#include "util/u_helpers.h"
#include "util/u_memory.h"
-static uint32_t cik_num_banks(struct si_screen *sscreen, unsigned bpe, unsigned tile_split)
+static void si_init_atom(struct r600_atom *atom, struct r600_atom **list_elem,
+ void (*emit)(struct si_context *ctx, struct r600_atom *state),
+ unsigned num_dw)
{
- if (sscreen->b.info.cik_macrotile_mode_array_valid) {
+ atom->emit = (void*)emit;
+ atom->num_dw = num_dw;
+ atom->dirty = false;
+ *list_elem = atom;
+}
+
+uint32_t si_num_banks(struct si_screen *sscreen, unsigned bpe, unsigned tile_split,
+ unsigned tile_mode_index)
+{
+ if ((sscreen->b.chip_class == CIK) &&
+ sscreen->b.info.cik_macrotile_mode_array_valid) {
unsigned index, tileb;
tileb = 8 * 8 * bpe;
for (index = 0; tileb > 64; index++) {
tileb >>= 1;
}
-
assert(index < 16);
return (sscreen->b.info.cik_macrotile_mode_array[index] >> 6) & 0x3;
}
+ if ((sscreen->b.chip_class == SI) &&
+ sscreen->b.info.si_tile_mode_array_valid) {
+ assert(tile_mode_index < 32);
+
+ return (sscreen->b.info.si_tile_mode_array[tile_mode_index] >> 20) & 0x3;
+ }
+
/* The old way. */
switch (sscreen->b.tiling_info.num_banks) {
case 2:
}
}
-static unsigned cik_tile_split(unsigned tile_split)
+unsigned cik_tile_split(unsigned tile_split)
{
switch (tile_split) {
case 64:
return tile_split;
}
-static unsigned cik_macro_tile_aspect(unsigned macro_tile_aspect)
+unsigned cik_macro_tile_aspect(unsigned macro_tile_aspect)
{
switch (macro_tile_aspect) {
default:
return macro_tile_aspect;
}
-static unsigned cik_bank_wh(unsigned bankwh)
+unsigned cik_bank_wh(unsigned bankwh)
{
switch (bankwh) {
default:
return bankwh;
}
-static unsigned cik_db_pipe_config(struct si_screen *sscreen, unsigned tile_mode)
+unsigned cik_db_pipe_config(struct si_screen *sscreen, unsigned tile_mode)
{
if (sscreen->b.info.si_tile_mode_array_valid) {
uint32_t gb_tile_mode = sscreen->b.info.si_tile_mode_array[tile_mode];
if (pm4 == NULL)
return;
- mask = (1ULL << ((unsigned)sctx->framebuffer.nr_cbufs * 4)) - 1;
+ mask = (1ULL << ((unsigned)sctx->framebuffer.state.nr_cbufs * 4)) - 1;
mask &= blend->cb_target_mask;
si_pm4_set_reg(pm4, R_028238_CB_TARGET_MASK, mask);
struct si_pm4_state *pm4;
float offset_units;
- if (!rs || !sctx->framebuffer.zsbuf)
+ if (!rs || !sctx->framebuffer.state.zsbuf)
return;
offset_units = sctx->queued.named.rasterizer->offset_units;
- switch (sctx->framebuffer.zsbuf->texture->format) {
+ switch (sctx->framebuffer.state.zsbuf->texture->format) {
case PIPE_FORMAT_S8_UINT_Z24_UNORM:
case PIPE_FORMAT_X8Z24_UNORM:
case PIPE_FORMAT_Z24X8_UNORM:
case PIPE_TEXTURE_3D:
return V_008F1C_SQ_RSRC_IMG_3D;
case PIPE_TEXTURE_CUBE:
+ case PIPE_TEXTURE_CUBE_ARRAY:
return V_008F1C_SQ_RSRC_IMG_CUBE;
}
}
if (type == UTIL_FORMAT_TYPE_FIXED)
return V_008F0C_BUF_DATA_FORMAT_INVALID;
+ if (desc->format == PIPE_FORMAT_R11G11B10_FLOAT)
+ return V_008F0C_BUF_DATA_FORMAT_10_11_11;
+
if (desc->nr_channels == 4 &&
desc->channel[0].size == 10 &&
desc->channel[1].size == 10 &&
const struct util_format_description *desc,
int first_non_void)
{
+ if (desc->format == PIPE_FORMAT_R11G11B10_FLOAT)
+ return V_008F0C_BUF_NUM_FORMAT_FLOAT;
+
switch (desc->channel[first_non_void].type) {
case UTIL_FORMAT_TYPE_SIGNED:
if (desc->channel[first_non_void].normalized)
return retval == usage;
}
-static unsigned si_tile_mode_index(struct r600_texture *rtex, unsigned level, bool stencil)
+unsigned si_tile_mode_index(struct r600_texture *rtex, unsigned level, bool stencil)
{
unsigned tile_mode_index = 0;
}
}
- if (rtex->cmask.size) {
- color_info |= S_028C70_FAST_CLEAR(1);
- }
-
offset += r600_resource_va(sctx->b.b.screen, surf->base.texture);
surf->cb_color_base = offset >> 8;
