return 0;
}
-static void *si_create_blend_state(struct pipe_context *ctx,
- const struct pipe_blend_state *state)
+static void *si_create_blend_state_mode(struct pipe_context *ctx,
+ const struct pipe_blend_state *state,
+ unsigned mode)
{
struct si_state_blend *blend = CALLOC_STRUCT(si_state_blend);
struct si_pm4_state *pm4 = &blend->pm4;
if (blend == NULL)
return NULL;
- color_control = S_028808_MODE(V_028808_CB_NORMAL);
+ blend->alpha_to_one = state->alpha_to_one;
+
+ color_control = S_028808_MODE(mode);
if (state->logicop_enable) {
color_control |= S_028808_ROP3(state->logicop_func | (state->logicop_func << 4));
} else {
}
si_pm4_set_reg(pm4, R_028808_CB_COLOR_CONTROL, color_control);
- si_pm4_set_reg(pm4, R_028C38_PA_SC_AA_MASK_X0Y0_X1Y0, ~0);
- si_pm4_set_reg(pm4, R_028C3C_PA_SC_AA_MASK_X0Y1_X1Y1, ~0);
+ si_pm4_set_reg(pm4, 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->cb_target_mask = 0;
for (int i = 0; i < 8; i++) {
return blend;
}
+static void *si_create_blend_state(struct pipe_context *ctx,
+ const struct pipe_blend_state *state)
+{
+ return si_create_blend_state_mode(ctx, state, V_028808_CB_NORMAL);
+}
+
static void si_bind_blend_state(struct pipe_context *ctx, void *state)
{
struct r600_context *rctx = (struct r600_context *)ctx;
}
rs->two_side = state->light_twoside;
+ rs->multisample_enable = state->multisample;
rs->clip_plane_enable = state->clip_plane_enable;
polygon_dual_mode = (state->fill_front != PIPE_POLYGON_MODE_FILL ||
tmp = (unsigned)state->line_width * 8;
si_pm4_set_reg(pm4, R_028A08_PA_SU_LINE_CNTL, S_028A08_WIDTH(tmp));
si_pm4_set_reg(pm4, R_028A48_PA_SC_MODE_CNTL_0,
- S_028A48_LINE_STIPPLE_ENABLE(state->line_stipple_enable));
+ S_028A48_LINE_STIPPLE_ENABLE(state->line_stipple_enable) |
+ S_028A48_MSAA_ENABLE(state->multisample));
- si_pm4_set_reg(pm4, R_028BDC_PA_SC_LINE_CNTL, 0x00000400);
si_pm4_set_reg(pm4, R_028BE4_PA_SU_VTX_CNTL,
- S_028BE4_PIX_CENTER(state->half_pixel_center));
+ S_028BE4_PIX_CENTER(state->half_pixel_center) |
+ S_028BE4_QUANT_MODE(V_028BE4_X_16_8_FIXED_POINT_1_256TH));
si_pm4_set_reg(pm4, R_028BE8_PA_CL_GB_VERT_CLIP_ADJ, 0x3F800000);
si_pm4_set_reg(pm4, R_028BEC_PA_CL_GB_VERT_DISC_ADJ, 0x3F800000);
si_pm4_set_reg(pm4, R_028BF0_PA_CL_GB_HORZ_CLIP_ADJ, 0x3F800000);
si_pm4_set_reg(pm4, R_028AC0_DB_SRESULTS_COMPARE_STATE0, 0x0);
si_pm4_set_reg(pm4, R_028AC4_DB_SRESULTS_COMPARE_STATE1, 0x0);
si_pm4_set_reg(pm4, R_028AC8_DB_PRELOAD_CONTROL, 0x0);
- si_pm4_set_reg(pm4, R_028B70_DB_ALPHA_TO_MASK, 0x0000AA00);
dsa->db_render_override = db_render_override;
return dsa;
}
static void *si_create_db_flush_dsa(struct r600_context *rctx, bool copy_depth,
- bool copy_stencil)
