/*
* Core functions
*/
static void cayman_gpu_init(struct radeon_device *rdev)
{
u32 gb_addr_config = 0;
u32 mc_shared_chmap, mc_arb_ramcfg;
u32 cgts_tcc_disable;
u32 sx_debug_1;
u32 smx_dc_ctl0;
u32 cgts_sm_ctrl_reg;
u32 hdp_host_path_cntl;
u32 tmp;
u32 disabled_rb_mask;
int i, j;
switch (rdev->family) {
case CHIP_CAYMAN:
rdev->config.cayman.max_shader_engines = 2;
rdev->config.cayman.max_pipes_per_simd = 4;
rdev->config.cayman.max_tile_pipes = 8;
rdev->config.cayman.max_simds_per_se = 12;
rdev->config.cayman.max_backends_per_se = 4;
rdev->config.cayman.max_texture_channel_caches = 8;
rdev->config.cayman.max_gprs = 256;
rdev->config.cayman.max_threads = 256;
rdev->config.cayman.max_gs_threads = 32;
rdev->config.cayman.max_stack_entries = 512;
rdev->config.cayman.sx_num_of_sets = 8;
rdev->config.cayman.sx_max_export_size = 256;
rdev->config.cayman.sx_max_export_pos_size = 64;
rdev->config.cayman.sx_max_export_smx_size = 192;
rdev->config.cayman.max_hw_contexts = 8;
rdev->config.cayman.sq_num_cf_insts = 2;
rdev->config.cayman.sc_prim_fifo_size = 0x100;
rdev->config.cayman.sc_hiz_tile_fifo_size = 0x30;
rdev->config.cayman.sc_earlyz_tile_fifo_size = 0x130;
gb_addr_config = CAYMAN_GB_ADDR_CONFIG_GOLDEN;
break;
case CHIP_ARUBA:
default:
rdev->config.cayman.max_shader_engines = 1;
rdev->config.cayman.max_pipes_per_simd = 4;
rdev->config.cayman.max_tile_pipes = 2;
if ((rdev->pdev->device == 0x9900) ||
(rdev->pdev->device == 0x9901) ||
(rdev->pdev->device == 0x9905) ||
(rdev->pdev->device == 0x9906) ||
(rdev->pdev->device == 0x9907) ||
(rdev->pdev->device == 0x9908) ||
(rdev->pdev->device == 0x9909) ||
(rdev->pdev->device == 0x9910) ||
(rdev->pdev->device == 0x9917)) {
rdev->config.cayman.max_simds_per_se = 6;
rdev->config.cayman.max_backends_per_se = 2;
} else if ((rdev->pdev->device == 0x9903) ||
(rdev->pdev->device == 0x9904) ||
(rdev->pdev->device == 0x990A) ||
(rdev->pdev->device == 0x9913) ||
(rdev->pdev->device == 0x9918)) {
rdev->config.cayman.max_simds_per_se = 4;
rdev->config.cayman.max_backends_per_se = 2;
} else if ((rdev->pdev->device == 0x9919) ||
(rdev->pdev->device == 0x9990) ||
(rdev->pdev->device == 0x9991) ||
(rdev->pdev->device == 0x9994) ||
(rdev->pdev->device == 0x99A0)) {
rdev->config.cayman.max_simds_per_se = 3;
rdev->config.cayman.max_backends_per_se = 1;
} else {
rdev->config.cayman.max_simds_per_se = 2;
rdev->config.cayman.max_backends_per_se = 1;
}
rdev->config.cayman.max_texture_channel_caches = 2;
rdev->config.cayman.max_gprs = 256;
rdev->config.cayman.max_threads = 256;
rdev->config.cayman.max_gs_threads = 32;
rdev->config.cayman.max_stack_entries = 512;
rdev->config.cayman.sx_num_of_sets = 8;
rdev->config.cayman.sx_max_export_size = 256;