surf->cb_color_info = color_info;
surf->cb_color_attrib = color_attrib;
- if (rtex->cmask.size) {
- surf->cb_color_cmask = (offset + rtex->cmask.offset) >> 8;
- surf->cb_color_cmask_slice = S_028C80_TILE_MAX(rtex->cmask.slice_tile_max);
- }
if (rtex->fmask.size) {
surf->cb_color_fmask = (offset + rtex->fmask.offset) >> 8;
surf->cb_color_fmask_slice = S_028C88_TILE_MAX(rtex->fmask.slice_tile_max);
+ } else {
+ /* This must be set for fast clear to work without FMASK. */
+ surf->cb_color_fmask = surf->cb_color_base;
+ surf->cb_color_fmask_slice = surf->cb_color_slice;
+ surf->cb_color_attrib |= S_028C74_FMASK_TILE_MODE_INDEX(tile_mode_index);
+
+ if (sctx->b.chip_class == SI) {
+ unsigned bankh = util_logbase2(rtex->surface.bankh);
+ surf->cb_color_attrib |= S_028C74_FMASK_BANK_HEIGHT(bankh);
+ }
+
+ if (sctx->b.chip_class >= CIK) {
+ surf->cb_color_pitch |= S_028C64_FMASK_TILE_MAX(pitch);
+ }
}
/* Determine pixel shader export format */
unsigned macro_aspect, tile_split, stile_split, bankh, bankw, nbanks, pipe_config;
uint32_t z_info, s_info, db_depth_info;
uint64_t z_offs, s_offs;
- uint32_t db_htile_data_base, db_htile_surface, pa_su_poly_offset_db_fmt_cntl;
+ uint32_t db_htile_data_base, db_htile_surface, pa_su_poly_offset_db_fmt_cntl = 0;
- switch (sctx->framebuffer.zsbuf->texture->format) {
+ switch (sctx->framebuffer.state.zsbuf->texture->format) {
case PIPE_FORMAT_S8_UINT_Z24_UNORM:
case PIPE_FORMAT_X8Z24_UNORM:
case PIPE_FORMAT_Z24X8_UNORM:
macro_aspect = cik_macro_tile_aspect(macro_aspect);
bankw = cik_bank_wh(bankw);
bankh = cik_bank_wh(bankh);
- nbanks = cik_num_banks(sscreen, rtex->surface.bpe, rtex->surface.tile_split);
+ nbanks = si_num_banks(sscreen, rtex->surface.bpe, rtex->surface.tile_split,
+ ~0);
tile_mode_index = si_tile_mode_index(rtex, level, false);
pipe_config = cik_db_pipe_config(sscreen, tile_mode_index);
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))
-
-/* 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;
-/* Cayman/SI 8xMSAA */
-static uint32_t cm_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 cm_max_dist_8x = 8;
-/* Cayman/SI 16xMSAA */
-static uint32_t cm_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 cm_max_dist_16x = 8;
-
-static void si_get_sample_position(struct pipe_context *ctx,
- unsigned sample_count,
- unsigned sample_index,
- float *out_value)
-{
- int offset, index;
- struct {
- int idx:4;
- } val;
- switch (sample_count) {
- case 1:
- default:
- out_value[0] = out_value[1] = 0.5;
- break;
- case 2:
- offset = 4 * (sample_index * 2);
- val.idx = (sample_locs_2x[0] >> offset) & 0xf;
- out_value[0] = (float)(val.idx + 8) / 16.0f;
- val.idx = (sample_locs_2x[0] >> (offset + 4)) & 0xf;
- out_value[1] = (float)(val.idx + 8) / 16.0f;
- break;
- case 4:
- offset = 4 * (sample_index * 2);
- val.idx = (sample_locs_4x[0] >> offset) & 0xf;
- out_value[0] = (float)(val.idx + 8) / 16.0f;
- val.idx = (sample_locs_4x[0] >> (offset + 4)) & 0xf;
- out_value[1] = (float)(val.idx + 8) / 16.0f;
- break;
- case 8:
- offset = 4 * (sample_index % 4 * 2);
- index = (sample_index / 4) * 4;
- val.idx = (cm_sample_locs_8x[index] >> offset) & 0xf;
- out_value[0] = (float)(val.idx + 8) / 16.0f;
- val.idx = (cm_sample_locs_8x[index] >> (offset + 4)) & 0xf;
- out_value[1] = (float)(val.idx + 8) / 16.0f;
- break;
- case 16:
- offset = 4 * (sample_index % 4 * 2);
- index = (sample_index / 4) * 4;
- val.idx = (cm_sample_locs_16x[index] >> offset) & 0xf;
- out_value[0] = (float)(val.idx + 8) / 16.0f;
- val.idx = (cm_sample_locs_16x[index] >> (offset + 4)) & 0xf;
- out_value[1] = (float)(val.idx + 8) / 16.0f;
- break;
- }
-}
-
