Merge branch 'drm-fixes-4.7' of git://people.freedesktop.org/~agd5f/linux into drm-fixes

Mostly memory leak and firmware leak fixes for amdgpu.  A bit bigger than
usual since this is several weeks worth of fixes.

* 'drm-fixes-4.7' of git://people.freedesktop.org/~agd5f/linux: (28 commits)
  drm/amd/powerplay: delete useless code as pptable changed in vbios.
  drm/amd/powerplay: fix bug visit array out of bounds
  drm/amdgpu: fix smu ucode memleak (v2)
  drm/amdgpu: add release firmware for cgs
  drm/amdgpu: fix tonga smu_fini mem leak
  drm/amdgpu: fix fiji smu fini mem leak
  drm/amdgpu: fix cik sdma ucode memleak
  drm/amdgpu: fix sdma24 ucode mem leak
  drm/amdgpu: fix sdma3 ucode mem leak
  drm/amdgpu: fix uvd fini mem leak
  drm/amdgpu: fix gfx 7 ucode mem leak
  drm/amdgpu: fix gfx8 ucode mem leak
  drm/amdgpu: fix missing free wb for cond_exec
  drm/amdgpu: fix memleak in pptable_init
  drm/amdgpu: fix mem leak in atombios
  drm/amdgpu: fix mem leak in pplib/hwmgr
  drm/amdgpu: fix mem leak in smumgr
  drm/amdgpu: add pipeline sync while vmid switch in same ctx
  drm/amdgpu: vBIOS post only call when mem_size zero
  drm/amdgpu: modify sdma start sequence
  ...

31 files changed, 249 insertions, 92 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 992f00b65be4..01c36b8d6222 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -799,6 +799,7 @@ struct amdgpu_ring {
 	unsigned		cond_exe_offs;
 	u64				cond_exe_gpu_addr;
 	volatile u32	*cond_exe_cpu_addr;
+	int                     vmid;
 };
 
 /*
@@ -936,7 +937,8 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring,
 		    unsigned vm_id, uint64_t pd_addr,
 		    uint32_t gds_base, uint32_t gds_size,
 		    uint32_t gws_base, uint32_t gws_size,
-		    uint32_t oa_base, uint32_t oa_size);
+		    uint32_t oa_base, uint32_t oa_size,
+		    bool vmid_switch);
 void amdgpu_vm_reset_id(struct amdgpu_device *adev, unsigned vm_id);
 uint64_t amdgpu_vm_map_gart(const dma_addr_t *pages_addr, uint64_t addr);
 int amdgpu_vm_update_page_directory(struct amdgpu_device *adev,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
index 199f76baf22c..8943099eb135 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cgs.c
@@ -696,6 +696,17 @@ static uint32_t fw_type_convert(struct cgs_device *cgs_device, uint32_t fw_type)
 	return result;
 }
 
+static int amdgpu_cgs_rel_firmware(struct cgs_device *cgs_device, enum cgs_ucode_id type)
+{
+	CGS_FUNC_ADEV;
+	if ((CGS_UCODE_ID_SMU == type) || (CGS_UCODE_ID_SMU_SK == type)) {
+		release_firmware(adev->pm.fw);
+		return 0;
+	}
+	/* cannot release other firmware because they are not created by cgs */
+	return -EINVAL;
+}
+
 static int amdgpu_cgs_get_firmware_info(struct cgs_device *cgs_device,
 					enum cgs_ucode_id type,
 					struct cgs_firmware_info *info)
@@ -1125,6 +1136,7 @@ static const struct cgs_ops amdgpu_cgs_ops = {
 	amdgpu_cgs_pm_query_clock_limits,
 	amdgpu_cgs_set_camera_voltages,
 	amdgpu_cgs_get_firmware_info,
+	amdgpu_cgs_rel_firmware,
 	amdgpu_cgs_set_powergating_state,
 	amdgpu_cgs_set_clockgating_state,
 	amdgpu_cgs_get_active_displays_info,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index bb8b149786d7..964f31404f17 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -827,8 +827,10 @@ static uint32_t cail_ioreg_read(struct card_info *info, uint32_t reg)
  */
 static void amdgpu_atombios_fini(struct amdgpu_device *adev)
 {
-	if (adev->mode_info.atom_context)
+	if (adev->mode_info.atom_context) {
 		kfree(adev->mode_info.atom_context->scratch);
+		kfree(adev->mode_info.atom_context->iio);
+	}
 	kfree(adev->mode_info.atom_context);
 	adev->mode_info.atom_context = NULL;
 	kfree(adev->mode_info.atom_card_info);
@@ -1325,6 +1327,11 @@ static int amdgpu_fini(struct amdgpu_device *adev)
 		adev->ip_block_status[i].valid = false;
 	}
 
