#include "util/u_pack_color.h"
#include "util/u_memory.h"
#include "util/u_framebuffer.h"
+#include "util/u_dual_blend.h"
static uint32_t r600_translate_blend_function(int blend_func)
{
return 0;
}
-static unsigned r600_tex_dim(unsigned dim)
+static unsigned r600_tex_dim(unsigned dim, unsigned nr_samples)
{
switch (dim) {
default:
return V_038000_SQ_TEX_DIM_1D_ARRAY;
case PIPE_TEXTURE_2D:
case PIPE_TEXTURE_RECT:
- return V_038000_SQ_TEX_DIM_2D;
+ return nr_samples > 1 ? V_038000_SQ_TEX_DIM_2D_MSAA :
+ V_038000_SQ_TEX_DIM_2D;
case PIPE_TEXTURE_2D_ARRAY:
- return V_038000_SQ_TEX_DIM_2D_ARRAY;
+ return nr_samples > 1 ? V_038000_SQ_TEX_DIM_2D_ARRAY_MSAA :
+ V_038000_SQ_TEX_DIM_2D_ARRAY;
case PIPE_TEXTURE_3D:
return V_038000_SQ_TEX_DIM_3D;
case PIPE_TEXTURE_CUBE:
case PIPE_FORMAT_R8SG8SB8UX8U_NORM:
case PIPE_FORMAT_X8B8G8R8_UNORM:
case PIPE_FORMAT_X8R8G8B8_UNORM:
- case PIPE_FORMAT_R8G8B8_UNORM:
case PIPE_FORMAT_R8G8B8A8_SINT:
case PIPE_FORMAT_R8G8B8A8_UINT:
return V_0280A0_COLOR_8_8_8_8;
case PIPE_FORMAT_R16G16B16A16_SNORM:
return V_0280A0_COLOR_16_16_16_16;
- case PIPE_FORMAT_R16G16B16_FLOAT:
case PIPE_FORMAT_R16G16B16A16_FLOAT:
return V_0280A0_COLOR_16_16_16_16_FLOAT;
case PIPE_FORMAT_L32A32_SINT:
return V_0280A0_COLOR_32_32;
- /* 96-bit buffers. */
- case PIPE_FORMAT_R32G32B32_FLOAT:
- return V_0280A0_COLOR_32_32_32_FLOAT;
-
/* 128-bit buffers. */
case PIPE_FORMAT_R32G32B32A32_FLOAT:
return V_0280A0_COLOR_32_32_32_32_FLOAT;
unsigned sample_count,
unsigned usage)
{
+ struct r600_screen *rscreen = (struct r600_screen*)screen;
unsigned retval = 0;
if (target >= PIPE_MAX_TEXTURE_TYPES) {
if (!util_format_is_supported(format, usage))
return FALSE;
- /* Multisample */
- if (sample_count > 1)
- return FALSE;
+ if (sample_count > 1) {
+ if (rscreen->info.drm_minor < 22)
+ return FALSE;
+
+ /* R11G11B10 is broken on R6xx. */
+ if (rscreen->chip_class == R600 &&
+ format == PIPE_FORMAT_R11G11B10_FLOAT)
+ return FALSE;
+
+ switch (sample_count) {
+ case 2:
+ case 4:
+ case 8:
+ break;
+ default:
+ return FALSE;
+ }
+
+ /* require render-target support for multisample resources */
+ if (util_format_is_depth_or_stencil(format)) {
+ usage |= PIPE_BIND_DEPTH_STENCIL;
+ } else if (util_format_is_pure_integer(format)) {
+ return FALSE; /* no integer textures */
+ } else {
+ usage |= PIPE_BIND_RENDER_TARGET;
+ }
+ }
if ((usage & PIPE_BIND_SAMPLER_VIEW) &&
r600_is_sampler_format_supported(screen, format)) {
float offset_units = rctx->rasterizer->offset_units;
unsigned offset_db_fmt_cntl = 0, depth;
- switch (rctx->framebuffer.zsbuf->texture->format) {
+ switch (rctx->framebuffer.zsbuf->format) {
case PIPE_FORMAT_Z24X8_UNORM:
case PIPE_FORMAT_Z24_UNORM_S8_UINT:
depth = -24;
offset_db_fmt_cntl |= S_028DF8_POLY_OFFSET_NEG_NUM_DB_BITS(depth);
r600_pipe_state_add_reg(&state,
R_028E00_PA_SU_POLY_OFFSET_FRONT_SCALE,
- fui(rctx->rasterizer->offset_scale), NULL, 0);
+ fui(rctx->rasterizer->offset_scale));
r600_pipe_state_add_reg(&state,
R_028E04_PA_SU_POLY_OFFSET_FRONT_OFFSET,
- fui(offset_units), NULL, 0);
+ fui(offset_units));
r600_pipe_state_add_reg(&state,
R_028E08_PA_SU_POLY_OFFSET_BACK_SCALE,
- fui(rctx->rasterizer->offset_scale), NULL, 0);
+ fui(rctx->rasterizer->offset_scale));
r600_pipe_state_add_reg(&state,
R_028E0C_PA_SU_POLY_OFFSET_BACK_OFFSET,
- fui(offset_units), NULL, 0);
+ fui(offset_units));
r600_pipe_state_add_reg(&state,
R_028DF8_PA_SU_POLY_OFFSET_DB_FMT_CNTL,
- offset_db_fmt_cntl, NULL, 0);
+ offset_db_fmt_cntl);
r600_context_pipe_state_set(rctx, &state);
}
}
-static void *r600_create_blend_state(struct pipe_context *ctx,
- const struct pipe_blend_state *state)
+static void *r600_create_blend_state_mode(struct pipe_context *ctx,
+ const struct pipe_blend_state *state,
+ int mode)
{
struct r600_context *rctx = (struct r600_context *)ctx;
struct r600_pipe_blend *blend = CALLOC_STRUCT(r600_pipe_blend);
struct r600_pipe_state *rstate;
- uint32_t color_control = 0, target_mask;
+ uint32_t color_control = 0, target_mask = 0;
if (blend == NULL) {
return NULL;
rstate->id = R600_PIPE_STATE_BLEND;
- target_mask = 0;
-
/* R600 does not support per-MRT blends */
if (rctx->family > CHIP_R600)
color_control |= S_028808_PER_MRT_BLEND(1);
+
if (state->logicop_enable) {
color_control |= (state->logicop_func << 16) | (state->logicop_func << 20);
} else {
target_mask |= (state->rt[0].colormask << (4 * i));
}
}
+
+ if (target_mask)
+ color_control |= S_028808_SPECIAL_OP(mode);
+ else
+ color_control |= S_028808_SPECIAL_OP(V_028808_DISABLE);
+
blend->cb_target_mask = target_mask;
blend->cb_color_control = color_control;
-
+ /* only MRT0 has dual src blend */
+ blend->dual_src_blend = util_blend_state_is_dual(state, 0);
for (int i = 0; i < 8; i++) {
/* state->rt entries > 0 only written if independent blending */
const int j = state->independent_blend_enable ? i : 0;
/* R600 does not support per-MRT blends */
if (rctx->family > CHIP_R600)
- r600_pipe_state_add_reg(rstate, R_028780_CB_BLEND0_CONTROL + i * 4, bc, NULL, 0);
+ r600_pipe_state_add_reg(rstate, R_028780_CB_BLEND0_CONTROL + i * 4, bc);
if (i == 0)
- r600_pipe_state_add_reg(rstate, R_028804_CB_BLEND_CONTROL, bc, NULL, 0);
+ r600_pipe_state_add_reg(rstate, R_028804_CB_BLEND_CONTROL, bc);
}
+
+ r600_pipe_state_add_reg(rstate, R_028D44_DB_ALPHA_TO_MASK,
+ S_028D44_ALPHA_TO_MASK_ENABLE(state->alpha_to_coverage) |
+ S_028D44_ALPHA_TO_MASK_OFFSET0(2) |
+ S_028D44_ALPHA_TO_MASK_OFFSET1(2) |
+ S_028D44_ALPHA_TO_MASK_OFFSET2(2) |
+ S_028D44_ALPHA_TO_MASK_OFFSET3(2));
+
+ blend->alpha_to_one = state->alpha_to_one;
return rstate;
}
+
+static void *r600_create_blend_state(struct pipe_context *ctx,
+ const struct pipe_blend_state *state)
+{
+ return r600_create_blend_state_mode(ctx, state, V_028808_SPECIAL_NORMAL);
+}
+
static void *r600_create_dsa_state(struct pipe_context *ctx,
const struct pipe_depth_stencil_alpha_state *state)
{
alpha_test_control |= S_028410_ALPHA_TEST_ENABLE(1);
alpha_ref = fui(state->alpha.ref_value);
}
+ dsa->sx_alpha_test_control = alpha_test_control & 0xff;
dsa->alpha_ref = alpha_ref;
- r600_pipe_state_add_reg(rstate, R_028410_SX_ALPHA_TEST_CONTROL, alpha_test_control, NULL, 0);
- r600_pipe_state_add_reg(rstate, R_028800_DB_DEPTH_CONTROL, db_depth_control, NULL, 0);
+ r600_pipe_state_add_reg(rstate, R_028800_DB_DEPTH_CONTROL, db_depth_control);
return rstate;
}
S_028810_ZCLIP_NEAR_DISABLE(!state->depth_clip) |
S_028810_ZCLIP_FAR_DISABLE(!state->depth_clip) |
S_028810_DX_LINEAR_ATTR_CLIP_ENA(1);
+ rs->multisample_enable = state->multisample;
/* offset */
rs->offset_units = state->offset_units;
tmp |= S_0286D4_PNT_SPRITE_TOP_1(1);
}
}
- r600_pipe_state_add_reg(rstate, R_0286D4_SPI_INTERP_CONTROL_0, tmp, NULL, 0);
+ r600_pipe_state_add_reg(rstate, R_0286D4_SPI_INTERP_CONTROL_0, tmp);
/* point size 12.4 fixed point */
tmp = r600_pack_float_12p4(state->point_size/2);
- r600_pipe_state_add_reg(rstate, R_028A00_PA_SU_POINT_SIZE, S_028A00_HEIGHT(tmp) | S_028A00_WIDTH(tmp), NULL, 0);
+ r600_pipe_state_add_reg(rstate, R_028A00_PA_SU_POINT_SIZE, S_028A00_HEIGHT(tmp) | S_028A00_WIDTH(tmp));
if (state->point_size_per_vertex) {
psize_min = util_get_min_point_size(state);
/* Divide by two, because 0.5 = 1 pixel. */
r600_pipe_state_add_reg(rstate, R_028A04_PA_SU_POINT_MINMAX,
S_028A04_MIN_SIZE(r600_pack_float_12p4(psize_min/2)) |
- S_028A04_MAX_SIZE(r600_pack_float_12p4(psize_max/2)),
- NULL, 0);
+ S_028A04_MAX_SIZE(r600_pack_float_12p4(psize_max/2)));
tmp = r600_pack_float_12p4(state->line_width/2);
- r600_pipe_state_add_reg(rstate, R_028A08_PA_SU_LINE_CNTL, S_028A08_WIDTH(tmp), NULL, 0);
+ r600_pipe_state_add_reg(rstate, R_028A08_PA_SU_LINE_CNTL, S_028A08_WIDTH(tmp));
if (rctx->chip_class >= R700) {
sc_mode_cntl =
+ S_028A4C_MSAA_ENABLE(state->multisample) |
S_028A4C_FORCE_EOV_CNTDWN_ENABLE(1) |
S_028A4C_FORCE_EOV_REZ_ENABLE(1) |
S_028A4C_R700_ZMM_LINE_OFFSET(1) |
S_028A4C_R700_VPORT_SCISSOR_ENABLE(state->scissor);
} else {
sc_mode_cntl =
+ S_028A4C_MSAA_ENABLE(state->multisample) |
S_028A4C_WALK_ALIGN8_PRIM_FITS_ST(1) |
S_028A4C_FORCE_EOV_CNTDWN_ENABLE(1);
rs->scissor_enable = state->scissor;
}
sc_mode_cntl |= S_028A4C_LINE_STIPPLE_ENABLE(state->line_stipple_enable);
- r600_pipe_state_add_reg(rstate, R_028A4C_PA_SC_MODE_CNTL, sc_mode_cntl,
- NULL, 0);
+ r600_pipe_state_add_reg(rstate, R_028A4C_PA_SC_MODE_CNTL, sc_mode_cntl);
r600_pipe_state_add_reg(rstate, R_028C08_PA_SU_VTX_CNTL,
- S_028C08_PIX_CENTER_HALF(state->gl_rasterization_rules),
- NULL, 0);
+ S_028C08_PIX_CENTER_HALF(state->gl_rasterization_rules) |
+ S_028C08_QUANT_MODE(V_028C08_X_1_256TH));
- r600_pipe_state_add_reg(rstate, R_028DFC_PA_SU_POLY_OFFSET_CLAMP, fui(state->offset_clamp), NULL, 0);
+ r600_pipe_state_add_reg(rstate, R_028DFC_PA_SU_POLY_OFFSET_CLAMP, fui(state->offset_clamp));
r600_pipe_state_add_reg(rstate, R_028814_PA_SU_SC_MODE_CNTL,
S_028814_PROVOKING_VTX_LAST(prov_vtx) |
S_028814_CULL_FRONT(state->cull_face & PIPE_FACE_FRONT ? 1 : 0) |
S_028814_POLY_OFFSET_PARA_ENABLE(state->offset_tri) |
