#ifndef __R300_CONTEXT_H__
#define __R300_CONTEXT_H__
-#include "tnl/t_vertex.h"
#include "drm.h"
#include "radeon_drm.h"
#include "dri_util.h"
-#include "texmem.h"
+#include "radeon_common.h"
-#include "macros.h"
-#include "mtypes.h"
-#include "colormac.h"
-
-#define USER_BUFFERS
-
-/* We don't handle 16 bits elts swapping yet */
-#ifdef MESA_BIG_ENDIAN
-#define FORCE_32BITS_ELTS
-#endif
-
-//#define OPTIMIZE_ELTS
+#include "main/mtypes.h"
+#include "shader/prog_instruction.h"
struct r300_context;
typedef struct r300_context r300ContextRec;
typedef struct r300_context *r300ContextPtr;
-#include "radeon_lock.h"
-#include "mm.h"
-/* From http://gcc.gnu.org/onlinedocs/gcc-3.2.3/gcc/Variadic-Macros.html .
+/* From http://gcc. gnu.org/onlinedocs/gcc-3.2.3/gcc/Variadic-Macros.html .
I suppose we could inline this and use macro to fetch out __LINE__ and stuff in case we run into trouble
with other compilers ... GLUE!
*/
}
#include "r300_vertprog.h"
-#include "r300_fragprog.h"
-
-/**
- * This function takes a float and packs it into a uint32_t
- */
-static __inline__ uint32_t r300PackFloat32(float fl)
-{
- union {
- float fl;
- uint32_t u;
- } u;
-
- u.fl = fl;
- return u.u;
-}
-
-/* This is probably wrong for some values, I need to test this
- * some more. Range checking would be a good idea also..
- *
- * But it works for most things. I'll fix it later if someone
- * else with a better clue doesn't
- */
-static __inline__ uint32_t r300PackFloat24(float f)
-{
- float mantissa;
- int exponent;
- uint32_t float24 = 0;
-
- if (f == 0.0)
- return 0;
-
- mantissa = frexpf(f, &exponent);
-
- /* Handle -ve */
- if (mantissa < 0) {
- float24 |= (1 << 23);
- mantissa = mantissa * -1.0;
- }
- /* Handle exponent, bias of 63 */
- exponent += 62;
- float24 |= (exponent << 16);
- /* Kill 7 LSB of mantissa */
- float24 |= (r300PackFloat32(mantissa) & 0x7FFFFF) >> 7;
-
- return float24;
-}
-
-/************ DMA BUFFERS **************/
-
-/* Need refcounting on dma buffers:
- */
-struct r300_dma_buffer {
- int refcount; /**< the number of retained regions in buf */
- drmBufPtr buf;
- int id;
-};
-#undef GET_START
-#ifdef USER_BUFFERS
-#define GET_START(rvb) (r300GartOffsetFromVirtual(rmesa, (rvb)->address+(rvb)->start))
-#else
-#define GET_START(rvb) (rmesa->radeon.radeonScreen->gart_buffer_offset + \
- (rvb)->address - rmesa->dma.buf0_address + \
- (rvb)->start)
-#endif
-/* A retained region, eg vertices for indexed vertices.
- */
-struct r300_dma_region {
- struct r300_dma_buffer *buf;
- char *address; /* == buf->address */
- int start, end, ptr; /* offsets from start of buf */
-
- int aos_offset; /* address in GART memory */
- int aos_stride; /* distance between elements, in dwords */
- int aos_size; /* number of components (1-4) */
- int aos_reg; /* VAP register assignment */
-};
-
-struct r300_dma {
- /* Active dma region. Allocations for vertices and retained
- * regions come from here. Also used for emitting random vertices,
- * these may be flushed by calling flush_current();
- */
- struct r300_dma_region current;
-
- void (*flush) (r300ContextPtr);
-
- char *buf0_address; /* start of buf[0], for index calcs */
-
- /* Number of "in-flight" DMA buffers, i.e. the number of buffers
- * for which a DISCARD command is currently queued in the command buffer.
- */
- GLuint nr_released_bufs;
-};
- /* Texture related */
-
-typedef struct r300_tex_obj r300TexObj, *r300TexObjPtr;
-
-/* Texture object in locally shared texture space.
- */
-struct r300_tex_obj {
- driTextureObject base;
-
- GLuint bufAddr; /* Offset to start of locally
- shared texture block */
-
- GLuint dirty_state; /* Flags (1 per texunit) for
- whether or not this texobj
- has dirty hardware state
- (pp_*) that needs to be
- brought into the
- texunit. */
-
- drm_radeon_tex_image_t image[6][RADEON_MAX_TEXTURE_LEVELS];
- /* Six, for the cube faces */
-
- GLuint pitch; /* this isn't sent to hardware just used in calculations */
- /* hardware register values */
- /* Note that R200 has 8 registers per texture and R300 only 7 */
- GLuint filter;
- GLuint filter_1;
- GLuint pitch_reg;
- GLuint size; /* npot only */
- GLuint format;
- GLuint offset; /* Image location in the card's address space.
