Merge branch 'drm-fixes-4.19' of git://people.freedesktop.org/~agd5f/linux into drm-fixes

Fixes for 4.19:
- SR-IOV fixes
- Kasan and page fault fix on device removal
- S3 stability fix for CZ/ST
- VCE regression fixes for CIK parts
- Avoid holding the mn_lock when allocating memory
- DC memory leak fix
- BO eviction fix
Signed-off-by: NDave Airlie <airlied@redhat.com>
From: Alex Deucher <alexdeucher@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20180829202555.2653-1-alexander.deucher@amd.com

drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c

@@ -1012,13 +1012,9 @@ static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
 		if (r)
 			return r;
 
-		if (chunk_ib->flags & AMDGPU_IB_FLAG_PREAMBLE) {
-			parser->job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT;
-			if (!parser->ctx->preamble_presented) {
-				parser->job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
-				parser->ctx->preamble_presented = true;
-			}
-		}
+		if (chunk_ib->flags & AMDGPU_IB_FLAG_PREAMBLE)
+			parser->job->preamble_status |=
+				AMDGPU_PREAMBLE_IB_PRESENT;
 
 		if (parser->ring && parser->ring != ring)
 			return -EINVAL;
@@ -1207,26 +1203,24 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 
 	int r;
 
+	job = p->job;
+	p->job = NULL;
+
+	r = drm_sched_job_init(&job->base, entity, p->filp);
+	if (r)
+		goto error_unlock;
+
+	/* No memory allocation is allowed while holding the mn lock */
 	amdgpu_mn_lock(p->mn);
 	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
 		struct amdgpu_bo *bo = e->robj;
 
 		if (amdgpu_ttm_tt_userptr_needs_pages(bo->tbo.ttm)) {
-			amdgpu_mn_unlock(p->mn);
-			return -ERESTARTSYS;
+			r = -ERESTARTSYS;
+			goto error_abort;
 		}
 	}
 
-	job = p->job;
-	p->job = NULL;
-
-	r = drm_sched_job_init(&job->base, entity, p->filp);
-	if (r) {
-		amdgpu_job_free(job);
-		amdgpu_mn_unlock(p->mn);
-		return r;
-	}
-
 	job->owner = p->filp;
 	p->fence = dma_fence_get(&job->base.s_fence->finished);
 
@@ -1241,6 +1235,12 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 
 	amdgpu_cs_post_dependencies(p);
 
+	if ((job->preamble_status & AMDGPU_PREAMBLE_IB_PRESENT) &&
+	    !p->ctx->preamble_presented) {
+		job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
+		p->ctx->preamble_presented = true;
+	}
+
 	cs->out.handle = seq;
 	job->uf_sequence = seq;
 
@@ -1258,6 +1258,15 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	amdgpu_mn_unlock(p->mn);
 
 	return 0;
+
+error_abort:
+	dma_fence_put(&job->base.s_fence->finished);
+	job->base.s_fence = NULL;
+
+error_unlock:
+	amdgpu_job_free(job);
+	amdgpu_mn_unlock(p->mn);
+	return r;
 }
 
 int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)

drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c

@@ -164,8 +164,10 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 		return r;
 	}
 
+	need_ctx_switch = ring->current_ctx != fence_ctx;
 	if (ring->funcs->emit_pipeline_sync && job &&
 	    ((tmp = amdgpu_sync_get_fence(&job->sched_sync, NULL)) ||
+	     (amdgpu_sriov_vf(adev) && need_ctx_switch) ||
 	     amdgpu_vm_need_pipeline_sync(ring, job))) {
 		need_pipe_sync = true;
 		dma_fence_put(tmp);
@@ -196,7 +198,6 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	}
 
 	skip_preamble = ring->current_ctx == fence_ctx;
-	need_ctx_switch = ring->current_ctx != fence_ctx;
 	if (job && ring->funcs->emit_cntxcntl) {
 		if (need_ctx_switch)
 			status |= AMDGPU_HAVE_CTX_SWITCH;

drivers/gpu/drm/amd/amdgpu/amdgpu_pm.c

@@ -1932,14 +1932,6 @@ void amdgpu_pm_compute_clocks(struct amdgpu_device *adev)
 			amdgpu_fence_wait_empty(ring);
 	}
 