+ bool copy_stencil, int sample)
{
struct pipe_depth_stencil_alpha_state dsa;
struct si_state_dsa *state;
si_pm4_set_reg(&state->pm4, R_028000_DB_RENDER_CONTROL,
S_028000_DEPTH_COPY(copy_depth) |
S_028000_STENCIL_COPY(copy_stencil) |
- S_028000_COPY_CENTROID(1));
+ S_028000_COPY_CENTROID(1) |
+ S_028000_COPY_SAMPLE(sample));
} else {
si_pm4_set_reg(&state->pm4, R_028000_DB_RENDER_CONTROL,
S_028000_DEPTH_COMPRESS_DISABLE(1) |
}
}
-static unsigned si_tex_dim(unsigned dim)
+static unsigned si_tex_dim(unsigned dim, unsigned nr_samples)
{
switch (dim) {
default:
return V_008F1C_SQ_RSRC_IMG_1D_ARRAY;
case PIPE_TEXTURE_2D:
case PIPE_TEXTURE_RECT:
- return V_008F1C_SQ_RSRC_IMG_2D;
+ return nr_samples > 1 ? V_008F1C_SQ_RSRC_IMG_2D_MSAA :
+ V_008F1C_SQ_RSRC_IMG_2D;
case PIPE_TEXTURE_2D_ARRAY:
- return V_008F1C_SQ_RSRC_IMG_2D_ARRAY;
+ return nr_samples > 1 ? V_008F1C_SQ_RSRC_IMG_2D_MSAA_ARRAY :
+ V_008F1C_SQ_RSRC_IMG_2D_ARRAY;
case PIPE_TEXTURE_3D:
return V_008F1C_SQ_RSRC_IMG_3D;
case PIPE_TEXTURE_CUBE:
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 (HAVE_LLVM < 0x0304 || rscreen->chip_class != SI)
+ return FALSE;
+
+ switch (sample_count) {
+ case 2:
+ case 4:
+ case 8:
+ break;
+ default:
+ return FALSE;
+ }
+ }
if ((usage & PIPE_BIND_SAMPLER_VIEW) &&
si_is_sampler_format_supported(screen, format)) {
return retval == usage;
}
-static unsigned si_tile_mode_index(struct r600_resource_texture *rtex, unsigned level, bool stencil)
+static unsigned si_tile_mode_index(struct r600_texture *rtex, unsigned level, bool stencil)
{
unsigned tile_mode_index = 0;
static void si_cb(struct r600_context *rctx, struct si_pm4_state *pm4,
const struct pipe_framebuffer_state *state, int cb)
{
- struct r600_resource_texture *rtex;
+ struct r600_texture *rtex;
struct r600_surface *surf;
unsigned level = state->cbufs[cb]->u.tex.level;
unsigned pitch, slice;
unsigned max_comp_size;
surf = (struct r600_surface *)state->cbufs[cb];
- rtex = (struct r600_resource_texture*)state->cbufs[cb]->texture;
+ rtex = (struct r600_texture*)state->cbufs[cb]->texture;
offset = rtex->surface.level[level].offset;
if (rtex->surface.level[level].mode < RADEON_SURF_MODE_1D) {
color_attrib = S_028C74_TILE_MODE_INDEX(tile_mode_index) |
S_028C74_FORCE_DST_ALPHA_1(desc->swizzle[3] == UTIL_FORMAT_SWIZZLE_1);
+ if (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);
+
+ if (rtex->fmask.size) {
+ color_info |= S_028C70_COMPRESSION(1);
+ unsigned fmask_bankh = util_logbase2(rtex->fmask.bank_height);
+
+ /* due to a bug in the hw, FMASK_BANK_HEIGHT must be set on SI too */
+ color_attrib |= S_028C74_FMASK_TILE_MODE_INDEX(rtex->fmask.tile_mode_index) |
+ S_028C74_FMASK_BANK_HEIGHT(fmask_bankh);