- All cube faces follow. */
- GLuint unknown4;
- GLuint unknown5;
- /* end hardware registers */
-
- /* registers computed by r200 code - keep them here to
- compare against what is actually written.
-
- to be removed later.. */
- GLuint pp_border_color;
- GLuint pp_cubic_faces; /* cube face 1,2,3,4 log2 sizes */
- GLuint format_x;
-
- GLboolean border_fallback;
-
- GLuint tile_bits; /* hw texture tile bits used on this texture */
-};
-
-struct r300_texture_env_state {
- r300TexObjPtr texobj;
- GLenum format;
- GLenum envMode;
-};
/* The blit width for texture uploads
*/
#define R300_BLIT_WIDTH_BYTES 1024
#define R300_MAX_TEXTURE_UNITS 8
-struct r300_texture_state {
- struct r300_texture_env_state unit[R300_MAX_TEXTURE_UNITS];
- int tc_count; /* number of incoming texture coordinates from VAP */
-};
-/**
- * A block of hardware state.
- *
- * When check returns non-zero, the returned number of dwords must be
- * copied verbatim into the command buffer in order to update a state atom
- * when it is dirty.
- */
-struct r300_state_atom {
- struct r300_state_atom *next, *prev;
- const char *name; /* for debug */
- int cmd_size; /* maximum size in dwords */
- GLuint idx; /* index in an array (e.g. textures) */
- uint32_t *cmd;
- GLboolean dirty;
-
- int (*check) (r300ContextPtr, struct r300_state_atom * atom);
-};
#define R300_VPT_CMD_0 0
#define R300_VPT_XSCALE 1
#define R300_GB_MISC_MSPOS_0 1
#define R300_GB_MISC_MSPOS_1 2
#define R300_GB_MISC_TILE_CONFIG 3
-#define R300_GB_MISC_SELECT 4
-#define R300_GB_MISC_AA_CONFIG 5
-#define R300_GB_MISC_CMDSIZE 6
+#define R300_GB_MISC_CMDSIZE 4
+#define R300_GB_MISC2_CMD_0 0
+#define R300_GB_MISC2_SELECT 1
+#define R300_GB_MISC2_AA_CONFIG 2
+#define R300_GB_MISC2_CMDSIZE 3
#define R300_TXE_CMD_0 0
#define R300_TXE_ENABLE 1
#define R300_RI_INTERP_7 8
#define R300_RI_CMDSIZE 9
+#define R500_RI_CMDSIZE 17
+
#define R300_RR_CMD_0 0 /* rr is variable size (at least 1) */
-#define R300_RR_ROUTE_0 1
-#define R300_RR_ROUTE_1 2
-#define R300_RR_ROUTE_2 3
-#define R300_RR_ROUTE_3 4
-#define R300_RR_ROUTE_4 5
-#define R300_RR_ROUTE_5 6
-#define R300_RR_ROUTE_6 7
-#define R300_RR_ROUTE_7 8
+#define R300_RR_INST_0 1
+#define R300_RR_INST_1 2
+#define R300_RR_INST_2 3
+#define R300_RR_INST_3 4
+#define R300_RR_INST_4 5
+#define R300_RR_INST_5 6
+#define R300_RR_INST_6 7
+#define R300_RR_INST_7 8
#define R300_RR_CMDSIZE 9
#define R300_FP_CMD_0 0
#define R300_FP_NODE3 8
#define R300_FP_CMDSIZE 9
+#define R500_FP_CMD_0 0
+#define R500_FP_CNTL 1
+#define R500_FP_PIXSIZE 2
+#define R500_FP_CMD_1 3
+#define R500_FP_CODE_ADDR 4
+#define R500_FP_CODE_RANGE 5
+#define R500_FP_CODE_OFFSET 6
+#define R500_FP_CMD_2 7
+#define R500_FP_FC_CNTL 8
+#define R500_FP_CMDSIZE 9
+
#define R300_FPT_CMD_0 0
#define R300_FPT_INSTR_0 1
#define R300_FPT_CMDSIZE 65
#define R300_FPI_CMD_0 0
#define R300_FPI_INSTR_0 1
#define R300_FPI_CMDSIZE 65
+/* R500 has space for 512 instructions - 6 dwords per instruction */
+#define R500_FPI_CMDSIZE (512*6+1)
#define R300_FPP_CMD_0 0
#define R300_FPP_PARAM_0 1
#define R300_FPP_CMDSIZE (32*4+1)
+/* R500 has spcae for 256 constants - 4 dwords per constant */
+#define R500_FPP_CMDSIZE (256*4+1)
#define R300_FOGS_CMD_0 0
#define R300_FOGS_STATE 1
#define R300_ZB_PITCH 2
#define R300_ZB_CMDSIZE 3
+#define R300_VAP_CNTL_FLUSH 0
+#define R300_VAP_CNTL_FLUSH_1 1
+#define R300_VAP_CNTL_CMD 2
+#define R300_VAP_CNTL_INSTR 3
+#define R300_VAP_CNTL_SIZE 4
+
#define R300_VPI_CMD_0 0
#define R300_VPI_INSTR_0 1
#define R300_VPI_CMDSIZE 1025 /* 256 16 byte instructions */
#define R300_VPP_PARAM_0 1