rdev->config.cayman.sx_max_export_pos_size = 64;
rdev->config.cayman.sx_max_export_smx_size = 192;
rdev->config.cayman.max_hw_contexts = 8;
rdev->config.cayman.sq_num_cf_insts = 2;
rdev->config.cayman.sc_prim_fifo_size = 0x40;
rdev->config.cayman.sc_hiz_tile_fifo_size = 0x30;
rdev->config.cayman.sc_earlyz_tile_fifo_size = 0x130;
gb_addr_config = ARUBA_GB_ADDR_CONFIG_GOLDEN;
break;
}
/* Initialize HDP */
for (i = 0, j = 0; i < 32; i++, j += 0x18) {
WREG32((0x2c14 + j), 0x00000000);
WREG32((0x2c18 + j), 0x00000000);
WREG32((0x2c1c + j), 0x00000000);
WREG32((0x2c20 + j), 0x00000000);
WREG32((0x2c24 + j), 0x00000000);
}
WREG32(GRBM_CNTL, GRBM_READ_TIMEOUT(0xff));
evergreen_fix_pci_max_read_req_size(rdev);
mc_shared_chmap = RREG32(MC_SHARED_CHMAP);
mc_arb_ramcfg = RREG32(MC_ARB_RAMCFG);
tmp = (mc_arb_ramcfg & NOOFCOLS_MASK) >> NOOFCOLS_SHIFT;
rdev->config.cayman.mem_row_size_in_kb = (4 * (1 << (8 + tmp))) / 1024;
if (rdev->config.cayman.mem_row_size_in_kb > 4)
rdev->config.cayman.mem_row_size_in_kb = 4;
/* XXX use MC settings? */
rdev->config.cayman.shader_engine_tile_size = 32;
rdev->config.cayman.num_gpus = 1;
rdev->config.cayman.multi_gpu_tile_size = 64;
tmp = (gb_addr_config & NUM_PIPES_MASK) >> NUM_PIPES_SHIFT;
rdev->config.cayman.num_tile_pipes = (1 << tmp);
tmp = (gb_addr_config & PIPE_INTERLEAVE_SIZE_MASK) >> PIPE_INTERLEAVE_SIZE_SHIFT;
rdev->config.cayman.mem_max_burst_length_bytes = (tmp + 1) * 256;
tmp = (gb_addr_config & NUM_SHADER_ENGINES_MASK) >> NUM_SHADER_ENGINES_SHIFT;
rdev->config.cayman.num_shader_engines = tmp + 1;
tmp = (gb_addr_config & NUM_GPUS_MASK) >> NUM_GPUS_SHIFT;
rdev->config.cayman.num_gpus = tmp + 1;
tmp = (gb_addr_config & MULTI_GPU_TILE_SIZE_MASK) >> MULTI_GPU_TILE_SIZE_SHIFT;
rdev->config.cayman.multi_gpu_tile_size = 1 << tmp;
tmp = (gb_addr_config & ROW_SIZE_MASK) >> ROW_SIZE_SHIFT;
rdev->config.cayman.mem_row_size_in_kb = 1 << tmp;
/* setup tiling info dword. gb_addr_config is not adequate since it does
* not have bank info, so create a custom tiling dword.
* bits 3:0 num_pipes
* bits 7:4 num_banks
* bits 11:8 group_size
* bits 15:12 row_size
*/
rdev->config.cayman.tile_config = 0;
switch (rdev->config.cayman.num_tile_pipes) {
case 1:
default:
rdev->config.cayman.tile_config |= (0 << 0);
break;
case 2:
rdev->config.cayman.tile_config |= (1 << 0);
break;
case 4:
rdev->config.cayman.tile_config |= (2 << 0);
break;
case 8:
rdev->config.cayman.tile_config |= (3 << 0);
break;
}
/* num banks is 8 on all fusion asics. 0 = 4, 1 = 8, 2 = 16 */
if (rdev->flags & RADEON_IS_IGP)
rdev->config.cayman.tile_config |= 1 << 4;
else {
switch ((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT) {
case 0: /* four banks */
rdev->config.cayman.tile_config |= 0 << 4;