+ }
+ }
+
+ if (rtex->cmask.size) {
+ color_info |= S_028C70_FAST_CLEAR(1);
+ }
+
offset += r600_resource_va(rctx->context.screen, state->cbufs[cb]->texture);
offset >>= 8;
si_pm4_set_reg(pm4, R_028C70_CB_COLOR0_INFO + cb * 0x3C, color_info);
si_pm4_set_reg(pm4, R_028C74_CB_COLOR0_ATTRIB + cb * 0x3C, color_attrib);
+ if (rtex->cmask.size) {
+ si_pm4_set_reg(pm4, R_028C7C_CB_COLOR0_CMASK + cb * 0x3C,
+ offset + (rtex->cmask.offset >> 8));
+ si_pm4_set_reg(pm4, R_028C80_CB_COLOR0_CMASK_SLICE + cb * 0x3C,
+ S_028C80_TILE_MAX(rtex->cmask.slice_tile_max));
+ }
+ if (rtex->fmask.size) {
+ si_pm4_set_reg(pm4, R_028C84_CB_COLOR0_FMASK + cb * 0x3C,
+ offset + (rtex->fmask.offset >> 8));
+ si_pm4_set_reg(pm4, R_028C88_CB_COLOR0_FMASK_SLICE + cb * 0x3C,
+ S_028C88_TILE_MAX(rtex->fmask.slice_tile_max));
+ }
+
/* set CB_COLOR1_INFO for possible dual-src blending */
if (state->nr_cbufs == 1) {
assert(cb == 0);
const struct pipe_framebuffer_state *state)
{
struct r600_screen *rscreen = rctx->screen;
- struct r600_resource_texture *rtex;
+ struct r600_texture *rtex;
struct r600_surface *surf;
unsigned level, pitch, slice, format, tile_mode_index, array_mode;
unsigned macro_aspect, tile_split, stile_split, bankh, bankw, nbanks, pipe_config;
surf = (struct r600_surface *)state->zsbuf;
level = surf->base.u.tex.level;
- rtex = (struct r600_resource_texture*)surf->base.texture;
+ rtex = (struct r600_texture*)surf->base.texture;
- format = si_translate_dbformat(rtex->real_format);
+ format = si_translate_dbformat(rtex->resource.b.b.format);
if (format == V_028040_Z_INVALID) {
- R600_ERR("Invalid DB format: %d, disabling DB.\n", rtex->real_format);
+ R600_ERR("Invalid DB format: %d, disabling DB.\n", rtex->resource.b.b.format);
}
assert(format != V_028040_Z_INVALID);
db_depth_info = S_02803C_ADDR5_SWIZZLE_MASK(1);
z_info = S_028040_FORMAT(format);
+ if (rtex->resource.b.b.nr_samples > 1) {
+ z_info |= S_028040_NUM_SAMPLES(util_logbase2(rtex->resource.b.b.nr_samples));
+ }
+
if (rtex->surface.flags & RADEON_SURF_SBUFFER)
s_info = S_028044_FORMAT(V_028044_STENCIL_8);
else
si_pm4_set_reg(pm4, R_02805C_DB_DEPTH_SLICE, S_02805C_SLICE_TILE_MAX(slice));
}
+#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 r600_context *rctx, 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 r600_context *rctx = (struct r600_context *)ctx;
struct si_pm4_state *pm4 = si_pm4_alloc_state(rctx);
uint32_t tl, br;
- int tl_x, tl_y, br_x, br_y;
+ int tl_x, tl_y, br_x, br_y, nr_samples, i;
if (pm4 == NULL)
return;
si_pm4_inval_fb_cache(pm4, state->nr_cbufs);
+ rctx->flush_and_inv_cb_meta = true;
if (state->zsbuf)
si_pm4_inval_zsbuf_cache(pm4);
/* build states */