#define R300_VPP_CMDSIZE 1025 /* 256 4-component parameters */
+#define R300_VPUCP_CMD_0 0
+#define R300_VPUCP_X 1
+#define R300_VPUCP_Y 2
+#define R300_VPUCP_Z 3
+#define R300_VPUCP_W 4
+#define R300_VPUCP_CMDSIZE 5 /* 256 4-component parameters */
+
#define R300_VPS_CMD_0 0
#define R300_VPS_ZERO_0 1
#define R300_VPS_ZERO_1 2
* Cache for hardware register state.
*/
struct r300_hw_state {
- struct r300_state_atom atomlist;
-
- GLboolean is_dirty;
- GLboolean all_dirty;
- int max_state_size; /* in dwords */
-
- struct r300_state_atom vpt; /* viewport (1D98) */
- struct r300_state_atom vap_cntl;
- struct r300_state_atom vof; /* VAP output format register 0x2090 */
- struct r300_state_atom vte; /* (20B0) */
- struct r300_state_atom unk2134; /* (2134) */
- struct r300_state_atom vap_cntl_status;
- struct r300_state_atom vir[2]; /* vap input route (2150/21E0) */
- struct r300_state_atom vic; /* vap input control (2180) */
- struct r300_state_atom unk21DC; /* (21DC) */
- struct r300_state_atom unk221C; /* (221C) */
- struct r300_state_atom unk2220; /* (2220) */
- struct r300_state_atom unk2288; /* (2288) */
- struct r300_state_atom pvs; /* pvs_cntl (22D0) */
- struct r300_state_atom gb_enable; /* (4008) */
- struct r300_state_atom gb_misc; /* Multisampling position shifts ? (4010) */
- struct r300_state_atom unk4200; /* (4200) */
- struct r300_state_atom unk4214; /* (4214) */
- struct r300_state_atom ps; /* pointsize (421C) */
- struct r300_state_atom unk4230; /* (4230) */
- struct r300_state_atom lcntl; /* line control */
- struct r300_state_atom unk4260; /* (4260) */
- struct r300_state_atom shade;
- struct r300_state_atom polygon_mode;
- struct r300_state_atom fogp; /* fog parameters (4294) */
- struct r300_state_atom unk429C; /* (429C) */
- struct r300_state_atom zbias_cntl;
- struct r300_state_atom zbs; /* zbias (42A4) */
- struct r300_state_atom occlusion_cntl;
- struct r300_state_atom cul; /* cull cntl (42B8) */
- struct r300_state_atom unk42C0; /* (42C0) */
- struct r300_state_atom rc; /* rs control (4300) */
- struct r300_state_atom ri; /* rs interpolators (4310) */
- struct r300_state_atom rr; /* rs route (4330) */
- struct r300_state_atom unk43A4; /* (43A4) */
- struct r300_state_atom unk43E8; /* (43E8) */
- struct r300_state_atom fp; /* fragment program cntl + nodes (4600) */
- struct r300_state_atom fpt; /* texi - (4620) */
- struct r300_state_atom unk46A4; /* (46A4) */
- struct r300_state_atom fpi[4]; /* fp instructions (46C0/47C0/48C0/49C0) */
- struct r300_state_atom fogs; /* fog state (4BC0) */
- struct r300_state_atom fogc; /* fog color (4BC8) */
- struct r300_state_atom at; /* alpha test (4BD4) */
- struct r300_state_atom unk4BD8; /* (4BD8) */
- struct r300_state_atom fpp; /* 0x4C00 and following */
- struct r300_state_atom unk4E00; /* (4E00) */
- struct r300_state_atom bld; /* blending (4E04) */
- struct r300_state_atom cmk; /* colormask (4E0C) */
- struct r300_state_atom blend_color; /* constant blend color */
- struct r300_state_atom cb; /* colorbuffer (4E28) */
- struct r300_state_atom unk4E50; /* (4E50) */
- struct r300_state_atom unk4E88; /* (4E88) */
- struct r300_state_atom unk4EA0; /* (4E88) I saw it only written on RV350 hardware.. */
- struct r300_state_atom zs; /* zstencil control (4F00) */
- struct r300_state_atom zstencil_format;
- struct r300_state_atom zb; /* z buffer (4F20) */
- struct r300_state_atom unk4F28; /* (4F28) */
- struct r300_state_atom unk4F30; /* (4F30) */
- struct r300_state_atom unk4F44; /* (4F44) */
- struct r300_state_atom unk4F54; /* (4F54) */
-