-	mutex_lock(&adev->pm.mutex);
-	/* update battery/ac status */
-	if (power_supply_is_system_supplied() > 0)
-		adev->pm.ac_power = true;
-	else
-		adev->pm.ac_power = false;
-	mutex_unlock(&adev->pm.mutex);
-
 	if (adev->powerplay.pp_funcs->dispatch_tasks) {
 		if (!amdgpu_device_has_dc_support(adev)) {
 			mutex_lock(&adev->pm.mutex);

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c

@@ -172,6 +172,7 @@ static void amdgpu_vm_bo_base_init(struct amdgpu_vm_bo_base *base,
 	 * is validated on next vm use to avoid fault.
 	 * */
 	list_move_tail(&base->vm_status, &vm->evicted);
+	base->moved = true;
 }
 
 /**
@@ -369,7 +370,6 @@ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev,
 	uint64_t addr;
 	int r;
 
-	addr = amdgpu_bo_gpu_offset(bo);
 	entries = amdgpu_bo_size(bo) / 8;
 
 	if (pte_support_ats) {
@@ -401,6 +401,7 @@ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev,
 	if (r)
 		goto error;
 
+	addr = amdgpu_bo_gpu_offset(bo);
 	if (ats_entries) {
 		uint64_t ats_value;
 
@@ -2483,28 +2484,52 @@ static uint32_t amdgpu_vm_get_block_size(uint64_t vm_size)
  * amdgpu_vm_adjust_size - adjust vm size, block size and fragment size
  *
  * @adev: amdgpu_device pointer
- * @vm_size: the default vm size if it's set auto
+ * @min_vm_size: the minimum vm size in GB if it's set auto
  * @fragment_size_default: Default PTE fragment size
  * @max_level: max VMPT level
  * @max_bits: max address space size in bits
  *
  */
-void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t vm_size,
+void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t min_vm_size,
 			   uint32_t fragment_size_default, unsigned max_level,
 			   unsigned max_bits)
 {
+	unsigned int max_size = 1 << (max_bits - 30);
+	unsigned int vm_size;
 	uint64_t tmp;
 
 	/* adjust vm size first */
 	if (amdgpu_vm_size != -1) {
-		unsigned max_size = 1 << (max_bits - 30);
-
 		vm_size = amdgpu_vm_size;
 		if (vm_size > max_size) {
 			dev_warn(adev->dev, "VM size (%d) too large, max is %u GB\n",
 				 amdgpu_vm_size, max_size);
 			vm_size = max_size;
 		}
+	} else {
+		struct sysinfo si;
+		unsigned int phys_ram_gb;
+
+		/* Optimal VM size depends on the amount of physical
+		 * RAM available. Underlying requirements and
+		 * assumptions:
+		 *
+		 *  - Need to map system memory and VRAM from all GPUs
+		 *     - VRAM from other GPUs not known here
+		 *     - Assume VRAM <= system memory
+		 *  - On GFX8 and older, VM space can be segmented for
+		 *    different MTYPEs
+		 *  - Need to allow room for fragmentation, guard pages etc.
+		 *
+		 * This adds up to a rough guess of system memory x3.
+		 * Round up to power of two to maximize the available
+		 * VM size with the given page table size.
+		 */
+		si_meminfo(&si);
+		phys_ram_gb = ((uint64_t)si.totalram * si.mem_unit +
+			       (1 << 30) - 1) >> 30;
+		vm_size = roundup_pow_of_two(
+			min(max(phys_ram_gb * 3, min_vm_size), max_size));
 	}
 
 	adev->vm_manager.max_pfn = (uint64_t)vm_size << 18;

drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h

@@ -321,7 +321,7 @@ struct amdgpu_bo_va_mapping *amdgpu_vm_bo_lookup_mapping(struct amdgpu_vm *vm,
 void amdgpu_vm_bo_trace_cs(struct amdgpu_vm *vm, struct ww_acquire_ctx *ticket);
 void amdgpu_vm_bo_rmv(struct amdgpu_device *adev,
 		      struct amdgpu_bo_va *bo_va);
-void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t vm_size,
+void amdgpu_vm_adjust_size(struct amdgpu_device *adev, uint32_t min_vm_size,
 			   uint32_t fragment_size_default, unsigned max_level,
 			   unsigned max_bits);
 int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp);

drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c

@@ -5664,6 +5664,11 @@ static int gfx_v8_0_set_powergating_state(void *handle,
 	if (amdgpu_sriov_vf(adev))
 		return 0;
 