rctx->export_16bpc = 0;
- for (int i = 0; i < state->nr_cbufs; i++) {
+ rctx->fb_compressed_cb_mask = 0;
+ for (i = 0; i < state->nr_cbufs; i++) {
+ struct r600_texture *rtex =
+ (struct r600_texture*)state->cbufs[i]->texture;
+
si_cb(rctx, pm4, state, i);
+
+ if (rtex->fmask.size || rtex->cmask.size) {
+ rctx->fb_compressed_cb_mask |= 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));
}
+
assert(!(rctx->export_16bpc & ~0xff));
si_db(rctx, pm4, state);
si_pm4_set_reg(pm4, R_028208_PA_SC_WINDOW_SCISSOR_BR, br);
si_pm4_set_reg(pm4, R_028200_PA_SC_WINDOW_OFFSET, 0x00000000);
si_pm4_set_reg(pm4, R_028230_PA_SC_EDGERULE, 0xAAAAAAAA);
- si_pm4_set_reg(pm4, R_028BE0_PA_SC_AA_CONFIG, 0x00000000);
+
+ 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;
+
+ si_set_msaa_state(rctx, pm4, nr_samples);
+ rctx->fb_log_samples = util_logbase2(nr_samples);
+ rctx->fb_cb0_is_integer = state->nr_cbufs &&
+ util_format_is_pure_integer(state->cbufs[0]->format);
si_pm4_set_state(rctx, framebuffer, pm4);
si_update_fb_rs_state(rctx);
for (i = 0; i < rctx->vertex_elements->count; ++i)
key->vs.instance_divisors[i] = rctx->vertex_elements->elements[i].instance_divisor;
+ if (rctx->queued.named.rasterizer->clip_plane_enable & 0xf0)
+ key->vs.ucps_enabled |= 0x2;
+ if (rctx->queued.named.rasterizer->clip_plane_enable & 0xf)
+ key->vs.ucps_enabled |= 0x1;
} else if (sel->type == PIPE_SHADER_FRAGMENT) {
if (sel->fs_write_all)
key->ps.nr_cbufs = rctx->framebuffer.nr_cbufs;
key->ps.export_16bpc = rctx->export_16bpc;
+
if (rctx->queued.named.rasterizer) {
key->ps.color_two_side = rctx->queued.named.rasterizer->two_side;
key->ps.flatshade = rctx->queued.named.rasterizer->flatshade;
+
+ if (rctx->queued.named.blend) {
+ key->ps.alpha_to_one = rctx->queued.named.blend->alpha_to_one &&
+ rctx->queued.named.rasterizer->multisample_enable &&
+ !rctx->fb_cb0_is_integer;
+ }
}
if (rctx->queued.named.dsa) {
key->ps.alpha_func = rctx->queued.named.dsa->alpha_func;
const struct pipe_sampler_view *state)
{
struct si_pipe_sampler_view *view = CALLOC_STRUCT(si_pipe_sampler_view);
- struct r600_resource_texture *tmp = (struct r600_resource_texture*)texture;
+ struct r600_texture *tmp = (struct r600_texture*)texture;
const struct util_format_description *desc;
unsigned format, num_format;
uint32_t pitch = 0;
S_008F1C_DST_SEL_Y(si_map_swizzle(swizzle[1])) |
S_008F1C_DST_SEL_Z(si_map_swizzle(swizzle[2])) |
S_008F1C_DST_SEL_W(si_map_swizzle(swizzle[3])) |
- S_008F1C_BASE_LEVEL(state->u.tex.first_level) |
- S_008F1C_LAST_LEVEL(state->u.tex.last_level) |
+ S_008F1C_BASE_LEVEL(texture->nr_samples > 1 ?
+ 0 : state->u.tex.first_level) |
+ S_008F1C_LAST_LEVEL(texture->nr_samples > 1 ?