- struct r300_state_atom vpi; /* vp instructions */
- struct r300_state_atom vpp; /* vp parameters */
- struct r300_state_atom vps; /* vertex point size (?) */
+ struct radeon_state_atom vpt; /* viewport (1D98) */
+ struct radeon_state_atom vap_cntl;
+ struct radeon_state_atom vap_index_offset; /* 0x208c r5xx only */
+ struct radeon_state_atom vof; /* VAP output format register 0x2090 */
+ struct radeon_state_atom vte; /* (20B0) */
+ struct radeon_state_atom vap_vf_max_vtx_indx; /* Maximum Vertex Indx Clamp (2134) */
+ struct radeon_state_atom vap_cntl_status;
+ struct radeon_state_atom vir[2]; /* vap input route (2150/21E0) */
+ struct radeon_state_atom vic; /* vap input control (2180) */
+ struct radeon_state_atom vap_psc_sgn_norm_cntl; /* Programmable Stream Control Signed Normalize Control (21DC) */
+ struct radeon_state_atom vap_clip_cntl;
+ struct radeon_state_atom vap_clip;
+ struct radeon_state_atom vap_pvs_vtx_timeout_reg; /* Vertex timeout register (2288) */
+ struct radeon_state_atom pvs; /* pvs_cntl (22D0) */
+ struct radeon_state_atom gb_enable; /* (4008) */
+ struct radeon_state_atom gb_misc; /* Multisampling position shifts ? (4010) */
+ struct radeon_state_atom gb_misc2; /* Multisampling position shifts ? (4010) */
+ struct radeon_state_atom ga_point_s0; /* S Texture Coordinate of Vertex 0 for Point texture stuffing (LLC) (4200) */
+ struct radeon_state_atom ga_triangle_stipple; /* (4214) */
+ struct radeon_state_atom ps; /* pointsize (421C) */
+ struct radeon_state_atom ga_point_minmax; /* (4230) */
+ struct radeon_state_atom lcntl; /* line control */
+ struct radeon_state_atom ga_line_stipple; /* (4260) */
+ struct radeon_state_atom shade;
+ struct radeon_state_atom shade2;
+ struct radeon_state_atom polygon_mode;
+ struct radeon_state_atom fogp; /* fog parameters (4294) */
+ struct radeon_state_atom ga_soft_reset; /* (429C) */
+ struct radeon_state_atom zbias_cntl;
+ struct radeon_state_atom zbs; /* zbias (42A4) */
+ struct radeon_state_atom occlusion_cntl;
+ struct radeon_state_atom cul; /* cull cntl (42B8) */
+ struct radeon_state_atom su_depth_scale; /* (42C0) */
+ struct radeon_state_atom rc; /* rs control (4300) */
+ struct radeon_state_atom ri; /* rs interpolators (4310) */
+ struct radeon_state_atom rr; /* rs route (4330) */
+ struct radeon_state_atom sc_hyperz; /* (43A4) */
+ struct radeon_state_atom sc_screendoor; /* (43E8) */
+ struct radeon_state_atom fp; /* fragment program cntl + nodes (4600) */
+ struct radeon_state_atom fpt; /* texi - (4620) */
+ struct radeon_state_atom us_out_fmt; /* (46A4) */
+ struct radeon_state_atom r500fp; /* r500 fp instructions */
+ struct radeon_state_atom r500fp_const; /* r500 fp constants */
+ struct radeon_state_atom fpi[4]; /* fp instructions (46C0/47C0/48C0/49C0) */
+ struct radeon_state_atom fogs; /* fog state (4BC0) */
+ struct radeon_state_atom fogc; /* fog color (4BC8) */
+ struct radeon_state_atom at; /* alpha test (4BD4) */
+ struct radeon_state_atom fg_depth_src; /* (4BD8) */
+ struct radeon_state_atom fpp; /* 0x4C00 and following */
+ struct radeon_state_atom rb3d_cctl; /* (4E00) */
+ struct radeon_state_atom bld; /* blending (4E04) */
+ struct radeon_state_atom cmk; /* colormask (4E0C) */
+ struct radeon_state_atom blend_color; /* constant blend color */
+ struct radeon_state_atom rop; /* ropcntl */
+ struct radeon_state_atom cb; /* colorbuffer (4E28) */
+ struct radeon_state_atom rb3d_dither_ctl; /* (4E50) */
+ struct radeon_state_atom rb3d_aaresolve_ctl; /* (4E88) */
+ struct radeon_state_atom rb3d_discard_src_pixel_lte_threshold; /* (4E88) I saw it only written on RV350 hardware.. */