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_SMG |
+				AMD_PG_SUPPORT_RLC_SMU_HS |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_GFX_DMG))
+		adev->gfx.rlc.funcs->enter_safe_mode(adev);
 	switch (adev->asic_type) {
 	case CHIP_CARRIZO:
 	case CHIP_STONEY:
@@ -5713,7 +5718,11 @@ static int gfx_v8_0_set_powergating_state(void *handle,
 	default:
 		break;
 	}
-
+	if (adev->pg_flags & (AMD_PG_SUPPORT_GFX_SMG |
+				AMD_PG_SUPPORT_RLC_SMU_HS |
+				AMD_PG_SUPPORT_CP |
+				AMD_PG_SUPPORT_GFX_DMG))
+		adev->gfx.rlc.funcs->exit_safe_mode(adev);
 	return 0;
 }
 

drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c

@@ -632,12 +632,6 @@ static void gmc_v6_0_gart_disable(struct amdgpu_device *adev)
 	amdgpu_gart_table_vram_unpin(adev);
 }
 
-static void gmc_v6_0_gart_fini(struct amdgpu_device *adev)
-{
-	amdgpu_gart_table_vram_free(adev);
-	amdgpu_gart_fini(adev);
-}
-
 static void gmc_v6_0_vm_decode_fault(struct amdgpu_device *adev,
 				     u32 status, u32 addr, u32 mc_client)
 {
@@ -935,8 +929,9 @@ static int gmc_v6_0_sw_fini(void *handle)
 
 	amdgpu_gem_force_release(adev);
 	amdgpu_vm_manager_fini(adev);
-	gmc_v6_0_gart_fini(adev);
+	amdgpu_gart_table_vram_free(adev);
 	amdgpu_bo_fini(adev);
+	amdgpu_gart_fini(adev);
 	release_firmware(adev->gmc.fw);
 	adev->gmc.fw = NULL;
 

drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c

@@ -746,19 +746,6 @@ static void gmc_v7_0_gart_disable(struct amdgpu_device *adev)
 	amdgpu_gart_table_vram_unpin(adev);
 }
 
-/**
- * gmc_v7_0_gart_fini - vm fini callback
- *
- * @adev: amdgpu_device pointer
- *
- * Tears down the driver GART/VM setup (CIK).
- */
-static void gmc_v7_0_gart_fini(struct amdgpu_device *adev)
-{
-	amdgpu_gart_table_vram_free(adev);
-	amdgpu_gart_fini(adev);
-}
-
 /**
  * gmc_v7_0_vm_decode_fault - print human readable fault info
  *
@@ -1095,8 +1082,9 @@ static int gmc_v7_0_sw_fini(void *handle)
 	amdgpu_gem_force_release(adev);
 	amdgpu_vm_manager_fini(adev);
 	kfree(adev->gmc.vm_fault_info);
-	gmc_v7_0_gart_fini(adev);
+	amdgpu_gart_table_vram_free(adev);
 	amdgpu_bo_fini(adev);
+	amdgpu_gart_fini(adev);
 	release_firmware(adev->gmc.fw);
 	adev->gmc.fw = NULL;
 

drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c

@@ -968,19 +968,6 @@ static void gmc_v8_0_gart_disable(struct amdgpu_device *adev)
 	amdgpu_gart_table_vram_unpin(adev);
 }
 
-/**
- * gmc_v8_0_gart_fini - vm fini callback
- *
- * @adev: amdgpu_device pointer
- *
- * Tears down the driver GART/VM setup (CIK).
- */
-static void gmc_v8_0_gart_fini(struct amdgpu_device *adev)
-{
-	amdgpu_gart_table_vram_free(adev);
-	amdgpu_gart_fini(adev);
-}
-
 /**
  * gmc_v8_0_vm_decode_fault - print human readable fault info
  *
@@ -1199,8 +1186,9 @@ static int gmc_v8_0_sw_fini(void *handle)
 	amdgpu_gem_force_release(adev);
 	amdgpu_vm_manager_fini(adev);
 	kfree(adev->gmc.vm_fault_info);
-	gmc_v8_0_gart_fini(adev);
+	amdgpu_gart_table_vram_free(adev);
 	amdgpu_bo_fini(adev);
+	amdgpu_gart_fini(adev);
 	release_firmware(adev->gmc.fw);
 	adev->gmc.fw = NULL;
 

drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c

@@ -942,26 +942,12 @@ static int gmc_v9_0_sw_init(void *handle)
 	return 0;
 }
 
-/**
- * gmc_v9_0_gart_fini - vm fini callback
- *
- * @adev: amdgpu_device pointer
- *
- * Tears down the driver GART/VM setup (CIK).
- */
-static void gmc_v9_0_gart_fini(struct amdgpu_device *adev)
-{
-	amdgpu_gart_table_vram_free(adev);
-	amdgpu_gart_fini(adev);
-}
-
 static int gmc_v9_0_sw_fini(void *handle)
 {
 	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 
 	amdgpu_gem_force_release(adev);
 	amdgpu_vm_manager_fini(adev);
-	gmc_v9_0_gart_fini(adev);
 