+	for (i = adev->num_ip_blocks - 1; i >= 0; i--) {
+		if (adev->ip_blocks[i].funcs->late_fini)
+			adev->ip_blocks[i].funcs->late_fini((void *)adev);
+	}
+
 	return 0;
 }
 
@@ -1513,8 +1520,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 		amdgpu_atombios_has_gpu_virtualization_table(adev);
 
 	/* Post card if necessary */
-	if (!amdgpu_card_posted(adev) ||
-	    adev->virtualization.supports_sr_iov) {
+	if (!amdgpu_card_posted(adev)) {
 		if (!adev->bios) {
 			dev_err(adev->dev, "Card not posted and no BIOS - ignoring\n");
 			return -EINVAL;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
index 34e35423b78e..7a0b1e50f293 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c
@@ -122,6 +122,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	bool skip_preamble, need_ctx_switch;
 	unsigned patch_offset = ~0;
 	struct amdgpu_vm *vm;
+	int vmid = 0, old_vmid = ring->vmid;
 	struct fence *hwf;
 	uint64_t ctx;
 
@@ -135,9 +136,11 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	if (job) {
 		vm = job->vm;
 		ctx = job->ctx;
+		vmid = job->vm_id;
 	} else {
 		vm = NULL;
 		ctx = 0;
+		vmid = 0;
 	}
 
 	if (!ring->ready) {
@@ -163,7 +166,8 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 		r = amdgpu_vm_flush(ring, job->vm_id, job->vm_pd_addr,
 				    job->gds_base, job->gds_size,
 				    job->gws_base, job->gws_size,
-				    job->oa_base, job->oa_size);
+				    job->oa_base, job->oa_size,
+				    (ring->current_ctx == ctx) && (old_vmid != vmid));
 		if (r) {
 			amdgpu_ring_undo(ring);
 			return r;
@@ -180,7 +184,6 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 	need_ctx_switch = ring->current_ctx != ctx;
 	for (i = 0; i < num_ibs; ++i) {
 		ib = &ibs[i];
-
 		/* drop preamble IBs if we don't have a context switch */
 		if ((ib->flags & AMDGPU_IB_FLAG_PREAMBLE) && skip_preamble)
 			continue;
@@ -188,6 +191,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 		amdgpu_ring_emit_ib(ring, ib, job ? job->vm_id : 0,
 				    need_ctx_switch);
 		need_ctx_switch = false;
+		ring->vmid = vmid;
 	}
 
 	if (ring->funcs->emit_hdp_invalidate)
@@ -198,6 +202,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned num_ibs,
 		dev_err(adev->dev, "failed to emit fence (%d)\n", r);
 		if (job && job->vm_id)
 			amdgpu_vm_reset_id(adev, job->vm_id);
+		ring->vmid = old_vmid;
 		amdgpu_ring_undo(ring);
 		return r;
 	}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c
index 6bd961fb43dc..10b1be51318b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_powerplay.c
@@ -183,13 +183,6 @@ static int amdgpu_pp_sw_fini(void *handle)
 	if (ret)
 		return ret;
 
-#ifdef CONFIG_DRM_AMD_POWERPLAY
-	if (adev->pp_enabled) {
-		amdgpu_pm_sysfs_fini(adev);
-		amd_powerplay_fini(adev->powerplay.pp_handle);
-	}
-#endif
-
 	return ret;
 }
 
@@ -223,6 +216,22 @@ static int amdgpu_pp_hw_fini(void *handle)
 	return ret;
 }
 
+static void amdgpu_pp_late_fini(void *handle)
+{
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+#ifdef CONFIG_DRM_AMD_POWERPLAY
+	if (adev->pp_enabled) {
+		amdgpu_pm_sysfs_fini(adev);
+		amd_powerplay_fini(adev->powerplay.pp_handle);
+	}
+
+	if (adev->powerplay.ip_funcs->late_fini)
+		adev->powerplay.ip_funcs->late_fini(
+			  adev->powerplay.pp_handle);
+#endif
+}
+
 static int amdgpu_pp_suspend(void *handle)
 {
 	int ret = 0;
@@ -311,6 +320,7 @@ const struct amd_ip_funcs amdgpu_pp_ip_funcs = {
 	.sw_fini = amdgpu_pp_sw_fini,
 	.hw_init = amdgpu_pp_hw_init,
 	.hw_fini = amdgpu_pp_hw_fini,
+	.late_fini = amdgpu_pp_late_fini,
 	.suspend = amdgpu_pp_suspend,
 	.resume = amdgpu_pp_resume,
 	.is_idle = amdgpu_pp_is_idle,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
index 3b02272db678..870f9494252c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
@@ -343,6 +343,7 @@ void amdgpu_ring_fini(struct amdgpu_ring *ring)
 	ring->ring = NULL;
 	ring->ring_obj = NULL;
 
