return 0;
}
-static unsigned r600_tex_dim(unsigned dim)
+static unsigned r600_tex_dim(unsigned dim, unsigned nr_samples)
{
switch (dim) {
default:
return V_030000_SQ_TEX_DIM_1D_ARRAY;
case PIPE_TEXTURE_2D:
case PIPE_TEXTURE_RECT:
- return V_030000_SQ_TEX_DIM_2D;
+ return nr_samples > 1 ? V_030000_SQ_TEX_DIM_2D_MSAA :
+ V_030000_SQ_TEX_DIM_2D;
case PIPE_TEXTURE_2D_ARRAY:
- return V_030000_SQ_TEX_DIM_2D_ARRAY;
+ return nr_samples > 1 ? V_030000_SQ_TEX_DIM_2D_ARRAY_MSAA :
+ V_030000_SQ_TEX_DIM_2D_ARRAY;
case PIPE_TEXTURE_3D:
return V_030000_SQ_TEX_DIM_3D;
case PIPE_TEXTURE_CUBE:
case PIPE_FORMAT_R16G16_SNORM:
case PIPE_FORMAT_R16G16_UINT:
case PIPE_FORMAT_R16G16_SINT:
- case PIPE_FORMAT_R16G16B16_FLOAT:
- case PIPE_FORMAT_R32G32B32_FLOAT:
return V_028C70_SWAP_STD;
/* 64-bit buffers. */
case PIPE_FORMAT_R16G16B16A16_SNORM:
return V_028C70_COLOR_16_16_16_16;
- case PIPE_FORMAT_R16G16B16_FLOAT:
case PIPE_FORMAT_R16G16B16A16_FLOAT:
return V_028C70_COLOR_16_16_16_16_FLOAT;
case PIPE_FORMAT_L32A32_SINT:
return V_028C70_COLOR_32_32;
- /* 96-bit buffers. */
- case PIPE_FORMAT_R32G32B32_FLOAT:
- return V_028C70_COLOR_32_32_32_FLOAT;
-
/* 128-bit buffers. */
case PIPE_FORMAT_R32G32B32A32_SNORM:
case PIPE_FORMAT_R32G32B32A32_UNORM:
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 < 19)
+ 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 {
+ usage |= PIPE_BIND_RENDER_TARGET;
+ }
+ }
if ((usage & PIPE_BIND_SAMPLER_VIEW) &&
r600_is_sampler_format_supported(screen, format)) {
return retval == usage;
}
-static void *evergreen_create_blend_state(struct pipe_context *ctx,
- const struct pipe_blend_state *state)
+static void *evergreen_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);
blend->cb_target_mask = target_mask;
if (target_mask)
- color_control |= S_028808_MODE(V_028808_CB_NORMAL);
+ color_control |= S_028808_MODE(mode);
else
color_control |= S_028808_MODE(V_028808_CB_DISABLE);
r600_pipe_state_add_reg(rstate, R_028780_CB_BLEND0_CONTROL + i * 4, blend_cntl[i]);
}
+ r600_pipe_state_add_reg(rstate, R_028B70_DB_ALPHA_TO_MASK,
+ S_028B70_ALPHA_TO_MASK_ENABLE(state->alpha_to_coverage) |
+ S_028B70_ALPHA_TO_MASK_OFFSET0(2) |
+ S_028B70_ALPHA_TO_MASK_OFFSET1(2) |
+ S_028B70_ALPHA_TO_MASK_OFFSET2(2) |
+ S_028B70_ALPHA_TO_MASK_OFFSET3(2));
+
+ blend->alpha_to_one = state->alpha_to_one;
return rstate;
}
+static void *evergreen_create_blend_state(struct pipe_context *ctx,
+ const struct pipe_blend_state *state)
+{
+
+ return evergreen_create_blend_state_mode(ctx, state, V_028808_CB_NORMAL);
+}
+
static void *evergreen_create_dsa_state(struct pipe_context *ctx,
const struct pipe_depth_stencil_alpha_state *state)
{
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 = (unsigned)state->line_width * 8;
r600_pipe_state_add_reg(rstate, R_028A08_PA_SU_LINE_CNTL, S_028A08_WIDTH(tmp));
r600_pipe_state_add_reg(rstate, R_028A48_PA_SC_MODE_CNTL_0,
+ S_028A48_MSAA_ENABLE(state->multisample) |
S_028A48_VPORT_SCISSOR_ENABLE(state->scissor) |
S_028A48_LINE_STIPPLE_ENABLE(state->line_stipple_enable));
S_028C08_PIX_CENTER_HALF(state->gl_rasterization_rules));
} else {
r600_pipe_state_add_reg(rstate, R_028C08_PA_SU_VTX_CNTL,
- S_028C08_PIX_CENTER_HALF(state->gl_rasterization_rules));
+ S_028C08_PIX_CENTER_HALF(state->gl_rasterization_rules) |
+ S_028C08_QUANT_MODE(V_028C08_X_1_4096TH));
}
r600_pipe_state_add_reg(rstate, R_028B7C_PA_SU_POLY_OFFSET_CLAMP, fui(state->offset_clamp));
r600_pipe_state_add_reg(rstate, R_028814_PA_SU_SC_MODE_CNTL,
static void *evergreen_create_sampler_state(struct pipe_context *ctx,
const struct pipe_sampler_state *state)
{
- struct r600_pipe_state *rstate = CALLOC_STRUCT(r600_pipe_state);
+ struct r600_pipe_sampler_state *ss = CALLOC_STRUCT(r600_pipe_sampler_state);
union util_color uc;
unsigned aniso_flag_offset = state->max_anisotropy > 1 ? 2 : 0;
- if (rstate == NULL) {
+ if (ss == NULL) {
return NULL;
}
- rstate->id = R600_PIPE_STATE_SAMPLER;
+ /* directly into sampler avoid r6xx code to emit useless reg */
+ ss->seamless_cube_map = 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), 8)) |
- S_03C004_MAX_LOD(S_FIXED(CLAMP(state->max_lod, 0, 15), 8)),
- NULL, 0);
- r600_pipe_state_add_reg_noblock(rstate, R_03C008_SQ_TEX_SAMPLER_WORD2_0,
- S_03C008_LOD_BIAS(S_FIXED(CLAMP(state->lod_bias, -16, 16), 8)) |
- (state->seamless_cube_map ? 0 : S_03C008_DISABLE_CUBE_WRAP(1)) |
- S_03C008_TYPE(1),
- NULL, 0);
-
+ ss->border_color_use = false;
+ /* 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), 8)) |
+ S_03C004_MAX_LOD(S_FIXED(CLAMP(state->max_lod, 0, 15), 8));
+ /* R_03C008_SQ_TEX_SAMPLER_WORD2_0 */
+ ss->tex_sampler_words[2] = S_03C008_LOD_BIAS(S_FIXED(CLAMP(state->lod_bias, -16, 16), 8)) |
+ (state->seamless_cube_map ? 0 : S_03C008_DISABLE_CUBE_WRAP(1)) |
+ S_03C008_TYPE(1);
if (uc.ui) {
- r600_pipe_state_add_reg_noblock(rstate, R_00A404_TD_PS_SAMPLER0_BORDER_RED, fui(state->border_color.f[0]), NULL, 0);
- r600_pipe_state_add_reg_noblock(rstate, R_00A408_TD_PS_SAMPLER0_BORDER_GREEN, fui(state->border_color.f[1]), NULL, 0);
- r600_pipe_state_add_reg_noblock(rstate, R_00A40C_TD_PS_SAMPLER0_BORDER_BLUE, fui(state->border_color.f[2]), NULL, 0);
- r600_pipe_state_add_reg_noblock(rstate, R_00A410_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 *evergreen_create_sampler_view(struct pipe_context *ctx,
{
struct r600_screen *rscreen = (struct r600_screen*)ctx->screen;
struct r600_pipe_sampler_view *view = CALLOC_STRUCT(r600_pipe_sampler_view);
- 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;
}
view->tex_resource = &tmp->resource;
- view->tex_resource_words[0] = (S_030000_DIM(r600_tex_dim(texture->target)) |
+ view->tex_resource_words[0] = (S_030000_DIM(r600_tex_dim(texture->target, texture->nr_samples)) |
S_030000_PITCH((pitch / 8) - 1) |
S_030000_TEX_WIDTH(width - 1));
if (rscreen->chip_class == CAYMAN)
view->tex_resource_words[1] = (S_030004_TEX_HEIGHT(height - 1) |
S_030004_TEX_DEPTH(depth - 1) |
S_030004_ARRAY_MODE(array_mode));
- view->tex_resource_words[2] = (tmp->offset[0] + r600_resource_va(ctx->screen, texture)) >> 8;
- if (state->u.tex.last_level) {
- view->tex_resource_words[3] = (tmp->offset[1] + r600_resource_va(ctx->screen, texture)) >> 8;
+ view->tex_resource_words[2] = (tmp->surface.level[0].offset + r600_resource_va(ctx->screen, texture)) >> 8;
+ if (state->u.tex.last_level && texture->nr_samples <= 1) {
+ view->tex_resource_words[3] = (tmp->surface.level[1].offset + r600_resource_va(ctx->screen, texture)) >> 8;
} else {
- view->tex_resource_words[3] = (tmp->offset[0] + r600_resource_va(ctx->screen, texture)) >> 8;
+ view->tex_resource_words[3] = (tmp->surface.level[0].offset + r600_resource_va(ctx->screen, texture)) >> 8;
}
view->tex_resource_words[4] = (word4 |
S_030010_SRF_MODE_ALL(V_030010_SRF_MODE_ZERO_CLAMP_MINUS_ONE) |
- S_030010_ENDIAN_SWAP(endian) |
- S_030010_BASE_LEVEL(state->u.tex.first_level));
- view->tex_resource_words[5] = (S_030014_LAST_LEVEL(state->u.tex.last_level) |
- S_030014_BASE_ARRAY(state->u.tex.first_layer) |
- S_030014_LAST_ARRAY(state->u.tex.last_layer));