S_028814_POLY_MODE(polygon_dual_mode) |
S_028814_POLYMODE_FRONT_PTYPE(r600_translate_fill(state->fill_front)) |
- S_028814_POLYMODE_BACK_PTYPE(r600_translate_fill(state->fill_back)),
- NULL, 0);
- r600_pipe_state_add_reg(rstate, R_028350_SX_MISC, S_028350_MULTIPASS(state->rasterizer_discard), NULL, 0);
+ S_028814_POLYMODE_BACK_PTYPE(r600_translate_fill(state->fill_back)));
+ r600_pipe_state_add_reg(rstate, R_028350_SX_MISC, S_028350_MULTIPASS(state->rasterizer_discard));
return rstate;
}
const struct pipe_sampler_state *state)
{
struct r600_pipe_sampler_state *ss = CALLOC_STRUCT(r600_pipe_sampler_state);
- struct r600_pipe_state *rstate;
union util_color uc;
unsigned aniso_flag_offset = state->max_anisotropy > 1 ? 4 : 0;
}
ss->seamless_cube_map = state->seamless_cube_map;
- rstate = &ss->rstate;
- rstate->id = R600_PIPE_STATE_SAMPLER;
+ ss->border_color_use = false;
util_pack_color(state->border_color.f, PIPE_FORMAT_B8G8R8A8_UNORM, &uc);
- r600_pipe_state_add_reg_noblock(rstate, R_03C000_SQ_TEX_SAMPLER_WORD0_0,
- S_03C000_CLAMP_X(r600_tex_wrap(state->wrap_s)) |
- S_03C000_CLAMP_Y(r600_tex_wrap(state->wrap_t)) |
- S_03C000_CLAMP_Z(r600_tex_wrap(state->wrap_r)) |
- S_03C000_XY_MAG_FILTER(r600_tex_filter(state->mag_img_filter) | aniso_flag_offset) |
- S_03C000_XY_MIN_FILTER(r600_tex_filter(state->min_img_filter) | aniso_flag_offset) |
- S_03C000_MIP_FILTER(r600_tex_mipfilter(state->min_mip_filter)) |
- S_03C000_MAX_ANISO(r600_tex_aniso_filter(state->max_anisotropy)) |
- S_03C000_DEPTH_COMPARE_FUNCTION(r600_tex_compare(state->compare_func)) |
- S_03C000_BORDER_COLOR_TYPE(uc.ui ? V_03C000_SQ_TEX_BORDER_COLOR_REGISTER : 0), NULL, 0);
- r600_pipe_state_add_reg_noblock(rstate, R_03C004_SQ_TEX_SAMPLER_WORD1_0,
- S_03C004_MIN_LOD(S_FIXED(CLAMP(state->min_lod, 0, 15), 6)) |
- S_03C004_MAX_LOD(S_FIXED(CLAMP(state->max_lod, 0, 15), 6)) |
- S_03C004_LOD_BIAS(S_FIXED(CLAMP(state->lod_bias, -16, 16), 6)), NULL, 0);
- r600_pipe_state_add_reg_noblock(rstate, R_03C008_SQ_TEX_SAMPLER_WORD2_0, S_03C008_TYPE(1), NULL, 0);
+ /* R_03C000_SQ_TEX_SAMPLER_WORD0_0 */
+ ss->tex_sampler_words[0] = S_03C000_CLAMP_X(r600_tex_wrap(state->wrap_s)) |
+ S_03C000_CLAMP_Y(r600_tex_wrap(state->wrap_t)) |
+ S_03C000_CLAMP_Z(r600_tex_wrap(state->wrap_r)) |
+ S_03C000_XY_MAG_FILTER(r600_tex_filter(state->mag_img_filter) | aniso_flag_offset) |
+ S_03C000_XY_MIN_FILTER(r600_tex_filter(state->min_img_filter) | aniso_flag_offset) |
+ S_03C000_MIP_FILTER(r600_tex_mipfilter(state->min_mip_filter)) |
+ S_03C000_MAX_ANISO(r600_tex_aniso_filter(state->max_anisotropy)) |
+ S_03C000_DEPTH_COMPARE_FUNCTION(r600_tex_compare(state->compare_func)) |
+ S_03C000_BORDER_COLOR_TYPE(uc.ui ? V_03C000_SQ_TEX_BORDER_COLOR_REGISTER : 0);
+ /* R_03C004_SQ_TEX_SAMPLER_WORD1_0 */
+ ss->tex_sampler_words[1] = S_03C004_MIN_LOD(S_FIXED(CLAMP(state->min_lod, 0, 15), 6)) |
+ S_03C004_MAX_LOD(S_FIXED(CLAMP(state->max_lod, 0, 15), 6)) |
+ S_03C004_LOD_BIAS(S_FIXED(CLAMP(state->lod_bias, -16, 16), 6));
+ /* R_03C008_SQ_TEX_SAMPLER_WORD2_0 */
+ ss->tex_sampler_words[2] = S_03C008_TYPE(1);
if (uc.ui) {
- r600_pipe_state_add_reg_noblock(rstate, R_00A400_TD_PS_SAMPLER0_BORDER_RED, fui(state->border_color.f[0]), NULL, 0);
- r600_pipe_state_add_reg_noblock(rstate, R_00A404_TD_PS_SAMPLER0_BORDER_GREEN, fui(state->border_color.f[1]), NULL, 0);
- r600_pipe_state_add_reg_noblock(rstate, R_00A408_TD_PS_SAMPLER0_BORDER_BLUE, fui(state->border_color.f[2]), NULL, 0);
- r600_pipe_state_add_reg_noblock(rstate, R_00A40C_TD_PS_SAMPLER0_BORDER_ALPHA, fui(state->border_color.f[3]), NULL, 0);
- }
- return rstate;
+ ss->border_color_use = true;
+ /* R_00A400_TD_PS_SAMPLER0_BORDER_RED */
+ ss->border_color[0] = fui(state->border_color.f[0]);
+ /* R_00A404_TD_PS_SAMPLER0_BORDER_GREEN */
+ ss->border_color[1] = fui(state->border_color.f[1]);
+ /* R_00A408_TD_PS_SAMPLER0_BORDER_BLUE */
+ ss->border_color[2] = fui(state->border_color.f[2]);
+ /* R_00A40C_TD_PS_SAMPLER0_BORDER_ALPHA */
+ ss->border_color[3] = fui(state->border_color.f[3]);
+ }
+ return ss;
}
static struct pipe_sampler_view *r600_create_sampler_view(struct pipe_context *ctx,
struct pipe_resource *texture,
const struct pipe_sampler_view *state)
{
- struct r600_screen *rscreen = (struct r600_screen*)ctx->screen;
struct r600_pipe_sampler_view *view = CALLOC_STRUCT(r600_pipe_sampler_view);
- struct r600_pipe_resource_state *rstate;
- struct r600_resource_texture *tmp = (struct r600_resource_texture*)texture;
+ struct r600_texture *tmp = (struct r600_texture*)texture;
unsigned format, endian;
uint32_t word4 = 0, yuv_format = 0, pitch = 0;
unsigned char swizzle[4], array_mode = 0, tile_type = 0;
if (view == NULL)
return NULL;
- rstate = &view->state;
/* initialize base object */
view->base = *state;
format = r600_translate_texformat(ctx->screen, state->format,
swizzle,
&word4, &yuv_format);
+ assert(format != ~0);
if (format == ~0) {
- format = 0;
+ FREE(view);
+ return NULL;
}
if (tmp->is_depth && !tmp->is_flushing_texture) {
- r600_texture_depth_flush(ctx, texture, TRUE);
+ if (!r600_init_flushed_depth_texture(ctx, texture, NULL)) {
+ FREE(view);
+ return NULL;
+ }
tmp = tmp->flushed_depth_texture;
}
offset_level = state->u.tex.first_level;
last_level = state->u.tex.last_level - offset_level;
- if (!rscreen->use_surface_alloc) {
- width = u_minify(texture->width0, offset_level);
- height = u_minify(texture->height0, offset_level);
- depth = u_minify(texture->depth0, offset_level);
-
- pitch = align(tmp->pitch_in_blocks[offset_level] *
- util_format_get_blockwidth(state->format), 8);
- array_mode = tmp->array_mode[offset_level];
- tile_type = tmp->tile_type;
-
- if (texture->target == PIPE_TEXTURE_1D_ARRAY) {
- height = 1;
- depth = texture->array_size;
- } else if (texture->target == PIPE_TEXTURE_2D_ARRAY) {
- depth = texture->array_size;
- }
+ width = tmp->surface.level[offset_level].npix_x;
+ height = tmp->surface.level[offset_level].npix_y;
+ depth = tmp->surface.level[offset_level].npix_z;
+ pitch = tmp->surface.level[offset_level].nblk_x * util_format_get_blockwidth(state->format);
+ tile_type = tmp->tile_type;
+
+ if (texture->target == PIPE_TEXTURE_1D_ARRAY) {
+ height = 1;
+ depth = texture->array_size;
+ } else if (texture->target == PIPE_TEXTURE_2D_ARRAY) {
+ depth = texture->array_size;
+ }
+ switch (tmp->surface.level[offset_level].mode) {
+ case RADEON_SURF_MODE_LINEAR_ALIGNED:
+ array_mode = V_038000_ARRAY_LINEAR_ALIGNED;
+ break;
+ case RADEON_SURF_MODE_1D:
+ array_mode = V_038000_ARRAY_1D_TILED_THIN1;
+ break;
+ case RADEON_SURF_MODE_2D:
+ array_mode = V_038000_ARRAY_2D_TILED_THIN1;
+ break;
+ case RADEON_SURF_MODE_LINEAR:
+ default:
+ array_mode = V_038000_ARRAY_LINEAR_GENERAL;
+ break;
+ }
- rstate->bo[0] = &tmp->resource;
- rstate->bo[1] = &tmp->resource;
- rstate->bo_usage[0] = RADEON_USAGE_READ;
- rstate->bo_usage[1] = RADEON_USAGE_READ;
-
- rstate->val[0] = (S_038000_DIM(r600_tex_dim(texture->target)) |
- S_038000_TILE_MODE(array_mode) |
- S_038000_TILE_TYPE(tile_type) |
- S_038000_PITCH((pitch / 8) - 1) |
- S_038000_TEX_WIDTH(width - 1));
- rstate->val[1] = (S_038004_TEX_HEIGHT(height - 1) |
- S_038004_TEX_DEPTH(depth - 1) |
- S_038004_DATA_FORMAT(format));
- rstate->val[2] = tmp->offset[offset_level] >> 8;
- rstate->val[3] = tmp->offset[offset_level+1] >> 8;
- rstate->val[4] = (word4 |
- S_038010_SRF_MODE_ALL(V_038010_SRF_MODE_ZERO_CLAMP_MINUS_ONE) |
- S_038010_REQUEST_SIZE(1) |
- S_038010_ENDIAN_SWAP(endian) |
- S_038010_BASE_LEVEL(0));
- rstate->val[5] = (S_038014_LAST_LEVEL(last_level) |
- S_038014_BASE_ARRAY(state->u.tex.first_layer) |
- S_038014_LAST_ARRAY(state->u.tex.last_layer));
- rstate->val[6] = (S_038018_TYPE(V_038010_SQ_TEX_VTX_VALID_TEXTURE) |
- S_038018_MAX_ANISO(4 /* max 16 samples */));
+ view->tex_resource = &tmp->resource;
+ view->tex_resource_words[0] = (S_038000_DIM(r600_tex_dim(texture->target, texture->nr_samples)) |
+ S_038000_TILE_MODE(array_mode) |
+ S_038000_TILE_TYPE(tile_type) |
+ S_038000_PITCH((pitch / 8) - 1) |
+ S_038000_TEX_WIDTH(width - 1));
+ view->tex_resource_words[1] = (S_038004_TEX_HEIGHT(height - 1) |
+ S_038004_TEX_DEPTH(depth - 1) |
+ S_038004_DATA_FORMAT(format));
+ view->tex_resource_words[2] = tmp->surface.level[offset_level].offset >> 8;
+ if (offset_level >= tmp->surface.last_level) {
+ view->tex_resource_words[3] = tmp->surface.level[offset_level].offset >> 8;
} else {
- width = tmp->surface.level[offset_level].npix_x;
- height = tmp->surface.level[offset_level].npix_y;
- depth = tmp->surface.level[offset_level].npix_z;
- pitch = tmp->surface.level[offset_level].nblk_x * util_format_get_blockwidth(state->format);
- tile_type = tmp->tile_type;
-
- if (texture->target == PIPE_TEXTURE_1D_ARRAY) {
- height = 1;
- depth = texture->array_size;
- } else if (texture->target == PIPE_TEXTURE_2D_ARRAY) {
- depth = texture->array_size;
- }
- switch (tmp->surface.level[offset_level].mode) {
- case RADEON_SURF_MODE_LINEAR_ALIGNED:
- array_mode = V_038000_ARRAY_LINEAR_ALIGNED;
- break;
- case RADEON_SURF_MODE_1D:
- array_mode = V_038000_ARRAY_1D_TILED_THIN1;
- break;
- case RADEON_SURF_MODE_2D:
- array_mode = V_038000_ARRAY_2D_TILED_THIN1;
- break;
- case RADEON_SURF_MODE_LINEAR:
- default:
- array_mode = V_038000_ARRAY_LINEAR_GENERAL;
- break;
- }
-
- rstate->bo[0] = &tmp->resource;
- rstate->bo[1] = &tmp->resource;
- rstate->bo_usage[0] = RADEON_USAGE_READ;
- rstate->bo_usage[1] = RADEON_USAGE_READ;
-
- rstate->val[0] = (S_038000_DIM(r600_tex_dim(texture->target)) |
- S_038000_TILE_MODE(array_mode) |
- S_038000_TILE_TYPE(tile_type) |
- S_038000_PITCH((pitch / 8) - 1) |
- S_038000_TEX_WIDTH(width - 1));
- rstate->val[1] = (S_038004_TEX_HEIGHT(height - 1) |
- S_038004_TEX_DEPTH(depth - 1) |
- S_038004_DATA_FORMAT(format));
- rstate->val[2] = tmp->surface.level[offset_level].offset >> 8;
- if (offset_level >= tmp->surface.last_level) {
- rstate->val[3] = tmp->surface.level[offset_level].offset >> 8;
- } else {
- rstate->val[3] = tmp->surface.level[offset_level + 1].offset >> 8;
- }
- rstate->val[4] = (word4 |
- S_038010_SRF_MODE_ALL(V_038010_SRF_MODE_ZERO_CLAMP_MINUS_ONE) |
- S_038010_REQUEST_SIZE(1) |
- S_038010_ENDIAN_SWAP(endian) |
- S_038010_BASE_LEVEL(0));
- rstate->val[5] = (S_038014_LAST_LEVEL(last_level) |
- S_038014_BASE_ARRAY(state->u.tex.first_layer) |
- S_038014_LAST_ARRAY(state->u.tex.last_layer));
- rstate->val[6] = (S_038018_TYPE(V_038010_SQ_TEX_VTX_VALID_TEXTURE) |
- S_038018_MAX_ANISO(4 /* max 16 samples */));
+ view->tex_resource_words[3] = tmp->surface.level[offset_level + 1].offset >> 8;
+ }
+ view->tex_resource_words[4] = (word4 |
+ S_038010_SRF_MODE_ALL(V_038010_SRF_MODE_ZERO_CLAMP_MINUS_ONE) |
+ S_038010_REQUEST_SIZE(1) |
+ S_038010_ENDIAN_SWAP(endian) |
+ S_038010_BASE_LEVEL(0));
+ view->tex_resource_words[5] = (S_038014_BASE_ARRAY(state->u.tex.first_layer) |
+ S_038014_LAST_ARRAY(state->u.tex.last_layer));
+ if (texture->nr_samples > 1) {
+ /* LAST_LEVEL holds log2(nr_samples) for multisample textures */
+ view->tex_resource_words[5] |= S_038014_LAST_LEVEL(util_logbase2(texture->nr_samples));
+ } else {
+ view->tex_resource_words[5] |= S_038014_LAST_LEVEL(last_level);
}
+ view->tex_resource_words[6] = (S_038018_TYPE(V_038010_SQ_TEX_VTX_VALID_TEXTURE) |
+ S_038018_MAX_ANISO(4 /* max 16 samples */));
return &view->base;
}
-static void r600_set_sampler_views(struct r600_context *rctx,
- struct r600_textures_info *dst,
- unsigned count,
- struct pipe_sampler_view **views,
- void (*set_resource)(struct r600_context*, struct r600_pipe_resource_state*, unsigned))
-{
- struct r600_pipe_sampler_view **rviews = (struct r600_pipe_sampler_view **)views;
- unsigned i;
-
- if (count)
- r600_inval_texture_cache(rctx);
-
- for (i = 0; i < count; i++) {
- if (rviews[i]) {
- if (((struct r600_resource_texture *)rviews[i]->base.texture)->is_depth)
- rctx->have_depth_texture = true;
-
- /* Changing from array to non-arrays textures and vice versa requires updating TEX_ARRAY_OVERRIDE. */
- if ((rviews[i]->base.texture->target == PIPE_TEXTURE_1D_ARRAY ||
- rviews[i]->base.texture->target == PIPE_TEXTURE_2D_ARRAY) != dst->is_array_sampler[i])
- dst->samplers_dirty = true;
-
- set_resource(rctx, &rviews[i]->state, i + R600_MAX_CONST_BUFFERS);
- } else {
- set_resource(rctx, NULL, i + R600_MAX_CONST_BUFFERS);
- }
-
- pipe_sampler_view_reference(
- (struct pipe_sampler_view **)&dst->views[i],
- views[i]);
- }
-
- for (i = count; i < dst->n_views; i++) {
- if (dst->views[i]) {
- set_resource(rctx, NULL, i + R600_MAX_CONST_BUFFERS);
- pipe_sampler_view_reference((struct pipe_sampler_view **)&dst->views[i], NULL);
- }
- }
-
- dst->n_views = count;
-}
-
-static void r600_set_vs_sampler_views(struct pipe_context *ctx, unsigned count,
- struct pipe_sampler_view **views)
-{
- struct r600_context *rctx = (struct r600_context *)ctx;
- r600_set_sampler_views(rctx, &rctx->vs_samplers, count, views,
- r600_context_pipe_state_set_vs_resource);
-}
-
-static void r600_set_ps_sampler_views(struct pipe_context *ctx, unsigned count,
- struct pipe_sampler_view **views)
-{
- struct r600_context *rctx = (struct r600_context *)ctx;
- r600_set_sampler_views(rctx, &rctx->ps_samplers, count, views,
- r600_context_pipe_state_set_ps_resource);
-}
-
-static void r600_set_seamless_cubemap(struct r600_context *rctx, boolean enable)
-{
- struct r600_pipe_state *rstate = CALLOC_STRUCT(r600_pipe_state);
- if (rstate == NULL)
- return;
-
- rstate->id = R600_PIPE_STATE_SEAMLESS_CUBEMAP;
- r600_pipe_state_add_reg(rstate, R_009508_TA_CNTL_AUX,
- (enable ? 0 : S_009508_DISABLE_CUBE_WRAP(1)) |
- S_009508_DISABLE_CUBE_ANISO(1) |
- S_009508_SYNC_GRADIENT(1) |
- S_009508_SYNC_WALKER(1) |
- S_009508_SYNC_ALIGNER(1),
- NULL, 0);
-
- free(rctx->states[R600_PIPE_STATE_SEAMLESS_CUBEMAP]);
- rctx->states[R600_PIPE_STATE_SEAMLESS_CUBEMAP] = rstate;
- r600_context_pipe_state_set(rctx, rstate);
-}
-
-static void r600_bind_samplers(struct r600_context *rctx,
- struct r600_textures_info *dst,
- unsigned count, void **states)
-{
- memcpy(dst->samplers, states, sizeof(void*) * count);
- dst->n_samplers = count;
- dst->samplers_dirty = true;
-}
-
-static void r600_bind_vs_samplers(struct pipe_context *ctx, unsigned count, void **states)
-{
- struct r600_context *rctx = (struct r600_context *)ctx;
- r600_bind_samplers(rctx, &rctx->vs_samplers, count, states);
-}
-
-static void r600_bind_ps_samplers(struct pipe_context *ctx, unsigned count, void **states)
-{
- struct r600_context *rctx = (struct r600_context *)ctx;
- r600_bind_samplers(rctx, &rctx->ps_samplers, count, states);
-}
-
-static void r600_update_samplers(struct r600_context *rctx,
- struct r600_textures_info *tex,
- void (*set_sampler)(struct r600_context*, struct r600_pipe_state*, unsigned))
-{
- unsigned i;
-
- if (tex->samplers_dirty) {
- int seamless = -1;
- for (i = 0; i < tex->n_samplers; i++) {
- if (!tex->samplers[i])
- continue;
-
- /* TEX_ARRAY_OVERRIDE must be set for array textures to disable
- * filtering between layers.
- * Don't update TEX_ARRAY_OVERRIDE if we don't have the sampler view. */
- if (tex->views[i]) {
- if (tex->views[i]->base.texture->target == PIPE_TEXTURE_1D_ARRAY ||
- tex->views[i]->base.texture->target == PIPE_TEXTURE_2D_ARRAY) {
- tex->samplers[i]->rstate.regs[0].value |= S_03C000_TEX_ARRAY_OVERRIDE(1);
- tex->is_array_sampler[i] = true;
- } else {
- tex->samplers[i]->rstate.regs[0].value &= C_03C000_TEX_ARRAY_OVERRIDE;
- tex->is_array_sampler[i] = false;
- }
- }
-
- set_sampler(rctx, &tex->samplers[i]->rstate, i);
-
- if (tex->samplers[i])
- seamless = tex->samplers[i]->seamless_cube_map;
- }
-
- if (seamless != -1)
- r600_set_seamless_cubemap(rctx, seamless);
-
- tex->samplers_dirty = false;
- }
-}
-
-void r600_update_sampler_states(struct r600_context *rctx)
-{
- r600_update_samplers(rctx, &rctx->vs_samplers,
- r600_context_pipe_state_set_vs_sampler);
- r600_update_samplers(rctx, &rctx->ps_samplers,
- r600_context_pipe_state_set_ps_sampler);
-}
-
static void r600_set_clip_state(struct pipe_context *ctx,
const struct pipe_clip_state *state)
{
struct r600_context *rctx = (struct r600_context *)ctx;
struct r600_pipe_state *rstate = CALLOC_STRUCT(r600_pipe_state);
- struct pipe_resource * cbuf;
+ struct pipe_constant_buffer cb;
if (rstate == NULL)
return;
for (int i = 0; i < 6; i++) {
r600_pipe_state_add_reg(rstate,
R_028E20_PA_CL_UCP0_X + i * 16,
- fui(state->ucp[i][0]), NULL, 0);
+ fui(state->ucp[i][0]));
r600_pipe_state_add_reg(rstate,
R_028E24_PA_CL_UCP0_Y + i * 16,
- fui(state->ucp[i][1]) , NULL, 0);
+ fui(state->ucp[i][1]) );
r600_pipe_state_add_reg(rstate,
R_028E28_PA_CL_UCP0_Z + i * 16,
- fui(state->ucp[i][2]), NULL, 0);
+ fui(state->ucp[i][2]));
r600_pipe_state_add_reg(rstate,
R_028E2C_PA_CL_UCP0_W + i * 16,
- fui(state->ucp[i][3]), NULL, 0);
+ fui(state->ucp[i][3]));
}
free(rctx->states[R600_PIPE_STATE_CLIP]);
rctx->states[R600_PIPE_STATE_CLIP] = rstate;
r600_context_pipe_state_set(rctx, rstate);
- cbuf = pipe_user_buffer_create(ctx->screen,
- state->ucp,
- 4*4*8, /* 8*4 floats */
- PIPE_BIND_CONSTANT_BUFFER);
- r600_set_constant_buffer(ctx, PIPE_SHADER_VERTEX, 1, cbuf);
- pipe_resource_reference(&cbuf, NULL);
+ cb.buffer = NULL;
+ cb.user_buffer = state->ucp;
+ cb.buffer_offset = 0;
+ cb.buffer_size = 4*4*8;
+ ctx->set_constant_buffer(ctx, PIPE_SHADER_VERTEX, 1, &cb);
+ pipe_resource_reference(&cb.buffer, NULL);
}
static void r600_set_polygon_stipple(struct pipe_context *ctx,
{
}
-static void r600_set_sample_mask(struct pipe_context *pipe, unsigned sample_mask)
-{
-}
-
void r600_set_scissor_state(struct r600_context *rctx,
const struct pipe_scissor_state *state)
{
tl = S_028240_TL_X(state->minx) | S_028240_TL_Y(state->miny) | S_028240_WINDOW_OFFSET_DISABLE(1);
br = S_028244_BR_X(state->maxx) | S_028244_BR_Y(state->maxy);
r600_pipe_state_add_reg(rstate,
- R_028250_PA_SC_VPORT_SCISSOR_0_TL, tl,
- NULL, 0);
+ R_028250_PA_SC_VPORT_SCISSOR_0_TL, tl);