-static void si_set_msaa_state(struct si_context *sctx, struct si_pm4_state *pm4, int nr_samples)
-{
- unsigned max_dist = 0;
-
- switch (nr_samples) {
- default:
- nr_samples = 0;
- break;
- case 2:
- si_pm4_set_reg(pm4, R_028BF8_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0, sample_locs_2x[0]);
- si_pm4_set_reg(pm4, R_028C08_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0, sample_locs_2x[1]);
- si_pm4_set_reg(pm4, R_028C18_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0, sample_locs_2x[2]);
- si_pm4_set_reg(pm4, R_028C28_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0, sample_locs_2x[3]);
- max_dist = max_dist_2x;
- break;
- case 4:
- si_pm4_set_reg(pm4, R_028BF8_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0, sample_locs_4x[0]);
- si_pm4_set_reg(pm4, R_028C08_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0, sample_locs_4x[1]);
- si_pm4_set_reg(pm4, R_028C18_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0, sample_locs_4x[2]);
- si_pm4_set_reg(pm4, R_028C28_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0, sample_locs_4x[3]);
- max_dist = max_dist_4x;
- break;
- case 8:
- si_pm4_set_reg(pm4, R_028BF8_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0, cm_sample_locs_8x[0]);
- si_pm4_set_reg(pm4, R_028BFC_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1, cm_sample_locs_8x[4]);
- si_pm4_set_reg(pm4, R_028C00_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2, 0);
- si_pm4_set_reg(pm4, R_028C04_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3, 0);
- si_pm4_set_reg(pm4, R_028C08_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0, cm_sample_locs_8x[1]);
- si_pm4_set_reg(pm4, R_028C0C_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1, cm_sample_locs_8x[5]);
- si_pm4_set_reg(pm4, R_028C10_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2, 0);
- si_pm4_set_reg(pm4, R_028C14_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3, 0);
- si_pm4_set_reg(pm4, R_028C18_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0, cm_sample_locs_8x[2]);
- si_pm4_set_reg(pm4, R_028C1C_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1, cm_sample_locs_8x[6]);
- si_pm4_set_reg(pm4, R_028C20_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2, 0);
- si_pm4_set_reg(pm4, R_028C24_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3, 0);
- si_pm4_set_reg(pm4, R_028C28_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0, cm_sample_locs_8x[3]);
- si_pm4_set_reg(pm4, R_028C2C_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1, cm_sample_locs_8x[7]);
- max_dist = cm_max_dist_8x;
- break;
- case 16:
- si_pm4_set_reg(pm4, R_028BF8_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_0, cm_sample_locs_16x[0]);
- si_pm4_set_reg(pm4, R_028BFC_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_1, cm_sample_locs_16x[4]);
- si_pm4_set_reg(pm4, R_028C00_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_2, cm_sample_locs_16x[8]);
- si_pm4_set_reg(pm4, R_028C04_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y0_3, cm_sample_locs_16x[12]);
- si_pm4_set_reg(pm4, R_028C08_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_0, cm_sample_locs_16x[1]);
- si_pm4_set_reg(pm4, R_028C0C_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_1, cm_sample_locs_16x[5]);
- si_pm4_set_reg(pm4, R_028C10_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_2, cm_sample_locs_16x[9]);
- si_pm4_set_reg(pm4, R_028C14_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y0_3, cm_sample_locs_16x[13]);
- si_pm4_set_reg(pm4, R_028C18_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_0, cm_sample_locs_16x[2]);