+ S_030010_ENDIAN_SWAP(endian));
+ view->tex_resource_words[5] = S_030014_BASE_ARRAY(state->u.tex.first_layer) |
+ S_030014_LAST_ARRAY(state->u.tex.last_layer);
+ if (texture->nr_samples > 1) {
+ unsigned log_samples = util_logbase2(texture->nr_samples);
+ if (rscreen->chip_class == CAYMAN) {
+ view->tex_resource_words[4] |= S_030010_LOG2_NUM_FRAGMENTS(log_samples);
+ }
+ /* LAST_LEVEL holds log2(nr_samples) for multisample textures */
+ view->tex_resource_words[5] |= S_030014_LAST_LEVEL(log_samples);
+ } else {
+ view->tex_resource_words[4] |= S_030010_BASE_LEVEL(state->u.tex.first_level);
+ view->tex_resource_words[5] |= S_030014_LAST_LEVEL(state->u.tex.last_level);
+ }
/* aniso max 16 samples */
view->tex_resource_words[6] = (S_030018_MAX_ANISO(4)) |
(S_030018_TILE_SPLIT(tile_split));
return &view->base;
}
-static void evergreen_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);
-}
-
-static void evergreen_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);
-}
-
-static void evergreen_bind_samplers(struct r600_context *rctx,
- struct r600_textures_info *dst,
- unsigned count, void **states,
- void (*set_sampler)(struct r600_context *ctx, struct r600_pipe_state *state, unsigned id))
-{
- struct r600_pipe_sampler_state **rstates = (struct r600_pipe_sampler_state**)states;
-
- for (int i = 0; i < count; i++) {
- if (rstates[i] != dst->samplers[i]) {
- set_sampler(rctx, &rstates[i]->rstate, i);
- }
- }
-
- memcpy(dst->samplers, states, sizeof(void*) * count);
- dst->n_samplers = count;
-}
-
-static void evergreen_bind_ps_samplers(struct pipe_context *ctx, unsigned count, void **states)
-{
- struct r600_context *rctx = (struct r600_context *)ctx;
- evergreen_bind_samplers(rctx, &rctx->ps_samplers, count, states,
- evergreen_context_pipe_state_set_ps_sampler);
-}
-
-static void evergreen_bind_vs_samplers(struct pipe_context *ctx, unsigned count, void **states)
-{
- struct r600_context *rctx = (struct r600_context *)ctx;
- evergreen_bind_samplers(rctx, &rctx->vs_samplers, count, states,
- evergreen_context_pipe_state_set_vs_sampler);
-}
-
static void evergreen_set_clip_state(struct pipe_context *ctx,
const struct pipe_clip_state *state)
{
cb.user_buffer = state->ucp;
cb.buffer_offset = 0;
cb.buffer_size = 4*4*8;
- r600_set_constant_buffer(ctx, PIPE_SHADER_VERTEX, 1, &cb);
+ ctx->set_constant_buffer(ctx, PIPE_SHADER_VERTEX, 1, &cb);
pipe_resource_reference(&cb.buffer, NULL);
}
{
}
-static void evergreen_set_sample_mask(struct pipe_context *pipe, unsigned sample_mask)
-{
-}
-
static void evergreen_get_scissor_rect(struct r600_context *rctx,
unsigned tl_x, unsigned tl_y, unsigned br_x, unsigned br_y,
uint32_t *tl, uint32_t *br)
r600_context_pipe_state_set(rctx, rstate);
}
-void evergreen_cb(struct r600_context *rctx, struct r600_pipe_state *rstate,
- const struct pipe_framebuffer_state *state, int cb)
+void evergreen_init_color_surface(struct r600_context *rctx,
+ struct r600_surface *surf)
{
struct r600_screen *rscreen = rctx->screen;
- struct r600_resource_texture *rtex;
- struct pipe_resource * pipe_tex;
- struct r600_surface *surf;
- unsigned level = state->cbufs[cb]->u.tex.level;
+ struct r600_texture *rtex = (struct r600_texture*)surf->base.texture;
+ struct pipe_resource *pipe_tex = surf->base.texture;
+ unsigned level = surf->base.u.tex.level;
unsigned pitch, slice;
unsigned color_info, color_attrib, color_dim = 0;
unsigned format, swap, ntype, endian;
- uint64_t offset;
- unsigned tile_type, macro_aspect, tile_split, bankh, bankw, nbanks;
+ uint64_t offset, base_offset;
+ unsigned tile_type, macro_aspect, tile_split, bankh, bankw, fmask_bankh, nbanks;
const struct util_format_description *desc;
int i;
- bool blend_clamp = 0, blend_bypass = 0, alphatest_bypass;
-
- surf = (struct r600_surface *)state->cbufs[cb];
- rtex = (struct r600_resource_texture*)state->cbufs[cb]->texture;
- pipe_tex = state->cbufs[cb]->texture;
+ bool blend_clamp = 0, blend_bypass = 0;
if (rtex->is_depth && !rtex->is_flushing_texture) {
- r600_init_flushed_depth_texture(&rctx->context,
- state->cbufs[cb]->texture, NULL);
+ r600_init_flushed_depth_texture(&rctx->context, pipe_tex, NULL);
rtex = rtex->flushed_depth_texture;
assert(rtex);
}
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;
+ 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;
macro_aspect = rtex->surface.mtilea;
bankw = rtex->surface.bankw;
bankh = rtex->surface.bankh;
+ fmask_bankh = rtex->fmask_bank_height;
tile_split = eg_tile_split(tile_split);
macro_aspect = eg_macro_tile_aspect(macro_aspect);
bankw = eg_bank_wh(bankw);
bankh = eg_bank_wh(bankh);
+ fmask_bankh = eg_bank_wh(fmask_bankh);
/* 128 bit formats require tile type = 1 */
if (rscreen->chip_class == CAYMAN) {
S_028C74_BANK_WIDTH(bankw) |
S_028C74_BANK_HEIGHT(bankh) |
S_028C74_MACRO_TILE_ASPECT(macro_aspect) |
- S_028C74_NON_DISP_TILING_ORDER(tile_type);
+ S_028C74_NON_DISP_TILING_ORDER(tile_type) |
+ S_028C74_FMASK_BANK_HEIGHT(fmask_bankh);
+
+ if (rctx->chip_class == CAYMAN && rtex->resource.b.b.nr_samples > 1) {
+ unsigned log_samples = util_logbase2(rtex->resource.b.b.nr_samples);
+ color_attrib |= S_028C74_NUM_SAMPLES(log_samples) |
+ S_028C74_NUM_FRAGMENTS(log_samples);
+ }
ntype = V_028C70_NUMBER_UNORM;
if (desc->colorspace == UTIL_FORMAT_COLORSPACE_SRGB)
blend_bypass = 1;
}
- /* Alpha-test is done on the first colorbuffer only. */
- if (cb == 0) {
- alphatest_bypass = ntype == V_028C70_NUMBER_UINT || ntype == V_028C70_NUMBER_SINT;
- if (rctx->alphatest_state.bypass != alphatest_bypass) {
- rctx->alphatest_state.bypass = alphatest_bypass;
- r600_atom_dirty(rctx, &rctx->alphatest_state.atom);
- }
- }
+ surf->alphatest_bypass = ntype == V_028C70_NUMBER_UINT || ntype == V_028C70_NUMBER_SINT;
color_info |= S_028C70_FORMAT(format) |
S_028C70_COMP_SWAP(swap) |
* - 11-bit or smaller UNORM/SNORM/SRGB
* - 16-bit or smaller FLOAT
*/
- /* XXX: This should probably be the same for all CBs if we want
- * useful alpha tests. */
if (desc->colorspace != UTIL_FORMAT_COLORSPACE_ZS &&
((desc->channel[i].size < 12 &&
desc->channel[i].type != UTIL_FORMAT_TYPE_FLOAT &&
(desc->channel[i].size < 17 &&
desc->channel[i].type == UTIL_FORMAT_TYPE_FLOAT))) {
color_info |= S_028C70_SOURCE_FORMAT(V_028C70_EXPORT_4C_16BPC);
- } else {
- rctx->export_16bpc = false;
- }
-
- /* Alpha-test is done on the first colorbuffer only. */
- if (cb == 0 && rctx->alphatest_state.cb0_export_16bpc != rctx->export_16bpc) {
- rctx->alphatest_state.cb0_export_16bpc = rctx->export_16bpc;
- r600_atom_dirty(rctx, &rctx->alphatest_state.atom);
+ surf->export_16bpc = true;
}
- /* for possible dual-src MRT */
- if (cb == 0 && rctx->framebuffer.nr_cbufs == 1 && !rtex->is_rat) {
- r600_pipe_state_add_reg_bo(rstate,
- R_028C70_CB_COLOR0_INFO + 1 * 0x3C,
- color_info, &rtex->resource, RADEON_USAGE_READWRITE);
+ if (rtex->fmask_size && rtex->cmask_size) {
+ color_info |= S_028C70_COMPRESSION(1) | S_028C70_FAST_CLEAR(1);
}
- offset += r600_resource_va(rctx->context.screen, state->cbufs[cb]->texture);
- offset >>= 8;
