1146    rmesa->hw.vtx.cmd[VTX_TCL_OUTPUT_COMPSEL] = (R200_OUTPUT_XYZW);
   1147    rmesa->hw.vtx.cmd[VTX_STATE_CNTL] = R200_VSC_UPDATE_USER_COLOR_0_ENABLE;
   1148 
   1149 
   1150    /* Matrix selection */
   1151    rmesa->hw.msl.cmd[MSL_MATRIX_SELECT_0] =
   1152       (R200_MTX_MV << R200_MODELVIEW_0_SHIFT);
   1153 
   1154    rmesa->hw.msl.cmd[MSL_MATRIX_SELECT_1] =
   1155        (R200_MTX_IMV << R200_IT_MODELVIEW_0_SHIFT);
   1156 
   1157    rmesa->hw.msl.cmd[MSL_MATRIX_SELECT_2] =
   1158       (R200_MTX_MVP << R200_MODELPROJECT_0_SHIFT);
   1159 
   1160    rmesa->hw.msl.cmd[MSL_MATRIX_SELECT_3] =
   1161       ((R200_MTX_TEX0 << R200_TEXMAT_0_SHIFT) |
   1162        (R200_MTX_TEX1 << R200_TEXMAT_1_SHIFT) |
   1163        (R200_MTX_TEX2 << R200_TEXMAT_2_SHIFT) |
   1164        (R200_MTX_TEX3 << R200_TEXMAT_3_SHIFT));
   1165 
   1166    rmesa->hw.msl.cmd[MSL_MATRIX_SELECT_4] =
   1167       ((R200_MTX_TEX4 << R200_TEXMAT_4_SHIFT) |
   1168        (R200_MTX_TEX5 << R200_TEXMAT_5_SHIFT));
   1169 
   1170 
   1171    /* General TCL state */
   1172    rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_0] =
   1173       (R200_SPECULAR_LIGHTS |
   1174        R200_DIFFUSE_SPECULAR_COMBINE |
   1175        R200_LOCAL_LIGHT_VEC_GL |
   1176        R200_LM0_SOURCE_MATERIAL_0 << R200_FRONT_SHININESS_SOURCE_SHIFT |
   1177        R200_LM0_SOURCE_MATERIAL_1 << R200_BACK_SHININESS_SOURCE_SHIFT);
   1178 
   1179    rmesa->hw.tcl.cmd[TCL_LIGHT_MODEL_CTL_1] =
   1180       ((R200_LM1_SOURCE_MATERIAL_0 << R200_FRONT_EMISSIVE_SOURCE_SHIFT) |
   1181        (R200_LM1_SOURCE_MATERIAL_0 << R200_FRONT_AMBIENT_SOURCE_SHIFT) |
   1182        (R200_LM1_SOURCE_MATERIAL_0 << R200_FRONT_DIFFUSE_SOURCE_SHIFT) |
   1183        (R200_LM1_SOURCE_MATERIAL_0 << R200_FRONT_SPECULAR_SOURCE_SHIFT) |
   1184        (R200_LM1_SOURCE_MATERIAL_1 << R200_BACK_EMISSIVE_SOURCE_SHIFT) |
   1185        (R200_LM1_SOURCE_MATERIAL_1 << R200_BACK_AMBIENT_SOURCE_SHIFT) |
   1186        (R200_LM1_SOURCE_MATERIAL_1 << R200_BACK_DIFFUSE_SOURCE_SHIFT) |
   1187        (R200_LM1_SOURCE_MATERIAL_1 << R200_BACK_SPECULAR_SOURCE_SHIFT));
   1188 
   1189    rmesa->hw.tcl.cmd[TCL_PER_LIGHT_CTL_0] = 0; /* filled in via callbacks */
   1190    rmesa->hw.tcl.cmd[TCL_PER_LIGHT_CTL_1] = 0;
   1191    rmesa->hw.tcl.cmd[TCL_PER_LIGHT_CTL_2] = 0;
   1192    rmesa->hw.tcl.cmd[TCL_PER_LIGHT_CTL_3] = 0;
   1193 
   1194    rmesa->hw.tcl.cmd[TCL_UCP_VERT_BLEND_CTL] =
   1195       (R200_UCP_IN_CLIP_SPACE |
   1196        R200_CULL_FRONT_IS_CCW);
   1197 
   1198    /* Texgen/Texmat state */