- si_pm4_set_reg(pm4, R_028C1C_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_1, cm_sample_locs_16x[6]);
- si_pm4_set_reg(pm4, R_028C20_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_2, cm_sample_locs_16x[10]);
- si_pm4_set_reg(pm4, R_028C24_PA_SC_AA_SAMPLE_LOCS_PIXEL_X0Y1_3, cm_sample_locs_16x[14]);
- si_pm4_set_reg(pm4, R_028C28_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_0, cm_sample_locs_16x[3]);
- si_pm4_set_reg(pm4, R_028C2C_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_1, cm_sample_locs_16x[7]);
- si_pm4_set_reg(pm4, R_028C30_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_2, cm_sample_locs_16x[11]);
- si_pm4_set_reg(pm4, R_028C34_PA_SC_AA_SAMPLE_LOCS_PIXEL_X1Y1_3, cm_sample_locs_16x[15]);
- max_dist = cm_max_dist_16x;
- break;
- }
-
- if (nr_samples > 1) {
- unsigned log_samples = util_logbase2(nr_samples);
-
- si_pm4_set_reg(pm4, R_028BDC_PA_SC_LINE_CNTL,
- S_028BDC_LAST_PIXEL(1) |
- S_028BDC_EXPAND_LINE_WIDTH(1));
- si_pm4_set_reg(pm4, 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));
-
- si_pm4_set_reg(pm4, 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 {
- si_pm4_set_reg(pm4, R_028BDC_PA_SC_LINE_CNTL, S_028BDC_LAST_PIXEL(1));
- si_pm4_set_reg(pm4, R_028BE0_PA_SC_AA_CONFIG, 0);
-
- si_pm4_set_reg(pm4, R_028804_DB_EQAA,
- S_028804_HIGH_QUALITY_INTERSECTIONS(1) |
- S_028804_STATIC_ANCHOR_ASSOCIATIONS(1));
- }
-}
-
static void si_set_framebuffer_state(struct pipe_context *ctx,
const struct pipe_framebuffer_state *state)
{
struct si_context *sctx = (struct si_context *)ctx;
struct r600_surface *surf = NULL;
struct r600_texture *rtex;
- struct si_pm4_state *pm4 = si_pm4_alloc_state(sctx);
- int nr_samples, i;
-
- if (pm4 == NULL)
- return;
+ int i;
- if (sctx->framebuffer.nr_cbufs) {
+ if (sctx->framebuffer.state.nr_cbufs) {
sctx->b.flags |= R600_CONTEXT_FLUSH_AND_INV_CB |
R600_CONTEXT_FLUSH_AND_INV_CB_META;
}
- if (sctx->framebuffer.zsbuf) {
+ if (sctx->framebuffer.state.zsbuf) {
sctx->b.flags |= R600_CONTEXT_FLUSH_AND_INV_DB |
R600_CONTEXT_FLUSH_AND_INV_DB_META;
}
- util_copy_framebuffer_state(&sctx->framebuffer, state);
+ util_copy_framebuffer_state(&sctx->framebuffer.state, state);
- /* build states */
- sctx->export_16bpc = 0;
- sctx->fb_compressed_cb_mask = 0;
+ sctx->framebuffer.export_16bpc = 0;
+ sctx->framebuffer.compressed_cb_mask = 0;
+ sctx->framebuffer.nr_samples = util_framebuffer_get_num_samples(state);
+ sctx->framebuffer.log_samples = util_logbase2(sctx->framebuffer.nr_samples);
+ sctx->framebuffer.cb0_is_integer = state->nr_cbufs && state->cbufs[0] &&
+ util_format_is_pure_integer(state->cbufs[0]->format);
for (i = 0; i < state->nr_cbufs; i++) {
- if (!state->cbufs[i]) {
- si_pm4_set_reg(pm4, R_028C70_CB_COLOR0_INFO + i * 0x3C,
- S_028C70_FORMAT(V_028C70_COLOR_INVALID));
+ if (!state->cbufs[i])
continue;
- }
surf = (struct r600_surface*)state->cbufs[i];
rtex = (struct r600_texture*)surf->base.texture;
}
if (surf->export_16bpc) {
- sctx->export_16bpc |= 1 << i;
+ sctx->framebuffer.export_16bpc |= 1 << i;
}
- if (rtex->fmask.size || rtex->cmask.size) {
- sctx->fb_compressed_cb_mask |= 1 << i;
+ if (rtex->fmask.size && rtex->cmask.size) {
+ sctx->framebuffer.compressed_cb_mask |= 1 << i;
}
-
- si_pm4_add_bo(pm4, &rtex->resource, RADEON_USAGE_READWRITE);
- si_pm4_set_reg(pm4, R_028C60_CB_COLOR0_BASE + i * 0x3C, surf->cb_color_base);
- si_pm4_set_reg(pm4, R_028C64_CB_COLOR0_PITCH + i * 0x3C, surf->cb_color_pitch);
- si_pm4_set_reg(pm4, R_028C68_CB_COLOR0_SLICE + i * 0x3C, surf->cb_color_slice);
- si_pm4_set_reg(pm4, R_028C6C_CB_COLOR0_VIEW + i * 0x3C, surf->cb_color_view);
- si_pm4_set_reg(pm4, R_028C70_CB_COLOR0_INFO + i * 0x3C, surf->cb_color_info);