r600_pipe_state_add_reg(rstate,
- R_028254_PA_SC_VPORT_SCISSOR_0_BR, br,
- NULL, 0);
+ R_028254_PA_SC_VPORT_SCISSOR_0_BR, br);
free(rctx->states[R600_PIPE_STATE_SCISSOR]);
rctx->states[R600_PIPE_STATE_SCISSOR] = rstate;
rctx->viewport = *state;
rstate->id = R600_PIPE_STATE_VIEWPORT;
- r600_pipe_state_add_reg(rstate, R_02843C_PA_CL_VPORT_XSCALE_0, fui(state->scale[0]), NULL, 0);
- r600_pipe_state_add_reg(rstate, R_028444_PA_CL_VPORT_YSCALE_0, fui(state->scale[1]), NULL, 0);
- r600_pipe_state_add_reg(rstate, R_02844C_PA_CL_VPORT_ZSCALE_0, fui(state->scale[2]), NULL, 0);
- r600_pipe_state_add_reg(rstate, R_028440_PA_CL_VPORT_XOFFSET_0, fui(state->translate[0]), NULL, 0);
- r600_pipe_state_add_reg(rstate, R_028448_PA_CL_VPORT_YOFFSET_0, fui(state->translate[1]), NULL, 0);
- r600_pipe_state_add_reg(rstate, R_028450_PA_CL_VPORT_ZOFFSET_0, fui(state->translate[2]), NULL, 0);
+ r600_pipe_state_add_reg(rstate, R_02843C_PA_CL_VPORT_XSCALE_0, fui(state->scale[0]));
+ r600_pipe_state_add_reg(rstate, R_028444_PA_CL_VPORT_YSCALE_0, fui(state->scale[1]));
+ r600_pipe_state_add_reg(rstate, R_02844C_PA_CL_VPORT_ZSCALE_0, fui(state->scale[2]));
+ r600_pipe_state_add_reg(rstate, R_028440_PA_CL_VPORT_XOFFSET_0, fui(state->translate[0]));
+ r600_pipe_state_add_reg(rstate, R_028448_PA_CL_VPORT_YOFFSET_0, fui(state->translate[1]));
+ r600_pipe_state_add_reg(rstate, R_028450_PA_CL_VPORT_ZOFFSET_0, fui(state->translate[2]));
free(rctx->states[R600_PIPE_STATE_VIEWPORT]);
rctx->states[R600_PIPE_STATE_VIEWPORT] = rstate;
r600_context_pipe_state_set(rctx, rstate);
}
-static void r600_cb(struct r600_context *rctx, struct r600_pipe_state *rstate,
- const struct pipe_framebuffer_state *state, int cb)
+static struct r600_resource *r600_buffer_create_helper(struct r600_screen *rscreen,
+ unsigned size, unsigned alignment)
+{
+ struct pipe_resource buffer;
+
+ memset(&buffer, 0, sizeof buffer);
+ buffer.target = PIPE_BUFFER;
+ buffer.format = PIPE_FORMAT_R8_UNORM;
+ buffer.bind = PIPE_BIND_CUSTOM;
+ buffer.usage = PIPE_USAGE_STATIC;
+ buffer.flags = 0;
+ buffer.width0 = size;
+ buffer.height0 = 1;
+ buffer.depth0 = 1;
+ buffer.array_size = 1;
+
+ return (struct r600_resource*)
+ r600_buffer_create(&rscreen->screen, &buffer, alignment);
+}
+
+static void r600_init_color_surface(struct r600_context *rctx,
+ struct r600_surface *surf,
+ bool force_cmask_fmask)
{
struct r600_screen *rscreen = rctx->screen;
- struct r600_resource_texture *rtex;
- struct r600_surface *surf;
- unsigned level = state->cbufs[cb]->u.tex.level;
+ struct r600_texture *rtex = (struct r600_texture*)surf->base.texture;
+ unsigned level = surf->base.u.tex.level;
unsigned pitch, slice;
unsigned color_info;
unsigned format, swap, ntype, endian;
unsigned offset;
const struct util_format_description *desc;
int i;
- unsigned blend_bypass = 0, blend_clamp = 1;
-
- surf = (struct r600_surface *)state->cbufs[cb];
- rtex = (struct r600_resource_texture*)state->cbufs[cb]->texture;
-
- if (rtex->is_depth)
- rctx->have_depth_fb = TRUE;
+ bool blend_bypass = 0, blend_clamp = 1;
if (rtex->is_depth && !rtex->is_flushing_texture) {
+ r600_init_flushed_depth_texture(&rctx->context, surf->base.texture, NULL);
rtex = rtex->flushed_depth_texture;
+ assert(rtex);
}
- /* XXX quite sure for dx10+ hw don't need any offset hacks */
- if (!rscreen->use_surface_alloc) {
- offset = r600_texture_get_offset(rtex,
- level, state->cbufs[cb]->u.tex.first_layer);
- pitch = rtex->pitch_in_blocks[level] / 8 - 1;
- slice = rtex->pitch_in_blocks[level] * surf->aligned_height / 64;
- if (slice) {
- slice = slice - 1;
- }
- color_info = S_0280A0_ARRAY_MODE(rtex->array_mode[level]);
- } else {
- offset = rtex->surface.level[level].offset;
- if (rtex->surface.level[level].mode < RADEON_SURF_MODE_1D) {
- offset += rtex->surface.level[level].slice_size *
- state->cbufs[cb]->u.tex.first_layer;
- }
- pitch = rtex->surface.level[level].nblk_x / 8 - 1;
- slice = (rtex->surface.level[level].nblk_x * rtex->surface.level[level].nblk_y) / 64;
- if (slice) {
- slice = slice - 1;
- }
- color_info = 0;
- switch (rtex->surface.level[level].mode) {
- case RADEON_SURF_MODE_LINEAR_ALIGNED:
- color_info = S_0280A0_ARRAY_MODE(V_038000_ARRAY_LINEAR_ALIGNED);
- break;
- case RADEON_SURF_MODE_1D:
- color_info = S_0280A0_ARRAY_MODE(V_038000_ARRAY_1D_TILED_THIN1);
- break;
- case RADEON_SURF_MODE_2D:
- color_info = S_0280A0_ARRAY_MODE(V_038000_ARRAY_2D_TILED_THIN1);
- break;
- case RADEON_SURF_MODE_LINEAR:
- default:
- color_info = S_0280A0_ARRAY_MODE(V_038000_ARRAY_LINEAR_GENERAL);
- break;
- }
+ offset = rtex->surface.level[level].offset;
+ if (rtex->surface.level[level].mode < RADEON_SURF_MODE_1D) {
+ offset += rtex->surface.level[level].slice_size *
+ surf->base.u.tex.first_layer;
+ }
+ pitch = rtex->surface.level[level].nblk_x / 8 - 1;
+ slice = (rtex->surface.level[level].nblk_x * rtex->surface.level[level].nblk_y) / 64;
+ if (slice) {
+ slice = slice - 1;
}
+ color_info = 0;
+ switch (rtex->surface.level[level].mode) {
+ case RADEON_SURF_MODE_LINEAR_ALIGNED:
+ color_info = S_0280A0_ARRAY_MODE(V_038000_ARRAY_LINEAR_ALIGNED);
+ break;
+ case RADEON_SURF_MODE_1D:
+ color_info = S_0280A0_ARRAY_MODE(V_038000_ARRAY_1D_TILED_THIN1);
+ break;
+ case RADEON_SURF_MODE_2D:
+ color_info = S_0280A0_ARRAY_MODE(V_038000_ARRAY_2D_TILED_THIN1);
+ break;
+ case RADEON_SURF_MODE_LINEAR:
+ default:
+ color_info = S_0280A0_ARRAY_MODE(V_038000_ARRAY_LINEAR_GENERAL);
+ break;
+ }
+
desc = util_format_description(surf->base.format);
for (i = 0; i < 4; i++) {
}
format = r600_translate_colorformat(surf->base.format);
+ assert(format != ~0);
+
swap = r600_translate_colorswap(surf->base.format);
- if(rtex->resource.b.b.b.usage == PIPE_USAGE_STAGING) {
+ assert(swap != ~0);
+
+ if (rtex->resource.b.b.usage == PIPE_USAGE_STAGING) {
endian = ENDIAN_NONE;
} else {
endian = r600_colorformat_endian_swap(format);
blend_bypass = 1;
}
+ surf->alphatest_bypass = ntype == V_0280A0_NUMBER_UINT || ntype == V_0280A0_NUMBER_SINT;
+
color_info |= S_0280A0_FORMAT(format) |
S_0280A0_COMP_SWAP(swap) |
S_0280A0_BLEND_BYPASS(blend_bypass) |
ntype != V_0280A0_NUMBER_UINT &&
ntype != V_0280A0_NUMBER_SINT) &&
G_0280A0_BLEND_CLAMP(color_info) &&
- !G_0280A0_BLEND_FLOAT32(color_info))
+ !G_0280A0_BLEND_FLOAT32(color_info)) {
color_info |= S_0280A0_SOURCE_FORMAT(V_0280A0_EXPORT_NORM);
+ surf->export_16bpc = true;
+ }
} else {
/* EXPORT_NORM can be enabled if:
* - 11-bit or smaller UNORM/SNORM/SRGB
desc->channel[i].type != UTIL_FORMAT_TYPE_FLOAT &&
ntype != V_0280A0_NUMBER_UINT && ntype != V_0280A0_NUMBER_SINT) ||
(desc->channel[i].size < 17 &&
- desc->channel[i].type == UTIL_FORMAT_TYPE_FLOAT)))
+ desc->channel[i].type == UTIL_FORMAT_TYPE_FLOAT))) {
color_info |= S_0280A0_SOURCE_FORMAT(V_0280A0_EXPORT_NORM);
+ surf->export_16bpc = true;
+ }
}
- r600_pipe_state_add_reg(rstate,
- R_028040_CB_COLOR0_BASE + cb * 4,
- offset >> 8, &rtex->resource, RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg(rstate,
- R_0280A0_CB_COLOR0_INFO + cb * 4,
- color_info, &rtex->resource, RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg(rstate,
- R_028060_CB_COLOR0_SIZE + cb * 4,
- S_028060_PITCH_TILE_MAX(pitch) |
- S_028060_SLICE_TILE_MAX(slice),
- NULL, 0);
- if (!rscreen->use_surface_alloc) {
- r600_pipe_state_add_reg(rstate,
- R_028080_CB_COLOR0_VIEW + cb * 4,
- 0x00000000, NULL, 0);
- } else {
- if (rtex->surface.level[level].mode < RADEON_SURF_MODE_1D) {
- r600_pipe_state_add_reg(rstate,
- R_028080_CB_COLOR0_VIEW + cb * 4,
- 0x00000000, NULL, 0);
- } else {
- r600_pipe_state_add_reg(rstate,
- R_028080_CB_COLOR0_VIEW + cb * 4,
- S_028080_SLICE_START(state->cbufs[cb]->u.tex.first_layer) |
- S_028080_SLICE_MAX(state->cbufs[cb]->u.tex.last_layer),
- NULL, 0);
+ /* These might not always be initialized to zero. */
+ surf->cb_color_base = offset >> 8;
+ surf->cb_color_size = S_028060_PITCH_TILE_MAX(pitch) |
+ S_028060_SLICE_TILE_MAX(slice);
+ surf->cb_color_fmask = surf->cb_color_base;
+ surf->cb_color_cmask = surf->cb_color_base;
+ surf->cb_color_mask = 0;
+
+ pipe_resource_reference((struct pipe_resource**)&surf->cb_buffer_cmask,
+ &rtex->resource.b.b);
+ pipe_resource_reference((struct pipe_resource**)&surf->cb_buffer_fmask,
+ &rtex->resource.b.b);
+
+ if (rtex->cmask_size) {
+ surf->cb_color_cmask = rtex->cmask_offset >> 8;
+ surf->cb_color_mask |= S_028100_CMASK_BLOCK_MAX(rtex->cmask_slice_tile_max);
+
+ if (rtex->fmask_size) {
+ color_info |= S_0280A0_TILE_MODE(V_0280A0_FRAG_ENABLE);
+ surf->cb_color_fmask = rtex->fmask_offset >> 8;
+ surf->cb_color_mask |= S_028100_FMASK_TILE_MAX(slice);
+ } else { /* cmask only */
+ color_info |= S_0280A0_TILE_MODE(V_0280A0_CLEAR_ENABLE);
+ }
+ } else if (force_cmask_fmask) {
+ /* Allocate dummy FMASK and CMASK if they aren't allocated already.
+ *
+ * R6xx needs FMASK and CMASK for the destination buffer of color resolve,
+ * otherwise it hangs. We don't have FMASK and CMASK pre-allocated,
+ * because it's not an MSAA buffer.
