diff options
-rw-r--r-- | src/gallium/drivers/radeonsi/si_shader.c | 12 | ||||
-rw-r--r-- | src/gallium/drivers/radeonsi/si_shader.h | 2 | ||||
-rw-r--r-- | src/gallium/drivers/radeonsi/si_state_shaders.c | 30 |
3 files changed, 28 insertions, 16 deletions
diff --git a/src/gallium/drivers/radeonsi/si_shader.c b/src/gallium/drivers/radeonsi/si_shader.c index f322c4e3f5a..85113c00f5b 100644 --- a/src/gallium/drivers/radeonsi/si_shader.c +++ b/src/gallium/drivers/radeonsi/si_shader.c @@ -1302,18 +1302,8 @@ static void si_llvm_init_export_args(struct lp_build_tgsi_context *bld_base, if (si_shader_ctx->type == TGSI_PROCESSOR_FRAGMENT) { int cbuf = target - V_008DFC_SQ_EXP_MRT; - if (cbuf >= 0 && cbuf < 8) { + if (cbuf >= 0 && cbuf < 8) compressed = (si_shader_ctx->shader->key.ps.export_16bpc >> cbuf) & 0x1; - - if (compressed) - si_shader_ctx->shader->spi_shader_col_format |= - V_028714_SPI_SHADER_FP16_ABGR << (4 * cbuf); - else - si_shader_ctx->shader->spi_shader_col_format |= - V_028714_SPI_SHADER_32_ABGR << (4 * cbuf); - - si_shader_ctx->shader->cb_shader_mask |= 0xf << (4 * cbuf); - } } /* Set COMPR flag */ diff --git a/src/gallium/drivers/radeonsi/si_shader.h b/src/gallium/drivers/radeonsi/si_shader.h index 3d14c79b803..b89d3b29e69 100644 --- a/src/gallium/drivers/radeonsi/si_shader.h +++ b/src/gallium/drivers/radeonsi/si_shader.h @@ -277,8 +277,6 @@ struct si_shader { unsigned spi_ps_input_ena; unsigned float_mode; unsigned scratch_bytes_per_wave; - unsigned spi_shader_col_format; - unsigned cb_shader_mask; union si_shader_key key; unsigned nparam; diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c b/src/gallium/drivers/radeonsi/si_state_shaders.c index 4b007ec8878..b08b035f62f 100644 --- a/src/gallium/drivers/radeonsi/si_state_shaders.c +++ b/src/gallium/drivers/radeonsi/si_state_shaders.c @@ -387,6 +387,8 @@ static void si_shader_ps(struct si_shader *shader) struct tgsi_shader_info *info = &shader->selector->info; struct si_pm4_state *pm4; unsigned i, spi_ps_in_control; + unsigned spi_shader_col_format = 0, cb_shader_mask = 0; + unsigned colors_written, export_16bpc; unsigned num_sgprs, num_user_sgprs; unsigned spi_baryc_cntl = 0; uint64_t va; @@ -422,12 +424,35 @@ static void si_shader_ps(struct si_shader *shader) } } + /* Find out what SPI_SHADER_COL_FORMAT and CB_SHADER_MASK should be. */ + colors_written = info->colors_written; + export_16bpc = shader->key.ps.export_16bpc; + + if (info->colors_written == 0x0) { + colors_written = 0x1; /* dummy export */ + export_16bpc = 0; + } else if (info->colors_written == 0x1 && + info->properties[TGSI_PROPERTY_FS_COLOR0_WRITES_ALL_CBUFS]) { + colors_written |= (1 << (shader->key.ps.last_cbuf + 1)) - 1; + } + + while (colors_written) { + i = u_bit_scan(&colors_written); + if (export_16bpc & (1 << i)) + spi_shader_col_format |= V_028714_SPI_SHADER_FP16_ABGR << (4 * i); + else + spi_shader_col_format |= V_028714_SPI_SHADER_32_ABGR << (4 * i); + cb_shader_mask |= 0xf << (4 * i); + } + + /* Set interpolation controls. */ has_centroid = G_0286CC_PERSP_CENTROID_ENA(shader->spi_ps_input_ena) || G_0286CC_LINEAR_CENTROID_ENA(shader->spi_ps_input_ena); spi_ps_in_control = S_0286D8_NUM_INTERP(shader->nparam) | S_0286D8_BC_OPTIMIZE_DISABLE(has_centroid); + /* Set registers. */ si_pm4_set_reg(pm4, R_0286E0_SPI_BARYC_CNTL, spi_baryc_cntl); si_pm4_set_reg(pm4, R_0286D8_SPI_PS_IN_CONTROL, spi_ps_in_control); @@ -437,9 +462,8 @@ static void si_shader_ps(struct si_shader *shader) info->writes_z ? V_028710_SPI_SHADER_32_R : V_028710_SPI_SHADER_ZERO); - si_pm4_set_reg(pm4, R_028714_SPI_SHADER_COL_FORMAT, - shader->spi_shader_col_format); - si_pm4_set_reg(pm4, R_02823C_CB_SHADER_MASK, shader->cb_shader_mask); + si_pm4_set_reg(pm4, R_028714_SPI_SHADER_COL_FORMAT, spi_shader_col_format); + si_pm4_set_reg(pm4, R_02823C_CB_SHADER_MASK, cb_shader_mask); va = shader->bo->gpu_address; si_pm4_add_bo(pm4, shader->bo, RADEON_USAGE_READ, RADEON_PRIO_USER_SHADER); |