diff options
Diffstat (limited to 'src/gallium/drivers/cell')
-rw-r--r-- | src/gallium/drivers/cell/common.h | 3 | ||||
-rw-r--r-- | src/gallium/drivers/cell/ppu/cell_context.h | 8 | ||||
-rw-r--r-- | src/gallium/drivers/cell/ppu/cell_draw_arrays.c | 20 | ||||
-rw-r--r-- | src/gallium/drivers/cell/ppu/cell_state_derived.c | 8 | ||||
-rw-r--r-- | src/gallium/drivers/cell/ppu/cell_state_emit.c | 2 | ||||
-rw-r--r-- | src/gallium/drivers/cell/spu/spu_command.c | 9 | ||||
-rw-r--r-- | src/gallium/drivers/cell/spu/spu_exec.c | 8 | ||||
-rw-r--r-- | src/gallium/drivers/cell/spu/spu_exec.h | 4 | ||||
-rw-r--r-- | src/gallium/drivers/cell/spu/spu_funcs.c | 2 | ||||
-rw-r--r-- | src/gallium/drivers/cell/spu/spu_main.h | 22 | ||||
-rw-r--r-- | src/gallium/drivers/cell/spu/spu_render.c | 2 | ||||
-rw-r--r-- | src/gallium/drivers/cell/spu/spu_vertex_fetch.c | 5 | ||||
-rw-r--r-- | src/gallium/drivers/cell/spu/spu_vertex_shader.c | 15 |
13 files changed, 55 insertions, 53 deletions
diff --git a/src/gallium/drivers/cell/common.h b/src/gallium/drivers/cell/common.h index d5f5c7bbba..aa29dcb394 100644 --- a/src/gallium/drivers/cell/common.h +++ b/src/gallium/drivers/cell/common.h @@ -358,6 +358,7 @@ struct cell_spu_function_info /** This is the object passed to spe_create_thread() */ +PIPE_ALIGN_TYPE(16, struct cell_init_info { unsigned id; @@ -370,7 +371,7 @@ struct cell_init_info uint *buffer_status; /**< points at cell_context->buffer_status */ struct cell_spu_function_info *spu_functions; -} ALIGN16_ATTRIB; +}); #endif /* CELL_COMMON_H */ diff --git a/src/gallium/drivers/cell/ppu/cell_context.h b/src/gallium/drivers/cell/ppu/cell_context.h index 1498fb665a..e402ed2922 100644 --- a/src/gallium/drivers/cell/ppu/cell_context.h +++ b/src/gallium/drivers/cell/ppu/cell_context.h @@ -89,7 +89,7 @@ struct cell_buffer_node; */ struct cell_buffer_list { - struct cell_fence fence ALIGN16_ATTRIB; + PIPE_ALIGN_VAR(16) struct cell_fence fence; struct cell_buffer_node *head; }; @@ -150,18 +150,18 @@ struct cell_context /** Mapped constant buffers */ void *mapped_constants[PIPE_SHADER_TYPES]; - struct cell_spu_function_info spu_functions ALIGN16_ATTRIB; + PIPE_ALIGN_VAR(16) struct cell_spu_function_info spu_functions; uint num_cells, num_spus; /** Buffers for command batches, vertex/index data */ uint buffer_size[CELL_NUM_BUFFERS]; - ubyte buffer[CELL_NUM_BUFFERS][CELL_BUFFER_SIZE] ALIGN16_ATTRIB; + PIPE_ALIGN_VAR(16) ubyte buffer[CELL_NUM_BUFFERS][CELL_BUFFER_SIZE]; int cur_batch; /**< which buffer is being filled w/ commands */ /** [4] to ensure 16-byte alignment for each status word */ - uint buffer_status[CELL_MAX_SPUS][CELL_NUM_BUFFERS][4] ALIGN16_ATTRIB; + PIPE_ALIGN_VAR(16) uint buffer_status[CELL_MAX_SPUS][CELL_NUM_BUFFERS][4]; /** Associated with each command/batch buffer is a list of pipe_buffers diff --git a/src/gallium/drivers/cell/ppu/cell_draw_arrays.c b/src/gallium/drivers/cell/ppu/cell_draw_arrays.c index 2016b21a40..0a4da8ecc8 100644 --- a/src/gallium/drivers/cell/ppu/cell_draw_arrays.c +++ b/src/gallium/drivers/cell/ppu/cell_draw_arrays.c @@ -59,7 +59,7 @@ cell_map_constant_buffers(struct cell_context *sp) } } - draw_set_mapped_constant_buffer(sp->draw, + draw_set_mapped_constant_buffer(sp->draw, PIPE_SHADER_VERTEX, sp->mapped_constants[PIPE_SHADER_VERTEX], sp->constants[PIPE_SHADER_VERTEX]->size); } @@ -85,7 +85,7 @@ cell_unmap_constant_buffers(struct cell_context *sp) * * XXX should the element buffer be specified/bound with a separate function? */ -static boolean +static void cell_draw_range_elements(struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, @@ -145,29 +145,27 @@ cell_draw_range_elements(struct pipe_context *pipe, /* Note: leave drawing surfaces mapped */ cell_unmap_constant_buffers(sp); - - return TRUE; } -static boolean +static void cell_draw_elements(struct pipe_context *pipe, struct pipe_buffer *indexBuffer, unsigned indexSize, unsigned mode, unsigned start, unsigned count) { - return cell_draw_range_elements( pipe, indexBuffer, - indexSize, - 0, 0xffffffff, - mode, start, count ); + cell_draw_range_elements( pipe, indexBuffer, + indexSize, + 0, 0xffffffff, + mode, start, count ); } -static boolean +static void cell_draw_arrays(struct pipe_context *pipe, unsigned mode, unsigned start, unsigned count) { - return cell_draw_elements(pipe, NULL, 0, mode, start, count); + cell_draw_elements(pipe, NULL, 0, mode, start, count); } diff --git a/src/gallium/drivers/cell/ppu/cell_state_derived.c b/src/gallium/drivers/cell/ppu/cell_state_derived.c index efc4f78364..b723e794e7 100644 --- a/src/gallium/drivers/cell/ppu/cell_state_derived.c +++ b/src/gallium/drivers/cell/ppu/cell_state_derived.c @@ -66,7 +66,7 @@ calculate_vertex_layout( struct cell_context *cell ) vinfo->num_attribs = 0; /* we always want to emit vertex pos */ - src = draw_find_vs_output(cell->draw, TGSI_SEMANTIC_POSITION, 0); + src = draw_find_shader_output(cell->draw, TGSI_SEMANTIC_POSITION, 0); assert(src >= 0); draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_POS, src); @@ -82,14 +82,14 @@ calculate_vertex_layout( struct cell_context *cell ) break; case TGSI_SEMANTIC_COLOR: - src = draw_find_vs_output(cell->draw, TGSI_SEMANTIC_COLOR, + src = draw_find_shader_output(cell->draw, TGSI_SEMANTIC_COLOR, fs->info.input_semantic_index[i]); assert(src >= 0); draw_emit_vertex_attr(vinfo, EMIT_4F, colorInterp, src); break; case TGSI_SEMANTIC_FOG: - src = draw_find_vs_output(cell->draw, TGSI_SEMANTIC_FOG, 0); + src = draw_find_shader_output(cell->draw, TGSI_SEMANTIC_FOG, 0); #if 1 if (src < 0) /* XXX temp hack, try demos/fogcoord.c with this */ src = 0; @@ -100,7 +100,7 @@ calculate_vertex_layout( struct cell_context *cell ) case TGSI_SEMANTIC_GENERIC: /* this includes texcoords and varying vars */ - src = draw_find_vs_output(cell->draw, TGSI_SEMANTIC_GENERIC, + src = draw_find_shader_output(cell->draw, TGSI_SEMANTIC_GENERIC, fs->info.input_semantic_index[i]); assert(src >= 0); draw_emit_vertex_attr(vinfo, EMIT_4F, INTERP_PERSPECTIVE, src); diff --git a/src/gallium/drivers/cell/ppu/cell_state_emit.c b/src/gallium/drivers/cell/ppu/cell_state_emit.c index ab55a24bb6..f1e1dcb9eb 100644 --- a/src/gallium/drivers/cell/ppu/cell_state_emit.c +++ b/src/gallium/drivers/cell/ppu/cell_state_emit.c @@ -331,7 +331,7 @@ cell_emit_state(struct cell_context *cell) const struct draw_context *const draw = cell->draw; struct cell_shader_info info; - info.num_outputs = draw_num_vs_outputs(draw); + info.num_outputs = draw_num_shader_outputs(draw); info.declarations = (uintptr_t) draw->vs.machine.Declarations; info.num_declarations = draw->vs.machine.NumDeclarations; info.instructions = (uintptr_t) draw->vs.machine.Instructions; diff --git a/src/gallium/drivers/cell/spu/spu_command.c b/src/gallium/drivers/cell/spu/spu_command.c index 5c0179d954..55bd85bde2 100644 --- a/src/gallium/drivers/cell/spu/spu_command.c +++ b/src/gallium/drivers/cell/spu/spu_command.c @@ -53,8 +53,7 @@ struct spu_vs_context draw; /** * Buffers containing dynamically generated SPU code: */ -static unsigned char attribute_fetch_code_buffer[136 * PIPE_MAX_ATTRIBS] - ALIGN16_ATTRIB; +PIPE_ALIGN_VAR(16) static unsigned char attribute_fetch_code_buffer[136 * PIPE_MAX_ATTRIBS]; @@ -405,8 +404,6 @@ cmd_state_sampler(const struct cell_command_sampler *sampler) case PIPE_TEX_FILTER_LINEAR: spu.min_sample_texture_2d[unit] = sample_texture_2d_bilinear; break; - case PIPE_TEX_FILTER_ANISO: - /* fall-through, for now */ case PIPE_TEX_FILTER_NEAREST: spu.min_sample_texture_2d[unit] = sample_texture_2d_nearest; break; @@ -418,8 +415,6 @@ cmd_state_sampler(const struct cell_command_sampler *sampler) case PIPE_TEX_FILTER_LINEAR: spu.mag_sample_texture_2d[unit] = sample_texture_2d_bilinear; break; - case PIPE_TEX_FILTER_ANISO: - /* fall-through, for now */ case PIPE_TEX_FILTER_NEAREST: spu.mag_sample_texture_2d[unit] = sample_texture_2d_nearest; break; @@ -547,7 +542,7 @@ cmd_batch(uint opcode) { const uint buf = (opcode >> 8) & 0xff; uint size = (opcode >> 16); - qword buffer[CELL_BUFFER_SIZE / 16] ALIGN16_ATTRIB; + PIPE_ALIGN_VAR(16) qword buffer[CELL_BUFFER_SIZE / 16]; const unsigned usize = ROUNDUP16(size) / sizeof(buffer[0]); uint pos; diff --git a/src/gallium/drivers/cell/spu/spu_exec.c b/src/gallium/drivers/cell/spu/spu_exec.c index 5ed330aa6e..d2166a4901 100644 --- a/src/gallium/drivers/cell/spu/spu_exec.c +++ b/src/gallium/drivers/cell/spu/spu_exec.c @@ -1681,7 +1681,7 @@ exec_instruction( } break; - case TGSI_OPCODE_SHR: + case TGSI_OPCODE_ISHR: FOR_EACH_ENABLED_CHANNEL( *inst, chan_index ) { FETCH( &r[0], 0, chan_index ); FETCH( &r[1], 1, chan_index ); @@ -1839,10 +1839,11 @@ spu_exec_machine_run( struct spu_exec_machine *mach ) /* execute declarations (interpolants) */ if( mach->Processor == TGSI_PROCESSOR_FRAGMENT ) { for (i = 0; i < mach->NumDeclarations; i++) { + PIPE_ALIGN_VAR(16) union { struct tgsi_full_declaration decl; qword buffer[ROUNDUP16(sizeof(struct tgsi_full_declaration)) / 16]; - } d ALIGN16_ATTRIB; + } d; unsigned ea = (unsigned) (mach->Declarations + pc); spu_dcache_fetch_unaligned(d.buffer, ea, sizeof(d.decl)); @@ -1853,10 +1854,11 @@ spu_exec_machine_run( struct spu_exec_machine *mach ) /* execute instructions, until pc is set to -1 */ while (pc != -1) { + PIPE_ALIGN_VAR(16) union { struct tgsi_full_instruction inst; qword buffer[ROUNDUP16(sizeof(struct tgsi_full_instruction)) / 16]; - } i ALIGN16_ATTRIB; + } i; unsigned ea = (unsigned) (mach->Instructions + pc); spu_dcache_fetch_unaligned(i.buffer, ea, sizeof(i.inst)); diff --git a/src/gallium/drivers/cell/spu/spu_exec.h b/src/gallium/drivers/cell/spu/spu_exec.h index 8605679940..0ca92af248 100644 --- a/src/gallium/drivers/cell/spu/spu_exec.h +++ b/src/gallium/drivers/cell/spu/spu_exec.h @@ -98,9 +98,9 @@ struct spu_exec_machine * 4 internal temporaries * 1 address */ + PIPE_ALIGN_VAR(16) struct