+	amdgpu_wb_free(ring->adev, ring->cond_exe_offs);
 	amdgpu_wb_free(ring->adev, ring->fence_offs);
 	amdgpu_wb_free(ring->adev, ring->rptr_offs);
 	amdgpu_wb_free(ring->adev, ring->wptr_offs);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
index 8bf84efafb04..48618ee324eb 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sa.c
@@ -115,6 +115,7 @@ int amdgpu_sa_bo_manager_start(struct amdgpu_device *adev,
 		return r;
 	}
 	r = amdgpu_bo_kmap(sa_manager->bo, &sa_manager->cpu_ptr);
+	memset(sa_manager->cpu_ptr, 0, sa_manager->size);
 	amdgpu_bo_unreserve(sa_manager->bo);
 	return r;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
index 01abfc21b4a2..e19520c4b4b6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c
@@ -253,19 +253,20 @@ int amdgpu_uvd_sw_fini(struct amdgpu_device *adev)
 {
 	int r;
 
-	if (adev->uvd.vcpu_bo == NULL)
-		return 0;
+	kfree(adev->uvd.saved_bo);
 
 	amd_sched_entity_fini(&adev->uvd.ring.sched, &adev->uvd.entity);
 
-	r = amdgpu_bo_reserve(adev->uvd.vcpu_bo, false);
-	if (!r) {
-		amdgpu_bo_kunmap(adev->uvd.vcpu_bo);
-		amdgpu_bo_unpin(adev->uvd.vcpu_bo);
-		amdgpu_bo_unreserve(adev->uvd.vcpu_bo);
-	}
+	if (adev->uvd.vcpu_bo) {
+		r = amdgpu_bo_reserve(adev->uvd.vcpu_bo, false);
+		if (!r) {
+			amdgpu_bo_kunmap(adev->uvd.vcpu_bo);
+			amdgpu_bo_unpin(adev->uvd.vcpu_bo);
+			amdgpu_bo_unreserve(adev->uvd.vcpu_bo);
+		}
 
-	amdgpu_bo_unref(&adev->uvd.vcpu_bo);
+		amdgpu_bo_unref(&adev->uvd.vcpu_bo);
+	}
 
 	amdgpu_ring_fini(&adev->uvd.ring);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 9f36ed30ba11..62a4c127620f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -298,7 +298,8 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring,
 		    unsigned vm_id, uint64_t pd_addr,
 		    uint32_t gds_base, uint32_t gds_size,
 		    uint32_t gws_base, uint32_t gws_size,
-		    uint32_t oa_base, uint32_t oa_size)
+		    uint32_t oa_base, uint32_t oa_size,
+		    bool vmid_switch)
 {
 	struct amdgpu_device *adev = ring->adev;
 	struct amdgpu_vm_id *id = &adev->vm_manager.ids[vm_id];
@@ -312,8 +313,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring,
 	int r;
 
 	if (ring->funcs->emit_pipeline_sync && (
-	    pd_addr != AMDGPU_VM_NO_FLUSH || gds_switch_needed ||
-		    ring->type == AMDGPU_RING_TYPE_COMPUTE))
+	    pd_addr != AMDGPU_VM_NO_FLUSH || gds_switch_needed || vmid_switch))
 		amdgpu_ring_emit_pipeline_sync(ring);
 
 	if (ring->funcs->emit_vm_flush &&
diff --git a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
index ea407db1fbcf..5ec1f1e9c983 100644
--- a/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/ci_dpm.c
@@ -6221,6 +6221,9 @@ static int ci_dpm_sw_fini(void *handle)
 	ci_dpm_fini(adev);
 	mutex_unlock(&adev->pm.mutex);
 
+	release_firmware(adev->pm.fw);
+	adev->pm.fw = NULL;
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
index 518dca43b133..9dc4e24e31e7 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_sdma.c
@@ -66,6 +66,16 @@ MODULE_FIRMWARE("radeon/mullins_sdma1.bin");
 
 u32 amdgpu_cik_gpu_check_soft_reset(struct amdgpu_device *adev);
 
+
+static void cik_sdma_free_microcode(struct amdgpu_device *adev)
+{
+	int i;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+			release_firmware(adev->sdma.instance[i].fw);
+			adev->sdma.instance[i].fw = NULL;
+	}
+}
+
 /*
  * sDMA - System DMA
  * Starting with CIK, the GPU has new asynchronous
@@ -419,6 +429,8 @@ static int cik_sdma_gfx_resume(struct amdgpu_device *adev)
 		/* Initialize the ring buffer's read and write pointers */
 		WREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i], 0);
 		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_OFFSET + sdma_offsets[i], 0);
 
