1 files changed, 71 insertions, 62 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index afe6af9c0138..c4a4e2fe6681 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -36,7 +36,9 @@
 #include <generated/utsrelease.h>
 #include <linux/pci-p2pdma.h>
 
+#include <drm/drm_aperture.h>
 #include <drm/drm_atomic_helper.h>
+#include <drm/drm_crtc_helper.h>
 #include <drm/drm_fb_helper.h>
 #include <drm/drm_probe_helper.h>
 #include <drm/amdgpu_drm.h>
@@ -90,6 +92,8 @@ MODULE_FIRMWARE("amdgpu/navi12_gpu_info.bin");
 #define AMDGPU_MAX_RETRY_LIMIT		2
 #define AMDGPU_RETRY_SRIOV_RESET(r) ((r) == -EBUSY || (r) == -ETIMEDOUT || (r) == -EINVAL)
 
+static const struct drm_driver amdgpu_kms_driver;
+
 const char *amdgpu_asic_name[] = {
 	"TAHITI",
 	"PITCAIRN",
@@ -160,7 +164,7 @@ static void amdgpu_device_get_pcie_info(struct amdgpu_device *adev);
  *
  * The amdgpu driver provides a sysfs API for reporting the product name
  * for the device
- * The file serial_number is used for this and returns the product name
+ * The file product_name is used for this and returns the product name
  * as returned from the FRU.
  * NOTE: This is only available for certain server cards
  */
@@ -182,7 +186,7 @@ static DEVICE_ATTR(product_name, S_IRUGO,
  *
  * The amdgpu driver provides a sysfs API for reporting the part number
  * for the device
- * The file serial_number is used for this and returns the part number
+ * The file product_number is used for this and returns the part number
  * as returned from the FRU.
  * NOTE: This is only available for certain server cards
  */
@@ -924,32 +928,33 @@ static int amdgpu_device_asic_init(struct amdgpu_device *adev)
 }
 
 /**
- * amdgpu_device_vram_scratch_init - allocate the VRAM scratch page
+ * amdgpu_device_mem_scratch_init - allocate the VRAM scratch page
  *
  * @adev: amdgpu_device pointer
  *
  * Allocates a scratch page of VRAM for use by various things in the
  * driver.
  */
-static int amdgpu_device_vram_scratch_init(struct amdgpu_device *adev)
+static int amdgpu_device_mem_scratch_init(struct amdgpu_device *adev)
 {
-	return amdgpu_bo_create_kernel(adev, AMDGPU_GPU_PAGE_SIZE,
-				       PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM,
-				       &adev->vram_scratch.robj,
-				       &adev->vram_scratch.gpu_addr,
-				       (void **)&adev->vram_scratch.ptr);
+	return amdgpu_bo_create_kernel(adev, AMDGPU_GPU_PAGE_SIZE, PAGE_SIZE,
+				       AMDGPU_GEM_DOMAIN_VRAM |
+				       AMDGPU_GEM_DOMAIN_GTT,
+				       &adev->mem_scratch.robj,
+				       &adev->mem_scratch.gpu_addr,
+				       (void **)&adev->mem_scratch.ptr);
 }
 
 /**
- * amdgpu_device_vram_scratch_fini - Free the VRAM scratch page
+ * amdgpu_device_mem_scratch_fini - Free the VRAM scratch page
  *
  * @adev: amdgpu_device pointer
  *
  * Frees the VRAM scratch page.
  */
-static void amdgpu_device_vram_scratch_fini(struct amdgpu_device *adev)
+static void amdgpu_device_mem_scratch_fini(struct amdgpu_device *adev)
 {
-	amdgpu_bo_free_kernel(&adev->vram_scratch.robj, NULL, NULL);
+	amdgpu_bo_free_kernel(&adev->mem_scratch.robj, NULL, NULL);
 }
 
 /**
@@ -1981,17 +1986,10 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev)
 	}
 
 	snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_gpu_info.bin", chip_name);
-	err = request_firmware(&adev->firmware.gpu_info_fw, fw_name, adev->dev);
+	err = amdgpu_ucode_request(adev, &adev->firmware.gpu_info_fw, fw_name);
 	if (err) {
 		dev_err(adev->dev,
-			"Failed to load gpu_info firmware \"%s\"\n",
-			fw_name);
-		goto out;
-	}
-	err = amdgpu_ucode_validate(adev->firmware.gpu_info_fw);
-	if (err) {
-		dev_err(adev->dev,
-			"Failed to validate gpu_info firmware \"%s\"\n",
+			"Failed to get gpu_info firmware \"%s\"\n",
 			fw_name);
 		goto out;
 	}
@@ -2078,6 +2076,7 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 	struct drm_device *dev = adev_to_drm(adev);
 	struct pci_dev *parent;
 	int i, r;
+	bool total;
 
 	amdgpu_device_enable_virtual_display(adev);
 