+ struct radeon_state_atom zs; /* zstencil control (4F00) */
+ struct radeon_state_atom zstencil_format;
+ struct radeon_state_atom zb; /* z buffer (4F20) */
+ struct radeon_state_atom zb_depthclearvalue; /* (4F28) */
+ struct radeon_state_atom zb_zmask; /* (4F30) */
+ struct radeon_state_atom zb_hiz_offset; /* (4F44) */
+ struct radeon_state_atom zb_hiz_pitch; /* (4F54) */
+
+ struct radeon_state_atom vpi; /* vp instructions */
+ struct radeon_state_atom vpp; /* vp parameters */
+ struct radeon_state_atom vps; /* vertex point size (?) */
+ struct radeon_state_atom vpucp[6]; /* vp user clip plane - 6 */
/* 8 texture units */
/* the state is grouped by function and not by
texture unit. This makes single unit updates
really awkward - we are much better off
updating the whole thing at once */
struct {
- struct r300_state_atom filter;
- struct r300_state_atom filter_1;
- struct r300_state_atom size;
- struct r300_state_atom format;
- struct r300_state_atom pitch;
- struct r300_state_atom offset;
- struct r300_state_atom chroma_key;
- struct r300_state_atom border_color;
+ struct radeon_state_atom filter;
+ struct radeon_state_atom filter_1;
+ struct radeon_state_atom size;
+ struct radeon_state_atom format;
+ struct radeon_state_atom pitch;
+ struct radeon_state_atom offset;
+ struct radeon_state_atom chroma_key;
+ struct radeon_state_atom border_color;
} tex;
- struct r300_state_atom txe; /* tex enable (4104) */
-};
+ struct radeon_state_atom txe; /* tex enable (4104) */
-/**
- * This structure holds the command buffer while it is being constructed.
- *
- * The first batch of commands in the buffer is always the state that needs
- * to be re-emitted when the context is lost. This batch can be skipped
- * otherwise.
- */
-struct r300_cmdbuf {
- int size; /* DWORDs allocated for buffer */
- uint32_t *cmd_buf;
- int count_used; /* DWORDs filled so far */
- int count_reemit; /* size of re-emission batch */
+ radeonTexObj *textures[R300_MAX_TEXTURE_UNITS];
};
/**
* State cache
*/
-struct r300_depthbuffer_state {
- GLfloat scale;
-};
-
-struct r300_stencilbuffer_state {
- GLuint clear;
- GLboolean hw_stencil;
-
-};
-
/* Vertex shader state */
/* Perhaps more if we store programs in vmem? */
union {
GLuint d[VSF_MAX_FRAGMENT_LENGTH];
float f[VSF_MAX_FRAGMENT_LENGTH];
- VERTEX_SHADER_INSTRUCTION i[VSF_MAX_FRAGMENT_LENGTH / 4];
+ GLuint i[VSF_MAX_FRAGMENT_LENGTH];
} body;
};
-#define VSF_DEST_PROGRAM 0x0
-#define VSF_DEST_MATRIX0 0x200
-#define VSF_DEST_MATRIX1 0x204
-#define VSF_DEST_MATRIX2 0x208
-#define VSF_DEST_VECTOR0 0x20c
-#define VSF_DEST_VECTOR1 0x20d
-#define VSF_DEST_UNKNOWN1 0x400
-#define VSF_DEST_UNKNOWN2 0x406
-
struct r300_vertex_shader_state {
struct r300_vertex_shader_fragment program;
-
- struct r300_vertex_shader_fragment unknown1;
- struct r300_vertex_shader_fragment unknown2;
-
- int program_start;
- int unknown_ptr1; /* pointer within program space */
- int program_end;
-
- int param_offset;
- int param_count;
-
- int unknown_ptr2; /* pointer within program space */
- int unknown_ptr3; /* pointer within program space */
};
-extern int hw_tcl_on;
+#define COLOR_IS_RGBA
+#define TAG(x) r300##x
+#include "tnl_dd/t_dd_vertex.h"
+#undef TAG
-//#define CURRENT_VERTEX_SHADER(ctx) (ctx->VertexProgram._Current)
#define CURRENT_VERTEX_SHADER(ctx) (R300_CONTEXT(ctx)->selected_vp)
-/* Should but doesnt work */
-//#define CURRENT_VERTEX_SHADER(ctx) (R300_CONTEXT(ctx)->curr_vp)
-
/* r300_vertex_shader_state and r300_vertex_program should probably be merged together someday.
