diff options
Diffstat (limited to 'drivers/gpu/drm/radeon')
25 files changed, 451 insertions, 388 deletions
diff --git a/drivers/gpu/drm/radeon/atombios_crtc.c b/drivers/gpu/drm/radeon/atombios_crtc.c index 01d77d1554f4..3904d7964a4b 100644 --- a/drivers/gpu/drm/radeon/atombios_crtc.c +++ b/drivers/gpu/drm/radeon/atombios_crtc.c @@ -1149,7 +1149,9 @@ static int dce4_crtc_do_set_base(struct drm_crtc *crtc, } if (tiling_flags & RADEON_TILING_MACRO) { - if (rdev->family >= CHIP_CAYMAN) + if (rdev->family >= CHIP_TAHITI) + tmp = rdev->config.si.tile_config; + else if (rdev->family >= CHIP_CAYMAN) tmp = rdev->config.cayman.tile_config; else tmp = rdev->config.evergreen.tile_config; @@ -1177,6 +1179,12 @@ static int dce4_crtc_do_set_base(struct drm_crtc *crtc, } else if (tiling_flags & RADEON_TILING_MICRO) fb_format |= EVERGREEN_GRPH_ARRAY_MODE(EVERGREEN_GRPH_ARRAY_1D_TILED_THIN1); + if ((rdev->family == CHIP_TAHITI) || + (rdev->family == CHIP_PITCAIRN)) + fb_format |= SI_GRPH_PIPE_CONFIG(SI_ADDR_SURF_P8_32x32_8x16); + else if (rdev->family == CHIP_VERDE) + fb_format |= SI_GRPH_PIPE_CONFIG(SI_ADDR_SURF_P4_8x16); + switch (radeon_crtc->crtc_id) { case 0: WREG32(AVIVO_D1VGA_CONTROL, 0); diff --git a/drivers/gpu/drm/radeon/atombios_encoders.c b/drivers/gpu/drm/radeon/atombios_encoders.c index e7b1ec5ae8c6..486ccdf4aacd 100644 --- a/drivers/gpu/drm/radeon/atombios_encoders.c +++ b/drivers/gpu/drm/radeon/atombios_encoders.c @@ -1926,7 +1926,9 @@ radeon_atom_encoder_mode_set(struct drm_encoder *encoder, if (atombios_get_encoder_mode(encoder) == ATOM_ENCODER_MODE_HDMI) { r600_hdmi_enable(encoder); - if (ASIC_IS_DCE4(rdev)) + if (ASIC_IS_DCE6(rdev)) + ; /* TODO (use pointers instead of if-s?) */ + else if (ASIC_IS_DCE4(rdev)) evergreen_hdmi_setmode(encoder, adjusted_mode); else r600_hdmi_setmode(encoder, adjusted_mode); diff --git a/drivers/gpu/drm/radeon/evergreen.c b/drivers/gpu/drm/radeon/evergreen.c index 01550d05e273..7fb3d2e0434c 100644 --- a/drivers/gpu/drm/radeon/evergreen.c +++ b/drivers/gpu/drm/radeon/evergreen.c @@ -1932,6 +1932,9 @@ static void evergreen_gpu_init(struct radeon_device *rdev) smx_dc_ctl0 |= NUMBER_OF_SETS(rdev->config.evergreen.sx_num_of_sets); WREG32(SMX_DC_CTL0, smx_dc_ctl0); + if (rdev->family <= CHIP_SUMO2) + WREG32(SMX_SAR_CTL0, 0x00010000); + WREG32(SX_EXPORT_BUFFER_SIZES, (COLOR_BUFFER_SIZE((rdev->config.evergreen.sx_max_export_size / 4) - 1) | POSITION_BUFFER_SIZE((rdev->config.evergreen.sx_max_export_pos_size / 4) - 1) | SMX_BUFFER_SIZE((rdev->config.evergreen.sx_max_export_smx_size / 4) - 1))); diff --git a/drivers/gpu/drm/radeon/evergreen_cs.c b/drivers/gpu/drm/radeon/evergreen_cs.c index 4e7dd2b4843d..c16554122ccd 100644 --- a/drivers/gpu/drm/radeon/evergreen_cs.c +++ b/drivers/gpu/drm/radeon/evergreen_cs.c @@ -52,6 +52,7 @@ struct evergreen_cs_track { u32 cb_color_view[12]; u32 cb_color_pitch[12]; u32 cb_color_slice[12]; + u32 cb_color_slice_idx[12]; u32 cb_color_attrib[12]; u32 cb_color_cmask_slice[8];/* unused */ u32 cb_color_fmask_slice[8];/* unused */ @@ -127,12 +128,14 @@ static void evergreen_cs_track_init(struct evergreen_cs_track *track) track->cb_color_info[i] = 0; track->cb_color_view[i] = 0xFFFFFFFF; track->cb_color_pitch[i] = 0; - track->cb_color_slice[i] = 0; + track->cb_color_slice[i] = 0xfffffff; + track->cb_color_slice_idx[i] = 0; } track->cb_target_mask = 0xFFFFFFFF; track->cb_shader_mask = 0xFFFFFFFF; track->cb_dirty = true; + track->db_depth_slice = 0xffffffff; track->db_depth_view = 0xFFFFC000; track->db_depth_size = 0xFFFFFFFF; track->db_depth_control = 0xFFFFFFFF; @@ -250,10 +253,9 @@ static int evergreen_surface_check_2d(struct radeon_cs_parser *p, { struct evergreen_cs_track *track = p->track; unsigned palign, halign, tileb, slice_pt; + unsigned mtile_pr, mtile_ps, mtileb; tileb = 64 * surf->bpe * surf->nsamples; - palign = track->group_size / (8 * surf->bpe * surf->nsamples); - palign = MAX(8, palign); slice_pt = 1; if (tileb > surf->tsplit) { slice_pt = tileb / surf->tsplit; @@ -262,7 +264,10 @@ static int evergreen_surface_check_2d(struct radeon_cs_parser *p, /* macro tile width & height */ palign = (8 * surf->bankw * track->npipes) * surf->mtilea; halign = (8 * surf->bankh * surf->nbanks) / surf->mtilea; - surf->layer_size = surf->nbx * surf->nby * surf->bpe * slice_pt; + mtileb = (palign / 8) * (halign / 8) * tileb;; + mtile_pr = surf->nbx / palign; + mtile_ps = (mtile_pr * surf->nby) / halign; + surf->layer_size = mtile_ps * mtileb * slice_pt; surf->base_align = (palign / 8) * (halign / 8) * tileb; surf->palign = palign; surf->halign = halign; @@ -434,6 +439,39 @@ static int evergreen_cs_track_validate_cb(struct radeon_cs_parser *p, unsigned i offset += surf.layer_size * mslice; if (offset > radeon_bo_size(track->cb_color_bo[id])) { + /* old ddx are broken they allocate bo with w*h*bpp but + * program slice with ALIGN(h, 8), catch this and patch + * command stream. + */ + if (!surf.mode) { + volatile u32 *ib = p->ib.ptr; + unsigned long tmp, nby, bsize, size, min = 0; + + /* find the height the ddx wants */ + if (surf.nby > 8) { + min = surf.nby - 8; + } + bsize = radeon_bo_size(track->cb_color_bo[id]); + tmp = track->cb_color_bo_offset[id] << 8; + for (nby = surf.nby; nby > min; nby--) { + size = nby * surf.nbx * surf.bpe * surf.nsamples; + if ((tmp + size * mslice) <= bsize) { + break; + } + } + if (nby > min) { + surf.nby = nby; + slice = ((nby * surf.nbx) / 64) - 1; + if (!evergreen_surface_check(p, &surf, "cb")) { + /* check if this one works */ + tmp += surf.layer_size * mslice; + if (tmp <= bsize) { + ib[track->cb_color_slice_idx[id]] = slice; + goto old_ddx_ok; + } + } + } + } dev_warn(p->dev, "%s:%d cb[%d] bo too small (layer size %d, " "offset %d, max layer %d, bo size %ld, slice %d)\n", __func__, __LINE__, id, surf.layer_size, @@ -446,6 +484,7 @@ static int evergreen_cs_track_validate_cb(struct radeon_cs_parser *p, unsigned i surf.tsplit, surf.mtilea); return -EINVAL; } +old_ddx_ok: return 0; } @@ -1532,6 +1571,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) case CB_COLOR7_SLICE: tmp = (reg - CB_COLOR0_SLICE) / 0x3c; track->cb_color_slice[tmp] = radeon_get_ib_value(p, idx); + track->cb_color_slice_idx[tmp] = idx; track->cb_dirty = true; break; case CB_COLOR8_SLICE: @@ -1540,6 +1580,7 @@ static int evergreen_cs_check_reg(struct radeon_cs_parser *p, u32 reg, u32 idx) case CB_COLOR11_SLICE: tmp = ((reg - CB_COLOR8_SLICE) / 0x1c) + 8; track->cb_color_slice[tmp] = radeon_get_ib_value(p, idx); + track->cb_color_slice_idx[tmp] = idx; track->cb_dirty = true; break; case CB_COLOR0_ATTRIB: diff --git a/drivers/gpu/drm/radeon/evergreen_hdmi.c b/drivers/gpu/drm/radeon/evergreen_hdmi.c index a51f880985f8..65c54160028b 100644 --- a/drivers/gpu/drm/radeon/evergreen_hdmi.c +++ b/drivers/gpu/drm/radeon/evergreen_hdmi.c @@ -156,9 +156,6 @@ void evergreen_hdmi_setmode(struct drm_encoder *encoder, struct drm_display_mode struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; uint32_t offset; - if (ASIC_IS_DCE5(rdev)) - return; - /* Silent, r600_hdmi_enable will raise WARN for us */ if (!dig->afmt->enabled) return; diff --git a/drivers/gpu/drm/radeon/evergreend.h b/drivers/gpu/drm/radeon/evergreend.h index 2773039b4902..b50b15c70498 100644 --- a/drivers/gpu/drm/radeon/evergreend.h +++ b/drivers/gpu/drm/radeon/evergreend.h @@ -503,6 +503,7 @@ #define SCRATCH_UMSK 0x8540 #define SCRATCH_ADDR 0x8544 +#define SMX_SAR_CTL0 0xA008 #define SMX_DC_CTL0 0xA020 #define USE_HASH_FUNCTION (1 << 0) #define NUMBER_OF_SETS(x) ((x) << 1) diff --git a/drivers/gpu/drm/radeon/ni.c b/drivers/gpu/drm/radeon/ni.c index 3df4efa11942..b7bf18e40215 100644 --- a/drivers/gpu/drm/radeon/ni.c +++ b/drivers/gpu/drm/radeon/ni.c @@ -460,15 +460,28 @@ static void cayman_gpu_init(struct radeon_device *rdev) 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 == 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 == 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 == 0x9990) || - (rdev->pdev->device == 0x9991)) { + } 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 { @@ -1290,6 +1303,10 @@ static int cayman_startup(struct radeon_device *rdev) if (r) return r; + r = r600_audio_init(rdev); + if (r) + return r; + return 0; } @@ -1316,6 +1333,7 @@ int cayman_resume(struct radeon_device *rdev) int cayman_suspend(struct radeon_device *rdev) { + r600_audio_fini(rdev); /* FIXME: we should wait for ring to be empty */ radeon_ib_pool_suspend(rdev); radeon_vm_manager_suspend(rdev); diff --git a/drivers/gpu/drm/radeon/r600.c b/drivers/gpu/drm/radeon/r600.c index 45cfcea63507..bff627293812 100644 --- a/drivers/gpu/drm/radeon/r600.c +++ b/drivers/gpu/drm/radeon/r600.c @@ -1839,6 +1839,7 @@ void r600_gpu_init(struct radeon_device *rdev) WREG32(PA_CL_ENHANCE, (CLIP_VTX_REORDER_ENA | NUM_CLIP_SEQ(3))); WREG32(PA_SC_ENHANCE, FORCE_EOV_MAX_CLK_CNT(4095)); + WREG32(VC_ENHANCE, 0); } @@ -2426,6 +2427,12 @@ int r600_startup(struct radeon_device *rdev) if (r) return r; + r = r600_audio_init(rdev); + if (r) { + DRM_ERROR("radeon: audio init failed\n"); + return r; + } + return 0; } @@ -2462,12 +2469,6 @@ int r600_resume(struct radeon_device *rdev) return r; } - r = r600_audio_init(rdev); - if (r) { - DRM_ERROR("radeon: audio resume failed\n"); - return r; - } - return r; } @@ -2577,9 +2578,6 @@ int r600_init(struct radeon_device *rdev) rdev->accel_working = false; } - r = r600_audio_init(rdev); - if (r) - return r; /* TODO error handling */ return 0; } diff --git a/drivers/gpu/drm/radeon/r600_audio.c b/drivers/gpu/drm/radeon/r600_audio.c index 7c4fa77f018f..79b55916cf90 100644 --- a/drivers/gpu/drm/radeon/r600_audio.c +++ b/drivers/gpu/drm/radeon/r600_audio.c @@ -57,7 +57,7 @@ static bool radeon_dig_encoder(struct drm_encoder *encoder) */ static int r600_audio_chipset_supported(struct radeon_device *rdev) { - return (rdev->family >= CHIP_R600 && !ASIC_IS_DCE5(rdev)) + return (rdev->family >= CHIP_R600 && !ASIC_IS_DCE6(rdev)) || rdev->family == CHIP_RS600 || rdev->family == CHIP_RS690 || rdev->family == CHIP_RS740; @@ -192,6 +192,7 @@ void r600_audio_set_clock(struct drm_encoder *encoder, int clock) struct radeon_device *rdev = dev->dev_private; struct radeon_encoder *radeon_encoder = to_radeon_encoder(encoder); struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; + struct radeon_crtc *radeon_crtc = to_radeon_crtc(encoder->crtc); int base_rate = 48000; switch (radeon_encoder->encoder_id) { @@ -217,8 +218,8 @@ void r600_audio_set_clock(struct drm_encoder *encoder, int clock) WREG32(EVERGREEN_AUDIO_PLL1_DIV, clock * 10); WREG32(EVERGREEN_AUDIO_PLL1_UNK, 0x00000071); - /* Some magic trigger or src sel? */ - WREG32_P(0x5ac, 0x01, ~0x77); + /* Select DTO source */ + WREG32(0x5ac, radeon_crtc->crtc_id); } else { switch (dig->dig_encoder) { case 0: diff --git a/drivers/gpu/drm/radeon/r600_cs.c b/drivers/gpu/drm/radeon/r600_cs.c index 0133f5f09bd6..ca87f7afaf23 100644 --- a/drivers/gpu/drm/radeon/r600_cs.c +++ b/drivers/gpu/drm/radeon/r600_cs.c @@ -2079,6 +2079,48 @@ static int r600_packet3_check(struct radeon_cs_parser *p, return -EINVAL; } break; + case PACKET3_STRMOUT_BASE_UPDATE: + if (p->family < CHIP_RV770) { + DRM_ERROR("STRMOUT_BASE_UPDATE only supported on 7xx\n"); + return -EINVAL; + } + if (pkt->count != 1) { + DRM_ERROR("bad STRMOUT_BASE_UPDATE packet count\n"); + return -EINVAL; + } + if (idx_value > 3) { + DRM_ERROR("bad STRMOUT_BASE_UPDATE index\n"); + return -EINVAL; + } + { + u64 offset; + + r = r600_cs_packet_next_reloc(p, &reloc); + if (r) { + DRM_ERROR("bad STRMOUT_BASE_UPDATE reloc\n"); + return -EINVAL; + } + + if (reloc->robj != track->vgt_strmout_bo[idx_value]) { + DRM_ERROR("bad STRMOUT_BASE_UPDATE, bo does not match\n"); + return -EINVAL; + } + + offset = radeon_get_ib_value(p, idx+1) << 8; + if (offset != track->vgt_strmout_bo_offset[idx_value]) { + DRM_ERROR("bad STRMOUT_BASE_UPDATE, bo offset does not match: 0x%llx, 0x%x\n", + offset, track->vgt_strmout_bo_offset[idx_value]); + return -EINVAL; + } + + if ((offset + 4) > radeon_bo_size(reloc->robj)) { + DRM_ERROR("bad STRMOUT_BASE_UPDATE bo too small: 0x%llx, 0x%lx\n", + offset + 4, radeon_bo_size(reloc->robj)); + return -EINVAL; + } + ib[idx+1] += (u32)((reloc->lobj.gpu_offset >> 8) & 0xffffffff); + } + break; case PACKET3_SURFACE_BASE_UPDATE: if (p->family >= CHIP_RV770 || p->family == CHIP_R600) { DRM_ERROR("bad SURFACE_BASE_UPDATE\n"); diff --git a/drivers/gpu/drm/radeon/r600_hdmi.c b/drivers/gpu/drm/radeon/r600_hdmi.c index 226379e00ac1..82a0a4c919c0 100644 --- a/drivers/gpu/drm/radeon/r600_hdmi.c +++ b/drivers/gpu/drm/radeon/r600_hdmi.c @@ -322,9 +322,6 @@ void r600_hdmi_setmode(struct drm_encoder *encoder, struct drm_display_mode *mod struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; uint32_t offset; - if (ASIC_IS_DCE5(rdev)) - return; - /* Silent, r600_hdmi_enable will raise WARN for us */ if (!dig->afmt->enabled) return; @@ -348,7 +345,6 @@ void r600_hdmi_setmode(struct drm_encoder *encoder, struct drm_display_mode *mod WREG32(HDMI0_AUDIO_PACKET_CONTROL + offset, HDMI0_AUDIO_SAMPLE_SEND | /* send audio packets */ HDMI0_AUDIO_DELAY_EN(1) | /* default audio delay */ - HDMI0_AUDIO_SEND_MAX_PACKETS | /* send NULL packets if no audio is available */ HDMI0_AUDIO_PACKETS_PER_LINE(3) | /* should be suffient for all audio modes and small enough for all hblanks */ HDMI0_60958_CS_UPDATE); /* allow 60958 channel status fields to be updated */ } @@ -484,7 +480,7 @@ void r600_hdmi_enable(struct drm_encoder *encoder) uint32_t offset; u32 hdmi; - if (ASIC_IS_DCE5(rdev)) + if (ASIC_IS_DCE6(rdev)) return; /* Silent, r600_hdmi_enable will raise WARN for us */ @@ -544,7 +540,7 @@ void r600_hdmi_disable(struct drm_encoder *encoder) struct radeon_encoder_atom_dig *dig = radeon_encoder->enc_priv; uint32_t offset; - if (ASIC_IS_DCE5(rdev)) + if (ASIC_IS_DCE6(rdev)) return; /* Called for ATOM_ENCODER_MODE_HDMI only */ diff --git a/drivers/gpu/drm/radeon/r600d.h b/drivers/gpu/drm/radeon/r600d.h index a0dbf1fe6a40..025fd5b6c08c 100644 --- a/drivers/gpu/drm/radeon/r600d.h +++ b/drivers/gpu/drm/radeon/r600d.h @@ -485,6 +485,7 @@ #define TC_L2_SIZE(x) ((x)<<5) #define L2_DISABLE_LATE_HIT (1<<9) +#define VC_ENHANCE 0x9714 #define VGT_CACHE_INVALIDATION 0x88C4 #define CACHE_INVALIDATION(x) ((x)<<0) @@ -1163,6 +1164,7 @@ #define PACKET3_SET_CTL_CONST 0x6F #define PACKET3_SET_CTL_CONST_OFFSET 0x0003cff0 #define PACKET3_SET_CTL_CONST_END 0x0003e200 +#define PACKET3_STRMOUT_BASE_UPDATE 0x72 /* r7xx */ #define PACKET3_SURFACE_BASE_UPDATE 0x73 diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h index 85dac33e3cce..fefcca55c1eb 100644 --- a/drivers/gpu/drm/radeon/radeon.h +++ b/drivers/gpu/drm/radeon/radeon.h @@ -1374,9 +1374,9 @@ struct cayman_asic { struct si_asic { unsigned max_shader_engines; - unsigned max_pipes_per_simd; unsigned max_tile_pipes; - unsigned max_simds_per_se; + unsigned max_cu_per_sh; + unsigned max_sh_per_se; unsigned max_backends_per_se; unsigned max_texture_channel_caches; unsigned max_gprs; @@ -1387,7 +1387,6 @@ struct si_asic { unsigned sc_hiz_tile_fifo_size; unsigned sc_earlyz_tile_fifo_size; - unsigned num_shader_engines; unsigned num_tile_pipes; unsigned num_backends_per_se; unsigned backend_disable_mask_per_asic; diff --git a/drivers/gpu/drm/radeon/radeon_drv.c b/drivers/gpu/drm/radeon/radeon_drv.c index f0bb2b543b13..2c4d53fd20c5 100644 --- a/drivers/gpu/drm/radeon/radeon_drv.c +++ b/drivers/gpu/drm/radeon/radeon_drv.c @@ -57,9 +57,11 @@ * 2.13.0 - virtual memory support, streamout * 2.14.0 - add evergreen tiling informations * 2.15.0 - add max_pipes query + * 2.16.0 - fix evergreen 2D tiled surface calculation + * 2.17.0 - add STRMOUT_BASE_UPDATE for r7xx */ #define KMS_DRIVER_MAJOR 2 -#define KMS_DRIVER_MINOR 15 +#define KMS_DRIVER_MINOR 17 #define KMS_DRIVER_PATCHLEVEL 0 int radeon_driver_load_kms(struct drm_device *dev, unsigned long flags); int radeon_driver_unload_kms(struct drm_device *dev); diff --git a/drivers/gpu/drm/radeon/radeon_gart.c b/drivers/gpu/drm/radeon/radeon_gart.c index 79db56e6c2ac..59d44937dd9f 100644 --- a/drivers/gpu/drm/radeon/radeon_gart.c +++ b/drivers/gpu/drm/radeon/radeon_gart.c @@ -476,12 +476,18 @@ int radeon_vm_bo_add(struct radeon_device *rdev, mutex_lock(&vm->mutex); if (last_pfn > vm->last_pfn) { - /* grow va space 32M by 32M */ - unsigned align = ((32 << 20) >> 12) - 1; + /* release mutex and lock in right order */ + mutex_unlock(&vm->mutex); radeon_mutex_lock(&rdev->cs_mutex); - radeon_vm_unbind_locked(rdev, vm); + mutex_lock(&vm->mutex); + /* and check again */ + if (last_pfn > vm->last_pfn) { + /* grow va space 32M by 32M */ + unsigned align = ((32 << 20) >> 12) - 