   1199    rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_2] = 0x00ffffff;
   1200    rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_3] =
   1201       ((0 << R200_TEXGEN_0_INPUT_TEX_SHIFT) |
   1202        (1 << R200_TEXGEN_1_INPUT_TEX_SHIFT) |
   1203        (2 << R200_TEXGEN_2_INPUT_TEX_SHIFT) |
   1204        (3 << R200_TEXGEN_3_INPUT_TEX_SHIFT) |
   1205        (4 << R200_TEXGEN_4_INPUT_TEX_SHIFT) |
   1206        (5 << R200_TEXGEN_5_INPUT_TEX_SHIFT));
   1207    rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_0] = 0;
   1208    rmesa->hw.tcg.cmd[TCG_TEX_PROC_CTL_1] =
   1209       ((0 << R200_TEXGEN_0_INPUT_SHIFT) |
   1210        (1 << R200_TEXGEN_1_INPUT_SHIFT) |
   1211        (2 << R200_TEXGEN_2_INPUT_SHIFT) |
   1212        (3 << R200_TEXGEN_3_INPUT_SHIFT) |
   1213        (4 << R200_TEXGEN_4_INPUT_SHIFT) |
   1214        (5 << R200_TEXGEN_5_INPUT_SHIFT));
   1215    rmesa->hw.tcg.cmd[TCG_TEX_CYL_WRAP_CTL] = 0;
   1216 
   1217 
   1218    for (i = 0 ; i < 8; i++) {
   1219       struct gl_light *l = &ctx->Light.Light[i];
   1220       GLenum p = GL_LIGHT0 + i;
   1221       *(float *)&(rmesa->hw.lit[i].cmd[LIT_RANGE_CUTOFF]) = FLT_MAX;
   1222 
   1223       ctx->Driver.Lightfv( ctx, p, GL_AMBIENT, l->Ambient );
   1224       ctx->Driver.Lightfv( ctx, p, GL_DIFFUSE, l->Diffuse );
   1225       ctx->Driver.Lightfv( ctx, p, GL_SPECULAR, l->Specular );
   1226       ctx->Driver.Lightfv( ctx, p, GL_POSITION, NULL );
   1227       ctx->Driver.Lightfv( ctx, p, GL_SPOT_DIRECTION, NULL );
   1228       ctx->Driver.Lightfv( ctx, p, GL_SPOT_EXPONENT, &l->SpotExponent );
   1229       ctx->Driver.Lightfv( ctx, p, GL_SPOT_CUTOFF, &l->SpotCutoff );
   1230       ctx->Driver.Lightfv( ctx, p, GL_CONSTANT_ATTENUATION,
   1231 			   &l->ConstantAttenuation );
   1232       ctx->Driver.Lightfv( ctx, p, GL_LINEAR_ATTENUATION,
   1233 			   &l->LinearAttenuation );
   1234       ctx->Driver.Lightfv( ctx, p, GL_QUADRATIC_ATTENUATION,
   1235 			   &l->QuadraticAttenuation );
   1236       *(float *)&(rmesa->hw.lit[i].cmd[LIT_ATTEN_XXX]) = 0.0;
   1237    }
   1238 
   1239    ctx->Driver.LightModelfv( ctx, GL_LIGHT_MODEL_AMBIENT,
   1240 			     ctx->Light.Model.Ambient );
   1241 
   1242    TNL_CONTEXT(ctx)->Driver.NotifyMaterialChange( ctx );
   1243 
   1244    for (i = 0 ; i < 6; i++) {
   1245       ctx->Driver.ClipPlane( ctx, GL_CLIP_PLANE0 + i, NULL );
   1246    }
   1247 
   1248    ctx->Driver.Fogfv( ctx, GL_FOG_MODE, NULL );
   1249    ctx->Driver.Fogfv( ctx, GL_FOG_DENSITY, &ctx->Fog.Density );
   1250    ctx->Driver.Fogfv( ctx, GL_FOG_START, &ctx->Fog.Start );