* Keeping them them seperate for now should ensure fixed pipeline keeps functioning properly.
*/
struct r300_vertex_program_key {
GLuint InputsRead;
GLuint OutputsWritten;
+ GLuint OutputsAdded;
};
struct r300_vertex_program {
struct r300_vertex_program *progs;
};
-#define PFS_MAX_ALU_INST 64
-#define PFS_MAX_TEX_INST 64
-#define PFS_MAX_TEX_INDIRECT 4
-#define PFS_NUM_TEMP_REGS 32
-#define PFS_NUM_CONST_REGS 16
+#define R300_PFS_MAX_ALU_INST 64
+#define R300_PFS_MAX_TEX_INST 32
+#define R300_PFS_MAX_TEX_INDIRECT 4
+#define R300_PFS_NUM_TEMP_REGS 32
+#define R300_PFS_NUM_CONST_REGS 32
-/* Mapping Mesa registers to R300 temporaries */
-struct reg_acc {
- int reg; /* Assigned hw temp */
- unsigned int refcount; /* Number of uses by mesa program */
-};
+#define R500_PFS_MAX_INST 512
+#define R500_PFS_NUM_TEMP_REGS 128
+#define R500_PFS_NUM_CONST_REGS 256
-/**
- * Describe the current lifetime information for an R300 temporary
- */
-struct reg_lifetime {
- /* Index of the first slot where this register is free in the sense
- that it can be used as a new destination register.
- This is -1 if the register has been assigned to a Mesa register
- and the last access to the register has not yet been emitted */
- int free;
-
- /* Index of the first slot where this register is currently reserved.
- This is used to stop e.g. a scalar operation from being moved
- before the allocation time of a register that was first allocated
- for a vector operation. */
- int reserved;
-
- /* Index of the first slot in which the register can be used as a
- source without losing the value that is written by the last
- emitted instruction that writes to the register */
- int vector_valid;
- int scalar_valid;
-
- /* Index to the slot where the register was last read.
- This is also the first slot in which the register may be written again */
- int vector_lastread;
- int scalar_lastread;
-};
+struct r300_pfs_compile_state;
+struct r500_pfs_compile_state;
/**
- * Store usage information about an ALU instruction slot during the
- * compilation of a fragment program.
+ * Stores state that influences the compilation of a fragment program.
*/
-#define SLOT_SRC_VECTOR (1<<0)
-#define SLOT_SRC_SCALAR (1<<3)
-#define SLOT_SRC_BOTH (SLOT_SRC_VECTOR | SLOT_SRC_SCALAR)
-#define SLOT_OP_VECTOR (1<<16)
-#define SLOT_OP_SCALAR (1<<17)
-#define SLOT_OP_BOTH (SLOT_OP_VECTOR | SLOT_OP_SCALAR)
-
-struct r300_pfs_compile_slot {
- /* Bitmask indicating which parts of the slot are used, using SLOT_ constants
- defined above */
- unsigned int used;
-
- /* Selected sources */
- int vsrc[3];
- int ssrc[3];
+struct r300_fragment_program_external_state {
+ struct {
+ /**
+ * If the sampler is used as a shadow sampler,
+ * this field is:
+ * 0 - GL_LUMINANCE
+ * 1 - GL_INTENSITY
+ * 2 - GL_ALPHA
+ * depending on the depth texture mode.
+ */
+ GLuint depth_texture_mode : 2;
+
+ /**
+ * If the sampler is used as a shadow sampler,
+ * this field is (texture_compare_func - GL_NEVER).
+ * [e.g. if compare function is GL_LEQUAL, this field is 3]
+ *
+ * Otherwise, this field is 0.
+ */
+ GLuint texture_compare_func : 3;
+ } unit[16];
};
-/**
- * Store information during compilation of fragment programs.
- */
-struct r300_pfs_compile_state {
- int nrslots; /* number of ALU slots used so far */
- /* Track which (parts of) slots are already filled with instructions */
- struct r300_pfs_compile_slot slot[PFS_MAX_ALU_INST];
-
- /* Track the validity of R300 temporaries */
- struct reg_lifetime hwtemps[PFS_NUM_TEMP_REGS];
-
- /* Used to map Mesa's inputs/temps onto hardware temps */
- int temp_in_use;
- struct reg_acc temps[PFS_NUM_TEMP_REGS];
- struct reg_acc inputs[32]; /* don't actually need 32... */
-
- /* Track usage of hardware temps, for register allocation,
- * indirection detection, etc. */
- GLuint used_in_node;
- GLuint dest_in_node;
+struct r300_fragment_program_node {
+ int tex_offset; /**< first tex instruction */
+ int tex_end; /**< last tex instruction, relative to tex_offset */
+ int alu_offset; /**< first ALU instruction */
+ int alu_end; /**< last ALU instruction, relative to alu_offset */
+ int flags;
};
/**
- * Store everything about a fragment program that is needed
- * to render with that program.