+ */
+ struct r600_cmask_info cmask;
+ struct r600_fmask_info fmask;
+
+ r600_texture_get_cmask_info(rscreen, rtex, &cmask);
+ r600_texture_get_fmask_info(rscreen, rtex, 8, &fmask);
+
+ /* CMASK. */
+ if (!rctx->dummy_cmask ||
+ rctx->dummy_cmask->buf->size < cmask.size ||
+ rctx->dummy_cmask->buf->alignment % cmask.alignment != 0) {
+ struct pipe_transfer *transfer;
+ void *ptr;
+
+ pipe_resource_reference((struct pipe_resource**)&rctx->dummy_cmask, NULL);
+ rctx->dummy_cmask = r600_buffer_create_helper(rscreen, cmask.size, cmask.alignment);
+
+ /* Set the contents to 0xCC. */
+ ptr = pipe_buffer_map(&rctx->context, &rctx->dummy_cmask->b.b, PIPE_TRANSFER_WRITE, &transfer);
+ memset(ptr, 0xCC, cmask.size);
+ pipe_buffer_unmap(&rctx->context, transfer);
+ }
+ pipe_resource_reference((struct pipe_resource**)&surf->cb_buffer_cmask,
+ &rctx->dummy_cmask->b.b);
+
+ /* FMASK. */
+ if (!rctx->dummy_fmask ||
+ rctx->dummy_fmask->buf->size < fmask.size ||
+ rctx->dummy_fmask->buf->alignment % fmask.alignment != 0) {
+ pipe_resource_reference((struct pipe_resource**)&rctx->dummy_fmask, NULL);
+ rctx->dummy_fmask = r600_buffer_create_helper(rscreen, fmask.size, fmask.alignment);
+
}
+ pipe_resource_reference((struct pipe_resource**)&surf->cb_buffer_fmask,
+ &rctx->dummy_fmask->b.b);
+
+ /* Init the registers. */
+ color_info |= S_0280A0_TILE_MODE(V_0280A0_FRAG_ENABLE);
+ surf->cb_color_cmask = 0;
+ surf->cb_color_fmask = 0;
+ surf->cb_color_mask = S_028100_CMASK_BLOCK_MAX(cmask.slice_tile_max) |
+ S_028100_FMASK_TILE_MAX(slice);
}
- r600_pipe_state_add_reg(rstate,
- R_0280E0_CB_COLOR0_FRAG + cb * 4,
- 0, &rtex->resource, RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg(rstate,
- R_0280C0_CB_COLOR0_TILE + cb * 4,
- 0, &rtex->resource, RADEON_USAGE_READWRITE);
+
+ surf->cb_color_info = color_info;
+
+ if (rtex->surface.level[level].mode < RADEON_SURF_MODE_1D) {
+ surf->cb_color_view = 0;
+ } else {
+ surf->cb_color_view = S_028080_SLICE_START(surf->base.u.tex.first_layer) |
+ S_028080_SLICE_MAX(surf->base.u.tex.last_layer);
+ }
+
+ surf->color_initialized = true;
}
-static void r600_db(struct r600_context *rctx, struct r600_pipe_state *rstate,
- const struct pipe_framebuffer_state *state)
+static void r600_init_depth_surface(struct r600_context *rctx,
+ struct r600_surface *surf)
{
- struct r600_screen *rscreen = rctx->screen;
- struct r600_resource_texture *rtex;
- struct r600_surface *surf;
+ struct r600_texture *rtex = (struct r600_texture*)surf->base.texture;
unsigned level, pitch, slice, format, offset, array_mode;
- if (state->zsbuf == NULL)
- return;
+ level = surf->base.u.tex.level;
+ offset = rtex->surface.level[level].offset;
+ pitch = rtex->surface.level[level].nblk_x / 8 - 1;
+ slice = (rtex->surface.level[level].nblk_x * rtex->surface.level[level].nblk_y) / 64;
+ if (slice) {
+ slice = slice - 1;
+ }
+ switch (rtex->surface.level[level].mode) {
+ case RADEON_SURF_MODE_2D:
+ array_mode = V_0280A0_ARRAY_2D_TILED_THIN1;
+ break;
+ case RADEON_SURF_MODE_1D:
+ case RADEON_SURF_MODE_LINEAR_ALIGNED:
+ case RADEON_SURF_MODE_LINEAR:
+ default:
+ array_mode = V_0280A0_ARRAY_1D_TILED_THIN1;
+ break;
+ }
+
+ format = r600_translate_dbformat(surf->base.format);
+ assert(format != ~0);
+
+ surf->db_depth_info = S_028010_ARRAY_MODE(array_mode) | S_028010_FORMAT(format);
+ surf->db_depth_base = offset >> 8;
+ surf->db_depth_view = S_028004_SLICE_START(surf->base.u.tex.first_layer) |
+ S_028004_SLICE_MAX(surf->base.u.tex.last_layer);
+ surf->db_depth_size = S_028000_PITCH_TILE_MAX(pitch) | S_028000_SLICE_TILE_MAX(slice);
+ surf->db_prefetch_limit = (rtex->surface.level[level].nblk_y / 8) - 1;
- level = state->zsbuf->u.tex.level;
+ surf->depth_initialized = true;
+}
- surf = (struct r600_surface *)state->zsbuf;
- rtex = (struct r600_resource_texture*)state->zsbuf->texture;
+#define FILL_SREG(s0x, s0y, s1x, s1y, s2x, s2y, s3x, s3y) \
+ (((s0x) & 0xf) | (((s0y) & 0xf) << 4) | \
+ (((s1x) & 0xf) << 8) | (((s1y) & 0xf) << 12) | \
+ (((s2x) & 0xf) << 16) | (((s2y) & 0xf) << 20) | \
+ (((s3x) & 0xf) << 24) | (((s3y) & 0xf) << 28))
- if (!rscreen->use_surface_alloc) {
- /* XXX remove this once tiling is properly supported */
- array_mode = rtex->array_mode[level] ? rtex->array_mode[level] :
- V_0280A0_ARRAY_1D_TILED_THIN1;
+static uint32_t r600_set_ms_pos(struct pipe_context *ctx, struct r600_pipe_state *rstate, int nsample)
+{
+ static uint32_t sample_locs_2x[] = {
+ FILL_SREG(-4, 4, 4, -4, -4, 4, 4, -4),
+ FILL_SREG(-4, 4, 4, -4, -4, 4, 4, -4),
+ };
+ static unsigned max_dist_2x = 4;
+ static uint32_t sample_locs_4x[] = {
+ FILL_SREG(-2, -2, 2, 2, -6, 6, 6, -6),
+ FILL_SREG(-2, -2, 2, 2, -6, 6, 6, -6),
+ };
+ static unsigned max_dist_4x = 6;
+ static uint32_t sample_locs_8x[] = {
+ FILL_SREG(-2, -5, 3, -4, -1, 5, -6, -2),
+ FILL_SREG( 6, 0, 0, 0, -5, 3, 4, 4),
+ };
+ static unsigned max_dist_8x = 8;
+ struct r600_context *rctx = (struct r600_context *)ctx;
- /* XXX quite sure for dx10+ hw don't need any offset hacks */
- offset = r600_texture_get_offset((struct r600_resource_texture *)state->zsbuf->texture,
- level, state->zsbuf->u.tex.first_layer);
- pitch = rtex->pitch_in_blocks[level] / 8 - 1;
- slice = rtex->pitch_in_blocks[level] * surf->aligned_height / 64;
- if (slice) {
- slice = slice - 1;
+ if (rctx->family == CHIP_R600) {
+ switch (nsample) {
+ case 0:
+ case 1:
+ return 0;
+ case 2:
+ r600_pipe_state_add_reg(rstate, R_008B40_PA_SC_AA_SAMPLE_LOCS_2S, sample_locs_2x[0]);
+ return max_dist_2x;
+ case 4:
+ r600_pipe_state_add_reg(rstate, R_008B44_PA_SC_AA_SAMPLE_LOCS_4S, sample_locs_4x[0]);
+ return max_dist_4x;
+ case 8:
+ r600_pipe_state_add_reg(rstate, R_008B48_PA_SC_AA_SAMPLE_LOCS_8S_WD0, sample_locs_8x[0]);
+ r600_pipe_state_add_reg(rstate, R_008B4C_PA_SC_AA_SAMPLE_LOCS_8S_WD1, sample_locs_8x[1]);
+ return max_dist_8x;
}
} else {
- offset = rtex->surface.level[level].offset;
- pitch = rtex->surface.level[level].nblk_x / 8 - 1;
- slice = (rtex->surface.level[level].nblk_x * rtex->surface.level[level].nblk_y) / 64;
- if (slice) {
- slice = slice - 1;
- }
- switch (rtex->surface.level[level].mode) {
- case RADEON_SURF_MODE_2D:
- array_mode = V_0280A0_ARRAY_2D_TILED_THIN1;
- break;
- case RADEON_SURF_MODE_1D:
- case RADEON_SURF_MODE_LINEAR_ALIGNED:
- case RADEON_SURF_MODE_LINEAR:
- default:
- array_mode = V_0280A0_ARRAY_1D_TILED_THIN1;
- break;
+ switch (nsample) {
+ case 0:
+ case 1:
+ r600_pipe_state_add_reg(rstate, R_028C1C_PA_SC_AA_SAMPLE_LOCS_MCTX, 0);
+ r600_pipe_state_add_reg(rstate, R_028C20_PA_SC_AA_SAMPLE_LOCS_8D_WD1_MCTX, 0);
+ return 0;
+ case 2:
+ r600_pipe_state_add_reg(rstate, R_028C1C_PA_SC_AA_SAMPLE_LOCS_MCTX, sample_locs_2x[0]);
+ r600_pipe_state_add_reg(rstate, R_028C20_PA_SC_AA_SAMPLE_LOCS_8D_WD1_MCTX, sample_locs_2x[1]);
+ return max_dist_2x;
+ case 4:
+ r600_pipe_state_add_reg(rstate, R_028C1C_PA_SC_AA_SAMPLE_LOCS_MCTX, sample_locs_4x[0]);
+ r600_pipe_state_add_reg(rstate, R_028C20_PA_SC_AA_SAMPLE_LOCS_8D_WD1_MCTX, sample_locs_4x[1]);
+ return max_dist_4x;
+ case 8:
+ r600_pipe_state_add_reg(rstate, R_028C1C_PA_SC_AA_SAMPLE_LOCS_MCTX, sample_locs_8x[0]);
+ r600_pipe_state_add_reg(rstate, R_028C20_PA_SC_AA_SAMPLE_LOCS_8D_WD1_MCTX, sample_locs_8x[1]);
+ return max_dist_8x;
}
}
-
- format = r600_translate_dbformat(state->zsbuf->texture->format);
-
- r600_pipe_state_add_reg(rstate, R_02800C_DB_DEPTH_BASE,
- offset >> 8, &rtex->resource, RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg(rstate, R_028000_DB_DEPTH_SIZE,
- S_028000_PITCH_TILE_MAX(pitch) | S_028000_SLICE_TILE_MAX(slice),
- NULL, 0);
- if (!rscreen->use_surface_alloc) {
- r600_pipe_state_add_reg(rstate, R_028004_DB_DEPTH_VIEW, 0x00000000, NULL, 0);
- } else {
- r600_pipe_state_add_reg(rstate, R_028004_DB_DEPTH_VIEW,
- S_028004_SLICE_START(state->zsbuf->u.tex.first_layer) |
- S_028004_SLICE_MAX(state->zsbuf->u.tex.last_layer),
- NULL, 0);
- }
- r600_pipe_state_add_reg(rstate, R_028010_DB_DEPTH_INFO,
- S_028010_ARRAY_MODE(array_mode) | S_028010_FORMAT(format),
- &rtex->resource, RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg(rstate, R_028D34_DB_PREFETCH_LIMIT,
- (surf->aligned_height / 8) - 1, NULL, 0);
+ R600_ERR("Invalid nr_samples %i\n", nsample);
+ return 0;
}
static void r600_set_framebuffer_state(struct pipe_context *ctx,
{
struct r600_context *rctx = (struct r600_context *)ctx;
struct r600_pipe_state *rstate = CALLOC_STRUCT(r600_pipe_state);
- uint32_t shader_mask, tl, br, shader_control;
+ struct r600_surface *surf;
+ struct r600_resource *res;
+ struct r600_texture *rtex;
+ uint32_t tl, br, i, nr_samples, max_dist;
+ bool is_resolve = state->nr_cbufs == 2 &&
+ state->cbufs[0]->texture->nr_samples > 1 &&
+ state->cbufs[1]->texture->nr_samples <= 1;
+ /* The resolve buffer must have CMASK and FMASK to prevent hardlocks on R6xx. */
+ bool cb1_force_cmask_fmask = rctx->chip_class == R600 && is_resolve;
if (rstate == NULL)
return;
- r600_flush_framebuffer(rctx, false);
+ if (rctx->framebuffer.nr_cbufs) {
+ rctx->flags |= R600_CONTEXT_CB_FLUSH;
+ }
+ if (rctx->framebuffer.zsbuf) {
+ rctx->flags |= R600_CONTEXT_DB_FLUSH;
+ }
+ /* R6xx errata */
+ if (rctx->chip_class == R600) {
+ rctx->flags |= R600_CONTEXT_FLUSH_AND_INV;
+ }
/* unreference old buffer and reference new one */
rstate->id = R600_PIPE_STATE_FRAMEBUFFER;
util_copy_framebuffer_state(&rctx->framebuffer, state);
- /* build states */
- rctx->have_depth_fb = 0;
- for (int i = 0; i < state->nr_cbufs; i++) {
- r600_cb(rctx, rstate, state, i);
+
+ /* Colorbuffers. */
+ rctx->export_16bpc = true;
+ rctx->nr_cbufs = state->nr_cbufs;
+ rctx->cb0_is_integer = state->nr_cbufs &&
+ util_format_is_pure_integer(state->cbufs[0]->format);
+ rctx->compressed_cb_mask = 0;
+
+ for (i = 0; i < state->nr_cbufs; i++) {
+ bool force_cmask_fmask = cb1_force_cmask_fmask && i == 1;
+ surf = (struct r600_surface*)state->cbufs[i];
+ res = (struct r600_resource*)surf->base.texture;
+ rtex = (struct r600_texture*)res;
+
+ if (!surf->color_initialized || force_cmask_fmask) {
+ r600_init_color_surface(rctx, surf, force_cmask_fmask);
+ if (force_cmask_fmask) {
+ /* re-initialize later without compression */
+ surf->color_initialized = false;
+ }
+ }
+
+ if (!surf->export_16bpc) {
+ rctx->export_16bpc = false;
+ }
+
+ r600_pipe_state_add_reg_bo(rstate, R_028040_CB_COLOR0_BASE + i * 4,
+ surf->cb_color_base, res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg_bo(rstate, R_0280A0_CB_COLOR0_INFO + i * 4,
+ surf->cb_color_info, res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg(rstate, R_028060_CB_COLOR0_SIZE + i * 4,
+ surf->cb_color_size);
+ r600_pipe_state_add_reg(rstate, R_028080_CB_COLOR0_VIEW + i * 4,
+ surf->cb_color_view);
+ r600_pipe_state_add_reg_bo(rstate, R_0280E0_CB_COLOR0_FRAG + i * 4,
+ surf->cb_color_fmask, surf->cb_buffer_fmask,
+ RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg_bo(rstate, R_0280C0_CB_COLOR0_TILE + i * 4,
+ surf->cb_color_cmask, surf->cb_buffer_cmask,
+ RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg(rstate, R_028100_CB_COLOR0_MASK + i * 4,
+ surf->cb_color_mask);
+
+ if (rtex->fmask_size && rtex->cmask_size) {
+ rctx->compressed_cb_mask |= 1 << i;
+ }
}
- if (state->zsbuf) {
- r600_db(rctx, rstate, state);
+ /* set CB_COLOR1_INFO for possible dual-src blending */
+ if (i == 1) {
+ r600_pipe_state_add_reg_bo(rstate, R_0280A0_CB_COLOR0_INFO + 1 * 4,
+ surf->cb_color_info, res, RADEON_USAGE_READWRITE);
+ i++;
+ }
+ for (; i < 8 ; i++) {
+ r600_pipe_state_add_reg(rstate, R_0280A0_CB_COLOR0_INFO + i * 4, 0);
}
- shader_mask = 0;
- shader_control = 0;
- for (int i = 0; i < state->nr_cbufs; i++) {
- shader_mask |= 0xf << (i * 4);
- shader_control |= 1 << i;
+ /* Update alpha-test state dependencies.