+ base_offset = r600_resource_va(rctx->context.screen, pipe_tex);
/* XXX handle enabling of CB beyond BASE8 which has different offset */
- r600_pipe_state_add_reg_bo(rstate,
- R_028C60_CB_COLOR0_BASE + cb * 0x3C,
- offset, &rtex->resource, RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg(rstate,
- R_028C78_CB_COLOR0_DIM + cb * 0x3C,
- color_dim);
- r600_pipe_state_add_reg_bo(rstate,
- R_028C70_CB_COLOR0_INFO + cb * 0x3C,
- color_info, &rtex->resource, RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg(rstate,
- R_028C64_CB_COLOR0_PITCH + cb * 0x3C,
- S_028C64_PITCH_TILE_MAX(pitch));
- r600_pipe_state_add_reg(rstate,
- R_028C68_CB_COLOR0_SLICE + cb * 0x3C,
- S_028C68_SLICE_TILE_MAX(slice));
+ surf->cb_color_base = (base_offset + offset) >> 8;
+ surf->cb_color_dim = color_dim;
+ surf->cb_color_info = color_info;
+ surf->cb_color_pitch = S_028C64_PITCH_TILE_MAX(pitch);
+ surf->cb_color_slice = S_028C68_SLICE_TILE_MAX(slice);
if (rtex->surface.level[level].mode < RADEON_SURF_MODE_1D) {
- r600_pipe_state_add_reg(rstate,
- R_028C6C_CB_COLOR0_VIEW + cb * 0x3C,
- 0x00000000);
+ surf->cb_color_view = 0;
} else {
- r600_pipe_state_add_reg(rstate,
- R_028C6C_CB_COLOR0_VIEW + cb * 0x3C,
- S_028C6C_SLICE_START(state->cbufs[cb]->u.tex.first_layer) |
- S_028C6C_SLICE_MAX(state->cbufs[cb]->u.tex.last_layer));
+ surf->cb_color_view = S_028C6C_SLICE_START(surf->base.u.tex.first_layer) |
+ S_028C6C_SLICE_MAX(surf->base.u.tex.last_layer);
}
- r600_pipe_state_add_reg_bo(rstate,
- R_028C74_CB_COLOR0_ATTRIB + cb * 0x3C,
- color_attrib,
- &rtex->resource, RADEON_USAGE_READWRITE);
+ surf->cb_color_attrib = color_attrib;
+ if (rtex->fmask_size && rtex->cmask_size) {
+ surf->cb_color_fmask = (base_offset + rtex->fmask_offset) >> 8;
+ surf->cb_color_cmask = (base_offset + rtex->cmask_offset) >> 8;
+ } else {
+ surf->cb_color_fmask = surf->cb_color_base;
+ surf->cb_color_cmask = surf->cb_color_base;
+ }
+ surf->cb_color_fmask_slice = S_028C88_TILE_MAX(slice);
+ surf->cb_color_cmask_slice = S_028C80_TILE_MAX(rtex->cmask_slice_tile_max);
+
+ surf->color_initialized = true;
}
-static void evergreen_db(struct r600_context *rctx, struct r600_pipe_state *rstate,
- const struct pipe_framebuffer_state *state)
+static void evergreen_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 pipe_screen *screen = &rscreen->screen;
+ struct r600_texture *rtex = (struct r600_texture*)surf->base.texture;
uint64_t offset;
unsigned level, pitch, slice, format, array_mode;
- unsigned macro_aspect, tile_split, bankh, bankw, z_info, nbanks;
-
- if (state->zsbuf == NULL)
- return;
+ unsigned macro_aspect, tile_split, bankh, bankw, nbanks;
- surf = (struct r600_surface *)state->zsbuf;
level = surf->base.u.tex.level;
- rtex = (struct r600_resource_texture*)surf->base.texture;
format = r600_translate_dbformat(surf->base.format);
assert(format != ~0);
- offset = r600_resource_va(rctx->context.screen, surf->base.texture);
+ offset = r600_resource_va(screen, surf->base.texture);
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;
nbanks = eg_num_banks(rscreen->tiling_info.num_banks);
offset >>= 8;
- z_info = S_028040_ARRAY_MODE(array_mode) |
- S_028040_FORMAT(format) |
- S_028040_TILE_SPLIT(tile_split)|
- S_028040_NUM_BANKS(nbanks) |
- S_028040_BANK_WIDTH(bankw) |
- S_028040_BANK_HEIGHT(bankh) |
- S_028040_MACRO_TILE_ASPECT(macro_aspect);
-
- r600_pipe_state_add_reg_bo(rstate, R_028048_DB_Z_READ_BASE,
- offset, &rtex->resource, RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg_bo(rstate, R_028050_DB_Z_WRITE_BASE,
- offset, &rtex->resource, RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg(rstate, R_028008_DB_DEPTH_VIEW,
- S_028008_SLICE_START(state->zsbuf->u.tex.first_layer) |
- S_028008_SLICE_MAX(state->zsbuf->u.tex.last_layer));
+ surf->db_depth_info = S_028040_ARRAY_MODE(array_mode) |
+ S_028040_FORMAT(format) |
+ S_028040_TILE_SPLIT(tile_split)|
+ S_028040_NUM_BANKS(nbanks) |
+ S_028040_BANK_WIDTH(bankw) |
+ S_028040_BANK_HEIGHT(bankh) |
+ S_028040_MACRO_TILE_ASPECT(macro_aspect);
+ if (rscreen->chip_class == CAYMAN && rtex->resource.b.b.nr_samples > 1) {
+ surf->db_depth_info |= S_028040_NUM_SAMPLES(util_logbase2(rtex->resource.b.b.nr_samples));
+ }
+ surf->db_depth_base = offset;
+ surf->db_depth_view = S_028008_SLICE_START(surf->base.u.tex.first_layer) |
+ S_028008_SLICE_MAX(surf->base.u.tex.last_layer);
+ surf->db_depth_size = S_028058_PITCH_TILE_MAX(pitch);
+ surf->db_depth_slice = S_02805C_SLICE_TILE_MAX(slice);
if (rtex->surface.flags & RADEON_SURF_SBUFFER) {
uint64_t stencil_offset = rtex->surface.stencil_offset;
unsigned stile_split = rtex->surface.stencil_tile_split;
stile_split = eg_tile_split(stile_split);
- stencil_offset += r600_resource_va(rctx->context.screen, surf->base.texture);
+ stencil_offset += r600_resource_va(screen, surf->base.texture);
stencil_offset += rtex->surface.level[level].offset / 4;
stencil_offset >>= 8;
- r600_pipe_state_add_reg_bo(rstate, R_02804C_DB_STENCIL_READ_BASE,
- stencil_offset, &rtex->resource,
- RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg_bo(rstate, R_028054_DB_STENCIL_WRITE_BASE,
- stencil_offset, &rtex->resource,
- RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg_bo(rstate, R_028044_DB_STENCIL_INFO,
- 1 | S_028044_TILE_SPLIT(stile_split),
- &rtex->resource,
- RADEON_USAGE_READWRITE);
+ surf->db_stencil_base = stencil_offset;
+ surf->db_stencil_info = 1 | S_028044_TILE_SPLIT(stile_split);
} else {
- r600_pipe_state_add_reg_bo(rstate, R_02804C_DB_STENCIL_READ_BASE,
- offset, &rtex->resource,
- RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg_bo(rstate, R_028054_DB_STENCIL_WRITE_BASE,
- offset, &rtex->resource,
- RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg_bo(rstate, R_028044_DB_STENCIL_INFO,
- 1, NULL, RADEON_USAGE_READWRITE);
- }
-
- r600_pipe_state_add_reg_bo(rstate, R_028040_DB_Z_INFO, z_info,
- &rtex->resource, RADEON_USAGE_READWRITE);
- r600_pipe_state_add_reg(rstate, R_028058_DB_DEPTH_SIZE,
- S_028058_PITCH_TILE_MAX(pitch));
- r600_pipe_state_add_reg(rstate, R_02805C_DB_DEPTH_SLICE,
- S_02805C_SLICE_TILE_MAX(slice));
+ surf->db_stencil_base = offset;
+ surf->db_stencil_info = 1;
+ }
+
+ surf->depth_initialized = true;
+}
+
+#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))
+
+static uint32_t evergreen_set_ms_pos(struct pipe_context *ctx, struct r600_pipe_state *rstate, int nsample)
+{
+ /* 2xMSAA
+ * There are two locations (-4, 4), (4, -4). */
+ 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),
+ 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;
+ /* 4xMSAA
+ * There are 4 locations: (-2, -2), (2, 2), (-6, 6), (6, -6). */
+ 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),
+ 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;
+ /* 8xMSAA */
+ 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),
+ FILL_SREG(-2, -5, 3, -4, -1, 5, -6, -2),
+ FILL_SREG( 6, 0, 0, 0, -5, 3, 4, 4),
+ FILL_SREG(-2, -5, 3, -4, -1, 5, -6, -2),
+ FILL_SREG( 6, 0, 0, 0, -5, 3, 4, 4),
+ 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;
+ unsigned i;