- si_pm4_set_reg(pm4, R_028C74_CB_COLOR0_ATTRIB + i * 0x3C, surf->cb_color_attrib);
- si_pm4_set_reg(pm4, R_028C7C_CB_COLOR0_CMASK + i * 0x3C, surf->cb_color_cmask);
- si_pm4_set_reg(pm4, R_028C80_CB_COLOR0_CMASK_SLICE + i * 0x3C, surf->cb_color_cmask_slice);
- si_pm4_set_reg(pm4, R_028C84_CB_COLOR0_FMASK + i * 0x3C, surf->cb_color_fmask);
- si_pm4_set_reg(pm4, R_028C88_CB_COLOR0_FMASK_SLICE + i * 0x3C, surf->cb_color_fmask_slice);
- }
- /* Set CB_COLOR1_INFO for possible dual-src blending. */
- if (i == 1 && surf) {
- si_pm4_set_reg(pm4, R_028C70_CB_COLOR0_INFO + 1 * 0x3C, surf->cb_color_info);
- /* Also set the 16BPC export. */
- if (surf->export_16bpc) {
- sctx->export_16bpc |= 1 << 1;
- }
- i++;
}
- for (; i < 8; i++) {
- si_pm4_set_reg(pm4, R_028C70_CB_COLOR0_INFO + i * 0x3C,
- S_028C70_FORMAT(V_028C70_COLOR_INVALID));
+ /* Set the 16BPC export for possible dual-src blending. */
+ if (i == 1 && surf && surf->export_16bpc) {
+ sctx->framebuffer.export_16bpc |= 1 << 1;
}
- assert(!(sctx->export_16bpc & ~0xff));
+ assert(!(sctx->framebuffer.export_16bpc & ~0xff));
if (state->zsbuf) {
surf = (struct r600_surface*)state->zsbuf;
- rtex = (struct r600_texture*)surf->base.texture;
if (!surf->depth_initialized) {
si_init_depth_surface(sctx, surf);
}
+ }
- if (surf->db_htile_data_base) {
- si_pm4_add_bo(pm4, rtex->htile_buffer, RADEON_USAGE_READWRITE);
+ si_update_fb_rs_state(sctx);
+ si_update_fb_blend_state(sctx);
+
+ sctx->framebuffer.atom.num_dw = state->nr_cbufs*15 + (8 - state->nr_cbufs)*3;
+ sctx->framebuffer.atom.num_dw += state->zsbuf ? 23 : 4;
+ sctx->framebuffer.atom.num_dw += 3; /* WINDOW_SCISSOR_BR */
+ sctx->framebuffer.atom.num_dw += 25; /* MSAA */
+ sctx->framebuffer.atom.dirty = true;
+}
+
+static void si_emit_framebuffer_state(struct si_context *sctx, struct r600_atom *atom)
+{
+ struct radeon_winsys_cs *cs = sctx->b.rings.gfx.cs;
+ struct pipe_framebuffer_state *state = &sctx->framebuffer.state;
+ unsigned i, nr_cbufs = state->nr_cbufs;
+ struct r600_texture *tex = NULL;
+ struct r600_surface *cb = NULL;
+
+ /* Colorbuffers. */
+ for (i = 0; i < nr_cbufs; i++) {
+ cb = (struct r600_surface*)state->cbufs[i];
+ if (!cb) {
+ r600_write_context_reg(cs, R_028C70_CB_COLOR0_INFO + i * 0x3C,
+ S_028C70_FORMAT(V_028C70_COLOR_INVALID));
+ continue;
}
- si_pm4_set_reg(pm4, R_028008_DB_DEPTH_VIEW, surf->db_depth_view);
- si_pm4_set_reg(pm4, R_028014_DB_HTILE_DATA_BASE, surf->db_htile_data_base);
-
- si_pm4_set_reg(pm4, R_02803C_DB_DEPTH_INFO, surf->db_depth_info);
- si_pm4_set_reg(pm4, R_028040_DB_Z_INFO, surf->db_z_info);
- si_pm4_set_reg(pm4, R_028044_DB_STENCIL_INFO, surf->db_stencil_info);
- si_pm4_add_bo(pm4, &rtex->resource, RADEON_USAGE_READWRITE);
- si_pm4_set_reg(pm4, R_028048_DB_Z_READ_BASE, surf->db_depth_base);
- si_pm4_set_reg(pm4, R_02804C_DB_STENCIL_READ_BASE, surf->db_stencil_base);
- si_pm4_set_reg(pm4, R_028050_DB_Z_WRITE_BASE, surf->db_depth_base);
- si_pm4_set_reg(pm4, R_028054_DB_STENCIL_WRITE_BASE, surf->db_stencil_base);
- si_pm4_set_reg(pm4, R_028058_DB_DEPTH_SIZE, surf->db_depth_size);
- si_pm4_set_reg(pm4, R_02805C_DB_DEPTH_SLICE, surf->db_depth_slice);
-
- si_pm4_set_reg(pm4, R_028ABC_DB_HTILE_SURFACE, surf->db_htile_surface);
- si_pm4_set_reg(pm4, R_028B78_PA_SU_POLY_OFFSET_DB_FMT_CNTL,
- surf->pa_su_poly_offset_db_fmt_cntl);
- } else {
- si_pm4_set_reg(pm4, R_028040_DB_Z_INFO, S_028040_FORMAT(V_028040_Z_INVALID));
- si_pm4_set_reg(pm4, R_028044_DB_STENCIL_INFO, S_028044_FORMAT(V_028044_STENCIL_INVALID));
+
+ tex = (struct r600_texture *)cb->base.texture;
+ r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx,
+ &tex->resource, RADEON_USAGE_READWRITE,
+ tex->surface.nsamples > 1 ?