 	/*
 	* TODO:
@@ -974,7 +960,9 @@ static int gmc_v9_0_sw_fini(void *handle)
 	*/
 	amdgpu_bo_free_kernel(&adev->stolen_vga_memory, NULL, NULL);
 
+	amdgpu_gart_table_vram_free(adev);
 	amdgpu_bo_fini(adev);
+	amdgpu_gart_fini(adev);
 
 	return 0;
 }

drivers/gpu/drm/amd/amdgpu/kv_dpm.c

@@ -65,8 +65,6 @@ static int kv_set_thermal_temperature_range(struct amdgpu_device *adev,
 					    int min_temp, int max_temp);
 static int kv_init_fps_limits(struct amdgpu_device *adev);
 
-static void kv_dpm_powergate_uvd(void *handle, bool gate);
-static void kv_dpm_powergate_vce(struct amdgpu_device *adev, bool gate);
 static void kv_dpm_powergate_samu(struct amdgpu_device *adev, bool gate);
 static void kv_dpm_powergate_acp(struct amdgpu_device *adev, bool gate);
 
@@ -1354,8 +1352,6 @@ static int kv_dpm_enable(struct amdgpu_device *adev)
 		return ret;
 	}
 
-	kv_update_current_ps(adev, adev->pm.dpm.boot_ps);
-
 	if (adev->irq.installed &&
 	    amdgpu_is_internal_thermal_sensor(adev->pm.int_thermal_type)) {
 		ret = kv_set_thermal_temperature_range(adev, KV_TEMP_RANGE_MIN, KV_TEMP_RANGE_MAX);
@@ -1374,6 +1370,8 @@ static int kv_dpm_enable(struct amdgpu_device *adev)
 
 static void kv_dpm_disable(struct amdgpu_device *adev)
 {
+	struct kv_power_info *pi = kv_get_pi(adev);
+
 	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
 		       AMDGPU_THERMAL_IRQ_LOW_TO_HIGH);
 	amdgpu_irq_put(adev, &adev->pm.dpm.thermal.irq,
@@ -1387,8 +1385,10 @@ static void kv_dpm_disable(struct amdgpu_device *adev)
 	/* powerup blocks */
 	kv_dpm_powergate_acp(adev, false);
 	kv_dpm_powergate_samu(adev, false);
-	kv_dpm_powergate_vce(adev, false);
-	kv_dpm_powergate_uvd(adev, false);
+	if (pi->caps_vce_pg) /* power on the VCE block */
+		amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_VCEPowerON);
+	if (pi->caps_uvd_pg) /* power on the UVD block */
+		amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_UVDPowerON);
 
 	kv_enable_smc_cac(adev, false);
 	kv_enable_didt(adev, false);
@@ -1551,7 +1551,6 @@ static int kv_update_vce_dpm(struct amdgpu_device *adev,
 	int ret;
 
 	if (amdgpu_new_state->evclk > 0 && amdgpu_current_state->evclk == 0) {
-		kv_dpm_powergate_vce(adev, false);
 		if (pi->caps_stable_p_state)
 			pi->vce_boot_level = table->count - 1;
 		else
@@ -1573,7 +1572,6 @@ static int kv_update_vce_dpm(struct amdgpu_device *adev,
 		kv_enable_vce_dpm(adev, true);
 	} else if (amdgpu_new_state->evclk == 0 && amdgpu_current_state->evclk > 0) {
 		kv_enable_vce_dpm(adev, false);
-		kv_dpm_powergate_vce(adev, true);
 	}
 
 	return 0;
@@ -1702,24 +1700,32 @@ static void kv_dpm_powergate_uvd(void *handle, bool gate)
 	}
 }
 