+
+ switch (nsample) {
+ case 2:
+ for (i = 0; i < Elements(sample_locs_2x); i++) {
+ r600_pipe_state_add_reg(rstate, R_028C1C_PA_SC_AA_SAMPLE_LOCS_0 + i*4,
+ sample_locs_2x[i]);
+ }
+ return max_dist_2x;
+ case 4:
+ for (i = 0; i < Elements(sample_locs_4x); i++) {
+ r600_pipe_state_add_reg(rstate, R_028C1C_PA_SC_AA_SAMPLE_LOCS_0 + i*4,
+ sample_locs_4x[i]);
+ }
+ return max_dist_4x;
+ case 8:
+ for (i = 0; i < Elements(sample_locs_8x); i++) {
+ r600_pipe_state_add_reg(rstate, R_028C1C_PA_SC_AA_SAMPLE_LOCS_0 + i*4,
+ sample_locs_8x[i]);
+ }
+ return max_dist_8x;
+ default:
+ R600_ERR("Invalid nr_samples %i\n", nsample);
+ return 0;
+ }
+}
+
+static uint32_t cayman_set_ms_pos(struct pipe_context *ctx, struct r600_pipe_state *rstate, int nsample)
+{
+ /* 2xMSAA
+ * There are two locations (-4, 4), (4, -4). */
+ 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),
+ 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;
+ /* 4xMSAA
+ * There are 4 locations: (-2, -2), (2, 2), (-6, 6), (6, -6). */
+ 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),
+ 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;
+ /* 8xMSAA */
+ static uint32_t sample_locs_8x[] = {
+ FILL_SREG(-2, -5, 3, -4, -1, 5, -6, -2),
+ FILL_SREG(-2, -5, 3, -4, -1, 5, -6, -2),
+ FILL_SREG(-2, -5, 3, -4, -1, 5, -6, -2),
+ FILL_SREG(-2, -5, 3, -4, -1, 5, -6, -2),
+ FILL_SREG( 6, 0, 0, 0, -5, 3, 4, 4),
+ FILL_SREG( 6, 0, 0, 0, -5, 3, 4, 4),
+ FILL_SREG( 6, 0, 0, 0, -5, 3, 4, 4),
+ FILL_SREG( 6, 0, 0, 0, -5, 3, 4, 4),
+ };
+ static unsigned max_dist_8x = 8;
+ /* 16xMSAA */
+ static uint32_t sample_locs_16x[] = {
+ FILL_SREG(-7, -3, 7, 3, 1, -5, -5, 5),
+ FILL_SREG(-7, -3, 7, 3, 1, -5, -5, 5),
+ FILL_SREG(-7, -3, 7, 3, 1, -5, -5, 5),
+ FILL_SREG(-7, -3, 7, 3, 1, -5, -5, 5),
+ FILL_SREG(-3, -7, 3, 7, 5, -1, -1, 1),
+ FILL_SREG(-3, -7, 3, 7, 5, -1, -1, 1),
+ FILL_SREG(-3, -7, 3, 7, 5, -1, -1, 1),
+ FILL_SREG(-3, -7, 3, 7, 5, -1, -1, 1),
+ FILL_SREG(-8, -6, 4, 2, 2, -8, -2, 6),
+ FILL_SREG(-8, -6, 4, 2, 2, -8, -2, 6),
+ FILL_SREG(-8, -6, 4, 2, 2, -8, -2, 6),
+ FILL_SREG(-8, -6, 4, 2, 2, -8, -2, 6),
+ FILL_SREG(-4, -2, 0, 4, 6, -4, -6, 0),
+ FILL_SREG(-4, -2, 0, 4, 6, -4, -6, 0),
+ FILL_SREG(-4, -2, 0, 4, 6, -4, -6, 0),
+ FILL_SREG(-4, -2, 0, 4, 6, -4, -6, 0),
+ };
+ static unsigned max_dist_16x = 8;
+ struct r600_context *rctx = (struct r600_context *)ctx;
+ uint32_t max_dist, num_regs, *sample_locs;
+
+ switch (nsample) {
+ case 2:
+ sample_locs = sample_locs_2x;
+ num_regs = Elements(sample_locs_2x);
+ max_dist = max_dist_2x;
+ break;
+ case 4:
+ sample_locs = sample_locs_4x;
+ num_regs = Elements(sample_locs_4x);
+ max_dist = max_dist_4x;
+ break;
+ case 8:
+ sample_locs = sample_locs_8x;
+ num_regs = Elements(sample_locs_8x);
+ max_dist = max_dist_8x;
+ break;
+ case 16:
+ sample_locs = sample_locs_16x;
+ num_regs = Elements(sample_locs_16x);
+ max_dist = max_dist_16x;
+ break;
+ default:
+ R600_ERR("Invalid nr_samples %i\n", nsample);
+ return 0;
+ }
+
+ r600_pipe_state_add_reg(rstate, CM_R_028BF8_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0, sample_locs[0]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C08_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0, sample_locs[1]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C18_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0, sample_locs[2]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C28_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0, sample_locs[3]);
+ if (num_regs <= 8) {
+ r600_pipe_state_add_reg(rstate, CM_R_028BFC_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1, sample_locs[4]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C0C_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1, sample_locs[5]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C1C_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1, sample_locs[6]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C2C_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1, sample_locs[7]);
+ }
+ if (num_regs <= 16) {
+ r600_pipe_state_add_reg(rstate, CM_R_028C00_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2, sample_locs[8]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C10_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2, sample_locs[9]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C20_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2, sample_locs[10]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C30_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2, sample_locs[11]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C04_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3, sample_locs[12]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C14_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3, sample_locs[13]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C24_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3, sample_locs[14]);
+ r600_pipe_state_add_reg(rstate, CM_R_028C34_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3, sample_locs[15]);
+ }
+ return max_dist;
}
static void evergreen_set_framebuffer_state(struct pipe_context *ctx,
- const struct pipe_framebuffer_state *state)
+ const struct pipe_framebuffer_state *state)
{
struct r600_context *rctx = (struct r600_context *)ctx;
struct r600_pipe_state *rstate = CALLOC_STRUCT(r600_pipe_state);
- uint32_t tl, br;
- int i;
+ struct r600_surface *surf;
+ struct r600_resource *res;
+ struct r600_texture *rtex;
+ uint32_t tl, br, i, nr_samples, log_samples;
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;
+ }
/* unreference old buffer and reference new one */
rstate->id = R600_PIPE_STATE_FRAMEBUFFER;
util_copy_framebuffer_state(&rctx->framebuffer, state);
- /* build states */
+ /* 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++) {
- evergreen_cb(rctx, rstate, state, i);
+ surf = (struct r600_surface*)state->cbufs[i];
+ res = (struct r600_resource*)surf->base.texture;
+ rtex = (struct r600_texture*)res;
+
+ if (!surf->color_initialized) {
+ evergreen_init_color_surface(rctx, surf);
+ }
+
+ if (!surf->export_16bpc) {
+ rctx->export_16bpc = false;
+ }
+
+ r600_pipe_state_add_reg_bo(rstate, R_028C60_CB_COLOR0_BASE + i * 0x3C,
+ surf->cb_color_base, res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg(rstate, R_028C78_CB_COLOR0_DIM + i * 0x3C,
+ surf->cb_color_dim);
+ r600_pipe_state_add_reg_bo(rstate, R_028C70_CB_COLOR0_INFO + i * 0x3C,
+ surf->cb_color_info, res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg(rstate, R_028C64_CB_COLOR0_PITCH + i * 0x3C,
+ surf->cb_color_pitch);
+ r600_pipe_state_add_reg(rstate, R_028C68_CB_COLOR0_SLICE + i * 0x3C,
+ surf->cb_color_slice);
+ r600_pipe_state_add_reg(rstate, R_028C6C_CB_COLOR0_VIEW + i * 0x3C,
+ surf->cb_color_view);
+ r600_pipe_state_add_reg_bo(rstate, R_028C74_CB_COLOR0_ATTRIB + i * 0x3C,
+ surf->cb_color_attrib, res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg_bo(rstate, R_028C7C_CB_COLOR0_CMASK + i * 0x3c,
+ surf->cb_color_cmask, res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg(rstate, R_028C80_CB_COLOR0_CMASK_SLICE + i * 0x3c,