1; + radeon_vm_unbind_locked(rdev, vm); + vm->last_pfn = (last_pfn + align) & ~align; + } radeon_mutex_unlock(&rdev->cs_mutex); - vm->last_pfn = (last_pfn + align) & ~align; } head = &vm->va; last_offset = 0; @@ -595,8 +601,8 @@ int radeon_vm_bo_rmv(struct radeon_device *rdev, if (bo_va == NULL) return 0; - mutex_lock(&vm->mutex); radeon_mutex_lock(&rdev->cs_mutex); + mutex_lock(&vm->mutex); radeon_vm_bo_update_pte(rdev, vm, bo, NULL); radeon_mutex_unlock(&rdev->cs_mutex); list_del(&bo_va->vm_list); @@ -641,9 +647,8 @@ void radeon_vm_fini(struct radeon_device *rdev, struct radeon_vm *vm) struct radeon_bo_va *bo_va, *tmp; int r; - mutex_lock(&vm->mutex); - radeon_mutex_lock(&rdev->cs_mutex); + mutex_lock(&vm->mutex); radeon_vm_unbind_locked(rdev, vm); radeon_mutex_unlock(&rdev->cs_mutex); diff --git a/drivers/gpu/drm/radeon/radeon_kms.c b/drivers/gpu/drm/radeon/radeon_kms.c index f1016a5820d1..5c58d7d90cb2 100644 --- a/drivers/gpu/drm/radeon/radeon_kms.c +++ b/drivers/gpu/drm/radeon/radeon_kms.c @@ -273,7 +273,7 @@ int radeon_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) break; case RADEON_INFO_MAX_PIPES: if (rdev->family >= CHIP_TAHITI) - value = rdev->config.si.max_pipes_per_simd; + value = rdev->config.si.max_cu_per_sh; else if (rdev->family >= CHIP_CAYMAN) value = rdev->config.cayman.max_pipes_per_simd; else if (rdev->family >= CHIP_CEDAR) diff --git a/drivers/gpu/drm/radeon/radeon_pm.c b/drivers/gpu/drm/radeon/radeon_pm.c index 08825548ee69..5b37e283ec38 100644 --- a/drivers/gpu/drm/radeon/radeon_pm.c +++ b/drivers/gpu/drm/radeon/radeon_pm.c @@ -801,9 +801,13 @@ static void radeon_dynpm_idle_work_handler(struct work_struct *work) int i; for (i = 0; i < RADEON_NUM_RINGS; ++i) { - not_processed += radeon_fence_count_emitted(rdev, i); - if (not_processed >= 3) - break; + struct radeon_ring *ring = &rdev->ring[i]; + + if (ring->ready) { + not_processed += radeon_fence_count_emitted(rdev, i); + if (not_processed >= 3) + break; + } } if (not_processed >= 3) { /* should upclock */ diff --git a/drivers/gpu/drm/radeon/radeon_prime.c b/drivers/gpu/drm/radeon/radeon_prime.c index 8ddab4c76710..6bef46ace831 100644 --- a/drivers/gpu/drm/radeon/radeon_prime.c +++ b/drivers/gpu/drm/radeon/radeon_prime.c @@ -169,11 +169,17 @@ struct dma_buf *radeon_gem_prime_export(struct drm_device *dev, struct radeon_bo *bo = gem_to_radeon_bo(obj); int ret = 0; + ret = radeon_bo_reserve(bo, false); + if (unlikely(ret != 0)) + return ERR_PTR(ret); + /* pin buffer into GTT */ ret = radeon_bo_pin(bo, RADEON_GEM_DOMAIN_GTT, NULL); - if (ret) + if (ret) { + radeon_bo_unreserve(bo); return ERR_PTR(ret); - + } + radeon_bo_unreserve(bo); return dma_buf_export(bo, &radeon_dmabuf_ops, obj->size, flags); } diff --git a/drivers/gpu/drm/radeon/rs600.c b/drivers/gpu/drm/radeon/rs600.c index 25f9eef12c42..e95c5e61d4e2 100644 --- a/drivers/gpu/drm/radeon/rs600.c +++ b/drivers/gpu/drm/radeon/rs600.c @@ -908,12 +908,6 @@ static int rs600_startup(struct radeon_device *rdev) return r; } - r = r600_audio_init(rdev); - if (r) { - dev_err(rdev->dev, "failed initializing audio\n"); - return r; - } - r = radeon_ib_pool_start(rdev); if (r) return r; @@ -922,6 +916,12 @@ static int rs600_startup(struct radeon_device *rdev) if (r) return r; + r = r600_audio_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing audio\n"); + return r; + } + return 0; } diff --git a/drivers/gpu/drm/radeon/rs690.c b/drivers/gpu/drm/radeon/rs690.c index 3277ddecfe9f..159b6a43fda0 100644 --- a/drivers/gpu/drm/radeon/rs690.c +++ b/drivers/gpu/drm/radeon/rs690.c @@ -637,12 +637,6 @@ static int rs690_startup(struct radeon_device *rdev) return r; } - r = r600_audio_init(rdev); - if (r) { - dev_err(rdev->dev, "failed initializing audio\n"); - return r; - } - r = radeon_ib_pool_start(rdev); if (r) return r; @@ -651,6 +645,12 @@ static int rs690_startup(struct radeon_device *rdev) if (r) return r; + r = r600_audio_init(rdev); + if (r) { + dev_err(rdev->dev, "failed initializing audio\n"); + return r; + } + return 0; } diff --git a/drivers/gpu/drm/radeon/rv770.c b/drivers/gpu/drm/radeon/rv770.c index 04ddc365a908..b4f51c569c36 100644 --- a/drivers/gpu/drm/radeon/rv770.c +++ b/drivers/gpu/drm/radeon/rv770.c @@ -616,6 +616,9 @@ static void rv770_gpu_init(struct radeon_device *rdev) ACK_FLUSH_CTL(3) | SYNC_FLUSH_CTL)); + if (rdev->family != CHIP_RV770) + WREG32(SMX_SAR_CTL0, 0x00003f3f); + db_debug3 = RREG32(DB_DEBUG3); db_debug3 &= ~DB_CLK_OFF_DELAY(0x1f); switch (rdev->family) { @@ -792,7 +795,7 @@ static void rv770_gpu_init(struct radeon_device *rdev) WREG32(PA_CL_ENHANCE, (CLIP_VTX_REORDER_ENA | NUM_CLIP_SEQ(3))); - + WREG32(VC_ENHANCE, 0); } void r700_vram_gtt_location(struct radeon_device *rdev, struct radeon_mc *mc) @@ -956,6 +959,12 @@ static int rv770_startup(struct radeon_device *rdev) if (r) return r; + r = r600_audio_init(rdev); + if (r) { + DRM_ERROR("radeon: audio init failed\n"); + return r; + } + return 0; } @@ -978,12 +987,6 @@ int rv770_resume(struct radeon_device *rdev) return r; } - r = r600_audio_init(rdev); - if (r) { - dev_err(rdev->dev, "radeon: audio init failed\n"); - return r; - } - return r; } @@ -1092,12 +1095,6 @@ int rv770_init(struct radeon_device *rdev) rdev->accel_working = false; } - r = r600_audio_init(rdev); - if (r) { - dev_err(rdev->dev, "radeon: audio init failed\n"); - return r; - } - return 0; } diff --git a/drivers/gpu/drm/radeon/rv770d.h b/drivers/gpu/drm/radeon/rv770d.h index fdc089896011..b0adfc595d75 100644 --- a/drivers/gpu/drm/radeon/rv770d.h +++ b/drivers/gpu/drm/radeon/rv770d.h @@ -211,6 +211,7 @@ #define SCRATCH_UMSK 0x8540 #define SCRATCH_ADDR 0x8544 +#define SMX_SAR_CTL0 0xA008 #define SMX_DC_CTL0 0xA020 #define USE_HASH_FUNCTION (1 << 0) #define CACHE_DEPTH(x) ((x) << 1) @@ -310,6 +311,8 @@ #define TCP_CNTL 0x9610 #define TCP_CHAN_STEER 0x9614 +#define VC_ENHANCE 0x9714 + #define VGT_CACHE_INVALIDATION 0x88C4 #define CACHE_INVALIDATION(x) ((x)<<0) #define VC_ONLY 0 diff --git a/drivers/gpu/drm/radeon/si.c b/drivers/gpu/drm/radeon/si.c index 549732e56ca9..c7b61f16ecfd 100644 --- a/drivers/gpu/drm/radeon/si.c +++ b/drivers/gpu/drm/radeon/si.c @@ -867,200 +867,6 @@ void dce6_bandwidth_update(struct radeon_device *rdev) /* * Core functions */ -static u32 si_get_tile_pipe_to_backend_map(struct radeon_device *rdev, - u32 num_tile_pipes, - u32 num_backends_per_asic, - u32 *backend_disable_mask_per_asic, - u32 num_shader_engines) -{ - u32 backend_map = 0; - u32 enabled_backends_mask = 0; - u32 enabled_backends_count = 0; - u32 num_backends_per_se; - u32 cur_pipe; - u32 swizzle_pipe[SI_MAX_PIPES]; - u32 cur_backend = 0; - u32 i; - bool force_no_swizzle; - - /* force legal values */ - if (num_tile_pipes < 1) - num_tile_pipes = 1; - if (num_tile_pipes > rdev->config.si.max_tile_pipes) - num_tile_pipes = rdev->config.si.max_tile_pipes; - if (num_shader_engines < 1) - num_shader_engines = 1; - if (num_shader_engines > rdev->config.si.max_shader_engines) - num_shader_engines = rdev->config.si.max_shader_engines; - if (num_backends_per_asic < num_shader_engines) - num_backends_per_asic = num_shader_engines; - if (num_backends_per_asic > (rdev->config.si.max_backends_per_se * num_shader_engines)) - num_backends_per_asic = rdev->config.si.max_backends_per_se * num_shader_engines; - - /* make sure we have the same number of backends per se */ - num_backends_per_asic = ALIGN(num_backends_per_asic, num_shader_engines); - /* set up the number of backends per se */ - num_backends_per_se = num_backends_per_asic / num_shader_engines; - if (num_backends_per_se > rdev->config.si.max_backends_per_se) { - num_backends_per_se = rdev->config.si.max_backends_per_se; - num_backends_per_asic = num_backends_per_se * num_shader_engines; - } - - /* create enable mask and count for enabled backends */ - for (i = 0; i < SI_MAX_BACKENDS; ++i) { - if (((*backend_disable_mask_per_asic >> i) & 1) == 0) { - enabled_backends_mask |= (1 << i); - ++enabled_backends_count; - } - if (enabled_backends_count == num_backends_per_asic) - break; - } - - /* force the backends mask to match the current number of backends */ - if (enabled_backends_count != num_backends_per_asic) { - u32 this_backend_enabled; - u32 shader_engine; - u32 backend_per_se; - - enabled_backends_mask = 0; - enabled_backends_count = 0; - *backend_disable_mask_per_asic = SI_MAX_BACKENDS_MASK; - for (i = 0; i < SI_MAX_BACKENDS; ++i) { - /* calc the current se */ - shader_engine = i / rdev->config.si.max_backends_per_se; - /* calc the backend per se */ - backend_per_se = i % rdev->config.si.max_backends_per_se; - /* default to not enabled */ - this_backend_enabled = 0; - if ((shader_engine < num_shader_engines) && - (backend_per_se < num_backends_per_se)) - this_backend_enabled = 1; - if (this_backend_enabled) { - enabled_backends_mask |= (1 << i); - *backend_disable_mask_per_asic &= ~(1 << i); - ++enabled_backends_count; - } - } - } - - - memset((uint8_t *)&swizzle_pipe[0], 0, sizeof(u32) * SI_MAX_PIPES); - switch (rdev->family) { - case CHIP_TAHITI: - case CHIP_PITCAIRN: - case CHIP_VERDE: - force_no_swizzle = true; - break; - default: - force_no_swizzle = false; - break; - } - if (force_no_swizzle) { - bool last_backend_enabled = false; - - force_no_swizzle = false; - for (i = 0; i < SI_MAX_BACKENDS; ++i) { - if (((enabled_backends_mask >> i) & 1) == 1) { - if (last_backend_enabled) - force_no_swizzle = true; - last_backend_enabled = true; - } else - last_backend_enabled = false; - } - } - - switch (num_tile_pipes) { - case 1: - case 3: - case 5: - case 7: - DRM_ERROR("odd number of pipes!\n"); - break; - case 2: - swizzle_pipe[0] = 0; - swizzle_pipe[1] = 1; - break; - case 4: - if (force_no_swizzle) { - swizzle_pipe[0] = 0; - swizzle_pipe[1] = 1; - swizzle_pipe[2] = 2; - swizzle_pipe[3] = 3; - } else { - swizzle_pipe[0] = 0; - swizzle_pipe[1] = 2; - swizzle_pipe[2] = 1; - swizzle_pipe[3] = 3; - } - break; - case 6: - if (force_no_swizzle) { - swizzle_pipe[0] = 0; - swizzle_pipe[1] = 1; - swizzle_pipe[2] = 2; - swizzle_pipe[3] = 3; - swizzle_pipe[4] = 4; - swizzle_pipe[5] = 5; - } else { - swizzle_pipe[0] = 0; - swizzle_pipe[1] = 2; - swizzle_pipe[2] = 4; - swizzle_pipe[3] = 1; - swizzle_pipe[4] = 3; - swizzle_pipe[5] = 5; - } - break; - case 8: - if (force_no_swizzle) { - swizzle_pipe[0] = 0; - swizzle_pipe[1] = 1; - swizzle_pipe[2] = 2; - swizzle_pipe[3] = 3; - swizzle_pipe[4] = 4; - swizzle_pipe[5] = 5; - swizzle_pipe[6] = 6; - swizzle_pipe[7] = 7; - } else { - swizzle_pipe[0] = 0; - swizzle_pipe[1] = 2; - swizzle_pipe[2] = 4; - swizzle_pipe[3] = 6; - swizzle_pipe[4] = 1; - swizzle_pipe[5] = 3; - swizzle_pipe[6] = 5; - swizzle_pipe[7] = 7; - } - break; - } - - for (cur_pipe = 0; cur_pipe < num_tile_pipes; ++cur_pipe) { - while (((1 << cur_backend) & enabled_backends_mask) == 0) - cur_backend = (cur_backend + 1) % SI_MAX_BACKENDS; - - backend_map |= (((cur_backend & 0xf) << (swizzle_pipe[cur_pipe] * 4))); - - cur_backend = (cur_backend + 1) % SI_MAX_BACKENDS; - } - - return backend_map; -} - -static u32 si_get_disable_mask_per_asic(struct radeon_device *rdev, - u32 disable_mask_per_se, - u32 max_disable_mask_per_se, - u32 