+ * Stores an R300 fragment program in its compiled-to-hardware form.
*/
-struct r300_fragment_program {
- struct gl_fragment_program mesa_program;
-
- GLcontext *ctx;
- GLboolean translated;
- GLboolean error;
- struct r300_pfs_compile_state *cs;
-
+struct r300_fragment_program_code {
struct {
- int length;
- GLuint inst[PFS_MAX_TEX_INST];
+ int length; /**< total # of texture instructions used */
+ GLuint inst[R300_PFS_MAX_TEX_INST];
} tex;
struct {
+ int length; /**< total # of ALU instructions used */
struct {
GLuint inst0;
GLuint inst1;
GLuint inst2;
GLuint inst3;
- } inst[PFS_MAX_ALU_INST];
+ } inst[R300_PFS_MAX_ALU_INST];
} alu;
- struct {
- int tex_offset;
- int tex_end;
- int alu_offset;
- int alu_end;
- int flags;
- } node[4];
+ struct r300_fragment_program_node node[4];
int cur_node;
int first_node_has_tex;
- int alu_offset;
- int alu_end;
- int tex_offset;
- int tex_end;
-
- /* Hardware constants.
- * Contains a pointer to the value. The destination of the pointer
- * is supposed to be updated when GL state changes.
- * Typically, this is either a pointer into
- * gl_program_parameter_list::ParameterValues, or a pointer to a
- * global constant (e.g. for sin/cos-approximation)
+ /**
+ * Remember which program register a given hardware constant
+ * belongs to.
*/
- const GLfloat *constant[PFS_NUM_CONST_REGS];
+ struct prog_src_register constant[R300_PFS_NUM_CONST_REGS];
int const_nr;
int max_temp_idx;
-
- GLuint optimization;
-};
-
-#define R300_MAX_AOS_ARRAYS 16
-
-#define AOS_FORMAT_USHORT 0
-#define AOS_FORMAT_FLOAT 1
-#define AOS_FORMAT_UBYTE 2
-#define AOS_FORMAT_FLOAT_COLOR 3
-
-#define REG_COORDS 0
-#define REG_COLOR0 1
-#define REG_TEX0 2
-
-struct dt {
- GLint size;
- GLenum type;
- GLsizei stride;
- void *data;
};
-struct radeon_vertex_buffer {
- int Count;
- void *Elts;
- int elt_size;
- int elt_min, elt_max; /* debug */
- struct dt AttribPtr[VERT_ATTRIB_MAX];
+struct r500_fragment_program_code {
+ struct {
+ GLuint inst0;
+ GLuint inst1;
+ GLuint inst2;
+ GLuint inst3;
+ GLuint inst4;
+ GLuint inst5;
+ } inst[R500_PFS_MAX_INST];
+
+ int inst_offset;
+ int inst_end;
+
+ /**
+ * Remember which program register a given hardware constant
+ * belongs to.
+ */
+ struct prog_src_register constant[R500_PFS_NUM_CONST_REGS];
+ int const_nr;
- const struct _mesa_prim *Primitive;
- GLuint PrimitiveCount;
- GLint LockFirst;
- GLsizei LockCount;
- int lock_uptodate;
+ int max_temp_idx;
};
-struct r300_aos_rec {
- GLuint offset;
- int element_size; /* in dwords */
- int stride; /* distance between elements, in dwords */
-
- int format;
+/**
+* Store everything about a fragment program that is needed
+* to render with that program.
+*/
+struct r300_fragment_program {
+ struct gl_fragment_program Base;
- int ncomponents; /* number of components - between 1 and 4, inclusive */
+ GLboolean translated;
+ GLboolean error;
- int reg; /* which register they are assigned to. */
+ struct r300_fragment_program_external_state state;
+ union rX00_fragment_program_code {
+ struct r300_fragment_program_code r300;
+ struct r500_fragment_program_code r500;
+ } code;
+ GLboolean writes_depth;
+ GLuint optimization;
};
-struct r300_state {
- struct r300_depthbuffer_state depth;
- struct r300_texture_state texture;
- int sw_tcl_inputs[VERT_ATTRIB_MAX];
- struct r300_vertex_shader_state vertex_shader;
- struct r300_pfs_compile_state pfs_compile;
- struct r300_dma_region aos[R300_MAX_AOS_ARRAYS];
- int aos_count;
- struct radeon_vertex_buffer VB;
+struct r300_fragment_program_compiler {
+ r300ContextPtr r300;
+ struct r300_fragment_program *fp;
+ union rX00_fragment_program_code *code;
+ struct gl_program *program;
+};
- GLuint *Elts;
- struct r300_dma_region elt_dma;
+#define R300_MAX_AOS_ARRAYS 16
- DECLARE_RENDERINPUTS(render_inputs_bitset); /* actual render inputs that R300 was configured for.