+ surf->cb_color_cmask_slice);
+ r600_pipe_state_add_reg_bo(rstate, R_028C84_CB_COLOR0_FMASK + i * 0x3c,
+ surf->cb_color_fmask, res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg(rstate, R_028C88_CB_COLOR0_FMASK_SLICE + i * 0x3c,
+ surf->cb_color_fmask_slice);
+
+ /* Cayman can fetch from a compressed MSAA colorbuffer,
+ * so it's pointless to track them. */
+ if (rctx->chip_class != CAYMAN && rtex->fmask_size && rtex->cmask_size) {
+ rctx->compressed_cb_mask |= 1 << i;
+ }
}
- /* CB_COLOR1_INFO is already initialized for possible dual-src blending */
- if (i == 1)
+ /* set CB_COLOR1_INFO for possible dual-src blending */
+ if (i == 1 && !((struct r600_texture*)res)->is_rat) {
+ r600_pipe_state_add_reg_bo(rstate, R_028C70_CB_COLOR0_INFO + 1 * 0x3C,
+ surf->cb_color_info, res, RADEON_USAGE_READWRITE);
i++;
+ }
for (; i < 8 ; i++) {
r600_pipe_state_add_reg(rstate, R_028C70_CB_COLOR0_INFO + i * 0x3C, 0);
}
+ /* 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);
+ }
+ if (rctx->alphatest_state.cb0_export_16bpc != surf->export_16bpc) {
+ rctx->alphatest_state.cb0_export_16bpc = surf->export_16bpc;
+ r600_atom_dirty(rctx, &rctx->alphatest_state.atom);
+ }
+ }
+
+ /* ZS buffer. */
if (state->zsbuf) {
- evergreen_db(rctx, rstate, state);
+ surf = (struct r600_surface*)state->zsbuf;
+ res = (struct r600_resource*)surf->base.texture;
+
+ if (!surf->depth_initialized) {
+ evergreen_init_depth_surface(rctx, surf);
+ }
+
+ r600_pipe_state_add_reg_bo(rstate, R_028048_DB_Z_READ_BASE, surf->db_depth_base,
+ res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg_bo(rstate, R_028050_DB_Z_WRITE_BASE, surf->db_depth_base,
+ res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg(rstate, R_028008_DB_DEPTH_VIEW, surf->db_depth_view);
+
+ r600_pipe_state_add_reg_bo(rstate, R_02804C_DB_STENCIL_READ_BASE, surf->db_stencil_base,
+ res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg_bo(rstate, R_028054_DB_STENCIL_WRITE_BASE, surf->db_stencil_base,
+ res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg_bo(rstate, R_028044_DB_STENCIL_INFO, surf->db_stencil_info,
+ res, RADEON_USAGE_READWRITE);
+
+ r600_pipe_state_add_reg_bo(rstate, R_028040_DB_Z_INFO, surf->db_depth_info,
+ res, RADEON_USAGE_READWRITE);
+ r600_pipe_state_add_reg(rstate, R_028058_DB_DEPTH_SIZE, surf->db_depth_size);
+ r600_pipe_state_add_reg(rstate, R_02805C_DB_DEPTH_SLICE, surf->db_depth_slice);
}
+ /* Framebuffer dimensions. */
evergreen_get_scissor_rect(rctx, 0, 0, state->width, state->height, &tl, &br);
r600_pipe_state_add_reg(rstate,
r600_pipe_state_add_reg(rstate,
R_028208_PA_SC_WINDOW_SCISSOR_BR, br);
+ /* 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;
+
+ if (nr_samples > 1) {
+ unsigned line_cntl = S_028C00_LAST_PIXEL(1) |
+ S_028C00_EXPAND_LINE_WIDTH(1);
+ log_samples = util_logbase2(nr_samples);
+
+ if (rctx->chip_class == CAYMAN) {
+ unsigned max_dist = cayman_set_ms_pos(ctx, rstate, nr_samples);
+
+ r600_pipe_state_add_reg(rstate, CM_R_028BDC_PA_SC_LINE_CNTL, line_cntl);
+ r600_pipe_state_add_reg(rstate, CM_R_028BE0_PA_SC_AA_CONFIG,
+ S_028BE0_MSAA_NUM_SAMPLES(log_samples) |
+ S_028BE0_MAX_SAMPLE_DIST(max_dist) |
+ S_028BE0_MSAA_EXPOSED_SAMPLES(log_samples));
+ r600_pipe_state_add_reg(rstate, CM_R_028804_DB_EQAA,
+ S_028804_MAX_ANCHOR_SAMPLES(log_samples) |
+ S_028804_PS_ITER_SAMPLES(log_samples) |
+ S_028804_MASK_EXPORT_NUM_SAMPLES(log_samples) |
+ S_028804_ALPHA_TO_MASK_NUM_SAMPLES(log_samples) |
+ S_028804_HIGH_QUALITY_INTERSECTIONS(1) |
+ S_028804_STATIC_ANCHOR_ASSOCIATIONS(1));
+ } else {
+ unsigned max_dist = evergreen_set_ms_pos(ctx, rstate, nr_samples);
+
+ r600_pipe_state_add_reg(rstate, R_028C00_PA_SC_LINE_CNTL, line_cntl);
+ 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 {
+ log_samples = 0;
+
+ if (rctx->chip_class == CAYMAN) {
+ r600_pipe_state_add_reg(rstate, CM_R_028BDC_PA_SC_LINE_CNTL, S_028C00_LAST_PIXEL(1));
+ r600_pipe_state_add_reg(rstate, CM_R_028BE0_PA_SC_AA_CONFIG, 0);
+ r600_pipe_state_add_reg(rstate, CM_R_028804_DB_EQAA,
+ S_028804_HIGH_QUALITY_INTERSECTIONS(1) |
+ S_028804_STATIC_ANCHOR_ASSOCIATIONS(1));
+
+ } 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;
r600_context_pipe_state_set(rctx, rstate);
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);
+ }
+
+ if (rctx->chip_class == CAYMAN && rctx->db_misc_state.log_samples != log_samples) {
+ rctx->db_misc_state.log_samples = log_samples;
+ r600_atom_dirty(rctx, &rctx->db_misc_state.atom);
+ }
}
static void evergreen_emit_cb_misc_state(struct r600_context *rctx, struct r600_atom *atom)
if (a->occlusion_query_enabled) {
db_count_control |= S_028004_PERFECT_ZPASS_COUNTS(1);
+ if (rctx->chip_class == CAYMAN) {
+ db_count_control |= S_028004_SAMPLE_RATE(a->log_samples);
+ }
db_render_override |= S_02800C_NOOP_CULL_DISABLE(1);
}
db_render_control |= S_028000_DEPTH_COPY_ENABLE(a->copy_depth) |
S_028000_STENCIL_COPY_ENABLE(a->copy_stencil) |
- S_028000_COPY_CENTROID(1);
+ S_028000_COPY_CENTROID(1) |
+ S_028000_COPY_SAMPLE(a->copy_sample);
}
r600_write_context_reg_seq(cs, R_028000_DB_RENDER_CONTROL, 2);
static void evergreen_emit_vs_constant_buffers(struct r600_context *rctx, struct r600_atom *atom)
{
- evergreen_emit_constant_buffers(rctx, &rctx->vs_constbuf_state, 176,
+ evergreen_emit_constant_buffers(rctx, &rctx->constbuf_state[PIPE_SHADER_VERTEX], 176,
R_028180_ALU_CONST_BUFFER_SIZE_VS_0,
R_028980_ALU_CONST_CACHE_VS_0);
}
static void evergreen_emit_ps_constant_buffers(struct r600_context *rctx, struct r600_atom *atom)
{
- evergreen_emit_constant_buffers(rctx, &rctx->ps_constbuf_state, 0,
+ evergreen_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);
}
evergreen_emit_sampler_views(rctx, &rctx->ps_samplers.views, R600_MAX_CONST_BUFFERS);
}
+static void evergreen_emit_sampler_states(struct r600_context *rctx,
+ struct r600_textures_info *texinfo,
+ unsigned resource_id_base,
+ unsigned border_index_reg)
+{
+ struct radeon_winsys_cs *cs = rctx->cs;
+ unsigned i;
+
+ for (i = 0; i < texinfo->n_samplers; i++) {
+
+ if (texinfo->samplers[i] == NULL) {
+ continue;
+ }
+ 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) {
+ r600_write_config_reg_seq(cs, border_index_reg, 5);
+ r600_write_value(cs, i);
+ r600_write_array(cs, 4, texinfo->samplers[i]->border_color);
+ }
+ }
+}
+
+static void evergreen_emit_vs_sampler_states(struct r600_context *rctx, struct r600_atom *atom)
+{
+ evergreen_emit_sampler_states(rctx, &rctx->vs_samplers, 18, R_00A414_TD_VS_SAMPLER0_BORDER_INDEX);
+}
+
+static void evergreen_emit_ps_sampler_states(struct r600_context *rctx, struct r600_atom *atom)
+{
+ evergreen_emit_sampler_states(rctx, &rctx->ps_samplers, 0, R_00A400_TD_PS_SAMPLER0_BORDER_INDEX);
+}
+
+static void evergreen_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_028C3C_PA_SC_AA_MASK,
+ mask | (mask << 8) | (mask << 16) | (mask << 24));
+}
+
+static void cayman_emit_sample_mask(struct r600_context *rctx, struct r600_atom *a)
+{
+ struct r600_sample_mask *s = (struct r600_sample_mask*)a;