+ RADEON_PRIO_COLOR_BUFFER_MSAA :
+ RADEON_PRIO_COLOR_BUFFER);
+
+ if (tex->cmask_buffer && tex->cmask_buffer != &tex->resource) {
+ r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx,
+ tex->cmask_buffer, RADEON_USAGE_READWRITE,
+ RADEON_PRIO_COLOR_META);
+ }
+
+ r600_write_context_reg_seq(cs, R_028C60_CB_COLOR0_BASE + i * 0x3C, 13);
+ radeon_emit(cs, cb->cb_color_base); /* R_028C60_CB_COLOR0_BASE */
+ radeon_emit(cs, cb->cb_color_pitch); /* R_028C64_CB_COLOR0_PITCH */
+ radeon_emit(cs, cb->cb_color_slice); /* R_028C68_CB_COLOR0_SLICE */
+ radeon_emit(cs, cb->cb_color_view); /* R_028C6C_CB_COLOR0_VIEW */
+ radeon_emit(cs, cb->cb_color_info | tex->cb_color_info); /* R_028C70_CB_COLOR0_INFO */
+ radeon_emit(cs, cb->cb_color_attrib); /* R_028C74_CB_COLOR0_ATTRIB */
+ radeon_emit(cs, 0); /* R_028C78 unused */
+ radeon_emit(cs, tex->cmask.base_address_reg); /* R_028C7C_CB_COLOR0_CMASK */
+ radeon_emit(cs, tex->cmask.slice_tile_max); /* R_028C80_CB_COLOR0_CMASK_SLICE */
+ radeon_emit(cs, cb->cb_color_fmask); /* R_028C84_CB_COLOR0_FMASK */
+ radeon_emit(cs, cb->cb_color_fmask_slice); /* R_028C88_CB_COLOR0_FMASK_SLICE */
+ radeon_emit(cs, tex->color_clear_value[0]); /* R_028C8C_CB_COLOR0_CLEAR_WORD0 */
+ radeon_emit(cs, tex->color_clear_value[1]); /* R_028C90_CB_COLOR0_CLEAR_WORD1 */
+ }
+ /* set CB_COLOR1_INFO for possible dual-src blending */
+ if (i == 1 && state->cbufs[0]) {
+ r600_write_context_reg(cs, R_028C70_CB_COLOR0_INFO + 1 * 0x3C,
+ cb->cb_color_info | tex->cb_color_info);
+ i++;
+ }
+ for (; i < 8 ; i++) {
+ r600_write_context_reg(cs, R_028C70_CB_COLOR0_INFO + i * 0x3C, 0);
}
- /* PA_SC_WINDOW_SCISSOR_TL is set in si_init_config() */
- si_pm4_set_reg(pm4, R_028208_PA_SC_WINDOW_SCISSOR_BR,
- S_028208_BR_X(state->width) | S_028208_BR_Y(state->height));
+ /* ZS buffer. */
+ if (state->zsbuf) {
+ struct r600_surface *zb = (struct r600_surface*)state->zsbuf;
+ struct r600_texture *rtex = (struct r600_texture*)zb->base.texture;
+
+ r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx,
+ &rtex->resource, RADEON_USAGE_READWRITE,
+ zb->base.texture->nr_samples > 1 ?
+ RADEON_PRIO_DEPTH_BUFFER_MSAA :
+ RADEON_PRIO_DEPTH_BUFFER);
+
+ if (zb->db_htile_data_base) {
+ r600_context_bo_reloc(&sctx->b, &sctx->b.rings.gfx,
+ rtex->htile_buffer, RADEON_USAGE_READWRITE,
+ RADEON_PRIO_DEPTH_META);
+ }
- nr_samples = util_framebuffer_get_num_samples(state);
+ r600_write_context_reg(cs, R_028008_DB_DEPTH_VIEW, zb->db_depth_view);
+ r600_write_context_reg(cs, R_028014_DB_HTILE_DATA_BASE, zb->db_htile_data_base);
+
+ r600_write_context_reg_seq(cs, R_02803C_DB_DEPTH_INFO, 9);
+ radeon_emit(cs, zb->db_depth_info); /* R_02803C_DB_DEPTH_INFO */
+ radeon_emit(cs, zb->db_z_info); /* R_028040_DB_Z_INFO */
+ radeon_emit(cs, zb->db_stencil_info); /* R_028044_DB_STENCIL_INFO */
+ radeon_emit(cs, zb->db_depth_base); /* R_028048_DB_Z_READ_BASE */
+ radeon_emit(cs, zb->db_stencil_base); /* R_02804C_DB_STENCIL_READ_BASE */
+ radeon_emit(cs, zb->db_depth_base); /* R_028050_DB_Z_WRITE_BASE */
+ radeon_emit(cs, zb->db_stencil_base); /* R_028054_DB_STENCIL_WRITE_BASE */
+ radeon_emit(cs, zb->db_depth_size); /* R_028058_DB_DEPTH_SIZE */
+ radeon_emit(cs, zb->db_depth_slice); /* R_02805C_DB_DEPTH_SLICE */
+
+ r600_write_context_reg(cs, R_028ABC_DB_HTILE_SURFACE, zb->db_htile_surface);
+ r600_write_context_reg(cs, R_028B78_PA_SU_POLY_OFFSET_DB_FMT_CNTL,
+ zb->pa_su_poly_offset_db_fmt_cntl);
+ } else {
+ r600_write_context_reg_seq(cs, R_028040_DB_Z_INFO, 2);
+ radeon_emit(cs, S_028040_FORMAT(V_028040_Z_INVALID)); /* R_028040_DB_Z_INFO */
+ radeon_emit(cs, S_028044_FORMAT(V_028044_STENCIL_INVALID)); /* R_028044_DB_STENCIL_INFO */
+ }
- si_set_msaa_state(sctx, pm4, nr_samples);