- They are the same as tnl->render_inputs for fixed pipeline */
- struct {
- int transform_offset; /* Transform matrix offset, -1 if none */
- } vap_param; /* vertex processor parameter allocation - tells where to write parameters */
+#define R300_FALLBACK_NONE 0
+#define R300_FALLBACK_TCL 1
+#define R300_FALLBACK_RAST 2
- struct r300_stencilbuffer_state stencil;
+/* r300_swtcl.c
+ */
+struct r300_swtcl_info {
+ /*
+ * Offset of the 4UB color data within a hardware (swtcl) vertex.
+ */
+ GLuint coloroffset;
+
+ /**
+ * Offset of the 3UB specular color data within a hardware (swtcl) vertex.
+ */
+ GLuint specoffset;
+
+ struct vertex_attribute{
+ GLuint attr;
+ GLubyte format;
+ GLubyte dst_loc;
+ GLuint swizzle;
+ GLubyte write_mask;
+ } vert_attrs[VERT_ATTRIB_MAX];
+
+ GLubyte vertex_attr_count;
+
+ int sw_tcl_inputs[VERT_ATTRIB_MAX];
+};
+struct r300_vtable {
+ void (* SetupRSUnit)(GLcontext *ctx);
+ void (* SetupFragmentShaderTextures)(GLcontext *ctx, int *tmu_mappings);
+ GLboolean (* FragmentProgramEmit)(struct r300_fragment_program_compiler *compiler);
+ void (* FragmentProgramDump)(union rX00_fragment_program_code *code);
+ GLboolean (* SetupPixelShader)(GLcontext *ctx);
};
-#define R300_FALLBACK_NONE 0
-#define R300_FALLBACK_TCL 1
-#define R300_FALLBACK_RAST 2
/**
* \brief R300 context structure.
struct r300_context {
struct radeon_context radeon; /* parent class, must be first */
+ struct r300_vtable vtbl;
+
struct r300_hw_state hw;
- struct r300_cmdbuf cmdbuf;
- struct r300_state state;
- struct gl_vertex_program *curr_vp;
+
+ struct r300_vertex_shader_state vertex_shader;
struct r300_vertex_program *selected_vp;
/* Vertex buffers
*/
- struct r300_dma dma;
- GLboolean save_on_next_unlock;
- GLuint NewGLState;
-
- /* Texture object bookkeeping
- */
- unsigned nr_heaps;
- driTexHeap *texture_heaps[RADEON_NR_TEX_HEAPS];
- driTextureObject swapped;
- int texture_depth;
- float initialMaxAnisotropy;
-
- /* Clientdata textures;
- */
- GLuint prefer_gart_client_texturing;
-
-#ifdef USER_BUFFERS
- struct r300_memory_manager *rmm;
GLvector4f dummy_attrib[_TNL_ATTRIB_MAX];
GLvector4f *temp_attrib[_TNL_ATTRIB_MAX];
-#endif
GLboolean disable_lowimpact_fallback;
-};
-struct r300_buffer_object {
- struct gl_buffer_object mesa_obj;
- int id;
+ struct r300_options {
+ uint32_t conformance_mode:1;
+ uint32_t hw_tcl_enabled:1;
+ uint32_t s3tc_force_enabled:1;
+ uint32_t s3tc_force_disabled:1;
+ uint32_t stencil_two_side_disabled:1;
+ } options;
+
+ struct r300_swtcl_info swtcl;
+ GLboolean vap_flush_needed;
+
+ DECLARE_RENDERINPUTS(render_inputs_bitset);
};
#define R300_CONTEXT(ctx) ((r300ContextPtr)(ctx->DriverCtx))
struct r300_vertex_program_cont *vp,
float *dst);
-#define RADEON_D_CAPTURE 0
-#define RADEON_D_PLAYBACK 1
-#define RADEON_D_PLAYBACK_RAW 2
-#define RADEON_D_T 3
+extern void r300InitShaderFunctions(r300ContextPtr r300);
+
+#define r300PackFloat32 radeonPackFloat32
+#define r300PackFloat24 radeonPackFloat24
#endif /* __R300_CONTEXT_H__ */