+ struct radeon_winsys_cs *cs = rctx->cs;
+ uint16_t mask = s->sample_mask;
+
+ r600_write_context_reg_seq(cs, CM_R_028C38_PA_SC_AA_MASK_X0Y0_X1Y0, 2);
+ r600_write_value(cs, mask | (mask << 16)); /* X0Y0_X1Y0 */
+ r600_write_value(cs, mask | (mask << 16)); /* X0Y1_X1Y1 */
+}
+
void evergreen_init_state_functions(struct r600_context *rctx)
{
- r600_init_atom(&rctx->cb_misc_state.atom, evergreen_emit_cb_misc_state, 0, 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++, evergreen_emit_vs_constant_buffers, 0);
+ r600_init_atom(rctx, &rctx->constbuf_state[PIPE_SHADER_FRAGMENT].atom, id++, evergreen_emit_ps_constant_buffers, 0);
+ /* shader program */
+ r600_init_atom(rctx, &rctx->cs_shader_state.atom, id++, evergreen_emit_cs_shader, 0);
+ /* sampler */
+ r600_init_atom(rctx, &rctx->vs_samplers.atom_sampler, id++, evergreen_emit_vs_sampler_states, 0);
+ r600_init_atom(rctx, &rctx->ps_samplers.atom_sampler, id++, evergreen_emit_ps_sampler_states, 0);
+ /* resources */
+ r600_init_atom(rctx, &rctx->vertex_buffer_state.atom, id++, evergreen_fs_emit_vertex_buffers, 0);
+ r600_init_atom(rctx, &rctx->cs_vertex_buffer_state.atom, id++, evergreen_cs_emit_vertex_buffers, 0);
+ r600_init_atom(rctx, &rctx->vs_samplers.views.atom, id++, evergreen_emit_vs_sampler_views, 0);
+ r600_init_atom(rctx, &rctx->ps_samplers.views.atom, id++, evergreen_emit_ps_sampler_views, 0);
+
+ if (rctx->chip_class == EVERGREEN) {
+ r600_init_atom(rctx, &rctx->sample_mask.atom, id++, evergreen_emit_sample_mask, 3);
+ } else {
+ r600_init_atom(rctx, &rctx->sample_mask.atom, id++, cayman_emit_sample_mask, 4);
+ }
+ rctx->sample_mask.sample_mask = ~0;
+ r600_atom_dirty(rctx, &rctx->sample_mask.atom);
+
+ r600_init_atom(rctx, &rctx->cb_misc_state.atom, id++, evergreen_emit_cb_misc_state, 0);
r600_atom_dirty(rctx, &rctx->cb_misc_state.atom);
- r600_init_atom(&rctx->db_misc_state.atom, evergreen_emit_db_misc_state, 7, 0);
+
+ 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++, evergreen_emit_db_misc_state, 7);
r600_atom_dirty(rctx, &rctx->db_misc_state.atom);
- r600_init_atom(&rctx->vertex_buffer_state.atom, evergreen_fs_emit_vertex_buffers, 0, 0);
- r600_init_atom(&rctx->cs_vertex_buffer_state.atom, evergreen_cs_emit_vertex_buffers, 0, 0);
- r600_init_atom(&rctx->vs_constbuf_state.atom, evergreen_emit_vs_constant_buffers, 0, 0);
- r600_init_atom(&rctx->ps_constbuf_state.atom, evergreen_emit_ps_constant_buffers, 0, 0);
- r600_init_atom(&rctx->vs_samplers.views.atom, evergreen_emit_vs_sampler_views, 0, 0);
- r600_init_atom(&rctx->ps_samplers.views.atom, evergreen_emit_ps_sampler_views, 0, 0);
- r600_init_atom(&rctx->cs_shader_state.atom, evergreen_emit_cs_shader, 0, 0);
rctx->context.create_blend_state = evergreen_create_blend_state;
rctx->context.create_depth_stencil_alpha_state = evergreen_create_dsa_state;
- rctx->context.create_fs_state = r600_create_shader_state_ps;
rctx->context.create_rasterizer_state = evergreen_create_rs_state;
rctx->context.create_sampler_state = evergreen_create_sampler_state;
rctx->context.create_sampler_view = evergreen_create_sampler_view;
- rctx->context.create_vertex_elements_state = r600_create_vertex_elements;
- rctx->context.create_vs_state = r600_create_shader_state_vs;
- 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 = evergreen_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 = evergreen_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 = evergreen_set_clip_state;
- rctx->context.set_constant_buffer = r600_set_constant_buffer;
- rctx->context.set_fragment_sampler_views = evergreen_set_ps_sampler_views;
rctx->context.set_framebuffer_state = evergreen_set_framebuffer_state;
rctx->context.set_polygon_stipple = evergreen_set_polygon_stipple;
- rctx->context.set_sample_mask = evergreen_set_sample_mask;
rctx->context.set_scissor_state = evergreen_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 = evergreen_set_vs_sampler_views;
rctx->context.set_viewport_state = evergreen_set_viewport_state;
- rctx->context.sampler_view_destroy = r600_sampler_view_destroy;
- 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;
evergreen_init_compute_state_functions(rctx);
}
{
struct r600_command_buffer *cb = &rctx->start_cs_cmd;
- r600_init_command_buffer(cb, 256, EMIT_EARLY);
+ r600_init_command_buffer(rctx, cb, 0, 256);
/* This must be first. */
r600_store_value(cb, PKT3(PKT3_CONTEXT_CONTROL, 1, 0));
r600_store_value(cb, 0); /* CM_R_0288E8_SQ_LDS_ALLOC */
r600_store_value(cb, 0); /* R_0288EC_SQ_LDS_ALLOC_PS */
- r600_store_context_reg(cb, CM_R_028804_DB_EQAA, 0x110000);
-
r600_store_context_reg_seq(cb, R_028380_SQ_VTX_SEMANTIC_0, 34);
r600_store_value(cb, 0); /* R_028380_SQ_VTX_SEMANTIC_0 */
r600_store_value(cb, 0);
r600_store_ctl_const(cb, R_03CFF0_SQ_VTX_BASE_VTX_LOC, 0);
- r600_store_context_reg_seq(cb, CM_R_028C38_PA_SC_AA_MASK_X0Y0_X1Y0, 2);
- r600_store_value(cb, ~0); /* CM_R_028C38_PA_SC_AA_MASK_X0Y0_X1Y0 */
- r600_store_value(cb, ~0); /* CM_R_028C3C_PA_SC_AA_MASK_X0Y1_X1Y1 */
-
r600_store_context_reg_seq(cb, R_028028_DB_STENCIL_CLEAR, 2);
r600_store_value(cb, 0); /* R_028028_DB_STENCIL_CLEAR */
r600_store_value(cb, 0x3F800000); /* R_02802C_DB_DEPTH_CLEAR */
r600_store_context_reg(cb, R_028230_PA_SC_EDGERULE, 0xAAAAAAAA);
r600_store_context_reg(cb, R_028818_PA_CL_VTE_CNTL, 0x0000043F);
r600_store_context_reg(cb, R_028820_PA_CL_NANINF_CNTL, 0);
- r600_store_context_reg(cb, R_028B70_DB_ALPHA_TO_MASK, 0x0000AA00);
-
- r600_store_context_reg_seq(cb, CM_R_028BDC_PA_SC_LINE_CNTL, 2);
- r600_store_value(cb, 0x00000400); /* CM_R_028BDC_PA_SC_LINE_CNTL */
- r600_store_value(cb, 0); /* CM_R_028BE0_PA_SC_AA_CONFIG */
r600_store_context_reg_seq(cb, CM_R_028BE8_PA_CL_GB_VERT_CLIP_ADJ, 4);
r600_store_value(cb, 0x3F800000); /* CM_R_028BE8_PA_CL_GB_VERT_CLIP_ADJ */
eg_store_loop_const(cb, R_03A200_SQ_LOOP_CONST_0 + (32 * 4), 0x01000FFF);
}
-void evergreen_init_atom_start_cs(struct r600_context *rctx)
+void evergreen_init_common_regs(struct r600_command_buffer *cb,
+ enum chip_class ctx_chip_class,
+ enum radeon_family ctx_family,
+ int ctx_drm_minor)
{
- struct r600_command_buffer *cb = &rctx->start_cs_cmd;
int ps_prio;
int vs_prio;
int gs_prio;
int es_prio;
- int hs_prio, cs_prio, ls_prio;
+
+ int hs_prio;
+ int cs_prio;
+ int ls_prio;
+
int num_ps_gprs;
int num_vs_gprs;
int num_gs_gprs;
int num_hs_gprs;
int num_ls_gprs;
int num_temp_gprs;
- int num_ps_threads;
- int num_vs_threads;
- int num_gs_threads;
- int num_es_threads;
- int num_hs_threads;
- int num_ls_threads;
- int num_ps_stack_entries;
- int num_vs_stack_entries;
- int num_gs_stack_entries;
- int num_es_stack_entries;
- int num_hs_stack_entries;
- int num_ls_stack_entries;
- enum radeon_family family;
- unsigned tmp;
-
- if (rctx->chip_class == CAYMAN) {
- cayman_init_atom_start_cs(rctx);
- return;
- }
- r600_init_command_buffer(cb, 256, EMIT_EARLY);
-
- /* This must be first. */
- r600_store_value(cb, PKT3(PKT3_CONTEXT_CONTROL, 1, 0));
- r600_store_value(cb, 0x80000000);
- r600_store_value(cb, 0x80000000);
+ unsigned tmp;