+ * Alpha-test is done on the first colorbuffer only. */
+ if (state->nr_cbufs) {
+ surf = (struct r600_surface*)state->cbufs[0];
+ if (rctx->alphatest_state.bypass != surf->alphatest_bypass) {
+ rctx->alphatest_state.bypass = surf->alphatest_bypass;
+ r600_atom_dirty(rctx, &rctx->alphatest_state.atom);
+ }
}
+
+ /* ZS buffer. */
+ if (state->zsbuf) {
+ surf = (struct r600_surface*)state->zsbuf;
+ res = (struct r600_resource*)surf->base.texture;
+
+ if (!surf->depth_initialized) {
+ r600_init_depth_surface(rctx, surf);
+ }
+
+ r600_pipe_state_add_reg_bo(rstate, R_02800C_DB_DEPTH_BASE, surf->db_depth_base,
+ res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg(rstate, R_028000_DB_DEPTH_SIZE, surf->db_depth_size);
+ r600_pipe_state_add_reg(rstate, R_028004_DB_DEPTH_VIEW, surf->db_depth_view);
+ r600_pipe_state_add_reg_bo(rstate, R_028010_DB_DEPTH_INFO, surf->db_depth_info,
+ res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg(rstate, R_028D34_DB_PREFETCH_LIMIT, surf->db_prefetch_limit);
+ }
+
+ /* Framebuffer dimensions. */
tl = S_028240_TL_X(0) | S_028240_TL_Y(0) | S_028240_WINDOW_OFFSET_DISABLE(1);
br = S_028244_BR_X(state->width) | S_028244_BR_Y(state->height);
r600_pipe_state_add_reg(rstate,
- R_028204_PA_SC_WINDOW_SCISSOR_TL, tl,
- NULL, 0);
+ R_028204_PA_SC_WINDOW_SCISSOR_TL, tl);
r600_pipe_state_add_reg(rstate,
- R_028208_PA_SC_WINDOW_SCISSOR_BR, br,
- NULL, 0);
+ R_028208_PA_SC_WINDOW_SCISSOR_BR, br);
- r600_pipe_state_add_reg(rstate, R_0287A0_CB_SHADER_CONTROL,
- shader_control, NULL, 0);
- r600_pipe_state_add_reg(rstate, R_02823C_CB_SHADER_MASK,
- shader_mask, NULL, 0);
+ /* If we're doing MSAA resolve... */
+ if (is_resolve) {
+ r600_pipe_state_add_reg(rstate, R_0287A0_CB_SHADER_CONTROL, 1);
+ } else {
+ /* Always enable the first colorbuffer in CB_SHADER_CONTROL. This
+ * will assure that the alpha-test will work even if there is
+ * no colorbuffer bound. */
+ r600_pipe_state_add_reg(rstate, R_0287A0_CB_SHADER_CONTROL,
+ (1ull << MAX2(state->nr_cbufs, 1)) - 1);
+ }
+
+ /* Multisampling */
+ if (state->nr_cbufs)
+ nr_samples = state->cbufs[0]->texture->nr_samples;
+ else if (state->zsbuf)
+ nr_samples = state->zsbuf->texture->nr_samples;
+ else
+ nr_samples = 0;
+
+ max_dist = r600_set_ms_pos(ctx, rstate, nr_samples);
+
+ if (nr_samples > 1) {
+ unsigned log_samples = util_logbase2(nr_samples);
+
+ r600_pipe_state_add_reg(rstate, R_028C00_PA_SC_LINE_CNTL,
+ S_028C00_LAST_PIXEL(1) |
+ S_028C00_EXPAND_LINE_WIDTH(1));
+ r600_pipe_state_add_reg(rstate, R_028C04_PA_SC_AA_CONFIG,
+ S_028C04_MSAA_NUM_SAMPLES(log_samples) |
+ S_028C04_MAX_SAMPLE_DIST(max_dist));
+ } else {
+ r600_pipe_state_add_reg(rstate, R_028C00_PA_SC_LINE_CNTL, S_028C00_LAST_PIXEL(1));
+ r600_pipe_state_add_reg(rstate, R_028C04_PA_SC_AA_CONFIG, 0);
+ }
free(rctx->states[R600_PIPE_STATE_FRAMEBUFFER]);
rctx->states[R600_PIPE_STATE_FRAMEBUFFER] = rstate;
if (state->zsbuf) {
r600_polygon_offset_update(rctx);
}
+
+ if (rctx->cb_misc_state.nr_cbufs != state->nr_cbufs) {
+ rctx->cb_misc_state.nr_cbufs = state->nr_cbufs;
+ r600_atom_dirty(rctx, &rctx->cb_misc_state.atom);
+ }
+
+ if (state->nr_cbufs == 0 && rctx->alphatest_state.bypass) {
+ rctx->alphatest_state.bypass = false;
+ r600_atom_dirty(rctx, &rctx->alphatest_state.atom);
+ }
+}
+
+static void r600_emit_cb_misc_state(struct r600_context *rctx, struct r600_atom *atom)
+{
+ struct radeon_winsys_cs *cs = rctx->cs;
+ struct r600_cb_misc_state *a = (struct r600_cb_misc_state*)atom;
+
+ if (G_028808_SPECIAL_OP(a->cb_color_control) == V_028808_SPECIAL_RESOLVE_BOX) {
+ r600_write_context_reg_seq(cs, R_028238_CB_TARGET_MASK, 2);
+ if (rctx->chip_class == R600) {
+ r600_write_value(cs, 0xff); /* R_028238_CB_TARGET_MASK */
+ r600_write_value(cs, 0xff); /* R_02823C_CB_SHADER_MASK */
+ } else {
+ r600_write_value(cs, 0xf); /* R_028238_CB_TARGET_MASK */
+ r600_write_value(cs, 0xf); /* R_02823C_CB_SHADER_MASK */
+ }
+ r600_write_context_reg(cs, R_028808_CB_COLOR_CONTROL, a->cb_color_control);
+ } else {
+ unsigned fb_colormask = (1ULL << ((unsigned)a->nr_cbufs * 4)) - 1;
+ unsigned ps_colormask = (1ULL << ((unsigned)a->nr_ps_color_outputs * 4)) - 1;
+ unsigned multiwrite = a->multiwrite && a->nr_cbufs > 1;
+
+ r600_write_context_reg_seq(cs, R_028238_CB_TARGET_MASK, 2);
+ r600_write_value(cs, a->blend_colormask & fb_colormask); /* R_028238_CB_TARGET_MASK */
+ /* Always enable the first color output to make sure alpha-test works even without one. */
+ r600_write_value(cs, 0xf | (multiwrite ? fb_colormask : ps_colormask)); /* R_02823C_CB_SHADER_MASK */
+ r600_write_context_reg(cs, R_028808_CB_COLOR_CONTROL,
+ a->cb_color_control |
+ S_028808_MULTIWRITE_ENABLE(multiwrite));
+ }
}
static void r600_emit_db_misc_state(struct r600_context *rctx, struct r600_atom *atom)
}
db_render_override |= S_028D10_NOOP_CULL_DISABLE(1);
}
- if (a->flush_depthstencil_enabled) {
- db_render_control |= S_028D0C_DEPTH_COPY_ENABLE(1) |
- S_028D0C_STENCIL_COPY_ENABLE(1) |
- S_028D0C_COPY_CENTROID(1);
+ if (a->flush_depthstencil_through_cb) {
+ assert(a->copy_depth || a->copy_stencil);
+
+ db_render_control |= S_028D0C_DEPTH_COPY_ENABLE(a->copy_depth) |
+ S_028D0C_STENCIL_COPY_ENABLE(a->copy_stencil) |
+ S_028D0C_COPY_CENTROID(1) |
+ S_028D0C_COPY_SAMPLE(a->copy_sample);
}
r600_write_context_reg_seq(cs, R_028D0C_DB_RENDER_CONTROL, 2);
static void r600_emit_vertex_buffers(struct r600_context *rctx, struct r600_atom *atom)
{
struct radeon_winsys_cs *cs = rctx->cs;
- struct pipe_vertex_buffer *vb = rctx->vbuf_mgr->real_vertex_buffer;
- unsigned count = rctx->vbuf_mgr->nr_real_vertex_buffers;
- unsigned i, offset;
+ uint32_t dirty_mask = rctx->vertex_buffer_state.dirty_mask;
- for (i = 0; i < count; i++) {
- struct r600_resource *rbuffer = (struct r600_resource*)vb[i].buffer;
+ while (dirty_mask) {
+ struct pipe_vertex_buffer *vb;
+ struct r600_resource *rbuffer;
+ unsigned offset;
+ unsigned buffer_index = u_bit_scan(&dirty_mask);
- if (!rbuffer) {
- continue;
- }
+ vb = &rctx->vertex_buffer_state.vb[buffer_index];
+ rbuffer = (struct r600_resource*)vb->buffer;
+ assert(rbuffer);
- offset = vb[i].buffer_offset;
+ offset = vb->buffer_offset;
/* fetch resources start at index 320 */
r600_write_value(cs, PKT3(PKT3_SET_RESOURCE, 7, 0));
- r600_write_value(cs, (320 + i) * 7);
+ r600_write_value(cs, (320 + buffer_index) * 7);
r600_write_value(cs, offset); /* RESOURCEi_WORD0 */
r600_write_value(cs, rbuffer->buf->size - offset - 1); /* RESOURCEi_WORD1 */
r600_write_value(cs, /* RESOURCEi_WORD2 */
S_038008_ENDIAN_SWAP(r600_endian_swap(32)) |
- S_038008_STRIDE(vb[i].stride));
+ S_038008_STRIDE(vb->stride));
r600_write_value(cs, 0); /* RESOURCEi_WORD3 */
r600_write_value(cs, 0); /* RESOURCEi_WORD4 */
r600_write_value(cs, 0); /* RESOURCEi_WORD5 */
}
}
+static void r600_emit_constant_buffers(struct r600_context *rctx,
+ struct r600_constbuf_state *state,
+ unsigned buffer_id_base,
+ unsigned reg_alu_constbuf_size,
+ unsigned reg_alu_const_cache)
+{
+ struct radeon_winsys_cs *cs = rctx->cs;
+ uint32_t dirty_mask = state->dirty_mask;
+
+ while (dirty_mask) {
+ struct pipe_constant_buffer *cb;
+ struct r600_resource *rbuffer;
+ unsigned offset;
+ unsigned buffer_index = ffs(dirty_mask) - 1;
+
+ cb = &state->cb[buffer_index];
+ rbuffer = (struct r600_resource*)cb->buffer;
+ assert(rbuffer);
+
+ offset = cb->buffer_offset;
+
+ r600_write_context_reg(cs, reg_alu_constbuf_size + buffer_index * 4,
+ ALIGN_DIVUP(cb->buffer_size >> 4, 16));
+ r600_write_context_reg(cs, reg_alu_const_cache + buffer_index * 4, offset >> 8);
+
+ r600_write_value(cs, PKT3(PKT3_NOP, 0, 0));
+ r600_write_value(cs, r600_context_bo_reloc(rctx, rbuffer, RADEON_USAGE_READ));
+
+ r600_write_value(cs, PKT3(PKT3_SET_RESOURCE, 7, 0));
+ r600_write_value(cs, (buffer_id_base + buffer_index) * 7);
+ r600_write_value(cs, offset); /* RESOURCEi_WORD0 */
+ r600_write_value(cs, rbuffer->buf->size - offset - 1); /* RESOURCEi_WORD1 */
+ r600_write_value(cs, /* RESOURCEi_WORD2 */
+ S_038008_ENDIAN_SWAP(r600_endian_swap(32)) |
+ S_038008_STRIDE(16));
+ r600_write_value(cs, 0); /* RESOURCEi_WORD3 */
+ r600_write_value(cs, 0); /* RESOURCEi_WORD4 */
+ r600_write_value(cs, 0); /* RESOURCEi_WORD5 */
+ r600_write_value(cs, 0xc0000000); /* RESOURCEi_WORD6 */
+
+ r600_write_value(cs, PKT3(PKT3_NOP, 0, 0));
+ r600_write_value(cs, r600_context_bo_reloc(rctx, rbuffer, RADEON_USAGE_READ));
+
+ dirty_mask &= ~(1 << buffer_index);
+ }
+ state->dirty_mask = 0;
+}
+
+static void r600_emit_vs_constant_buffers(struct r600_context *rctx, struct r600_atom *atom)
+{
+ r600_emit_constant_buffers(rctx, &rctx->constbuf_state[PIPE_SHADER_VERTEX], 160,
+ R_028180_ALU_CONST_BUFFER_SIZE_VS_0,
+ R_028980_ALU_CONST_CACHE_VS_0);
+}
+
+static void r600_emit_ps_constant_buffers(struct r600_context *rctx, struct r600_atom *atom)
+{
+ r600_emit_constant_buffers(rctx, &rctx->constbuf_state[PIPE_SHADER_FRAGMENT], 0,
+ R_028140_ALU_CONST_BUFFER_SIZE_PS_0,
+ R_028940_ALU_CONST_CACHE_PS_0);
+}
+
+static void r600_emit_sampler_views(struct r600_context *rctx,
+ struct r600_samplerview_state *state,
+ unsigned resource_id_base)
+{
+ struct radeon_winsys_cs *cs = rctx->cs;
+ uint32_t dirty_mask = state->dirty_mask;
+
+ while (dirty_mask) {
+ struct r600_pipe_sampler_view *rview;
+ unsigned resource_index = u_bit_scan(&dirty_mask);
+ unsigned reloc;
+
+ rview = state->views[resource_index];
+ assert(rview);
+
+ r600_write_value(cs, PKT3(PKT3_SET_RESOURCE, 7, 0));
+ r600_write_value(cs, (resource_id_base + resource_index) * 7);
+ r600_write_array(cs, 7, rview->tex_resource_words);
+
+ /* XXX The kernel needs two relocations. This is stupid. */
+ reloc = r600_context_bo_reloc(rctx, rview->tex_resource,
+ RADEON_USAGE_READ);
+ r600_write_value(cs, PKT3(PKT3_NOP, 0, 0));
+ r600_write_value(cs, reloc);
+ r600_write_value(cs, PKT3(PKT3_NOP, 0, 0));
+ r600_write_value(cs, reloc);
+ }
+ state->dirty_mask = 0;
+}
+
+static void r600_emit_vs_sampler_views(struct r600_context *rctx, struct r600_atom *atom)
+{
+ r600_emit_sampler_views(rctx, &rctx->vs_samplers.views, 160 + R600_MAX_CONST_BUFFERS);
+}
+
+static void r600_emit_ps_sampler_views(struct r600_context *rctx, struct r600_atom *atom)
+{
+ r600_emit_sampler_views(rctx, &rctx->ps_samplers.views, R600_MAX_CONST_BUFFERS);
+}
+
+static void r600_emit_sampler_states(struct r600_context *rctx,
+ struct r600_textures_info *texinfo,
+ unsigned resource_id_base,
+ unsigned border_color_reg)
+{
+ struct radeon_winsys_cs *cs = rctx->cs;
+ unsigned i;
+
+ for (i = 0; i < texinfo->n_samplers; i++) {
+
+ if (texinfo->samplers[i] == NULL) {
+ continue;
+ }
+
+ /* TEX_ARRAY_OVERRIDE must be set for array textures to disable
+ * filtering between layers.