+ util_logbase2(texture->nr_samples) :
+ state->u.tex.last_level) |
S_008F1C_TILING_INDEX(si_tile_mode_index(tmp, 0, false)) |
S_008F1C_POW2_PAD(texture->last_level > 0) |
- S_008F1C_TYPE(si_tex_dim(texture->target)));
+ S_008F1C_TYPE(si_tex_dim(texture->target, texture->nr_samples)));
view->state[4] = (S_008F20_DEPTH(depth - 1) | S_008F20_PITCH(pitch - 1));
view->state[5] = (S_008F24_BASE_ARRAY(state->u.tex.first_layer) |
S_008F24_LAST_ARRAY(state->u.tex.last_layer));
view->state[6] = 0;
view->state[7] = 0;
+ /* Initialize the sampler view for FMASK. */
+ if (tmp->fmask.size) {
+ uint64_t va = r600_resource_va(ctx->screen, texture) + tmp->fmask.offset;
+ uint32_t fmask_format;
+
+ switch (texture->nr_samples) {
+ case 2:
+ fmask_format = V_008F14_IMG_DATA_FORMAT_FMASK8_S2_F2;
+ break;
+ case 4:
+ fmask_format = V_008F14_IMG_DATA_FORMAT_FMASK8_S4_F4;
+ break;
+ case 8:
+ fmask_format = V_008F14_IMG_DATA_FORMAT_FMASK32_S8_F8;
+ break;
+ default:
+ assert(0);
+ fmask_format = V_008F14_IMG_DATA_FORMAT_INVALID;
+ }
+
+ view->fmask_state[0] = va >> 8;
+ view->fmask_state[1] = S_008F14_BASE_ADDRESS_HI(va >> 40) |
+ S_008F14_DATA_FORMAT(fmask_format) |
+ S_008F14_NUM_FORMAT(V_008F14_IMG_NUM_FORMAT_UINT);
+ view->fmask_state[2] = S_008F18_WIDTH(width - 1) |
+ S_008F18_HEIGHT(height - 1);
+ view->fmask_state[3] = S_008F1C_DST_SEL_X(V_008F1C_SQ_SEL_X) |
+ S_008F1C_DST_SEL_Y(V_008F1C_SQ_SEL_X) |
+ S_008F1C_DST_SEL_Z(V_008F1C_SQ_SEL_X) |
+ S_008F1C_DST_SEL_W(V_008F1C_SQ_SEL_X) |
+ S_008F1C_TILING_INDEX(tmp->fmask.tile_mode_index) |
+ S_008F1C_TYPE(si_tex_dim(texture->target, 0));
+ view->fmask_state[4] = S_008F20_PITCH(tmp->fmask.pitch - 1);
+ view->fmask_state[5] = S_008F24_BASE_ARRAY(state->u.tex.first_layer) |
+ S_008F24_LAST_ARRAY(state->u.tex.last_layer);
+ view->fmask_state[6] = 0;
+ view->fmask_state[7] = 0;
+ }
+
return &view->base;
}
return rstate;
}
-static struct si_pm4_state *si_set_sampler_view(struct r600_context *rctx,
- unsigned count,
- struct pipe_sampler_view **views,
- struct r600_textures_info *samplers,
- unsigned user_data_reg)
+/* XXX consider moving this function to si_descriptors.c for gcc to inline
+ * the si_set_sampler_view calls. LTO might help too. */
+static struct si_pm4_state *si_set_sampler_views(struct r600_context *rctx,
+ unsigned shader, unsigned count,
+ struct pipe_sampler_view **views)
{
- struct si_pipe_sampler_view **resource = (struct si_pipe_sampler_view **)views;
+ struct r600_textures_info *samplers = &rctx->samplers[shader];
+ struct si_pipe_sampler_view **rviews = (struct si_pipe_sampler_view **)views;
struct si_pm4_state *pm4 = si_pm4_alloc_state(rctx);
- int i, j;
-
- if (!count)
- goto out;
+ int i;
si_pm4_inval_texture_cache(pm4);
- si_pm4_sh_data_begin(pm4);
for (i = 0; i < count; i++) {
- pipe_sampler_view_reference(
- (struct pipe_sampler_view **)&samplers->views[i],
- views[i]);
-
if (views[i]) {
- struct r600_resource_texture *rtex =
- (struct r600_resource_texture*)views[i]->texture;
+ struct r600_texture *rtex =
+ (struct r600_texture*)views[i]->texture;
if (rtex->is_depth && !rtex->is_flushing_texture) {
samplers->depth_texture_mask |= 1 << i;
} else {
samplers->depth_texture_mask &= ~(1 << i);
}
+ if (rtex->cmask.size || rtex->fmask.size) {
