diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 103 |
1 files changed, 65 insertions, 38 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 2616f1b59bbd..a5492e375f29 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -48,8 +48,15 @@ #include "amdgpu_ras.h" -#include "sdma0/sdma0_4_0_offset.h" -#include "sdma1/sdma1_4_0_offset.h" +#include "sdma0/sdma0_4_2_offset.h" +#include "sdma1/sdma1_4_2_offset.h" +#include "sdma2/sdma2_4_2_2_offset.h" +#include "sdma3/sdma3_4_2_2_offset.h" +#include "sdma4/sdma4_4_2_2_offset.h" +#include "sdma5/sdma5_4_2_2_offset.h" +#include "sdma6/sdma6_4_2_2_offset.h" +#include "sdma7/sdma7_4_2_2_offset.h" + #define GFX9_NUM_GFX_RINGS 1 #define GFX9_MEC_HPD_SIZE 4096 #define RLCG_UCODE_LOADING_START_ADDRESS 0x00002000L @@ -981,8 +988,7 @@ static void gfx_v9_0_check_fw_write_wait(struct amdgpu_device *adev) (adev->gfx.mec_feature_version < 46) || (adev->gfx.pfp_fw_version < 0x000000b7) || (adev->gfx.pfp_feature_version < 46)) - DRM_WARN_ONCE("Warning: check cp_fw_version and update it to realize \ - GRBM requires 1-cycle delay in cp firmware\n"); + DRM_WARN_ONCE("CP firmware version too old, please update!"); switch (adev->asic_type) { case CHIP_VEGA10: @@ -1042,17 +1048,10 @@ static void gfx_v9_0_check_if_need_gfxoff(struct amdgpu_device *adev) case CHIP_VEGA20: break; case CHIP_RAVEN: - /* Disable GFXOFF on original raven. There are combinations - * of sbios and platforms that are not stable. - */ - if (!(adev->rev_id >= 0x8 || adev->pdev->device == 0x15d8)) - adev->pm.pp_feature &= ~PP_GFXOFF_MASK; - else if (!(adev->rev_id >= 0x8 || adev->pdev->device == 0x15d8) - &&((adev->gfx.rlc_fw_version != 106 && - adev->gfx.rlc_fw_version < 531) || - (adev->gfx.rlc_fw_version == 53815) || - (adev->gfx.rlc_feature_version < 1) || - !adev->gfx.rlc.is_rlc_v2_1)) + if (!(adev->rev_id >= 0x8 || + adev->pdev->device == 0x15d8) && + (adev->pm.fw_version < 0x41e2b || /* not raven1 fresh */ + !adev->gfx.rlc.is_rlc_v2_1)) /* without rlc save restore ucodes */ adev->pm.pp_feature &= ~PP_GFXOFF_MASK; if (adev->pm.pp_feature & PP_GFXOFF_MASK) @@ -3933,43 +3932,58 @@ static const u32 sgpr_init_compute_shader[] = 0xbe800080, 0xbf810000, }; +/* When below register arrays changed, please update gpr_reg_size, + and sec_ded_counter_reg_size in function gfx_v9_0_do_edc_gpr_workarounds, + to cover all gfx9 ASICs */ static const struct soc15_reg_entry vgpr_init_regs[] = { - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff }, - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff }, - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff }, - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 4 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x3f }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x400000 }, /* 64KB LDS */ + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0xffffffff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0xffffffff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0xffffffff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0xffffffff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0xffffffff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0xffffffff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0xffffffff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0xffffffff }, }; static const struct soc15_reg_entry sgpr1_init_regs[] = { - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x000000ff }, - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x000000ff }, - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x000000ff }, - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x000000ff }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 8 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x240 }, /* (80 GPRS) */ { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x0 }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x000000ff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x000000ff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x000000ff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x000000ff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0x000000ff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0x000000ff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0x000000ff }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0x000000ff }, }; static const struct soc15_reg_entry sgpr2_init_regs[] = { - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x0000ff00 }, - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x0000ff00 }, - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x0000ff00 }, - { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x0000ff00 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_RESOURCE_LIMITS), 0x0000000 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_X), 0x40 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Y), 8 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_NUM_THREAD_Z), 1 }, { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC1), 0x240 }, /* (80 GPRS) */ { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_PGM_RSRC2), 0x0 }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE0), 0x0000ff00 }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE1), 0x0000ff00 }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE2), 0x0000ff00 }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE3), 0x0000ff00 }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE4), 0x0000ff00 }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE5), 0x0000ff00 }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE6), 0x0000ff00 }, + { SOC15_REG_ENTRY(GC, 0, mmCOMPUTE_STATIC_THREAD_MGMT_SE7), 0x0000ff00 }, }; static const struct soc15_reg_entry sec_ded_counter_registers[] = { @@ -4006,9 +4020,15 @@ static const struct soc15_reg_entry sec_ded_counter_registers[] = { { SOC15_REG_ENTRY(GC, 0, mmTCC_EDC_CNT2), 0, 1, 16}, { SOC15_REG_ENTRY(GC, 0, mmTCA_EDC_CNT), 0, 1, 2}, { SOC15_REG_ENTRY(GC, 0, mmSQC_EDC_CNT3), 0, 4, 6}, + { SOC15_REG_ENTRY(HDP, 0, mmHDP_EDC_CNT), 0, 1, 1}, { SOC15_REG_ENTRY(SDMA0, 0, mmSDMA0_EDC_COUNTER), 0, 1, 1}, { SOC15_REG_ENTRY(SDMA1, 0, mmSDMA1_EDC_COUNTER), 0, 1, 1}, - { SOC15_REG_ENTRY(HDP, 0, mmHDP_EDC_CNT), 0, 1, 1}, + { SOC15_REG_ENTRY(SDMA2, 0, mmSDMA2_EDC_COUNTER), 0, 1, 1}, + { SOC15_REG_ENTRY(SDMA3, 0, mmSDMA3_EDC_COUNTER), 0, 1, 1}, + { SOC15_REG_ENTRY(SDMA4, 0, mmSDMA4_EDC_COUNTER), 0, 1, 1}, + { SOC15_REG_ENTRY(SDMA5, 0, mmSDMA5_EDC_COUNTER), 0, 1, 1}, + { SOC15_REG_ENTRY(SDMA6, 0, mmSDMA6_EDC_COUNTER), 0, 1, 1}, + { SOC15_REG_ENTRY(SDMA7, 0, mmSDMA7_EDC_COUNTER), 0, 1, 1}, }; static int gfx_v9_0_do_edc_gds_workarounds(struct amdgpu_device *adev) @@ -4067,6 +4087,13 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) unsigned total_size, vgpr_offset, sgpr_offset; u64 gpu_addr; + int compute_dim_x = adev->gfx.config.max_shader_engines * + adev->gfx.config.max_cu_per_sh * + adev->gfx.config.max_sh_per_se; + int sgpr_work_group_size = 5; + int gpr_reg_size = compute_dim_x / 16 + 6; + int sec_ded_counter_reg_size = adev->sdma.num_instances + 34; + /* only support when RAS is enabled */ if (!amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__GFX)) return 0; @@ -4076,11 +4103,11 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) return 0; total_size = - ((ARRAY_SIZE(vgpr_init_regs) * 3) + 4 + 5 + 2) * 4; + (gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* VGPRS */ total_size += - ((ARRAY_SIZE(sgpr1_init_regs) * 3) + 4 + 5 + 2) * 4; + (gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* SGPRS1 */ total_size += - ((ARRAY_SIZE(sgpr2_init_regs) * 3) + 4 + 5 + 2) * 4; + (gpr_reg_size * 3 + 4 + 5 + 2) * 4; /* SGPRS2 */ total_size = ALIGN(total_size, 256); vgpr_offset = total_size; total_size += ALIGN(sizeof(vgpr_init_compute_shader), 256); @@ -4107,7 +4134,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) /* VGPR */ /* write the register state for the compute dispatch */ - for (i = 0; i < ARRAY_SIZE(vgpr_init_regs); i++) { + for (i = 0; i < gpr_reg_size; i++) { ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1); ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(vgpr_init_regs[i]) - PACKET3_SET_SH_REG_START; @@ -4123,7 +4150,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) /* write dispatch packet */ ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3); - ib.ptr[ib.length_dw++] = 0x40*2; /* x */ + ib.ptr[ib.length_dw++] = compute_dim_x; /* x */ ib.ptr[ib.length_dw++] = 1; /* y */ ib.ptr[ib.length_dw++] = 1; /* z */ ib.ptr[ib.length_dw++] = @@ -4135,7 +4162,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) /* SGPR1 */ /* write the register state for the compute dispatch */ - for (i = 0; i < ARRAY_SIZE(sgpr1_init_regs); i++) { + for (i = 0; i < gpr_reg_size; i++) { ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1); ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(sgpr1_init_regs[i]) - PACKET3_SET_SH_REG_START; @@ -4151,7 +4178,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) /* write dispatch packet */ ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3); - ib.ptr[ib.length_dw++] = 0xA0*2; /* x */ + ib.ptr[ib.length_dw++] = compute_dim_x / 2 * sgpr_work_group_size; /* x */ ib.ptr[ib.length_dw++] = 1; /* y */ ib.ptr[ib.length_dw++] = 1; /* z */ ib.ptr[ib.length_dw++] = @@ -4163,7 +4190,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) /* SGPR2 */ /* write the register state for the compute dispatch */ - for (i = 0; i < ARRAY_SIZE(sgpr2_init_regs); i++) { + for (i = 0; i < gpr_reg_size; i++) { ib.ptr[ib.length_dw++] = PACKET3(PACKET3_SET_SH_REG, 1); ib.ptr[ib.length_dw++] = SOC15_REG_ENTRY_OFFSET(sgpr2_init_regs[i]) - PACKET3_SET_SH_REG_START; @@ -4179,7 +4206,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) /* write dispatch packet */ ib.ptr[ib.length_dw++] = PACKET3(PACKET3_DISPATCH_DIRECT, 3); - ib.ptr[ib.length_dw++] = 0xA0*2; /* x */ + ib.ptr[ib.length_dw++] = compute_dim_x / 2 * sgpr_work_group_size; /* x */ ib.ptr[ib.length_dw++] = 1; /* y */ ib.ptr[ib.length_dw++] = 1; /* z */ ib.ptr[ib.length_dw++] = @@ -4205,7 +4232,7 @@ static int gfx_v9_0_do_edc_gpr_workarounds(struct amdgpu_device *adev) /* read back registers to clear the counters */ mutex_lock(&adev->grbm_idx_mutex); - for (i = 0; i < ARRAY_SIZE(sec_ded_counter_registers); i++) { + for (i = 0; i < sec_ded_counter_reg_size; i++) { for (j = 0; j < sec_ded_counter_registers[i].se_num; j++) { for (k = 0; k < sec_ded_counter_registers[i].instance; k++) { gfx_v9_0_select_se_sh(adev, j, 0x0, k); |