+ * Don't update TEX_ARRAY_OVERRIDE if we don't have the sampler view.
+ */
+ if (texinfo->views.views[i]) {
+ if (texinfo->views.views[i]->base.texture->target == PIPE_TEXTURE_1D_ARRAY ||
+ texinfo->views.views[i]->base.texture->target == PIPE_TEXTURE_2D_ARRAY) {
+ texinfo->samplers[i]->tex_sampler_words[0] |= S_03C000_TEX_ARRAY_OVERRIDE(1);
+ texinfo->is_array_sampler[i] = true;
+ } else {
+ texinfo->samplers[i]->tex_sampler_words[0] &= C_03C000_TEX_ARRAY_OVERRIDE;
+ texinfo->is_array_sampler[i] = false;
+ }
+ }
+
+ r600_write_value(cs, PKT3(PKT3_SET_SAMPLER, 3, 0));
+ r600_write_value(cs, (resource_id_base + i) * 3);
+ r600_write_array(cs, 3, texinfo->samplers[i]->tex_sampler_words);
+
+ if (texinfo->samplers[i]->border_color_use) {
+ unsigned offset;
+
+ offset = border_color_reg;
+ offset += i * 16;
+ r600_write_config_reg_seq(cs, offset, 4);
+ r600_write_array(cs, 4, texinfo->samplers[i]->border_color);
+ }
+ }
+}
+
+static void r600_emit_vs_sampler_states(struct r600_context *rctx, struct r600_atom *atom)
+{
+ r600_emit_sampler_states(rctx, &rctx->vs_samplers, 18, R_00A600_TD_VS_SAMPLER0_BORDER_RED);
+}
+
+static void r600_emit_ps_sampler_states(struct r600_context *rctx, struct r600_atom *atom)
+{
+ r600_emit_sampler_states(rctx, &rctx->ps_samplers, 0, R_00A400_TD_PS_SAMPLER0_BORDER_RED);
+}
+
+static void r600_emit_seamless_cube_map(struct r600_context *rctx, struct r600_atom *atom)
+{
+ struct radeon_winsys_cs *cs = rctx->cs;
+ unsigned tmp;
+
+ tmp = S_009508_DISABLE_CUBE_ANISO(1) |
+ S_009508_SYNC_GRADIENT(1) |
+ S_009508_SYNC_WALKER(1) |
+ S_009508_SYNC_ALIGNER(1);
+ if (!rctx->seamless_cube_map.enabled) {
+ tmp |= S_009508_DISABLE_CUBE_WRAP(1);
+ }
+ r600_write_config_reg(cs, R_009508_TA_CNTL_AUX, tmp);
+}
+
+static void r600_emit_sample_mask(struct r600_context *rctx, struct r600_atom *a)
+{
+ struct r600_sample_mask *s = (struct r600_sample_mask*)a;
+ uint8_t mask = s->sample_mask;
+
+ r600_write_context_reg(rctx->cs, R_028C48_PA_SC_AA_MASK,
+ mask | (mask << 8) | (mask << 16) | (mask << 24));
+}
+
void r600_init_state_functions(struct r600_context *rctx)
{
- r600_init_atom(&rctx->db_misc_state.atom, r600_emit_db_misc_state, 4, 0);
+ unsigned id = 4;
+
+ /* !!!
+ * To avoid GPU lockup registers must be emited in a specific order
+ * (no kidding ...). The order below is important and have been
+ * partialy infered from analyzing fglrx command stream.
+ *
+ * Don't reorder atom without carefully checking the effect (GPU lockup
+ * or piglit regression).
+ * !!!
+ */
+
+ /* shader const */
+ r600_init_atom(rctx, &rctx->constbuf_state[PIPE_SHADER_VERTEX].atom, id++, r600_emit_vs_constant_buffers, 0);
+ r600_init_atom(rctx, &rctx->constbuf_state[PIPE_SHADER_FRAGMENT].atom, id++, r600_emit_ps_constant_buffers, 0);
+
+ /* sampler must be emited before TA_CNTL_AUX otherwise DISABLE_CUBE_WRAP change
+ * does not take effect (TA_CNTL_AUX emited by r600_emit_seamless_cube_map)
+ */
+ r600_init_atom(rctx, &rctx->vs_samplers.atom_sampler, id++, r600_emit_vs_sampler_states, 0);
+ r600_init_atom(rctx, &rctx->ps_samplers.atom_sampler, id++, r600_emit_ps_sampler_states, 0);
+ /* resource */
+ r600_init_atom(rctx, &rctx->vs_samplers.views.atom, id++, r600_emit_vs_sampler_views, 0);
+ r600_init_atom(rctx, &rctx->ps_samplers.views.atom, id++, r600_emit_ps_sampler_views, 0);
+ r600_init_atom(rctx, &rctx->vertex_buffer_state.atom, id++, r600_emit_vertex_buffers, 0);
+
+ r600_init_atom(rctx, &rctx->seamless_cube_map.atom, id++, r600_emit_seamless_cube_map, 3);
+ r600_atom_dirty(rctx, &rctx->seamless_cube_map.atom);
+
+ r600_init_atom(rctx, &rctx->sample_mask.atom, id++, r600_emit_sample_mask, 3);
+ rctx->sample_mask.sample_mask = ~0;
+ r600_atom_dirty(rctx, &rctx->sample_mask.atom);
+
+ r600_init_atom(rctx, &rctx->cb_misc_state.atom, id++, r600_emit_cb_misc_state, 0);
+ r600_atom_dirty(rctx, &rctx->cb_misc_state.atom);
+
+ r600_init_atom(rctx, &rctx->alphatest_state.atom, id++, r600_emit_alphatest_state, 6);
+ r600_atom_dirty(rctx, &rctx->alphatest_state.atom);
+
+ r600_init_atom(rctx, &rctx->db_misc_state.atom, id++, r600_emit_db_misc_state, 4);
r600_atom_dirty(rctx, &rctx->db_misc_state.atom);
- r600_init_atom(&rctx->vertex_buffer_state, r600_emit_vertex_buffers, 0, 0);
rctx->context.create_blend_state = r600_create_blend_state;
rctx->context.create_depth_stencil_alpha_state = r600_create_dsa_state;
- rctx->context.create_fs_state = r600_create_shader_state;
rctx->context.create_rasterizer_state = r600_create_rs_state;
rctx->context.create_sampler_state = r600_create_sampler_state;
rctx->context.create_sampler_view = r600_create_sampler_view;
- rctx->context.create_vertex_elements_state = r600_create_vertex_elements;
- rctx->context.create_vs_state = r600_create_shader_state;
- rctx->context.bind_blend_state = r600_bind_blend_state;
- rctx->context.bind_depth_stencil_alpha_state = r600_bind_dsa_state;
- rctx->context.bind_fragment_sampler_states = r600_bind_ps_samplers;
- rctx->context.bind_fs_state = r600_bind_ps_shader;
- rctx->context.bind_rasterizer_state = r600_bind_rs_state;
- rctx->context.bind_vertex_elements_state = r600_bind_vertex_elements;
- rctx->context.bind_vertex_sampler_states = r600_bind_vs_samplers;
- rctx->context.bind_vs_state = r600_bind_vs_shader;
- rctx->context.delete_blend_state = r600_delete_state;
- rctx->context.delete_depth_stencil_alpha_state = r600_delete_state;
- rctx->context.delete_fs_state = r600_delete_ps_shader;
- rctx->context.delete_rasterizer_state = r600_delete_rs_state;
- rctx->context.delete_sampler_state = r600_delete_state;
- rctx->context.delete_vertex_elements_state = r600_delete_vertex_element;
- rctx->context.delete_vs_state = r600_delete_vs_shader;
- rctx->context.set_blend_color = r600_set_blend_color;
rctx->context.set_clip_state = r600_set_clip_state;
- rctx->context.set_constant_buffer = r600_set_constant_buffer;
- rctx->context.set_fragment_sampler_views = r600_set_ps_sampler_views;
rctx->context.set_framebuffer_state = r600_set_framebuffer_state;
rctx->context.set_polygon_stipple = r600_set_polygon_stipple;
- rctx->context.set_sample_mask = r600_set_sample_mask;
rctx->context.set_scissor_state = r600_pipe_set_scissor_state;
- rctx->context.set_stencil_ref = r600_set_pipe_stencil_ref;
- rctx->context.set_vertex_buffers = r600_set_vertex_buffers;
- rctx->context.set_index_buffer = r600_set_index_buffer;
- rctx->context.set_vertex_sampler_views = r600_set_vs_sampler_views;
rctx->context.set_viewport_state = r600_set_viewport_state;
- rctx->context.sampler_view_destroy = r600_sampler_view_destroy;
- rctx->context.redefine_user_buffer = u_default_redefine_user_buffer;
- rctx->context.texture_barrier = r600_texture_barrier;
- rctx->context.create_stream_output_target = r600_create_so_target;
- rctx->context.stream_output_target_destroy = r600_so_target_destroy;
- rctx->context.set_stream_output_targets = r600_set_so_targets;
}
+/* Adjust GPR allocation on R6xx/R7xx */
void r600_adjust_gprs(struct r600_context *rctx)
{
struct r600_pipe_state rstate;
unsigned tmp;
int diff;
- if (rctx->chip_class >= EVERGREEN)
- return;
-
- if (!rctx->ps_shader || !rctx->vs_shader)
- return;
-
- if (rctx->ps_shader->shader.bc.ngpr > rctx->default_ps_gprs)
- {
- diff = rctx->ps_shader->shader.bc.ngpr - rctx->default_ps_gprs;
+ if (rctx->ps_shader->current->shader.bc.ngpr > rctx->default_ps_gprs) {
+ diff = rctx->ps_shader->current->shader.bc.ngpr - rctx->default_ps_gprs;
num_vs_gprs -= diff;
num_ps_gprs += diff;
}
- if (rctx->vs_shader->shader.bc.ngpr > rctx->default_vs_gprs)
+ if (rctx->vs_shader->current->shader.bc.ngpr > rctx->default_vs_gprs)
{
- diff = rctx->vs_shader->shader.bc.ngpr - rctx->default_vs_gprs;
+ diff = rctx->vs_shader->current->shader.bc.ngpr - rctx->default_vs_gprs;
num_ps_gprs -= diff;
num_vs_gprs += diff;
}
tmp |= S_008C04_NUM_VS_GPRS(num_vs_gprs);
tmp |= S_008C04_NUM_CLAUSE_TEMP_GPRS(rctx->r6xx_num_clause_temp_gprs);
rstate.nregs = 0;
- r600_pipe_state_add_reg(&rstate, R_008C04_SQ_GPR_RESOURCE_MGMT_1, tmp, NULL, 0);
+ r600_pipe_state_add_reg(&rstate, R_008C04_SQ_GPR_RESOURCE_MGMT_1, tmp);
r600_context_pipe_state_set(rctx, &rstate);
}
enum radeon_family family;
struct r600_command_buffer *cb = &rctx->start_cs_cmd;
uint32_t tmp;
- unsigned i;
- r600_init_command_buffer(cb, 256, EMIT_EARLY);
+ r600_init_command_buffer(rctx, cb, 0, 256);
/* R6xx requires this packet at the start of each command buffer */
if (rctx->chip_class == R600) {
r600_store_value(cb, 0); /* R_0288C4_SQ_REDUC_RING_ITEMSIZE */
r600_store_value(cb, 0); /* R_0288C8_SQ_GS_VERT_ITEMSIZE */
+ /* to avoid GPU doing any preloading of constant from random address */
+ r600_store_context_reg_seq(cb, R_028140_ALU_CONST_BUFFER_SIZE_PS_0, 8);
+ r600_store_value(cb, 0); /* R_028140_ALU_CONST_BUFFER_SIZE_PS_0 */
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_context_reg_seq(cb, R_028180_ALU_CONST_BUFFER_SIZE_VS_0, 8);
+ r600_store_value(cb, 0); /* R_028180_ALU_CONST_BUFFER_SIZE_VS_0 */
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+ r600_store_value(cb, 0);
+
r600_store_context_reg_seq(cb, R_028A10_VGT_OUTPUT_PATH_CNTL, 13);
r600_store_value(cb, 0); /* R_028A10_VGT_OUTPUT_PATH_CNTL */
r600_store_value(cb, 0); /* R_028A14_VGT_HOS_CNTL */
r600_store_value(cb, 0); /* R_028D2C_DB_SRESULTS_COMPARE_STATE1 */
r600_store_value(cb, 0); /* R_028D30_DB_PRELOAD_CONTROL */
- r600_store_context_reg(cb, R_028D44_DB_ALPHA_TO_MASK, 0xAA00);
-
r600_store_context_reg(cb, R_028820_PA_CL_NANINF_CNTL, 0);
r600_store_context_reg(cb, R_028A48_PA_SC_MPASS_PS_CNTL, 0);
- r600_store_context_reg_seq(cb, R_028C00_PA_SC_LINE_CNTL, 2);