num_shader_engines) -{ - u32 disable_field_width_per_se = r600_count_pipe_bits(disable_mask_per_se); - u32 disable_mask_per_asic = disable_mask_per_se & max_disable_mask_per_se; - - if (num_shader_engines == 1) - return disable_mask_per_asic; - else if (num_shader_engines == 2) - return disable_mask_per_asic | (disable_mask_per_asic << disable_field_width_per_se); - else - return 0xffffffff; -} - static void si_tiling_mode_table_init(struct radeon_device *rdev) { const u32 num_tile_mode_states = 32; @@ -1562,18 +1368,151 @@ static void si_tiling_mode_table_init(struct radeon_device *rdev) DRM_ERROR("unknown asic: 0x%x\n", rdev->family); } +static void si_select_se_sh(struct radeon_device *rdev, + u32 se_num, u32 sh_num) +{ + u32 data = INSTANCE_BROADCAST_WRITES; + + if ((se_num == 0xffffffff) && (sh_num == 0xffffffff)) + data = SH_BROADCAST_WRITES | SE_BROADCAST_WRITES; + else if (se_num == 0xffffffff) + data |= SE_BROADCAST_WRITES | SH_INDEX(sh_num); + else if (sh_num == 0xffffffff) + data |= SH_BROADCAST_WRITES | SE_INDEX(se_num); + else + data |= SH_INDEX(sh_num) | SE_INDEX(se_num); + WREG32(GRBM_GFX_INDEX, data); +} + +static u32 si_create_bitmask(u32 bit_width) +{ + u32 i, mask = 0; + + for (i = 0; i < bit_width; i++) { + mask <<= 1; + mask |= 1; + } + return mask; +} + +static u32 si_get_cu_enabled(struct radeon_device *rdev, u32 cu_per_sh) +{ + u32 data, mask; + + data = RREG32(CC_GC_SHADER_ARRAY_CONFIG); + if (data & 1) + data &= INACTIVE_CUS_MASK; + else + data = 0; + data |= RREG32(GC_USER_SHADER_ARRAY_CONFIG); + + data >>= INACTIVE_CUS_SHIFT; + + mask = si_create_bitmask(cu_per_sh); + + return ~data & mask; +} + +static void si_setup_spi(struct radeon_device *rdev, + u32 se_num, u32 sh_per_se, + u32 cu_per_sh) +{ + int i, j, k; + u32 data, mask, active_cu; + + for (i = 0; i < se_num; i++) { + for (j = 0; j < sh_per_se; j++) { + si_select_se_sh(rdev, i, j); + data = RREG32(SPI_STATIC_THREAD_MGMT_3); + active_cu = si_get_cu_enabled(rdev, cu_per_sh); + + mask = 1; + for (k = 0; k < 16; k++) { + mask <<= k; + if (active_cu & mask) { + data &= ~mask; + WREG32(SPI_STATIC_THREAD_MGMT_3, data); + break; + } + } + } + } + si_select_se_sh(rdev, 0xffffffff, 0xffffffff); +} + +static u32 si_get_rb_disabled(struct radeon_device *rdev, + u32 max_rb_num, u32 se_num, + u32 sh_per_se) +{ + u32 data, mask; + + data = RREG32(CC_RB_BACKEND_DISABLE); + if (data & 1) + data &= BACKEND_DISABLE_MASK; + else + data = 0; + data |= RREG32(GC_USER_RB_BACKEND_DISABLE); + + data >>= BACKEND_DISABLE_SHIFT; + + mask = si_create_bitmask(max_rb_num / se_num / sh_per_se); + + return data & mask; +} + +static void si_setup_rb(struct radeon_device *rdev, + u32 se_num, u32 sh_per_se, + u32 max_rb_num) +{ + int i, j; + u32 data, mask; + u32 disabled_rbs = 0; + u32 enabled_rbs = 0; + + for (i = 0; i < se_num; i++) { + for (j = 0; j < sh_per_se; j++) { + si_select_se_sh(rdev, i, j); + data = si_get_rb_disabled(rdev, max_rb_num, se_num, sh_per_se); + disabled_rbs |= data << ((i * sh_per_se + j) * TAHITI_RB_BITMAP_WIDTH_PER_SH); + } + } + si_select_se_sh(rdev, 0xffffffff, 0xffffffff); + + mask = 1; + for (i = 0; i < max_rb_num; i++) { + if (!(disabled_rbs & mask)) + enabled_rbs |= mask; + mask <<= 1; + } + + for (i = 0; i < se_num; i++) { + si_select_se_sh(rdev, i, 0xffffffff); + data = 0; + for (j = 0; j < sh_per_se; j++) { + switch (enabled_rbs & 3) { + case 1: + data |= (RASTER_CONFIG_RB_MAP_0 << (i * sh_per_se + j) * 2); + break; + case 2: + data |= (RASTER_CONFIG_RB_MAP_3 << (i * sh_per_se + j) * 2); + break; + case 3: + default: + data |= (RASTER_CONFIG_RB_MAP_2 << (i * sh_per_se + j) * 2); + break; + } + enabled_rbs >>= 2; + } + WREG32(PA_SC_RASTER_CONFIG, data); + } + si_select_se_sh(rdev, 0xffffffff, 0xffffffff); +} + static void si_gpu_init(struct radeon_device *rdev) { - u32 cc_rb_backend_disable = 0; - u32 cc_gc_shader_array_config; u32 gb_addr_config = 0; u32 mc_shared_chmap, mc_arb_ramcfg; - u32 gb_backend_map; - u32 cgts_tcc_disable; u32 sx_debug_1; - u32 gc_user_shader_array_config; - u32 gc_user_rb_backend_disable; - u32 cgts_user_tcc_disable; u32 hdp_host_path_cntl; u32 tmp; int i, j; @@ -1581,9 +1520,9 @@ static void si_gpu_init(struct radeon_device *rdev) switch (rdev->family) { case CHIP_TAHITI: rdev->config.si.max_shader_engines = 2; - rdev->config.si.max_pipes_per_simd = 4; rdev->config.si.max_tile_pipes = 12; - rdev->config.si.max_simds_per_se = 8; + rdev->config.si.max_cu_per_sh = 8; + rdev->config.si.max_sh_per_se = 2; rdev->config.si.max_backends_per_se = 4; rdev->config.si.max_texture_channel_caches = 12; rdev->config.si.max_gprs = 256; @@ -1594,12 +1533,13 @@ static void si_gpu_init(struct radeon_device *rdev) rdev->config.si.sc_prim_fifo_size_backend = 0x100; rdev->config.si.sc_hiz_tile_fifo_size = 0x30; rdev->config.si.sc_earlyz_tile_fifo_size = 0x130; + gb_addr_config = TAHITI_GB_ADDR_CONFIG_GOLDEN; break; case CHIP_PITCAIRN: rdev->config.si.max_shader_engines = 2; - rdev->config.si.max_pipes_per_simd = 4; rdev->config.si.max_tile_pipes = 8; - rdev->config.si.max_simds_per_se = 5; + rdev->config.si.max_cu_per_sh = 5; + rdev->config.si.max_sh_per_se = 2; rdev->config.si.max_backends_per_se = 4; rdev->config.si.max_texture_channel_caches = 8; rdev->config.si.max_gprs = 256; @@ -1610,13 +1550,14 @@ static void si_gpu_init(struct radeon_device *rdev) rdev->config.si.sc_prim_fifo_size_backend = 0x100; rdev->config.si.sc_hiz_tile_fifo_size = 0x30; rdev->config.si.sc_earlyz_tile_fifo_size = 0x130; + gb_addr_config = TAHITI_GB_ADDR_CONFIG_GOLDEN; break; case