- sctx->fb_log_samples = util_logbase2(nr_samples);
- sctx->fb_cb0_is_integer = state->nr_cbufs && state->cbufs[0] &&
- util_format_is_pure_integer(state->cbufs[0]->format);
+ /* Framebuffer dimensions. */
+ /* PA_SC_WINDOW_SCISSOR_TL is set in si_init_config() */
+ r600_write_context_reg(cs, R_028208_PA_SC_WINDOW_SCISSOR_BR,
+ S_028208_BR_X(state->width) | S_028208_BR_Y(state->height));
- si_pm4_set_state(sctx, framebuffer, pm4);
- si_update_fb_rs_state(sctx);
- si_update_fb_blend_state(sctx);
+ cayman_emit_msaa_state(cs, sctx->framebuffer.nr_samples);
}
/*
key->vs.as_es = sctx->gs_shader != NULL;
} else if (sel->type == PIPE_SHADER_FRAGMENT) {
if (sel->fs_write_all)
- key->ps.nr_cbufs = sctx->framebuffer.nr_cbufs;
- key->ps.export_16bpc = sctx->export_16bpc;
+ key->ps.nr_cbufs = sctx->framebuffer.state.nr_cbufs;
+ key->ps.export_16bpc = sctx->framebuffer.export_16bpc;
if (sctx->queued.named.rasterizer) {
key->ps.color_two_side = sctx->queued.named.rasterizer->two_side;
if (sctx->queued.named.blend) {
key->ps.alpha_to_one = sctx->queued.named.blend->alpha_to_one &&
sctx->queued.named.rasterizer->multisample_enable &&
- !sctx->fb_cb0_is_integer;
+ !sctx->framebuffer.cb0_is_integer;
}
}
if (sctx->queued.named.dsa) {
key->ps.alpha_func = sctx->queued.named.dsa->alpha_func;
/* Alpha-test should be disabled if colorbuffer 0 is integer. */
- if (sctx->framebuffer.nr_cbufs &&
- sctx->framebuffer.cbufs[0] &&
- util_format_is_pure_integer(sctx->framebuffer.cbufs[0]->texture->format))
+ if (sctx->framebuffer.cb0_is_integer)
key->ps.alpha_func = PIPE_FUNC_ALWAYS;
} else {
key->ps.alpha_func = PIPE_FUNC_ALWAYS;
return si_create_shader_state(ctx, state, PIPE_SHADER_FRAGMENT);
}
-#if HAVE_LLVM >= 0x0305
+#if LLVM_SUPPORTS_GEOM_SHADERS
static void *si_create_gs_state(struct pipe_context *ctx,
const struct pipe_shader_state *state)
sctx->vs_shader = sel;
}
-#if HAVE_LLVM >= 0x0305
+#if LLVM_SUPPORTS_GEOM_SHADERS
static void si_bind_gs_shader(struct pipe_context *ctx, void *state)
{
struct si_context *sctx = (struct si_context *)ctx;
struct si_pipe_shader_selector *sel = state;
+ /* skip if supplied shader is one already in use */
if (sctx->ps_shader == sel)
return;
+ /* use dummy shader if supplied shader is corrupt */
if (!sel || !sel->current)
sel = sctx->dummy_pixel_shader;
while (p) {
c = p->next_variant;
- si_pm4_delete_state(sctx, vs, p->pm4);
+ if (sel->type == PIPE_SHADER_GEOMETRY)
+ si_pm4_delete_state(sctx, gs, p->pm4);
+ else if (sel->type == PIPE_SHADER_FRAGMENT)
+ si_pm4_delete_state(sctx, ps, p->pm4);
+ else if (p->key.vs.as_es)
+ si_pm4_delete_state(sctx, es, p->pm4);
+ else
+ si_pm4_delete_state(sctx, vs, p->pm4);
si_pipe_shader_destroy(ctx, p);
free(p);
p = c;
si_delete_shader_selector(ctx, sel);
}
-#if HAVE_LLVM >= 0x0305
+#if LLVM_SUPPORTS_GEOM_SHADERS
static void si_delete_gs_shader(struct pipe_context *ctx, void *state)
{
depth = texture->array_size;
} else if (texture->target == PIPE_TEXTURE_2D_ARRAY) {
depth = texture->array_size;
- }
+ } else if (texture->target == PIPE_TEXTURE_CUBE_ARRAY)
+ depth = texture->array_size / 6;
va = r600_resource_va(ctx->screen, texture);
va += surflevel[0].offset;
rstate->val[0] = (S_008F30_CLAMP_X(si_tex_wrap(state->wrap_s)) |
S_008F30_CLAMP_Y(si_tex_wrap(state->wrap_t)) |
S_008F30_CLAMP_Z(si_tex_wrap(state->wrap_r)) |
- (state->max_anisotropy & 0x7) << 9 | /* XXX */
+ r600_tex_aniso_filter(state->max_anisotropy) << 9 |
S_008F30_DEPTH_COMPARE_FUNC(si_tex_compare(state->compare_func)) |
S_008F30_FORCE_UNNORMALIZED(!state->normalized_coords) |
- aniso_flag_offset << 16 | /* XXX */
S_008F30_DISABLE_CUBE_WRAP(!state->seamless_cube_map));
rstate->val[1] = (S_008F34_MIN_LOD(S_FIXED(CLAMP(state->min_lod, 0, 15), 8)) |