- family = rctx->family;
ps_prio = 0;
vs_prio = 1;
gs_prio = 2;
ls_prio = 0;
cs_prio = 0;
- switch (family) {
+ switch (ctx_family) {
case CHIP_CEDAR:
default:
num_ps_gprs = 93;
num_es_gprs = 31;
num_hs_gprs = 23;
num_ls_gprs = 23;
- num_ps_threads = 96;
- num_vs_threads = 16;
- num_gs_threads = 16;
- num_es_threads = 16;
- num_hs_threads = 16;
- num_ls_threads = 16;
- num_ps_stack_entries = 42;
- num_vs_stack_entries = 42;
- num_gs_stack_entries = 42;
- num_es_stack_entries = 42;
- num_hs_stack_entries = 42;
- num_ls_stack_entries = 42;
break;
case CHIP_REDWOOD:
num_ps_gprs = 93;
num_es_gprs = 31;
num_hs_gprs = 23;
num_ls_gprs = 23;
- num_ps_threads = 128;
- num_vs_threads = 20;
- num_gs_threads = 20;
- num_es_threads = 20;
- num_hs_threads = 20;
- num_ls_threads = 20;
- num_ps_stack_entries = 42;
- num_vs_stack_entries = 42;
- num_gs_stack_entries = 42;
- num_es_stack_entries = 42;
- num_hs_stack_entries = 42;
- num_ls_stack_entries = 42;
break;
case CHIP_JUNIPER:
num_ps_gprs = 93;
num_es_gprs = 31;
num_hs_gprs = 23;
num_ls_gprs = 23;
- num_ps_threads = 128;
- num_vs_threads = 20;
- num_gs_threads = 20;
- num_es_threads = 20;
- num_hs_threads = 20;
- num_ls_threads = 20;
- num_ps_stack_entries = 85;
- num_vs_stack_entries = 85;
- num_gs_stack_entries = 85;
- num_es_stack_entries = 85;
- num_hs_stack_entries = 85;
- num_ls_stack_entries = 85;
break;
case CHIP_CYPRESS:
case CHIP_HEMLOCK:
num_es_gprs = 31;
num_hs_gprs = 23;
num_ls_gprs = 23;
- num_ps_threads = 128;
- num_vs_threads = 20;
- num_gs_threads = 20;
- num_es_threads = 20;
- num_hs_threads = 20;
- num_ls_threads = 20;
- num_ps_stack_entries = 85;
- num_vs_stack_entries = 85;
- num_gs_stack_entries = 85;
- num_es_stack_entries = 85;
- num_hs_stack_entries = 85;
- num_ls_stack_entries = 85;
break;
case CHIP_PALM:
num_ps_gprs = 93;
num_es_gprs = 31;
num_hs_gprs = 23;
num_ls_gprs = 23;
- num_ps_threads = 96;
- num_vs_threads = 16;
- num_gs_threads = 16;
- num_es_threads = 16;
- num_hs_threads = 16;
- num_ls_threads = 16;
- num_ps_stack_entries = 42;
- num_vs_stack_entries = 42;
- num_gs_stack_entries = 42;
- num_es_stack_entries = 42;
- num_hs_stack_entries = 42;
- num_ls_stack_entries = 42;
break;
case CHIP_SUMO:
num_ps_gprs = 93;
num_es_gprs = 31;
num_hs_gprs = 23;
num_ls_gprs = 23;
- num_ps_threads = 96;
- num_vs_threads = 25;
- num_gs_threads = 25;
- num_es_threads = 25;
- num_hs_threads = 25;
- num_ls_threads = 25;
- num_ps_stack_entries = 42;
- num_vs_stack_entries = 42;
- num_gs_stack_entries = 42;
- num_es_stack_entries = 42;
- num_hs_stack_entries = 42;
- num_ls_stack_entries = 42;
break;
case CHIP_SUMO2:
num_ps_gprs = 93;
num_es_gprs = 31;
num_hs_gprs = 23;
num_ls_gprs = 23;
- num_ps_threads = 96;
- num_vs_threads = 25;
- num_gs_threads = 25;
- num_es_threads = 25;
- num_hs_threads = 25;
- num_ls_threads = 25;
- num_ps_stack_entries = 85;
- num_vs_stack_entries = 85;
- num_gs_stack_entries = 85;
- num_es_stack_entries = 85;
- num_hs_stack_entries = 85;
- num_ls_stack_entries = 85;
break;
case CHIP_BARTS:
num_ps_gprs = 93;
num_es_gprs = 31;
num_hs_gprs = 23;
num_ls_gprs = 23;
- num_ps_threads = 128;
- num_vs_threads = 20;
- num_gs_threads = 20;
- num_es_threads = 20;
- num_hs_threads = 20;
- num_ls_threads = 20;
- num_ps_stack_entries = 85;
- num_vs_stack_entries = 85;
- num_gs_stack_entries = 85;
- num_es_stack_entries = 85;
- num_hs_stack_entries = 85;
- num_ls_stack_entries = 85;
break;
case CHIP_TURKS:
num_ps_gprs = 93;
num_es_gprs = 31;
num_hs_gprs = 23;
num_ls_gprs = 23;
- num_ps_threads = 128;
- num_vs_threads = 20;
- num_gs_threads = 20;
- num_es_threads = 20;
- num_hs_threads = 20;
- num_ls_threads = 20;
- num_ps_stack_entries = 42;
- num_vs_stack_entries = 42;
- num_gs_stack_entries = 42;
- num_es_stack_entries = 42;
- num_hs_stack_entries = 42;
- num_ls_stack_entries = 42;
break;
case CHIP_CAICOS:
num_ps_gprs = 93;
num_es_gprs = 31;
num_hs_gprs = 23;
num_ls_gprs = 23;
- num_ps_threads = 128;
- num_vs_threads = 10;
- num_gs_threads = 10;
- num_es_threads = 10;
- num_hs_threads = 10;
- num_ls_threads = 10;
- num_ps_stack_entries = 42;
- num_vs_stack_entries = 42;
- num_gs_stack_entries = 42;
- num_es_stack_entries = 42;
- num_hs_stack_entries = 42;
- num_ls_stack_entries = 42;
break;
}
tmp = 0;
- switch (family) {
+ switch (ctx_family) {
case CHIP_CEDAR:
case CHIP_PALM:
case CHIP_SUMO:
tmp |= S_008C00_ES_PRIO(es_prio);
/* enable dynamic GPR resource management */
- if (rctx->screen->info.drm_minor >= 7) {
+ if (ctx_drm_minor >= 7) {
r600_store_config_reg_seq(cb, R_008C00_SQ_CONFIG, 2);
r600_store_value(cb, tmp); /* R_008C00_SQ_CONFIG */
/* always set temp clauses */
r600_store_value(cb, tmp); /* R_008C0C_SQ_GPR_RESOURCE_MGMT_3 */
}
+ r600_store_config_reg(cb, R_008E2C_SQ_LDS_RESOURCE_MGMT,
+ S_008E2C_NUM_PS_LDS(0x1000) | S_008E2C_NUM_LS_LDS(0x1000));
+
+ r600_store_context_reg(cb, R_028A4C_PA_SC_MODE_CNTL_1, 0);
+
+ r600_store_context_reg_seq(cb, R_028B94_VGT_STRMOUT_CONFIG, 2);
+ r600_store_value(cb, 0); /* R_028B94_VGT_STRMOUT_CONFIG */
+ r600_store_value(cb, 0); /* R_028B98_VGT_STRMOUT_BUFFER_CONFIG */
+
+ r600_store_context_reg(cb, R_028230_PA_SC_EDGERULE, 0xAAAAAAAA);
+
+ 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_value(cb, 0x3F800000); /* R_0282D4_PA_SC_VPORT_ZMAX_0 */
+
+ r600_store_context_reg_seq(cb, R_028AC0_DB_SRESULTS_COMPARE_STATE0, 3);
+ r600_store_value(cb, 0); /* R_028AC0_DB_SRESULTS_COMPARE_STATE0 */
+ r600_store_value(cb, 0); /* R_028AC4_DB_SRESULTS_COMPARE_STATE1 */
+ r600_store_value(cb, 0); /* R_028AC8_DB_PRELOAD_CONTROL */
+
+ r600_store_context_reg(cb, R_028848_SQ_PGM_RESOURCES_2_PS, S_028848_SINGLE_ROUND(V_SQ_ROUND_NEAREST_EVEN));
+ r600_store_context_reg(cb, R_028864_SQ_PGM_RESOURCES_2_VS, S_028864_SINGLE_ROUND(V_SQ_ROUND_NEAREST_EVEN));
+
+ /* 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(cb, R_028354_SX_SURFACE_SYNC, S_028354_SURFACE_SYNC_MASK(0xf));
+
+ return;
+}
+
+void evergreen_init_atom_start_cs(struct r600_context *rctx)
+{
+ struct r600_command_buffer *cb = &rctx->start_cs_cmd;
+ int num_ps_threads;
+ int num_vs_threads;
+ int num_gs_threads;
+ int num_es_threads;
+ int num_hs_threads;
+ int num_ls_threads;
+
+ int num_ps_stack_entries;
+ int num_vs_stack_entries;
+ int num_gs_stack_entries;
+ int num_es_stack_entries;
+ int num_hs_stack_entries;
+ int num_ls_stack_entries;
+ enum radeon_family family;
+ unsigned tmp;
+
+ if (rctx->chip_class == CAYMAN) {
+ cayman_init_atom_start_cs(rctx);
+ return;
+ }
+
+ r600_init_command_buffer(rctx, cb, 0, 256);
+
+ /* This must be first. */
+ r600_store_value(cb, PKT3(PKT3_CONTEXT_CONTROL, 1, 0));
+ r600_store_value(cb, 0x80000000);
+ r600_store_value(cb, 0x80000000);
+
+ evergreen_init_common_regs(cb, rctx->chip_class
+ , rctx->family, rctx->screen->info.drm_minor);
+
+ family = rctx->family;
+ switch (family) {
+ case CHIP_CEDAR:
+ default:
+ num_ps_threads = 96;
+ num_vs_threads = 16;
+ num_gs_threads = 16;
+ num_es_threads = 16;
+ num_hs_threads = 16;
+ num_ls_threads = 16;
+ num_ps_stack_entries = 42;
+ num_vs_stack_entries = 42;