+ samplers->compressed_colortex_mask |= 1 << i;
+ } else {
+ samplers->compressed_colortex_mask &= ~(1 << i);
+ }
+
+ si_set_sampler_view(rctx, shader, i, views[i], rviews[i]->state);
- si_pm4_add_bo(pm4, resource[i]->resource, RADEON_USAGE_READ);
+ if (rtex->fmask.size) {
+ si_set_sampler_view(rctx, shader, FMASK_TEX_OFFSET + i,
+ views[i], rviews[i]->fmask_state);
+ } else {
+ si_set_sampler_view(rctx, shader, FMASK_TEX_OFFSET + i,
+ NULL, NULL);
+ }
} else {
samplers->depth_texture_mask &= ~(1 << i);
- }
-
- for (j = 0; j < Elements(resource[i]->state); ++j) {
- si_pm4_sh_data_add(pm4, resource[i] ? resource[i]->state[j] : 0);
+ samplers->compressed_colortex_mask &= ~(1 << i);
+ si_set_sampler_view(rctx, shader, i, NULL, NULL);
+ si_set_sampler_view(rctx, shader, FMASK_TEX_OFFSET + i,
+ NULL, NULL);
}
}
-
- for (i = count; i < NUM_TEX_UNITS; i++) {
- if (samplers->views[i])
- pipe_sampler_view_reference((struct pipe_sampler_view **)&samplers->views[i], NULL);
+ for (; i < samplers->n_views; i++) {
+ samplers->depth_texture_mask &= ~(1 << i);
+ samplers->compressed_colortex_mask &= ~(1 << i);
+ si_set_sampler_view(rctx, shader, i, NULL, NULL);
+ si_set_sampler_view(rctx, shader, FMASK_TEX_OFFSET + i,
+ NULL, NULL);
}
- si_pm4_sh_data_end(pm4, user_data_reg, SI_SGPR_RESOURCE);
-
-out:
- rctx->ps_samplers.n_views = count;
+ samplers->n_views = count;
return pm4;
}
-static void si_set_vs_sampler_view(struct pipe_context *ctx, unsigned count,
- struct pipe_sampler_view **views)
+static void si_set_vs_sampler_views(struct pipe_context *ctx, unsigned count,
+ struct pipe_sampler_view **views)
{
struct r600_context *rctx = (struct r600_context *)ctx;
struct si_pm4_state *pm4;
- pm4 = si_set_sampler_view(rctx, count, views, &rctx->vs_samplers,
- R_00B130_SPI_SHADER_USER_DATA_VS_0);
+ pm4 = si_set_sampler_views(rctx, PIPE_SHADER_VERTEX, count, views);
si_pm4_set_state(rctx, vs_sampler_views, pm4);
}
-static void si_set_ps_sampler_view(struct pipe_context *ctx, unsigned count,
- struct pipe_sampler_view **views)
+static void si_set_ps_sampler_views(struct pipe_context *ctx, unsigned count,
+ struct pipe_sampler_view **views)
{
struct r600_context *rctx = (struct r600_context *)ctx;
struct si_pm4_state *pm4;
- pm4 = si_set_sampler_view(rctx, count, views, &rctx->ps_samplers,
- R_00B030_SPI_SHADER_USER_DATA_PS_0);
+ pm4 = si_set_sampler_views(rctx, PIPE_SHADER_FRAGMENT, count, views);
si_pm4_set_state(rctx, ps_sampler_views, pm4);
}
-static struct si_pm4_state *si_bind_sampler(struct r600_context *rctx, unsigned count,
- void **states,
- struct r600_textures_info *samplers,
- unsigned user_data_reg)
+static struct si_pm4_state *si_bind_sampler_states(struct r600_context *rctx, unsigned count,
+ void **states,
+ struct r600_textures_info *samplers,
+ unsigned user_data_reg)
{
struct si_pipe_sampler_state **rstates = (struct si_pipe_sampler_state **)states;
struct si_pm4_state *pm4 = si_pm4_alloc_state(rctx);
return pm4;
}
-static void si_bind_vs_sampler(struct pipe_context *ctx, unsigned count, void **states)
+static void si_bind_vs_sampler_states(struct pipe_context *ctx, unsigned count, void **states)
{
struct r600_context *rctx = (struct r600_context *)ctx;
struct si_pm4_state *pm4;