S_008F34_MAX_LOD(S_FIXED(CLAMP(state->max_lod, 0, 15), 8)));
rstate->val[2] = (S_008F38_LOD_BIAS(S_FIXED(CLAMP(state->lod_bias, -16, 16), 8)) |
- S_008F38_XY_MAG_FILTER(si_tex_filter(state->mag_img_filter)) |
- S_008F38_XY_MIN_FILTER(si_tex_filter(state->min_img_filter)) |
+ S_008F38_XY_MAG_FILTER(si_tex_filter(state->mag_img_filter) | aniso_flag_offset) |
+ S_008F38_XY_MIN_FILTER(si_tex_filter(state->min_img_filter) | aniso_flag_offset) |
S_008F38_MIP_FILTER(si_tex_mipfilter(state->min_mip_filter)));
rstate->val[3] = S_008F3C_BORDER_COLOR_TYPE(border_color_type);
if (sctx->b.chip_class >= CIK)
si_pm4_set_reg(pm4, R_028084_TA_BC_BASE_ADDR_HI, va_offset >> 40);
sctx->b.ws->buffer_unmap(sctx->border_color_table->cs_buf);
- si_pm4_add_bo(pm4, sctx->border_color_table, RADEON_USAGE_READ);
+ si_pm4_add_bo(pm4, sctx->border_color_table, RADEON_USAGE_READ,
+ RADEON_PRIO_SHADER_DATA);
}
memcpy(samplers->samplers, states, sizeof(void*) * count);
si_set_sampler_states(sctx, pm4, count, states,
&sctx->samplers[PIPE_SHADER_VERTEX],
R_00B130_SPI_SHADER_USER_DATA_VS_0);
-#if HAVE_LLVM >= 0x0305
+#if LLVM_SUPPORTS_GEOM_SHADERS
si_set_sampler_states(sctx, pm4, count, states,
&sctx->samplers[PIPE_SHADER_VERTEX],
R_00B330_SPI_SHADER_USER_DATA_ES_0);
return si_create_blend_state_mode(&sctx->b.b, &blend, mode);
}
-static boolean si_dma_copy(struct pipe_context *ctx,
- struct pipe_resource *dst,
- unsigned dst_level,
- unsigned dst_x, unsigned dst_y, unsigned dst_z,
- struct pipe_resource *src,
- unsigned src_level,
- const struct pipe_box *src_box)
-{
- /* XXX implement this or share evergreen_dma_blit with r600g */
- return FALSE;
-}
-
static void si_set_occlusion_query_state(struct pipe_context *ctx, bool enable)
{
/* XXX Turn this into a proper state. Right now the queries are
{
int i;
+ si_init_atom(&sctx->framebuffer.atom, &sctx->atoms.s.framebuffer, si_emit_framebuffer_state, 0);
+
sctx->b.b.create_blend_state = si_create_blend_state;
sctx->b.b.bind_blend_state = si_bind_blend_state;
sctx->b.b.delete_blend_state = si_delete_blend_state;
sctx->custom_dsa_flush_inplace = si_create_db_flush_dsa(sctx, false, false, 0);
sctx->custom_blend_resolve = si_create_blend_custom(sctx, V_028808_CB_RESOLVE);
sctx->custom_blend_decompress = si_create_blend_custom(sctx, V_028808_CB_FMASK_DECOMPRESS);
+ sctx->custom_blend_fastclear = si_create_blend_custom(sctx, V_028808_CB_ELIMINATE_FAST_CLEAR);
sctx->b.b.set_clip_state = si_set_clip_state;
sctx->b.b.set_scissor_states = si_set_scissor_states;
sctx->b.b.set_stencil_ref = si_set_pipe_stencil_ref;
sctx->b.b.set_framebuffer_state = si_set_framebuffer_state;
- sctx->b.b.get_sample_position = si_get_sample_position;
+ sctx->b.b.get_sample_position = cayman_get_sample_position;
sctx->b.b.create_vs_state = si_create_vs_state;
sctx->b.b.create_fs_state = si_create_fs_state;
sctx->b.b.bind_fs_state = si_bind_ps_shader;
sctx->b.b.delete_vs_state = si_delete_vs_shader;
sctx->b.b.delete_fs_state = si_delete_ps_shader;
-#if HAVE_LLVM >= 0x0305
+#if LLVM_SUPPORTS_GEOM_SHADERS
sctx->b.b.create_gs_state = si_create_gs_state;
sctx->b.b.bind_gs_state = si_bind_gs_shader;
sctx->b.b.delete_gs_state = si_delete_gs_shader;
si_pm4_set_reg(pm4, R_028B68_VGT_GS_VERT_ITEMSIZE_3, 0);
si_pm4_set_reg(pm4, R_028B90_VGT_GS_INSTANCE_CNT, 0);
- si_pm4_set_reg(pm4, R_028B94_VGT_STRMOUT_CONFIG, 0x0);
si_pm4_set_reg(pm4, R_028B98_VGT_STRMOUT_BUFFER_CONFIG, 0x0);
if (sctx->b.chip_class == SI) {
si_pm4_set_reg(pm4, R_028AA8_IA_MULTI_VGT_PARAM,
/* XXX todo */
case CHIP_KABINI:
/* XXX todo */
+ case CHIP_MULLINS:
+ /* XXX todo */
default:
si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x00000000);
si_pm4_set_reg(pm4, R_028354_PA_SC_RASTER_CONFIG_1, 0x00000000);
projects / mesa.git / blobdiff