CHIP_VERDE: default: rdev->config.si.max_shader_engines = 1; - rdev->config.si.max_pipes_per_simd = 4; rdev->config.si.max_tile_pipes = 4; - rdev->config.si.max_simds_per_se = 2; + rdev->config.si.max_cu_per_sh = 2; + rdev->config.si.max_sh_per_se = 2; rdev->config.si.max_backends_per_se = 4; rdev->config.si.max_texture_channel_caches = 4; rdev->config.si.max_gprs = 256; @@ -1627,6 +1568,7 @@ static void si_gpu_init(struct radeon_device *rdev) rdev->config.si.sc_prim_fifo_size_backend = 0x40; rdev->config.si.sc_hiz_tile_fifo_size = 0x30; rdev->config.si.sc_earlyz_tile_fifo_size = 0x130; + gb_addr_config = VERDE_GB_ADDR_CONFIG_GOLDEN; break; } @@ -1648,31 +1590,7 @@ static void si_gpu_init(struct radeon_device *rdev) mc_shared_chmap = RREG32(MC_SHARED_CHMAP); mc_arb_ramcfg = RREG32(MC_ARB_RAMCFG); - cc_rb_backend_disable = RREG32(CC_RB_BACKEND_DISABLE); - cc_gc_shader_array_config = RREG32(CC_GC_SHADER_ARRAY_CONFIG); - cgts_tcc_disable = 0xffff0000; - for (i = 0; i < rdev->config.si.max_texture_channel_caches; i++) - cgts_tcc_disable &= ~(1 << (16 + i)); - gc_user_rb_backend_disable = RREG32(GC_USER_RB_BACKEND_DISABLE); - gc_user_shader_array_config = RREG32(GC_USER_SHADER_ARRAY_CONFIG); - cgts_user_tcc_disable = RREG32(CGTS_USER_TCC_DISABLE); - - rdev->config.si.num_shader_engines = rdev->config.si.max_shader_engines; rdev->config.si.num_tile_pipes = rdev->config.si.max_tile_pipes; - tmp = ((~gc_user_rb_backend_disable) & BACKEND_DISABLE_MASK) >> BACKEND_DISABLE_SHIFT; - rdev->config.si.num_backends_per_se = r600_count_pipe_bits(tmp); - tmp = (gc_user_rb_backend_disable & BACKEND_DISABLE_MASK) >> BACKEND_DISABLE_SHIFT; - rdev->config.si.backend_disable_mask_per_asic = - si_get_disable_mask_per_asic(rdev, tmp, SI_MAX_BACKENDS_PER_SE_MASK, - rdev->config.si.num_shader_engines); - rdev->config.si.backend_map = - si_get_tile_pipe_to_backend_map(rdev, rdev->config.si.num_tile_pipes, - rdev->config.si.num_backends_per_se * - rdev->config.si.num_shader_engines, - &rdev->config.si.backend_disable_mask_per_asic, - rdev->config.si.num_shader_engines); - tmp = ((~cgts_user_tcc_disable) & TCC_DISABLE_MASK) >> TCC_DISABLE_SHIFT; - rdev->config.si.num_texture_channel_caches = r600_count_pipe_bits(tmp); rdev->config.si.mem_max_burst_length_bytes = 256; tmp = (mc_arb_ramcfg & NOOFCOLS_MASK) >> NOOFCOLS_SHIFT; rdev->config.si.mem_row_size_in_kb = (4 * (1 << (8 + tmp))) / 1024; @@ -1683,55 +1601,8 @@ static void si_gpu_init(struct radeon_device *rdev) rdev->config.si.num_gpus = 1; rdev->config.si.multi_gpu_tile_size = 64; - gb_addr_config = 0; - switch (rdev->config.si.num_tile_pipes) { - case 1: - gb_addr_config |= NUM_PIPES(0); - break; - case 2: - gb_addr_config |= NUM_PIPES(1); - break; - case 4: - gb_addr_config |= NUM_PIPES(2); - break; - case 8: - default: - gb_addr_config |= NUM_PIPES(3); - break; - } - - tmp = (rdev->config.si.mem_max_burst_length_bytes / 256) - 1; - gb_addr_config |= PIPE_INTERLEAVE_SIZE(tmp); - gb_addr_config |= NUM_SHADER_ENGINES(rdev->config.si.num_shader_engines - 1); - tmp = (rdev->config.si.shader_engine_tile_size / 16) - 1; - gb_addr_config |= SHADER_ENGINE_TILE_SIZE(tmp); - switch (rdev->config.si.num_gpus) { - case 1: - default: - gb_addr_config |= NUM_GPUS(0); - break; - case 2: - gb_addr_config |= NUM_GPUS(1); - break; - case 4: - gb_addr_config |= NUM_GPUS(2); - break; - } - switch (rdev->config.si.multi_gpu_tile_size) { - case 16: - gb_addr_config |= MULTI_GPU_TILE_SIZE(0); - break; - case 32: - default: - gb_addr_config |= MULTI_GPU_TILE_SIZE(1); - break; - case 64: - gb_addr_config |= MULTI_GPU_TILE_SIZE(2); - break; - case 128: - gb_addr_config |= MULTI_GPU_TILE_SIZE(3); - break; - } + /* fix up row size */ + gb_addr_config &= ~ROW_SIZE_MASK; switch (rdev->config.si.mem_row_size_in_kb) { case 1: default: @@ -1745,26 +1616,6 @@ static void si_gpu_init(struct radeon_device *rdev) break; } - tmp = (gb_addr_config & NUM_PIPES_MASK) >> NUM_PIPES_SHIFT; - rdev->config.si.num_tile_pipes = (1 << tmp); - tmp = (gb_addr_config & PIPE_INTERLEAVE_SIZE_MASK) >> PIPE_INTERLEAVE_SIZE_SHIFT; - rdev->config.si.mem_max_burst_length_bytes = (tmp + 1) * 256; - tmp = (gb_addr_config & NUM_SHADER_ENGINES_MASK) >> NUM_SHADER_ENGINES_SHIFT; - rdev->config.si.num_shader_engines = tmp + 1; - tmp = (gb_addr_config & NUM_GPUS_MASK) >> NUM_GPUS_SHIFT; - rdev->config.si.num_gpus = tmp + 1; - tmp = (gb_addr_config & MULTI_GPU_TILE_SIZE_MASK) >> MULTI_GPU_TILE_SIZE_SHIFT; - rdev->config.si.multi_gpu_tile_size = 1 << tmp; - tmp = (gb_addr_config & ROW_SIZE_MASK) >> ROW_SIZE_SHIFT; - rdev->config.si.mem_row_size_in_kb = 1 << tmp; - - gb_backend_map = - si_get_tile_pipe_to_backend_map(rdev, rdev->config.si.num_tile_pipes, - rdev->config.si.num_backends_per_se * - rdev->config.si.num_shader_engines, - &rdev->config.si.backend_disable_mask_per_asic, - rdev->config.si.num_shader_engines); - /* 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 @@ -1789,33 +1640,29 @@ static void si_gpu_init(struct radeon_device *rdev) rdev->config.si.tile_config |= (3 << 0); break; } - rdev->config.si.tile_config |= - ((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT) << 4; + if ((mc_arb_ramcfg & NOOFBANK_MASK) >> NOOFBANK_SHIFT) + rdev->config.si.tile_config |= 1 << 4; + else + rdev->config.si.tile_config |= 0 << 4; rdev->config.si.tile_config |= ((gb_addr_config & PIPE_INTERLEAVE_SIZE_MASK) >> PIPE_INTERLEAVE_SIZE_SHIFT) << 8; rdev->config.si.tile_config |= ((gb_addr_config & ROW_SIZE_MASK) >> ROW_SIZE_SHIFT) << 12; - rdev->config.si.backend_map = gb_backend_map; WREG32(GB_ADDR_CONFIG, gb_addr_config); WREG32(DMIF_ADDR_CONFIG, gb_addr_config); WREG32(HDP_ADDR_CONFIG, gb_addr_config); - /* primary versions */ - WREG32(CC_RB_BACKEND_DISABLE, cc_rb_backend_disable); - WREG32(CC_SYS_RB_BACKEND_DISABLE, cc_rb_backend_disable); - WREG32(CC_GC_SHADER_ARRAY_CONFIG, cc_gc_shader_array_config); - - WREG32(CGTS_TCC_DISABLE, cgts_tcc_disable); + si_tiling_mode_table_init(rdev); - /* user versions */ - WREG32(GC_USER_RB_BACKEND_DISABLE, cc_rb_backend_disable); - WREG32(GC_USER_SYS_RB_BACKEND_DISABLE, cc_rb_backend_disable); - WREG32(GC_USER_SHADER_ARRAY_CONFIG, cc_gc_shader_array_config); + si_setup_rb(rdev, rdev->config.si.max_shader_engines, + rdev->config.si.max_sh_per_se, + rdev->config.si.max_backends_per_se); - WREG32(CGTS_USER_TCC_DISABLE, cgts_tcc_disable); + si_setup_spi(rdev, rdev->config.si.max_shader_engines, + rdev->config.si.max_sh_per_se, + rdev->config.si.max_cu_per_sh); - si_tiling_mode_table_init(rdev); /* set HW defaults for 3D engine */ WREG32(CP_QUEUE_THRESHOLDS, (ROQ_IB1_START(0x16) | diff --git a/drivers/gpu/drm/radeon/si_reg.h b/drivers/gpu/drm/radeon/si_reg.h index eda938a7cb6e..501f9d431d57 100644 --- a/drivers/gpu/drm/radeon/si_reg.h +++ b/drivers/gpu/drm/radeon/si_reg.h @@ -30,4 +30,76 @@ #define SI_DC_GPIO_HPD_EN 0x65b8 #define SI_DC_GPIO_HPD_Y 0x65bc +#define SI_GRPH_CONTROL 0x6804 +# define SI_GRPH_DEPTH(x) (((x) & 0x3) << 0) +# define SI_GRPH_DEPTH_8BPP 0 +# define SI_GRPH_DEPTH_16BPP 1 +# define SI_GRPH_DEPTH_32BPP 2 +# define SI_GRPH_NUM_BANKS(x) (((x) & 0x3) << 2) +# define SI_ADDR_SURF_2_BANK 0 +# define SI_ADDR_SURF_4_BANK 1 +# define SI_ADDR_SURF_8_BANK 2 +# define SI_ADDR_SURF_16_BANK 3 +# define SI_GRPH_Z(x) (((x) & 0x3) << 4) +# define SI_GRPH_BANK_WIDTH(x) (((x) & 0x3) << 6) +# define SI_ADDR_SURF_BANK_WIDTH_1 0 +# define SI_ADDR_SURF_BANK_WIDTH_2 1 +# define SI_ADDR_SURF_BANK_WIDTH_4 2 +# define SI_ADDR_SURF_BANK_WIDTH_8 3 +# define SI_GRPH_FORMAT(x) (((x) & 0x7) << 8) +/* 8 BPP */ +# define SI_GRPH_FORMAT_INDEXED 0 +/* 16 BPP */ +# define SI_GRPH_FORMAT_ARGB1555 0 +# define SI_GRPH_FORMAT_ARGB565 1 +# define SI_GRPH_FORMAT_ARGB4444 2 +# define SI_GRPH_FORMAT_AI88 3 +# define SI_GRPH_FORMAT_MONO16 4 +# define SI_GRPH_FORMAT_BGRA5551 5 +/* 32 BPP */ +# define SI_GRPH_FORMAT_ARGB8888 0 +# define SI_GRPH_FORMAT_ARGB2101010 1 +# define SI_GRPH_FORMAT_32BPP_DIG 2 +# define SI_GRPH_FORMAT_8B_ARGB2101010 3 +# define SI_GRPH_FORMAT_BGRA1010102 4 +# define SI_GRPH_FORMAT_8B_BGRA1010102 5 +# define SI_GRPH_FORMAT_RGB111110 6 +# define SI_GRPH_FORMAT_BGR101111 7 +# define SI_GRPH_BANK_HEIGHT(x) (((x) & 0x3) << 11) +# define SI_ADDR_SURF_BANK_HEIGHT_1 0 +# define SI_ADDR_SURF_BANK_HEIGHT_2 1 +# define SI_ADDR_SURF_BANK_HEIGHT_4 2 +# define SI_ADDR_SURF_BANK_HEIGHT_8 3 +# define SI_GRPH_TILE_SPLIT(x) (((x) & 0x7) << 13) +# define SI_ADDR_SURF_TILE_SPLIT_64B 0 +# define SI_ADDR_SURF_TILE_SPLIT_128B 1 +# define SI_ADDR_SURF_TILE_SPLIT_256B 2 +# define SI_ADDR_SURF_TILE_SPLIT_512B 3 +# define SI_ADDR_SURF_TILE_SPLIT_1KB 4 +# define SI_ADDR_SURF_TILE_SPLIT_2KB 5 +# define SI_ADDR_SURF_TILE_SPLIT_4KB 6 +# define SI_GRPH_MACRO_TILE_ASPECT(x) (((x) & 0x3) << 18) +# define SI_ADDR_SURF_MACRO_TILE_ASPECT_1 0 +# define SI_ADDR_SURF_MACRO_TILE_ASPECT_2 1 +# define SI_ADDR_SURF_MACRO_TILE_ASPECT_4 2 +# define SI_ADDR_SURF_MACRO_TILE_ASPECT_8 3 +# define SI_GRPH_ARRAY_MODE(x) (((x) & 0x7) << 20) +# define SI_GRPH_ARRAY_LINEAR_GENERAL 0 +# define SI_GRPH_ARRAY_LINEAR_ALIGNED 1 +# define SI_GRPH_ARRAY_1D_TILED_THIN1 2 +# define SI_GRPH_ARRAY_2D_TILED_THIN1 4 +# define SI_GRPH_PIPE_CONFIG(x) (((x) & 0x1f) << 24) +# define SI_ADDR_SURF_P2 0 +# define SI_ADDR_SURF_P4_8x16 4 +# define SI_ADDR_SURF_P4_16x16 5 +# define SI_ADDR_SURF_P4_16x32 6 +# define SI_ADDR_SURF_P4_32x32 7 +# define SI_ADDR_SURF_P8_16x16_8x16 8 +# define SI_ADDR_SURF_P8_16x32_8x16 9 +# define SI_ADDR_SURF_P8_32x32_8x16 10 +# define SI_ADDR_SURF_P8_16x32_16x16 11 +# define SI_ADDR_SURF_P8_32x32_16x16 12 +# define SI_ADDR_SURF_P8_32x32_16x32 13 +# define SI_ADDR_SURF_P8_32x64_32x32 14 + #endif diff --git a/drivers/gpu/drm/radeon/sid.h b/drivers/gpu/drm/radeon/sid.h index 53ea2c42dbd6..db4067962868 100644 --- a/drivers/gpu/drm/radeon/sid.h +++ b/drivers/gpu/drm/radeon/sid.h @@ -24,6 +24,11 @@ #ifndef SI_H #define SI_H +#define TAHITI_RB_BITMAP_WIDTH_PER_SH 2 + +#define TAHITI_GB_ADDR_CONFIG_GOLDEN 0x12011003 +#define VERDE_GB_ADDR_CONFIG_GOLDEN 0x12010002 + #define CG_MULT_THERMAL_STATUS 0x714 #define ASIC_MAX_TEMP(x) ((x) << 0) #define ASIC_MAX_TEMP_MASK 0x000001ff @@ -408,6 +413,12 @@ #define SOFT_RESET_IA (1 << 15) #define GRBM_GFX_INDEX 0x802C +#define INSTANCE_INDEX(x) ((x) << 0) +#define SH_INDEX(x) ((x) << 8) +#define SE_INDEX(x) ((x) << 16) +#define SH_BROADCAST_WRITES (1 << 29) +#define INSTANCE_BROADCAST_WRITES (1 << 30) +#define SE_BROADCAST_WRITES (1 << 31) #define GRBM_INT_CNTL 0x8060 # define RDERR_INT_ENABLE (1 << 0) @@ -480,6 +491,8 @@ #define VGT_TF_MEMORY_BASE 0x89B8 #define CC_GC_SHADER_ARRAY_CONFIG 0x89bc +#define INACTIVE_CUS_MASK 0xFFFF0000 +#define INACTIVE_CUS_SHIFT 16 #define GC_USER_SHADER_ARRAY_CONFIG 0x89c0 #define PA_CL_ENHANCE 0x8A14 @@ -688,6 +701,12 @@ #define RLC_MC_CNTL 0xC344 #define RLC_UCODE_CNTL 0xC348 +#define PA_SC_RASTER_CONFIG 0x28350 +# define RASTER_CONFIG_RB_MAP_0 0 +# define RASTER_CONFIG_RB_MAP_1 1 +# define RASTER_CONFIG_RB_MAP_2 2 +# define RASTER_CONFIG_RB_MAP_3 3 + #define VGT_EVENT_INITIATOR 0x28a90 # define SAMPLE_STREAMOUTSTATS1 (1 << 0) # define SAMPLE_STREAMOUTSTATS2 (2 << 0) |