- pm4 = si_bind_sampler(rctx, count, states, &rctx->vs_samplers,
+ pm4 = si_bind_sampler_states(rctx, count, states, &rctx->samplers[PIPE_SHADER_VERTEX],
R_00B130_SPI_SHADER_USER_DATA_VS_0);
si_pm4_set_state(rctx, vs_sampler, pm4);
}
-static void si_bind_ps_sampler(struct pipe_context *ctx, unsigned count, void **states)
+static void si_bind_ps_sampler_states(struct pipe_context *ctx, unsigned count, void **states)
{
struct r600_context *rctx = (struct r600_context *)ctx;
struct si_pm4_state *pm4;
- pm4 = si_bind_sampler(rctx, count, states, &rctx->ps_samplers,
+ pm4 = si_bind_sampler_states(rctx, count, states, &rctx->samplers[PIPE_SHADER_FRAGMENT],
R_00B030_SPI_SHADER_USER_DATA_PS_0);
si_pm4_set_state(rctx, ps_sampler, pm4);
}
-static void si_set_sample_mask(struct pipe_context *pipe, unsigned sample_mask)
+static void si_set_sample_mask(struct pipe_context *ctx, unsigned sample_mask)
{
+ struct r600_context *rctx = (struct r600_context *)ctx;
+ struct si_pm4_state *pm4 = si_pm4_alloc_state(rctx);
+ uint16_t mask = sample_mask;
+
+ if (pm4 == NULL)
+ return;
+
+ si_pm4_set_reg(pm4, R_028C38_PA_SC_AA_MASK_X0Y0_X1Y0, mask | (mask << 16));
+ si_pm4_set_reg(pm4, R_028C3C_PA_SC_AA_MASK_X0Y1_X1Y1, mask | (mask << 16));
+
+ si_pm4_set_state(rctx, sample_mask, pm4);
}
static void si_delete_sampler_state(struct pipe_context *ctx, void *state)
si_pm4_set_state(rctx, texture_barrier, pm4);
}
+static void *si_create_blend_custom(struct r600_context *rctx, unsigned mode)
+{
+ struct pipe_blend_state blend;
+
+ memset(&blend, 0, sizeof(blend));
+ blend.independent_blend_enable = true;
+ blend.rt[0].colormask = 0xf;
+ return si_create_blend_state_mode(&rctx->context, &blend, mode);
+}
+
void si_init_state_functions(struct r600_context *rctx)
{
+ int i;
+
rctx->context.create_blend_state = si_create_blend_state;
rctx->context.bind_blend_state = si_bind_blend_state;
rctx->context.delete_blend_state = si_delete_blend_state;
rctx->context.create_depth_stencil_alpha_state = si_create_dsa_state;
rctx->context.bind_depth_stencil_alpha_state = si_bind_dsa_state;
rctx->context.delete_depth_stencil_alpha_state = si_delete_dsa_state;
- rctx->custom_dsa_flush_depth_stencil = si_create_db_flush_dsa(rctx, true, true);
- rctx->custom_dsa_flush_depth = si_create_db_flush_dsa(rctx, true, false);
- rctx->custom_dsa_flush_stencil = si_create_db_flush_dsa(rctx, false, true);
- rctx->custom_dsa_flush_inplace = si_create_db_flush_dsa(rctx, false, false);
+
+ for (i = 0; i < 8; i++) {
+ rctx->custom_dsa_flush_depth_stencil[i] = si_create_db_flush_dsa(rctx, true, true, i);
+ rctx->custom_dsa_flush_depth[i] = si_create_db_flush_dsa(rctx, true, false, i);
+ rctx->custom_dsa_flush_stencil[i] = si_create_db_flush_dsa(rctx, false, true, i);
+ }
+ rctx->custom_dsa_flush_inplace = si_create_db_flush_dsa(rctx, false, false, 0);
+ rctx->custom_blend_resolve = si_create_blend_custom(rctx, V_028808_CB_RESOLVE);
+ rctx->custom_blend_decompress = si_create_blend_custom(rctx, V_028808_CB_FMASK_DECOMPRESS);
rctx->context.set_clip_state = si_set_clip_state;
rctx->context.set_scissor_states = si_set_scissor_states;
rctx->context.set_stencil_ref = si_set_pipe_stencil_ref;
rctx->context.set_framebuffer_state = si_set_framebuffer_state;