break;
case 1: /* eight banks */
rdev->config.cayman.tile_config |= 1 << 4;
break;
case 2: /* sixteen banks */
default:
rdev->config.cayman.tile_config |= 2 << 4;
break;
}
}
rdev->config.cayman.tile_config |=
((gb_addr_config & PIPE_INTERLEAVE_SIZE_MASK) >> PIPE_INTERLEAVE_SIZE_SHIFT) << 8;
rdev->config.cayman.tile_config |=
((gb_addr_config & ROW_SIZE_MASK) >> ROW_SIZE_SHIFT) << 12;
tmp = 0;
for (i = (rdev->config.cayman.max_shader_engines - 1); i >= 0; i--) {
u32 rb_disable_bitmap;
WREG32(GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES | SE_INDEX(i));
WREG32(RLC_GFX_INDEX, INSTANCE_BROADCAST_WRITES | SE_INDEX(i));
rb_disable_bitmap = (RREG32(CC_RB_BACKEND_DISABLE) & 0x00ff0000) >> 16;
tmp <<= 4;
tmp |= rb_disable_bitmap;
}
/* enabled rb are just the one not disabled :) */
disabled_rb_mask = tmp;
WREG32(GRBM_GFX_INDEX, INSTANCE_BROADCAST_WRITES | SE_BROADCAST_WRITES);
WREG32(RLC_GFX_INDEX, INSTANCE_BROADCAST_WRITES | SE_BROADCAST_WRITES);
WREG32(GB_ADDR_CONFIG, gb_addr_config);
WREG32(DMIF_ADDR_CONFIG, gb_addr_config);
WREG32(HDP_ADDR_CONFIG, gb_addr_config);
tmp = gb_addr_config & NUM_PIPES_MASK;
tmp = r6xx_remap_render_backend(rdev, tmp,
rdev->config.cayman.max_backends_per_se *
rdev->config.cayman.max_shader_engines,
CAYMAN_MAX_BACKENDS, disabled_rb_mask);
WREG32(GB_BACKEND_MAP, tmp);
cgts_tcc_disable = 0xffff0000;
for (i = 0; i < rdev->config.cayman.max_texture_channel_caches; i++)
cgts_tcc_disable &= ~(1 << (16 + i));
WREG32(CGTS_TCC_DISABLE, cgts_tcc_disable);
WREG32(CGTS_SYS_TCC_DISABLE, cgts_tcc_disable);
WREG32(CGTS_USER_SYS_TCC_DISABLE, cgts_tcc_disable);
WREG32(CGTS_USER_TCC_DISABLE, cgts_tcc_disable);
/* reprogram the shader complex */
cgts_sm_ctrl_reg = RREG32(CGTS_SM_CTRL_REG);
for (i = 0; i < 16; i++)
WREG32(CGTS_SM_CTRL_REG, OVERRIDE);
WREG32(CGTS_SM_CTRL_REG, cgts_sm_ctrl_reg);
/* set HW defaults for 3D engine */
WREG32(CP_MEQ_THRESHOLDS, MEQ1_START(0x30) | MEQ2_START(0x60));
sx_debug_1 = RREG32(SX_DEBUG_1);
sx_debug_1 |= ENABLE_NEW_SMX_ADDRESS;
WREG32(SX_DEBUG_1, sx_debug_1);
smx_dc_ctl0 = RREG32(SMX_DC_CTL0);
smx_dc_ctl0 &= ~NUMBER_OF_SETS(0x1ff);
smx_dc_ctl0 |= NUMBER_OF_SETS(rdev->config.cayman.sx_num_of_sets);
WREG32(SMX_DC_CTL0, smx_dc_ctl0);
WREG32(SPI_CONFIG_CNTL_1, VTX_DONE_DELAY(4) | CRC_SIMD_ID_WADDR_DISABLE);
/* need to be explicitly zero-ed */
WREG32(VGT_OFFCHIP_LDS_BASE, 0);
WREG32(SQ_LSTMP_RING_BASE, 0);
WREG32(SQ_HSTMP_RING_BASE, 0);
WREG32(SQ_ESTMP_RING_BASE, 0);
WREG32(SQ_GSTMP_RING_BASE, 0);
WREG32(SQ_VSTMP_RING_BASE, 0);
WREG32(SQ_PSTMP_RING_BASE, 0);
WREG32(TA_CNTL_AUX, DISABLE_CUBE_ANISO);
WREG32(SX_EXPORT_BUFFER_SIZES, (COLOR_BUFFER_SIZE((rdev->config.cayman.sx_max_export_size / 4) - 1) |