@@ -2161,6 +2160,7 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 	if (amdgpu_sriov_vf(adev) && adev->asic_type == CHIP_SIENNA_CICHLID)
 		adev->pm.pp_feature &= ~PP_OVERDRIVE_MASK;
 
+	total = true;
 	for (i = 0; i < adev->num_ip_blocks; i++) {
 		if ((amdgpu_ip_block_mask & (1 << i)) == 0) {
 			DRM_ERROR("disabled ip block: %d <%s>\n",
@@ -2174,7 +2174,7 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 				} else if (r) {
 					DRM_ERROR("early_init of IP block <%s> failed %d\n",
 						  adev->ip_blocks[i].version->funcs->name, r);
-					return r;
+					total = false;
 				} else {
 					adev->ip_blocks[i].status.valid = true;
 				}
@@ -2205,6 +2205,8 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev)
 
 		}
 	}
+	if (!total)
+		return -ENODEV;
 
 	adev->cg_flags &= amdgpu_cg_mask;
 	adev->pg_flags &= amdgpu_pg_mask;
@@ -2390,9 +2392,9 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
 			if (amdgpu_sriov_vf(adev))
 				amdgpu_virt_exchange_data(adev);
 
-			r = amdgpu_device_vram_scratch_init(adev);
+			r = amdgpu_device_mem_scratch_init(adev);
 			if (r) {
-				DRM_ERROR("amdgpu_vram_scratch_init failed %d\n", r);
+				DRM_ERROR("amdgpu_mem_scratch_init failed %d\n", r);
 				goto init_failed;
 			}
 			r = adev->ip_blocks[i].version->funcs->hw_init((void *)adev);
@@ -2410,8 +2412,9 @@ static int amdgpu_device_ip_init(struct amdgpu_device *adev)
 			/* right after GMC hw init, we create CSA */
 			if (amdgpu_mcbp) {
 				r = amdgpu_allocate_static_csa(adev, &adev->virt.csa_obj,
-								AMDGPU_GEM_DOMAIN_VRAM,
-								AMDGPU_CSA_SIZE);
+							       AMDGPU_GEM_DOMAIN_VRAM |
+							       AMDGPU_GEM_DOMAIN_GTT,
+							       AMDGPU_CSA_SIZE);
 				if (r) {
 					DRM_ERROR("allocate CSA failed %d\n", r);
 					goto init_failed;
@@ -2581,9 +2584,10 @@ int amdgpu_device_set_cg_state(struct amdgpu_device *adev,
 		i = state == AMD_CG_STATE_GATE ? j : adev->num_ip_blocks - j - 1;
 		if (!adev->ip_blocks[i].status.late_initialized)
 			continue;
-		/* skip CG for GFX on S0ix */
+		/* skip CG for GFX, SDMA on S0ix */
 		if (adev->in_s0ix &&
-		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX)
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX ||
+		     adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SDMA))
 			continue;
 		/* skip CG for VCE/UVD, it's handled specially */
 		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD &&
@@ -2617,9 +2621,10 @@ int amdgpu_device_set_pg_state(struct amdgpu_device *adev,
 		i = state == AMD_PG_STATE_GATE ? j : adev->num_ip_blocks - j - 1;
 		if (!adev->ip_blocks[i].status.late_initialized)
 			continue;
-		/* skip PG for GFX on S0ix */
+		/* skip PG for GFX, SDMA on S0ix */
 		if (adev->in_s0ix &&
-		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX)
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX ||
+		     adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SDMA))
 			continue;
 		/* skip CG for VCE/UVD, it's handled specially */
 		if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD &&
@@ -2871,7 +2876,7 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev)
 			amdgpu_ucode_free_bo(adev);
 			amdgpu_free_static_csa(&adev->virt.csa_obj);
 			amdgpu_device_wb_fini(adev);
-			amdgpu_device_vram_scratch_fini(adev);
+			amdgpu_device_mem_scratch_fini(adev);
 			amdgpu_ib_pool_fini(adev);
 		}
 
@@ -3027,6 +3032,24 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
 		     adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_MES))
 			continue;
 
+		/* SDMA 5.x+ is part of GFX power domain so it's covered by GFXOFF */
+		if (adev->in_s0ix &&
+		    (adev->ip_versions[SDMA0_HWIP][0] >= IP_VERSION(5, 0, 0)) &&
+		    (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SDMA))
+			continue;
+
+		/* Once swPSP provides the IMU, RLC FW binaries to TOS during cold-boot.
+		 * These are in TMR, hence are expected to be reused by PSP-TOS to reload
+		 * from this location and RLC Autoload automatically also gets loaded
+		 * from here based on PMFW -> PSP message during re-init sequence.
+		 * Therefore, the psp suspend & resume should be skipped to avoid destroy
+		 * the TMR and reload FWs again for IMU enabled APU ASICs.
+		 */
+		if (amdgpu_in_reset(adev) &&
+		    (adev->flags & AMD_IS_APU) && adev->gfx.imu.funcs &&
+		    adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP)
+			continue;
+
 		/* XXX handle errors */
 		r = adev->ip_blocks[i].version->funcs->suspend(adev);
 		/* XXX handle errors */
@@ -3227,15 +3250,6 @@ static int amdgpu_device_ip_resume_phase2(struct amdgpu_device *adev)
 			return r;
 		}
 		adev->ip_blocks[i].status.hw = true;
-
-		if (adev->in_s0ix && adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SMC) {
-			/* disable gfxoff for IP resume. The gfxoff will be re-enabled in
-			 * amdgpu_device_resume() after IP resume.
-			 */
-			amdgpu_gfx_off_ctrl(adev, false);
-			DRM_DEBUG("will disable gfxoff for re-initializing other blocks\n");
-		}
-
 	}
 