+ rctx->context.get_sample_position = si_get_sample_position;
rctx->context.create_vs_state = si_create_vs_state;
rctx->context.create_fs_state = si_create_fs_state;
rctx->context.delete_fs_state = si_delete_ps_shader;
rctx->context.create_sampler_state = si_create_sampler_state;
- rctx->context.bind_vertex_sampler_states = si_bind_vs_sampler;
- rctx->context.bind_fragment_sampler_states = si_bind_ps_sampler;
+ rctx->context.bind_vertex_sampler_states = si_bind_vs_sampler_states;
+ rctx->context.bind_fragment_sampler_states = si_bind_ps_sampler_states;
rctx->context.delete_sampler_state = si_delete_sampler_state;
rctx->context.create_sampler_view = si_create_sampler_view;
- rctx->context.set_vertex_sampler_views = si_set_vs_sampler_view;
- rctx->context.set_fragment_sampler_views = si_set_ps_sampler_view;
+ rctx->context.set_vertex_sampler_views = si_set_vs_sampler_views;
+ rctx->context.set_fragment_sampler_views = si_set_ps_sampler_views;
rctx->context.sampler_view_destroy = si_sampler_view_destroy;
rctx->context.set_sample_mask = si_set_sample_mask;
S_028AA8_PRIMGROUP_SIZE(63));
si_pm4_set_reg(pm4, R_028AB4_VGT_REUSE_OFF, 0x00000000);
si_pm4_set_reg(pm4, R_028AB8_VGT_VTX_CNT_EN, 0x0);
- si_pm4_set_reg(pm4, R_008A14_PA_CL_ENHANCE, S_008A14_NUM_CLIP_SEQ(3) |
- S_008A14_CLIP_VTX_REORDER_ENA(1));
+ if (rctx->chip_class < CIK)
+ si_pm4_set_reg(pm4, R_008A14_PA_CL_ENHANCE, S_008A14_NUM_CLIP_SEQ(3) |
+ S_008A14_CLIP_VTX_REORDER_ENA(1));
si_pm4_set_reg(pm4, R_028B54_VGT_SHADER_STAGES_EN, 0);
si_pm4_set_reg(pm4, R_028BD4_PA_SC_CENTROID_PRIORITY_0, 0x76543210);
si_pm4_set_reg(pm4, R_028BD8_PA_SC_CENTROID_PRIORITY_1, 0xfedcba98);
- si_pm4_set_reg(pm4, R_028804_DB_EQAA, 0x110000);
-
si_pm4_set_reg(pm4, R_02882C_PA_SU_PRIM_FILTER_CNTL, 0);
- switch (rctx->screen->family) {
- case CHIP_TAHITI:
- case CHIP_PITCAIRN:
- si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x2a00126a);
- break;
- case CHIP_VERDE:
- si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x0000124a);
- break;
- case CHIP_OLAND:
- si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x00000082);
- break;
- case CHIP_HAINAN:
- si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x00000000);
- break;
- default:
- si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x00000000);
- break;
+ if (rctx->chip_class >= CIK) {
+ switch (rctx->screen->family) {
+ case CHIP_BONAIRE:
+ si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x16000012);
+ si_pm4_set_reg(pm4, R_028354_PA_SC_RASTER_CONFIG_1, 0x00000000);
+ break;
+ case CHIP_KAVERI:
+ /* XXX todo */
+ case CHIP_KABINI:
+ /* 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);
+ break;
+ }
+ } else {
+ switch (rctx->screen->family) {
+ case CHIP_TAHITI:
+ case CHIP_PITCAIRN:
+ si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x2a00126a);
+ break;
+ case CHIP_VERDE:
+ si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x0000124a);
+ break;
+ case CHIP_OLAND:
+ si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x00000082);
+ break;
+ case CHIP_HAINAN:
+ si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x00000000);
+ break;
+ default:
+ si_pm4_set_reg(pm4, R_028350_PA_SC_RASTER_CONFIG, 0x00000000);
+ break;
+ }
}
si_pm4_set_state(rctx, init, pm4);
projects / mesa.git / blobdiff