- r600_store_value(cb, 0x400); /* R_028C00_PA_SC_LINE_CNTL */
- r600_store_value(cb, 0); /* R_028C04_PA_SC_AA_CONFIG */
-
- r600_store_context_reg_seq(cb, R_028C0C_PA_CL_GB_VERT_CLIP_ADJ, 6);
+ r600_store_context_reg_seq(cb, R_028C0C_PA_CL_GB_VERT_CLIP_ADJ, 4);
r600_store_value(cb, 0x3F800000); /* R_028C0C_PA_CL_GB_VERT_CLIP_ADJ */
r600_store_value(cb, 0x3F800000); /* R_028C10_PA_CL_GB_VERT_DISC_ADJ */
r600_store_value(cb, 0x3F800000); /* R_028C14_PA_CL_GB_HORZ_CLIP_ADJ */
r600_store_value(cb, 0x3F800000); /* R_028C18_PA_CL_GB_HORZ_DISC_ADJ */
- r600_store_value(cb, 0); /* R_028C1C_PA_SC_AA_SAMPLE_LOCS_MCTX */
- r600_store_value(cb, 0); /* R_028C20_PA_SC_AA_SAMPLE_LOCS_8S_WD1_MCTX */
r600_store_context_reg_seq(cb, R_0282D0_PA_SC_VPORT_ZMIN_0, 2);
r600_store_value(cb, 0); /* R_0282D0_PA_SC_VPORT_ZMIN_0 */
r600_store_context_reg(cb, R_028818_PA_CL_VTE_CNTL, 0x43F);
- r600_store_context_reg_seq(cb, R_028100_CB_COLOR0_MASK, 8);
- for (i = 0; i < 8; i++) {
- r600_store_value(cb, 0);
- }
-
r600_store_context_reg(cb, R_028200_PA_SC_WINDOW_OFFSET, 0);
r600_store_context_reg(cb, R_02820C_PA_SC_CLIPRECT_RULE, 0xFFFF);
r600_store_value(cb, 0xFF); /* R_028C38_CB_CLRCMP_DST */
r600_store_value(cb, 0xFFFFFFFF); /* R_028C3C_CB_CLRCMP_MSK */
- r600_store_context_reg(cb, R_028C48_PA_SC_AA_MASK, 0xFFFFFFFF);
-
r600_store_context_reg_seq(cb, R_028030_PA_SC_SCREEN_SCISSOR_TL, 2);
r600_store_value(cb, 0); /* R_028030_PA_SC_SCREEN_SCISSOR_TL */
r600_store_value(cb, S_028034_BR_X(8192) | S_028034_BR_Y(8192)); /* R_028034_PA_SC_SCREEN_SCISSOR_BR */
r600_store_context_reg(cb, R_0288A4_SQ_PGM_RESOURCES_FS, 0);
r600_store_context_reg(cb, R_0288DC_SQ_PGM_CF_OFFSET_FS, 0);
- if (rctx->chip_class == R700 && rctx->screen->info.r600_has_streamout)
+ if (rctx->chip_class == R700 && rctx->screen->has_streamout)
r600_store_context_reg(cb, R_028354_SX_SURFACE_SYNC, S_028354_SURFACE_SYNC_MASK(0xf));
r600_store_context_reg(cb, R_028800_DB_DEPTH_CONTROL, 0);
+ if (rctx->screen->has_streamout) {
+ r600_store_context_reg(cb, R_028B28_VGT_STRMOUT_DRAW_OPAQUE_OFFSET, 0);
+ }
r600_store_loop_const(cb, R_03E200_SQ_LOOP_CONST_0, 0x1000FFF);
r600_store_loop_const(cb, R_03E200_SQ_LOOP_CONST_0 + (32 * 4), 0x1000FFF);
int pos_index = -1, face_index = -1;
unsigned tmp, sid, ufi = 0;
int need_linear = 0;
+ unsigned z_export = 0, stencil_export = 0;
rstate->nregs = 0;
}
r600_pipe_state_add_reg(rstate, R_028644_SPI_PS_INPUT_CNTL_0 + i * 4,
- tmp, NULL, 0);
+ tmp);
}
db_shader_control = S_02880C_Z_ORDER(V_02880C_EARLY_Z_THEN_LATE_Z);
for (i = 0; i < rshader->noutput; i++) {
if (rshader->output[i].name == TGSI_SEMANTIC_POSITION)
- db_shader_control |= S_02880C_Z_EXPORT_ENABLE(1);
+ z_export = 1;
if (rshader->output[i].name == TGSI_SEMANTIC_STENCIL)
- db_shader_control |= S_02880C_STENCIL_REF_EXPORT_ENABLE(1);
+ stencil_export = 1;
}
+ db_shader_control |= S_02880C_Z_EXPORT_ENABLE(z_export);
+ db_shader_control |= S_02880C_STENCIL_REF_EXPORT_ENABLE(stencil_export);
if (rshader->uses_kill)
db_shader_control |= S_02880C_KILL_ENABLE(1);
exports_ps = 0;
- num_cout = 0;
for (i = 0; i < rshader->noutput; i++) {
if (rshader->output[i].name == TGSI_SEMANTIC_POSITION ||
- rshader->output[i].name == TGSI_SEMANTIC_STENCIL)
+ rshader->output[i].name == TGSI_SEMANTIC_STENCIL) {
exports_ps |= 1;
- else if (rshader->output[i].name == TGSI_SEMANTIC_COLOR) {
- num_cout++;
}
}
+ num_cout = rshader->nr_ps_color_exports;
exports_ps |= S_028854_EXPORT_COLORS(num_cout);
if (!exports_ps) {
/* always at least export 1 component per pixel */
exports_ps = 2;
}
+ shader->nr_ps_color_outputs = num_cout;
+
spi_ps_in_control_0 = S_0286CC_NUM_INTERP(rshader->ninput) |
S_0286CC_PERSP_GRADIENT_ENA(1)|
S_0286CC_LINEAR_GRADIENT_ENA(need_linear);
if (rctx->family == CHIP_R600)
ufi = 1;
- r600_pipe_state_add_reg(rstate, R_0286CC_SPI_PS_IN_CONTROL_0, spi_ps_in_control_0, NULL, 0);
- r600_pipe_state_add_reg(rstate, R_0286D0_SPI_PS_IN_CONTROL_1, spi_ps_in_control_1, NULL, 0);
- r600_pipe_state_add_reg(rstate, R_0286D8_SPI_INPUT_Z, spi_input_z, NULL, 0);
- r600_pipe_state_add_reg(rstate,
- R_028840_SQ_PGM_START_PS,
- 0, shader->bo, RADEON_USAGE_READ);
+ r600_pipe_state_add_reg(rstate, R_0286CC_SPI_PS_IN_CONTROL_0, spi_ps_in_control_0);
+ r600_pipe_state_add_reg(rstate, R_0286D0_SPI_PS_IN_CONTROL_1, spi_ps_in_control_1);
+ r600_pipe_state_add_reg(rstate, R_0286D8_SPI_INPUT_Z, spi_input_z);
+ r600_pipe_state_add_reg_bo(rstate,
+ R_028840_SQ_PGM_START_PS,
+ 0, shader->bo, RADEON_USAGE_READ);
r600_pipe_state_add_reg(rstate,
R_028850_SQ_PGM_RESOURCES_PS,
S_028850_NUM_GPRS(rshader->bc.ngpr) |
S_028850_STACK_SIZE(rshader->bc.nstack) |
- S_028850_UNCACHED_FIRST_INST(ufi),
- NULL, 0);
+ S_028850_UNCACHED_FIRST_INST(ufi));
r600_pipe_state_add_reg(rstate,
R_028854_SQ_PGM_EXPORTS_PS,
- exports_ps, NULL, 0);
+ exports_ps);
/* only set some bits here, the other bits are set in the dsa state */
- r600_pipe_state_add_reg(rstate, R_02880C_DB_SHADER_CONTROL,
- db_shader_control,
- NULL, 0);
+ shader->db_shader_control = db_shader_control;
+ shader->ps_depth_export = z_export | stencil_export;
shader->sprite_coord_enable = rctx->sprite_coord_enable;
if (rctx->rasterizer)
for (i = 0; i < 10; i++) {
r600_pipe_state_add_reg(rstate,
R_028614_SPI_VS_OUT_ID_0 + i * 4,
- spi_vs_out_id[i], NULL, 0);
+ spi_vs_out_id[i]);
}
/* Certain attributes (position, psize, etc.) don't count as params.
nparams = 1;
r600_pipe_state_add_reg(rstate,
- R_0286C4_SPI_VS_OUT_CONFIG,
- S_0286C4_VS_EXPORT_COUNT(nparams - 1),
- NULL, 0);
- r600_pipe_state_add_reg(rstate,
- R_028868_SQ_PGM_RESOURCES_VS,
- S_028868_NUM_GPRS(rshader->bc.ngpr) |
- S_028868_STACK_SIZE(rshader->bc.nstack),
- NULL, 0);
+ R_0286C4_SPI_VS_OUT_CONFIG,
+ S_0286C4_VS_EXPORT_COUNT(nparams - 1));
r600_pipe_state_add_reg(rstate,
+ R_028868_SQ_PGM_RESOURCES_VS,
+ S_028868_NUM_GPRS(rshader->bc.ngpr) |
+ S_028868_STACK_SIZE(rshader->bc.nstack));
+ r600_pipe_state_add_reg_bo(rstate,
R_028858_SQ_PGM_START_VS,
0, shader->bo, RADEON_USAGE_READ);
rstate = &ve->rstate;
rstate->id = R600_PIPE_STATE_FETCH_SHADER;
rstate->nregs = 0;
- r600_pipe_state_add_reg(rstate, R_028894_SQ_PGM_START_FS,
+ r600_pipe_state_add_reg_bo(rstate, R_028894_SQ_PGM_START_FS,
0,
ve->fetch_shader, RADEON_USAGE_READ);
}
+void *r600_create_resolve_blend(struct r600_context *rctx)
+{
+ struct pipe_blend_state blend;
+ struct r600_pipe_state *rstate;
+ unsigned i;
+
+ memset(&blend, 0, sizeof(blend));
+ blend.independent_blend_enable = true;
+ for (i = 0; i < 2; i++) {
+ blend.rt[i].colormask = 0xf;
+ blend.rt[i].blend_enable = 1;
+ blend.rt[i].rgb_func = PIPE_BLEND_ADD;
+ blend.rt[i].alpha_func = PIPE_BLEND_ADD;
+ blend.rt[i].rgb_src_factor = PIPE_BLENDFACTOR_ZERO;
+ blend.rt[i].rgb_dst_factor = PIPE_BLENDFACTOR_ZERO;
+ blend.rt[i].alpha_src_factor = PIPE_BLENDFACTOR_ZERO;
+ blend.rt[i].alpha_dst_factor = PIPE_BLENDFACTOR_ZERO;
+ }
+ rstate = r600_create_blend_state_mode(&rctx->context, &blend, V_028808_SPECIAL_RESOLVE_BOX);
+ return rstate;
+}
+
+void *r700_create_resolve_blend(struct r600_context *rctx)
+{
+ struct pipe_blend_state blend;
+ struct r600_pipe_state *rstate;
+
+ memset(&blend, 0, sizeof(blend));
+ blend.independent_blend_enable = true;
+ blend.rt[0].colormask = 0xf;
+ rstate = r600_create_blend_state_mode(&rctx->context, &blend, V_028808_SPECIAL_RESOLVE_BOX);
+ return rstate;
+}
+
+void *r600_create_decompress_blend(struct r600_context *rctx)
+{
+ struct pipe_blend_state blend;
+ struct r600_pipe_state *rstate;
+
+ memset(&blend, 0, sizeof(blend));
+ blend.independent_blend_enable = true;
+ blend.rt[0].colormask = 0xf;
+ rstate = r600_create_blend_state_mode(&rctx->context, &blend, V_028808_SPECIAL_EXPAND_SAMPLES);
+ return rstate;
+}
+
void *r600_create_db_flush_dsa(struct r600_context *rctx)
{
struct pipe_depth_stencil_alpha_state dsa;
- struct r600_pipe_state *rstate;
- struct r600_pipe_dsa *dsa_state;
boolean quirk = false;
if (rctx->family == CHIP_RV610 || rctx->family == CHIP_RV630 ||
dsa.stencil[0].writemask = 0xff;
}
- rstate = rctx->context.create_depth_stencil_alpha_state(&rctx->context, &dsa);
- dsa_state = (struct r600_pipe_dsa*)rstate;
- dsa_state->is_flush = true;
- return rstate;
+ return rctx->context.create_depth_stencil_alpha_state(&rctx->context, &dsa);
}
-void r600_pipe_init_buffer_resource(struct r600_context *rctx,
- struct r600_pipe_resource_state *rstate)
+void r600_update_dual_export_state(struct r600_context * rctx)
{
- rstate->id = R600_PIPE_STATE_RESOURCE;
-
- rstate->bo[0] = NULL;
- rstate->val[0] = 0;
- rstate->val[1] = 0;
- rstate->val[2] = 0;
- rstate->val[3] = 0;
- rstate->val[4] = 0;
- rstate->val[5] = 0;
- rstate->val[6] = 0xc0000000;
-}
+ unsigned dual_export = rctx->export_16bpc && rctx->nr_cbufs &&
+ !rctx->ps_shader->current->ps_depth_export;
+ unsigned db_shader_control = rctx->ps_shader->current->db_shader_control |
+ S_02880C_DUAL_EXPORT_ENABLE(dual_export);
-void r600_pipe_mod_buffer_resource(struct r600_pipe_resource_state *rstate,
- struct r600_resource *rbuffer,
- unsigned offset, unsigned stride,
- enum radeon_bo_usage usage)
-{
- rstate->val[0] = offset;
- rstate->bo[0] = rbuffer;
- rstate->bo_usage[0] = usage;
- rstate->val[1] = rbuffer->buf->size - offset - 1;
- rstate->val[2] = S_038008_ENDIAN_SWAP(r600_endian_swap(32)) |
- S_038008_STRIDE(stride);
+ if (db_shader_control != rctx->db_shader_control) {
+ struct r600_pipe_state rstate;
+
+ rctx->db_shader_control = db_shader_control;
+ rstate.nregs = 0;
+ r600_pipe_state_add_reg(&rstate, R_02880C_DB_SHADER_CONTROL, db_shader_control);
+ r600_context_pipe_state_set(rctx, &rstate);
+ }
}