 	return 0;
@@ -3687,6 +3701,11 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	if (r)
 		return r;
 
+	/* Get rid of things like offb */
+	r = drm_aperture_remove_conflicting_pci_framebuffers(adev->pdev, &amdgpu_kms_driver);
+	if (r)
+		return r;
+
 	/* Enable TMZ based on IP_VERSION */
 	amdgpu_gmc_tmz_set(adev);
 
@@ -3989,10 +4008,8 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev)
 	}
 	amdgpu_fence_driver_hw_fini(adev);
 
-	if (adev->mman.initialized) {
-		flush_delayed_work(&adev->mman.bdev.wq);
-		ttm_bo_lock_delayed_workqueue(&adev->mman.bdev);
-	}
+	if (adev->mman.initialized)
+		drain_workqueue(adev->mman.bdev.wq);
 
 	if (adev->pm_sysfs_en)
 		amdgpu_pm_sysfs_fini(adev);
@@ -4014,7 +4031,8 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev)
 
 	amdgpu_gart_dummy_page_fini(adev);
 
-	amdgpu_device_unmap_mmio(adev);
+	if (drm_dev_is_unplugged(adev_to_drm(adev)))
+		amdgpu_device_unmap_mmio(adev);
 
 }
 
@@ -4024,8 +4042,7 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev)
 
 	amdgpu_fence_driver_sw_fini(adev);
 	amdgpu_device_ip_fini(adev);
-	release_firmware(adev->firmware.gpu_info_fw);
-	adev->firmware.gpu_info_fw = NULL;
+	amdgpu_ucode_release(&adev->firmware.gpu_info_fw);
 	adev->accel_working = false;
 	dma_fence_put(rcu_dereference_protected(adev->gang_submit, true));
 
@@ -4223,13 +4240,6 @@ exit:
 	/* Make sure IB tests flushed */
 	flush_delayed_work(&adev->delayed_init_work);
 
-	if (adev->in_s0ix) {
-		/* re-enable gfxoff after IP resume. This re-enables gfxoff after
-		 * it was disabled for IP resume in amdgpu_device_ip_resume_phase2().
-		 */
-		amdgpu_gfx_off_ctrl(adev, true);
-		DRM_DEBUG("will enable gfxoff for the mission mode\n");
-	}
 	if (fbcon)
 		drm_fb_helper_set_suspend_unlocked(adev_to_drm(adev)->fb_helper, false);
 
@@ -4260,6 +4270,9 @@ exit:
 	}
 	adev->in_suspend = false;
 
+	if (adev->enable_mes)
+		amdgpu_mes_self_test(adev);
+
 	if (amdgpu_acpi_smart_shift_update(dev, AMDGPU_SS_DEV_D0))
 		DRM_WARN("smart shift update failed\n");
 
@@ -4610,11 +4623,6 @@ bool amdgpu_device_should_recover_gpu(struct amdgpu_device *adev)
 	if (!amdgpu_ras_is_poison_mode_supported(adev))
 		return true;
 
-	if (!amdgpu_device_ip_check_soft_reset(adev)) {
-		dev_info(adev->dev,"Timeout, but no hardware hang detected.\n");
-		return false;
-	}
-
 	if (amdgpu_sriov_vf(adev))
 		return true;
 
@@ -4739,7 +4747,8 @@ int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev,
 		if (!need_full_reset)
 			need_full_reset = amdgpu_device_ip_need_full_reset(adev);
 
-		if (!need_full_reset && amdgpu_gpu_recovery) {
+		if (!need_full_reset && amdgpu_gpu_recovery &&
+		    amdgpu_device_ip_check_soft_reset(adev)) {
 			amdgpu_device_ip_pre_soft_reset(adev);
 			r = amdgpu_device_ip_soft_reset(adev);
 			amdgpu_device_ip_post_soft_reset(adev);
@@ -5865,8 +5874,8 @@ void amdgpu_device_invalidate_hdp(struct amdgpu_device *adev,
 int amdgpu_in_reset(struct amdgpu_device *adev)
 {
 	return atomic_read(&adev->reset_domain->in_gpu_reset);
-	}
-	
+}
+
 /**
  * amdgpu_device_halt() - bring hardware to some kind of halt state
  *