+ num_gs_stack_entries = 42;
+ num_es_stack_entries = 42;
+ num_hs_stack_entries = 42;
+ num_ls_stack_entries = 42;
+ break;
+ case CHIP_REDWOOD:
+ num_ps_threads = 128;
+ num_vs_threads = 20;
+ num_gs_threads = 20;
+ num_es_threads = 20;
+ num_hs_threads = 20;
+ num_ls_threads = 20;
+ num_ps_stack_entries = 42;
+ num_vs_stack_entries = 42;
+ num_gs_stack_entries = 42;
+ num_es_stack_entries = 42;
+ num_hs_stack_entries = 42;
+ num_ls_stack_entries = 42;
+ break;
+ case CHIP_JUNIPER:
+ num_ps_threads = 128;
+ num_vs_threads = 20;
+ num_gs_threads = 20;
+ num_es_threads = 20;
+ num_hs_threads = 20;
+ num_ls_threads = 20;
+ num_ps_stack_entries = 85;
+ num_vs_stack_entries = 85;
+ num_gs_stack_entries = 85;
+ num_es_stack_entries = 85;
+ num_hs_stack_entries = 85;
+ num_ls_stack_entries = 85;
+ break;
+ case CHIP_CYPRESS:
+ case CHIP_HEMLOCK:
+ num_ps_threads = 128;
+ num_vs_threads = 20;
+ num_gs_threads = 20;
+ num_es_threads = 20;
+ num_hs_threads = 20;
+ num_ls_threads = 20;
+ num_ps_stack_entries = 85;
+ num_vs_stack_entries = 85;
+ num_gs_stack_entries = 85;
+ num_es_stack_entries = 85;
+ num_hs_stack_entries = 85;
+ num_ls_stack_entries = 85;
+ break;
+ case CHIP_PALM:
+ num_ps_threads = 96;
+ num_vs_threads = 16;
+ num_gs_threads = 16;
+ num_es_threads = 16;
+ num_hs_threads = 16;
+ num_ls_threads = 16;
+ num_ps_stack_entries = 42;
+ num_vs_stack_entries = 42;
+ num_gs_stack_entries = 42;
+ num_es_stack_entries = 42;
+ num_hs_stack_entries = 42;
+ num_ls_stack_entries = 42;
+ break;
+ case CHIP_SUMO:
+ num_ps_threads = 96;
+ num_vs_threads = 25;
+ num_gs_threads = 25;
+ num_es_threads = 25;
+ num_hs_threads = 25;
+ num_ls_threads = 25;
+ num_ps_stack_entries = 42;
+ num_vs_stack_entries = 42;
+ num_gs_stack_entries = 42;
+ num_es_stack_entries = 42;
+ num_hs_stack_entries = 42;
+ num_ls_stack_entries = 42;
+ break;
+ case CHIP_SUMO2:
+ num_ps_threads = 96;
+ num_vs_threads = 25;
+ num_gs_threads = 25;
+ num_es_threads = 25;
+ num_hs_threads = 25;
+ num_ls_threads = 25;
+ num_ps_stack_entries = 85;
+ num_vs_stack_entries = 85;
+ num_gs_stack_entries = 85;
+ num_es_stack_entries = 85;
+ num_hs_stack_entries = 85;
+ num_ls_stack_entries = 85;
+ break;
+ case CHIP_BARTS:
+ num_ps_threads = 128;
+ num_vs_threads = 20;
+ num_gs_threads = 20;
+ num_es_threads = 20;
+ num_hs_threads = 20;
+ num_ls_threads = 20;
+ num_ps_stack_entries = 85;
+ num_vs_stack_entries = 85;
+ num_gs_stack_entries = 85;
+ num_es_stack_entries = 85;
+ num_hs_stack_entries = 85;
+ num_ls_stack_entries = 85;
+ break;
+ case CHIP_TURKS:
+ num_ps_threads = 128;
+ num_vs_threads = 20;
+ num_gs_threads = 20;
+ num_es_threads = 20;
+ num_hs_threads = 20;
+ num_ls_threads = 20;
+ num_ps_stack_entries = 42;
+ num_vs_stack_entries = 42;
+ num_gs_stack_entries = 42;
+ num_es_stack_entries = 42;
+ num_hs_stack_entries = 42;
+ num_ls_stack_entries = 42;
+ break;
+ case CHIP_CAICOS:
+ num_ps_threads = 128;
+ num_vs_threads = 10;
+ num_gs_threads = 10;
+ num_es_threads = 10;
+ num_hs_threads = 10;
+ num_ls_threads = 10;
+ num_ps_stack_entries = 42;
+ num_vs_stack_entries = 42;
+ num_gs_stack_entries = 42;
+ num_es_stack_entries = 42;
+ num_hs_stack_entries = 42;
+ num_ls_stack_entries = 42;
+ break;
+ }
+
tmp = S_008C18_NUM_PS_THREADS(num_ps_threads);
tmp |= S_008C18_NUM_VS_THREADS(num_vs_threads);
tmp |= S_008C18_NUM_GS_THREADS(num_gs_threads);
tmp |= S_008C18_NUM_ES_THREADS(num_es_threads);
+
r600_store_config_reg_seq(cb, R_008C18_SQ_THREAD_RESOURCE_MGMT_1, 5);
r600_store_value(cb, tmp); /* R_008C18_SQ_THREAD_RESOURCE_MGMT_1 */
tmp |= S_008C28_NUM_LS_STACK_ENTRIES(num_ls_stack_entries);
r600_store_value(cb, tmp); /* R_008C28_SQ_STACK_RESOURCE_MGMT_3 */
- r600_store_config_reg(cb, R_008E2C_SQ_LDS_RESOURCE_MGMT,
- S_008E2C_NUM_PS_LDS(0x1000) | S_008E2C_NUM_LS_LDS(0x1000));
-
r600_store_config_reg(cb, R_009100_SPI_CONFIG_CNTL, 0);
r600_store_config_reg(cb, R_00913C_SPI_CONFIG_CNTL_1, S_00913C_VTX_DONE_DELAY(4));
- r600_store_context_reg(cb, R_028A4C_PA_SC_MODE_CNTL_1, 0);
-
r600_store_context_reg_seq(cb, R_028900_SQ_ESGS_RING_ITEMSIZE, 6);
r600_store_value(cb, 0); /* R_028900_SQ_ESGS_RING_ITEMSIZE */
r600_store_value(cb, 0); /* R_028904_SQ_GSVS_RING_ITEMSIZE */
r600_store_value(cb, 0); /* R_028A3C_VGT_GROUP_VECT_1_FMT_CNTL */
r600_store_value(cb, 0); /* R_028A40_VGT_GS_MODE */
- r600_store_context_reg_seq(cb, R_028B94_VGT_STRMOUT_CONFIG, 2);
- r600_store_value(cb, 0); /* R_028B94_VGT_STRMOUT_CONFIG */
- r600_store_value(cb, 0); /* R_028B98_VGT_STRMOUT_BUFFER_CONFIG */
-
r600_store_context_reg_seq(cb, R_028AB4_VGT_REUSE_OFF, 2);
r600_store_value(cb, 0); /* R_028AB4_VGT_REUSE_OFF */
r600_store_value(cb, 0); /* R_028AB8_VGT_VTX_CNT_EN */
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_context_reg(cb, R_028230_PA_SC_EDGERULE, 0xAAAAAAAA);
-
- 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_value(cb, 0x3F800000); /* R_0282D4_PA_SC_VPORT_ZMAX_0 */
r600_store_context_reg(cb, R_0286DC_SPI_FOG_CNTL, 0);
r600_store_context_reg(cb, R_028818_PA_CL_VTE_CNTL, 0x0000043F);
r600_store_value(cb, 0); /* R_028AC4_DB_SRESULTS_COMPARE_STATE1 */
r600_store_value(cb, 0); /* R_028AC8_DB_PRELOAD_CONTROL */
- r600_store_context_reg(cb, R_028B70_DB_ALPHA_TO_MASK, 0x0000AA00);
-
- r600_store_context_reg_seq(cb, R_028C00_PA_SC_LINE_CNTL, 2);
- r600_store_value(cb, 0x00000400); /* 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, 5);
+ 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_0 */
-
- r600_store_context_reg(cb, R_028C3C_PA_SC_AA_MASK, ~0);
r600_store_context_reg_seq(cb, R_028240_PA_SC_GENERIC_SCISSOR_TL, 2);
r600_store_value(cb, 0); /* R_028240_PA_SC_GENERIC_SCISSOR_TL */
r600_store_value(cb, 0); /* R_028030_PA_SC_SCREEN_SCISSOR_TL */
r600_store_value(cb, S_028034_BR_X(16384) | S_028034_BR_Y(16384)); /* R_028034_PA_SC_SCREEN_SCISSOR_BR */
- r600_store_context_reg(cb, R_028848_SQ_PGM_RESOURCES_2_PS, S_028848_SINGLE_ROUND(V_SQ_ROUND_NEAREST_EVEN));
- r600_store_context_reg(cb, R_028864_SQ_PGM_RESOURCES_2_VS, S_028864_SINGLE_ROUND(V_SQ_ROUND_NEAREST_EVEN));
r600_store_context_reg(cb, R_0288A8_SQ_PGM_RESOURCES_FS, 0);
- 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);
ve->fetch_shader, RADEON_USAGE_READ);
}
+void *evergreen_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 = evergreen_create_blend_state_mode(&rctx->context, &blend, V_028808_CB_RESOLVE);
+ return rstate;
+}
+
+void *evergreen_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 = evergreen_create_blend_state_mode(&rctx->context, &blend, V_028808_CB_DECOMPRESS);
+ return rstate;
+}
+
void *evergreen_create_db_flush_dsa(struct r600_context *rctx)
{
struct pipe_depth_stencil_alpha_state dsa = {{0}};
unsigned db_shader_control = rctx->ps_shader->current->db_shader_control |
S_02880C_DUAL_EXPORT_ENABLE(dual_export) |
- S_02880C_DB_SOURCE_FORMAT(db_source_format);
+ S_02880C_DB_SOURCE_FORMAT(db_source_format) |
+ S_02880C_ALPHA_TO_MASK_DISABLE(rctx->cb0_is_integer);
if (db_shader_control != rctx->db_shader_control) {
struct r600_pipe_state rstate;
projects / mesa.git / blobdiff