-static void kv_dpm_powergate_vce(struct amdgpu_device *adev, bool gate)
+static void kv_dpm_powergate_vce(void *handle, bool gate)
 {
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
 	struct kv_power_info *pi = kv_get_pi(adev);
-
-	if (pi->vce_power_gated == gate)
-		return;
+	int ret;
 
 	pi->vce_power_gated = gate;
 
-	if (!pi->caps_vce_pg)
-		return;
-
-	if (gate)
-		amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_VCEPowerOFF);
-	else
-		amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_VCEPowerON);
+	if (gate) {
+		/* stop the VCE block */
+		ret = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							     AMD_PG_STATE_GATE);
+		kv_enable_vce_dpm(adev, false);
+		if (pi->caps_vce_pg) /* power off the VCE block */
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_VCEPowerOFF);
+	} else {
+		if (pi->caps_vce_pg) /* power on the VCE block */
+			amdgpu_kv_notify_message_to_smu(adev, PPSMC_MSG_VCEPowerON);
+		kv_enable_vce_dpm(adev, true);
+		/* re-init the VCE block */
+		ret = amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE,
+							     AMD_PG_STATE_UNGATE);
+	}
 }
 
+
 static void kv_dpm_powergate_samu(struct amdgpu_device *adev, bool gate)
 {
 	struct kv_power_info *pi = kv_get_pi(adev);
@@ -3061,7 +3067,7 @@ static int kv_dpm_hw_init(void *handle)
 	else
 		adev->pm.dpm_enabled = true;
 	mutex_unlock(&adev->pm.mutex);
-
+	amdgpu_pm_compute_clocks(adev);
 	return ret;
 }
 
@@ -3313,6 +3319,9 @@ static int kv_set_powergating_by_smu(void *handle,
 	case AMD_IP_BLOCK_TYPE_UVD:
 		kv_dpm_powergate_uvd(handle, gate);
 		break;
+	case AMD_IP_BLOCK_TYPE_VCE:
+		kv_dpm_powergate_vce(handle, gate);
+		break;
 	default:
 		break;
 	}

drivers/gpu/drm/amd/amdgpu/si_dpm.c

@@ -6887,7 +6887,6 @@ static int si_dpm_enable(struct amdgpu_device *adev)
 
 	si_enable_auto_throttle_source(adev, AMDGPU_DPM_AUTO_THROTTLE_SRC_THERMAL, true);
 	si_thermal_start_thermal_controller(adev);
-	ni_update_current_ps(adev, boot_ps);
 
 	return 0;
 }
@@ -7763,7 +7762,7 @@ static int si_dpm_hw_init(void *handle)
 	else
 		adev->pm.dpm_enabled = true;
 	mutex_unlock(&adev->pm.mutex);
-
+	amdgpu_pm_compute_clocks(adev);
 	return ret;
 }
 

drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_pp_smu.c

@@ -480,12 +480,20 @@ void pp_rv_set_display_requirement(struct pp_smu *pp,
 {
 	struct dc_context *ctx = pp->ctx;
 	struct amdgpu_device *adev = ctx->driver_context;
+	void *pp_handle = adev->powerplay.pp_handle;
 	const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs;
+	struct pp_display_clock_request clock = {0};
 
-	if (!pp_funcs || !pp_funcs->display_configuration_changed)
+	if (!pp_funcs || !pp_funcs->display_clock_voltage_request)
 		return;
 
-	amdgpu_dpm_display_configuration_changed(adev);
+	clock.clock_type = amd_pp_dcf_clock;
+	clock.clock_freq_in_khz = req->hard_min_dcefclk_khz;
+	pp_funcs->display_clock_voltage_request(pp_handle, &clock);
+
+	clock.clock_type = amd_pp_f_clock;
+	clock.clock_freq_in_khz = req->hard_min_fclk_khz;
+	pp_funcs->display_clock_voltage_request(pp_handle, &clock);
 }
 
 void pp_rv_set_wm_ranges(struct pp_smu *pp,

drivers/gpu/drm/amd/display/dc/core/dc_link.c

@@ -754,8 +754,12 @@ bool dc_link_detect(struct dc_link *link, enum dc_detect_reason reason)
 			 * fail-safe mode
 			 */
 			if (dc_is_hdmi_signal(link->connector_signal) ||
-			    dc_is_dvi_signal(link->connector_signal))
+			    dc_is_dvi_signal(link->connector_signal)) {
+				if (prev_sink != NULL)
+					dc_sink_release(prev_sink);
+
 				return false;
+			}
 		default:
 			break;
 		}