spu_exec_vector Temps[TGSI_EXEC_NUM_TEMPS - + TGSI_EXEC_NUM_TEMP_EXTRAS + 1] - ALIGN16_ATTRIB; + + TGSI_EXEC_NUM_TEMP_EXTRAS + 1]; struct spu_exec_vector *Addrs; diff --git a/src/gallium/drivers/cell/spu/spu_funcs.c b/src/gallium/drivers/cell/spu/spu_funcs.c index ff3d609d25..98919c43ff 100644 --- a/src/gallium/drivers/cell/spu/spu_funcs.c +++ b/src/gallium/drivers/cell/spu/spu_funcs.c @@ -144,7 +144,7 @@ export_func(struct cell_spu_function_info *spu_functions, void return_function_info(void) { - struct cell_spu_function_info funcs ALIGN16_ATTRIB; + PIPE_ALIGN_VAR(16) struct cell_spu_function_info funcs; int tag = TAG_MISC; ASSERT(sizeof(funcs) == 256); /* must be multiple of 16 bytes */ diff --git a/src/gallium/drivers/cell/spu/spu_main.h b/src/gallium/drivers/cell/spu/spu_main.h index 33767e7c51..b18f4c22ef 100644 --- a/src/gallium/drivers/cell/spu/spu_main.h +++ b/src/gallium/drivers/cell/spu/spu_main.h @@ -93,6 +93,7 @@ typedef vector unsigned int (*spu_fragment_program_func)(vector float *inputs, vector float *constants); +PIPE_ALIGN_TYPE(16, struct spu_framebuffer { void *color_start; /**< addr of color surface in main memory */ @@ -107,10 +108,11 @@ struct spu_framebuffer uint zsize; /**< 0, 2 or 4 bytes per Z */ float zscale; /**< 65535.0, 2^24-1 or 2^32-1 */ -} ALIGN16_ATTRIB; +}); /** per-texture level info */ +PIPE_ALIGN_TYPE(16, struct spu_texture_level { void *start; @@ -123,20 +125,22 @@ struct spu_texture_level vector signed int mask_s, mask_t, mask_r; /** texcoord clamp limits */ vector signed int max_s, max_t, max_r; -} ALIGN16_ATTRIB; +}); +PIPE_ALIGN_TYPE(16, struct spu_texture { struct spu_texture_level level[CELL_MAX_TEXTURE_LEVELS]; uint max_level; uint target; /**< PIPE_TEXTURE_x */ -} ALIGN16_ATTRIB; +}); /** * All SPU global/context state will be in a singleton object of this type: */ +PIPE_ALIGN_TYPE(16, struct spu_global { /** One-time init/constant info */ @@ -155,8 +159,8 @@ struct spu_global struct vertex_info vertex_info; /** Current color and Z tiles */ - tile_t ctile ALIGN16_ATTRIB; - tile_t ztile ALIGN16_ATTRIB; + PIPE_ALIGN_VAR(16) tile_t ctile; + PIPE_ALIGN_VAR(16) tile_t ztile; /** Read depth/stencil tiles? */ boolean read_depth_stencil; @@ -165,8 +169,8 @@ struct spu_global ubyte cur_ctile_status, cur_ztile_status; /** Status of all tiles in framebuffer */ - ubyte ctile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; - ubyte ztile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE] ALIGN16_ATTRIB; + PIPE_ALIGN_VAR(16) ubyte ctile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE]; + PIPE_ALIGN_VAR(16) ubyte ztile_status[CELL_MAX_HEIGHT/TILE_SIZE][CELL_MAX_WIDTH/TILE_SIZE]; /** Current fragment ops machine code, at 8-byte boundary */ uint *fragment_ops_code; @@ -175,7 +179,7 @@ struct spu_global spu_fragment_ops_func fragment_ops[2]; /** Current fragment program machine code, at 8-byte boundary */ - uint fragment_program_code[SPU_MAX_FRAGMENT_PROGRAM_INSTS] ALIGN8_ATTRIB; + PIPE_ALIGN_VAR(8) uint fragment_program_code[SPU_MAX_FRAGMENT_PROGRAM_INSTS]; /** Current fragment ops function */ spu_fragment_program_func fragment_program; @@ -187,7 +191,7 @@ struct spu_global /** Fragment program constants */ vector float constants[4 * CELL_MAX_CONSTANTS]; -} ALIGN16_ATTRIB; +}); extern