   1251    ctx->Driver.Fogfv( ctx, GL_FOG_END, &ctx->Fog.End );
   1252    ctx->Driver.Fogfv( ctx, GL_FOG_COLOR, ctx->Fog.Color );
   1253    ctx->Driver.Fogfv( ctx, GL_FOG_COORDINATE_SOURCE_EXT, NULL );
   1254 
   1255    rmesa->hw.grd.cmd[GRD_VERT_GUARD_CLIP_ADJ] = IEEE_ONE;
   1256    rmesa->hw.grd.cmd[GRD_VERT_GUARD_DISCARD_ADJ] = IEEE_ONE;
   1257    rmesa->hw.grd.cmd[GRD_HORZ_GUARD_CLIP_ADJ] = IEEE_ONE;
   1258    rmesa->hw.grd.cmd[GRD_HORZ_GUARD_DISCARD_ADJ] = IEEE_ONE;
   1259 
   1260    rmesa->hw.eye.cmd[EYE_X] = 0;
   1261    rmesa->hw.eye.cmd[EYE_Y] = 0;
   1262    rmesa->hw.eye.cmd[EYE_Z] = IEEE_ONE;
   1263    rmesa->hw.eye.cmd[EYE_RESCALE_FACTOR] = IEEE_ONE;
   1264 
   1265    rmesa->hw.spr.cmd[SPR_POINT_SPRITE_CNTL] =
   1266       R200_PS_SE_SEL_STATE | R200_PS_MULT_CONST;
   1267 
   1268    /* ptp_eye is presumably used to calculate the attenuation wrt a different
   1269       location? In any case, since point attenuation triggers _needeyecoords,
   1270       it is constant. Probably ignored as long as R200_PS_USE_MODEL_EYE_VEC
   1271       isn't set */
   1272    rmesa->hw.ptp.cmd[PTP_EYE_X] = 0;
   1273    rmesa->hw.ptp.cmd[PTP_EYE_Y] = 0;
   1274    rmesa->hw.ptp.cmd[PTP_EYE_Z] = IEEE_ONE | 0x80000000; /* -1.0 */
   1275    rmesa->hw.ptp.cmd[PTP_EYE_3] = 0;
   1276    /* no idea what the ptp_vport_scale values are good for, except the
   1277       PTSIZE one - hopefully doesn't matter */
   1278    rmesa->hw.ptp.cmd[PTP_VPORT_SCALE_0] = IEEE_ONE;
   1279    rmesa->hw.ptp.cmd[PTP_VPORT_SCALE_1] = IEEE_ONE;
   1280    rmesa->hw.ptp.cmd[PTP_VPORT_SCALE_PTSIZE] = IEEE_ONE;
   1281    rmesa->hw.ptp.cmd[PTP_VPORT_SCALE_3] = IEEE_ONE;
   1282    rmesa->hw.ptp.cmd[PTP_ATT_CONST_QUAD] = 0;
   1283    rmesa->hw.ptp.cmd[PTP_ATT_CONST_LIN] = 0;
   1284    rmesa->hw.ptp.cmd[PTP_ATT_CONST_CON] = IEEE_ONE;
   1285    rmesa->hw.ptp.cmd[PTP_ATT_CONST_3] = 0;
   1286    rmesa->hw.ptp.cmd[PTP_CLAMP_MIN] = IEEE_ONE;
   1287    rmesa->hw.ptp.cmd[PTP_CLAMP_MAX] = 0x44ffe000; /* 2047 */
   1288    rmesa->hw.ptp.cmd[PTP_CLAMP_2] = 0;
   1289    rmesa->hw.ptp.cmd[PTP_CLAMP_3] = 0;
   1290 
   1291    r200LightingSpaceChange( ctx );
   1292 
   1293    radeon_init_query_stateobj(&rmesa->radeon, R200_QUERYOBJ_CMDSIZE);
   1294    rmesa->radeon.query.queryobj.cmd[R200_QUERYOBJ_CMD_0] = CP_PACKET0(RADEON_RB3D_ZPASS_DATA, 0);
   1295    rmesa->radeon.query.queryobj.cmd[R200_QUERYOBJ_DATA_0] = 0;
   1296 
   1297    rmesa->radeon.hw.all_dirty = GL_TRUE;
   1298 
   1299    rcommonInitCmdBuf(&rmesa->radeon);
   1300 }
   1301