POSITION_BUFFER_SIZE((rdev->config.cayman.sx_max_export_pos_size / 4) - 1) |
SMX_BUFFER_SIZE((rdev->config.cayman.sx_max_export_smx_size / 4) - 1)));
WREG32(PA_SC_FIFO_SIZE, (SC_PRIM_FIFO_SIZE(rdev->config.cayman.sc_prim_fifo_size) |
SC_HIZ_TILE_FIFO_SIZE(rdev->config.cayman.sc_hiz_tile_fifo_size) |
SC_EARLYZ_TILE_FIFO_SIZE(rdev->config.cayman.sc_earlyz_tile_fifo_size)));
WREG32(VGT_NUM_INSTANCES, 1);
WREG32(CP_PERFMON_CNTL, 0);
WREG32(SQ_MS_FIFO_SIZES, (CACHE_FIFO_SIZE(16 * rdev->config.cayman.sq_num_cf_insts) |
FETCH_FIFO_HIWATER(0x4) |
DONE_FIFO_HIWATER(0xe0) |
ALU_UPDATE_FIFO_HIWATER(0x8)));
WREG32(SQ_GPR_RESOURCE_MGMT_1, NUM_CLAUSE_TEMP_GPRS(4));
WREG32(SQ_CONFIG, (VC_ENABLE |
EXPORT_SRC_C |
GFX_PRIO(0) |
CS1_PRIO(0) |
CS2_PRIO(1)));
WREG32(SQ_DYN_GPR_CNTL_PS_FLUSH_REQ, DYN_GPR_ENABLE);
WREG32(PA_SC_FORCE_EOV_MAX_CNTS, (FORCE_EOV_MAX_CLK_CNT(4095) |
FORCE_EOV_MAX_REZ_CNT(255)));
WREG32(VGT_CACHE_INVALIDATION, CACHE_INVALIDATION(VC_AND_TC) |
AUTO_INVLD_EN(ES_AND_GS_AUTO));
WREG32(VGT_GS_VERTEX_REUSE, 16);
WREG32(PA_SC_LINE_STIPPLE_STATE, 0);
WREG32(CB_PERF_CTR0_SEL_0, 0);
WREG32(CB_PERF_CTR0_SEL_1, 0);
WREG32(CB_PERF_CTR1_SEL_0, 0);
WREG32(CB_PERF_CTR1_SEL_1, 0);
WREG32(CB_PERF_CTR2_SEL_0, 0);
WREG32(CB_PERF_CTR2_SEL_1, 0);
WREG32(CB_PERF_CTR3_SEL_0, 0);
WREG32(CB_PERF_CTR3_SEL_1, 0);
tmp = RREG32(HDP_MISC_CNTL);
tmp |= HDP_FLUSH_INVALIDATE_CACHE;
WREG32(HDP_MISC_CNTL, tmp);
hdp_host_path_cntl = RREG32(HDP_HOST_PATH_CNTL);
WREG32(HDP_HOST_PATH_CNTL, hdp_host_path_cntl);
WREG32(PA_CL_ENHANCE, CLIP_VTX_REORDER_ENA | NUM_CLIP_SEQ(3));
udelay(50);
}
static void
si_write_harvested_raster_configs(struct radv_physical_device *physical_device,
struct radeon_winsys_cs *cs,
unsigned raster_config,
unsigned raster_config_1)
{
unsigned sh_per_se = MAX2(physical_device->rad_info.max_sh_per_se, 1);
unsigned num_se = MAX2(physical_device->rad_info.max_se, 1);
unsigned rb_mask = physical_device->rad_info.enabled_rb_mask;
unsigned num_rb = MIN2(physical_device->rad_info.num_render_backends, 16);
unsigned rb_per_pkr = MIN2(num_rb / num_se / sh_per_se, 2);
unsigned rb_per_se = num_rb / num_se;
unsigned se_mask[4];
unsigned se;
se_mask[0] = ((1 << rb_per_se) - 1) & rb_mask;
se_mask[1] = (se_mask[0] << rb_per_se) & rb_mask;
se_mask[2] = (se_mask[1] << rb_per_se) & rb_mask;
se_mask[3] = (se_mask[2] << rb_per_se) & rb_mask;
assert(num_se == 1 || num_se == 2 || num_se == 4);
assert(sh_per_se == 1 || sh_per_se == 2);
assert(rb_per_pkr == 1 || rb_per_pkr == 2);
/* XXX: I can't figure out what the *_XSEL and *_YSEL
* fields are for, so I'm leaving them as their default
* values. */
if ((num_se > 2) && ((!se_mask[0] && !se_mask[1]) ||
(!se_mask[2] && !se_mask[3]))) {
raster_config_1 &= C_028354_SE_PAIR_MAP;
if (!se_mask[0] && !se_mask[1]) {