struct spu_global spu; diff --git a/src/gallium/drivers/cell/spu/spu_render.c b/src/gallium/drivers/cell/spu/spu_render.c index 5ffb7073ab..14987e3c3a 100644 --- a/src/gallium/drivers/cell/spu/spu_render.c +++ b/src/gallium/drivers/cell/spu/spu_render.c @@ -169,7 +169,7 @@ void cmd_render(const struct cell_command_render *render, uint *pos_incr) { /* we'll DMA into these buffers */ - ubyte vertex_data[CELL_BUFFER_SIZE] ALIGN16_ATTRIB; + PIPE_ALIGN_VAR(16) ubyte vertex_data[CELL_BUFFER_SIZE]; const uint vertex_size = render->vertex_size; /* in bytes */ /*const*/ uint total_vertex_bytes = render->num_verts * vertex_size; uint index_bytes; diff --git a/src/gallium/drivers/cell/spu/spu_vertex_fetch.c b/src/gallium/drivers/cell/spu/spu_vertex_fetch.c index 03375d84a5..087963960d 100644 --- a/src/gallium/drivers/cell/spu/spu_vertex_fetch.c +++ b/src/gallium/drivers/cell/spu/spu_vertex_fetch.c @@ -43,7 +43,8 @@ typedef void (*spu_fetch_func)(qword *out, const qword *in, const qword *shuffle_data); -static const qword fetch_shuffle_data[5] ALIGN16_ATTRIB = { +PIPE_ALIGN_VAR(16) static const qword +fetch_shuffle_data[5] = { /* Shuffle used by CVT_64_FLOAT */ { @@ -110,7 +111,7 @@ static void generic_vertex_fetch(struct spu_vs_context *draw, unsigned idx; const unsigned bytes_per_entry = draw->vertex_fetch.size[attr]; const unsigned quads_per_entry = (bytes_per_entry + 15) / 16; - qword in[2 * 4] ALIGN16_ATTRIB; + PIPE_ALIGN_VAR(16) qword in[2 * 4]; /* Fetch four attributes for four vertices. diff --git a/src/gallium/drivers/cell/spu/spu_vertex_shader.c b/src/gallium/drivers/cell/spu/spu_vertex_shader.c index fbe5b34d39..3e9804bf8e 100644 --- a/src/gallium/drivers/cell/spu/spu_vertex_shader.c +++ b/src/gallium/drivers/cell/spu/spu_vertex_shader.c @@ -107,8 +107,8 @@ run_vertex_program(struct spu_vs_context *draw, struct spu_exec_machine *machine = &draw->machine; unsigned int j; - ALIGN16_DECL(struct spu_exec_vector, inputs, PIPE_MAX_ATTRIBS); - ALIGN16_DECL(struct spu_exec_vector, outputs, PIPE_MAX_ATTRIBS); + PIPE_ALIGN_VAR(16) struct spu_exec_vector inputs[PIPE_MAX_ATTRIBS]; + PIPE_ALIGN_VAR(16) struct spu_exec_vector outputs[PIPE_MAX_ATTRIBS]; const float *scale = draw->viewport.scale; const float *trans = draw->viewport.translate; @@ -119,8 +119,8 @@ run_vertex_program(struct spu_vs_context *draw, ASSERT_ALIGN16(draw->constants); machine->Consts = (float (*)[4]) draw->constants; - machine->Inputs = ALIGN16_ASSIGN(inputs); - machine->Outputs = ALIGN16_ASSIGN(outputs); + machine->Inputs = inputs; + machine->Outputs = outputs; spu_vertex_fetch( draw, machine, elts, count ); @@ -132,8 +132,9 @@ run_vertex_program(struct spu_vs_context *draw, for (j = 0; j < count; j++) { unsigned slot; float x, y, z, w; + PIPE_ALIGN_VAR(16) unsigned char buffer[sizeof(struct vertex_header) - + MAX_VERTEX_SIZE] ALIGN16_ATTRIB; + + MAX_VERTEX_SIZE]; struct vertex_header *const tmpOut = (struct vertex_header *) buffer; const unsigned vert_size = ROUNDUP16(sizeof(struct vertex_header) @@ -186,8 +187,8 @@ run_vertex_program(struct spu_vs_context *draw, } -unsigned char immediates[(sizeof(float) * 4 * TGSI_EXEC_NUM_IMMEDIATES) + 32] - ALIGN16_ATTRIB; +PIPE_ALIGN_VAR(16) unsigned char +immediates[(sizeof(float) * 4 * TGSI_EXEC_NUM_IMMEDIATES) + 32]); void |