raster_config_1 |=
S_028354_SE_PAIR_MAP(V_028354_RASTER_CONFIG_SE_PAIR_MAP_3);
} else {
raster_config_1 |=
S_028354_SE_PAIR_MAP(V_028354_RASTER_CONFIG_SE_PAIR_MAP_0);
}
}
for (se = 0; se < num_se; se++) {
unsigned raster_config_se = raster_config;
unsigned pkr0_mask = ((1 << rb_per_pkr) - 1) << (se * rb_per_se);
unsigned pkr1_mask = pkr0_mask << rb_per_pkr;
int idx = (se / 2) * 2;
if ((num_se > 1) && (!se_mask[idx] || !se_mask[idx + 1])) {
raster_config_se &= C_028350_SE_MAP;
if (!se_mask[idx]) {
raster_config_se |=
S_028350_SE_MAP(V_028350_RASTER_CONFIG_SE_MAP_3);
} else {
raster_config_se |=
S_028350_SE_MAP(V_028350_RASTER_CONFIG_SE_MAP_0);
}
}
pkr0_mask &= rb_mask;
pkr1_mask &= rb_mask;
if (rb_per_se > 2 && (!pkr0_mask || !pkr1_mask)) {
raster_config_se &= C_028350_PKR_MAP;
if (!pkr0_mask) {
raster_config_se |=
S_028350_PKR_MAP(V_028350_RASTER_CONFIG_PKR_MAP_3);
} else {
raster_config_se |=
S_028350_PKR_MAP(V_028350_RASTER_CONFIG_PKR_MAP_0);
}
}
if (rb_per_se >= 2) {
unsigned rb0_mask = 1 << (se * rb_per_se);
unsigned rb1_mask = rb0_mask << 1;
rb0_mask &= rb_mask;
rb1_mask &= rb_mask;
if (!rb0_mask || !rb1_mask) {
raster_config_se &= C_028350_RB_MAP_PKR0;
if (!rb0_mask) {
raster_config_se |=
S_028350_RB_MAP_PKR0(V_028350_RASTER_CONFIG_RB_MAP_3);
} else {
raster_config_se |=
S_028350_RB_MAP_PKR0(V_028350_RASTER_CONFIG_RB_MAP_0);
}
}
if (rb_per_se > 2) {
rb0_mask = 1 << (se * rb_per_se + rb_per_pkr);
rb1_mask = rb0_mask << 1;
rb0_mask &= rb_mask;
rb1_mask &= rb_mask;
if (!rb0_mask || !rb1_mask) {
raster_config_se &= C_028350_RB_MAP_PKR1;
if (!rb0_mask) {
raster_config_se |=
S_028350_RB_MAP_PKR1(V_028350_RASTER_CONFIG_RB_MAP_3);
} else {
raster_config_se |=
S_028350_RB_MAP_PKR1(V_028350_RASTER_CONFIG_RB_MAP_0);
}
}
}
}
/* GRBM_GFX_INDEX has a different offset on SI and CI+ */
if (physical_device->rad_info.chip_class < CIK)
radeon_set_config_reg(cs, GRBM_GFX_INDEX,
SE_INDEX(se) | SH_BROADCAST_WRITES |
INSTANCE_BROADCAST_WRITES);
else
radeon_set_uconfig_reg(cs, R_030800_GRBM_GFX_INDEX,
S_030800_SE_INDEX(se) | S_030800_SH_BROADCAST_WRITES(1) |
S_030800_INSTANCE_BROADCAST_WRITES(1));
radeon_set_context_reg(cs, R_028350_PA_SC_RASTER_CONFIG, raster_config_se);
if (physical_device->rad_info.chip_class >= CIK)
radeon_set_context_reg(cs, R_028354_PA_SC_RASTER_CONFIG_1, raster_config_1);
}
/* GRBM_GFX_INDEX has a different offset on SI and CI+ */
if (physical_device->rad_info.chip_class < CIK)
radeon_set_config_reg(cs, GRBM_GFX_INDEX,
SE_BROADCAST_WRITES | SH_BROADCAST_WRITES |
INSTANCE_BROADCAST_WRITES);
else
radeon_set_uconfig_reg(cs, R_030800_GRBM_GFX_INDEX,
S_030800_SE_BROADCAST_WRITES(1) | S_030800_SH_BROADCAST_WRITES(1) |
S_030800_INSTANCE_BROADCAST_WRITES(1));
}