 		/* set the wb address whether it's enabled or not */
 		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
@@ -446,7 +458,12 @@ static int cik_sdma_gfx_resume(struct amdgpu_device *adev)
 		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
 
 		ring->ready = true;
+	}
+
+	cik_sdma_enable(adev, true);
 
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
 		r = amdgpu_ring_test_ring(ring);
 		if (r) {
 			ring->ready = false;
@@ -529,8 +546,8 @@ static int cik_sdma_start(struct amdgpu_device *adev)
 	if (r)
 		return r;
 
-	/* unhalt the MEs */
-	cik_sdma_enable(adev, true);
+	/* halt the engine before programing */
+	cik_sdma_enable(adev, false);
 
 	/* start the gfx rings and rlc compute queues */
 	r = cik_sdma_gfx_resume(adev);
@@ -998,6 +1015,7 @@ static int cik_sdma_sw_fini(void *handle)
 	for (i = 0; i < adev->sdma.num_instances; i++)
 		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
 
+	cik_sdma_free_microcode(adev);
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c b/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c
index 245cabf06575..ed03b75175d4 100644
--- a/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/fiji_dpm.c
@@ -72,6 +72,11 @@ static int fiji_dpm_sw_init(void *handle)
 
 static int fiji_dpm_sw_fini(void *handle)
 {
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	release_firmware(adev->pm.fw);
+	adev->pm.fw = NULL;
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
index 7f18a53ab53a..8c6ad1e72f02 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c
@@ -991,6 +991,22 @@ out:
 	return err;
 }
 
+static void gfx_v7_0_free_microcode(struct amdgpu_device *adev)
+{
+	release_firmware(adev->gfx.pfp_fw);
+	adev->gfx.pfp_fw = NULL;
+	release_firmware(adev->gfx.me_fw);
+	adev->gfx.me_fw = NULL;
+	release_firmware(adev->gfx.ce_fw);
+	adev->gfx.ce_fw = NULL;
+	release_firmware(adev->gfx.mec_fw);
+	adev->gfx.mec_fw = NULL;
+	release_firmware(adev->gfx.mec2_fw);
+	adev->gfx.mec2_fw = NULL;
+	release_firmware(adev->gfx.rlc_fw);
+	adev->gfx.rlc_fw = NULL;
+}
+
 /**
  * gfx_v7_0_tiling_mode_table_init - init the hw tiling table
  *
@@ -4489,6 +4505,7 @@ static int gfx_v7_0_sw_fini(void *handle)
 	gfx_v7_0_cp_compute_fini(adev);
 	gfx_v7_0_rlc_fini(adev);
 	gfx_v7_0_mec_fini(adev);
+	gfx_v7_0_free_microcode(adev);
 
 	return 0;
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index f19bab68fd83..9f6f8669edc3 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -836,6 +836,26 @@ err1:
 	return r;
 }
 
+
+static void gfx_v8_0_free_microcode(struct amdgpu_device *adev) {
+	release_firmware(adev->gfx.pfp_fw);
+	adev->gfx.pfp_fw = NULL;
+	release_firmware(adev->gfx.me_fw);
+	adev->gfx.me_fw = NULL;
+	release_firmware(adev->gfx.ce_fw);
+	adev->gfx.ce_fw = NULL;
+	release_firmware(adev->gfx.rlc_fw);
+	adev->gfx.rlc_fw = NULL;
+	release_firmware(adev->gfx.mec_fw);
+	adev->gfx.mec_fw = NULL;
+	if ((adev->asic_type != CHIP_STONEY) &&
+	    (adev->asic_type != CHIP_TOPAZ))
+		release_firmware(adev->gfx.mec2_fw);
+	adev->gfx.mec2_fw = NULL;
+
+	kfree(adev->gfx.rlc.register_list_format);
+}
+
 static int gfx_v8_0_init_microcode(struct amdgpu_device *adev)
 {
 	const char *chip_name;
@@ -1983,7 +2003,7 @@ static int gfx_v8_0_sw_fini(void *handle)
 
 	gfx_v8_0_rlc_fini(adev);
 
-	kfree(adev->gfx.rlc.register_list_format);
+	gfx_v8_0_free_microcode(adev);
 
 	return 0;
 }
@@ -3974,11 +3994,15 @@ static int gfx_v8_0_cp_gfx_start(struct amdgpu_device *adev)
 		amdgpu_ring_write(ring, 0x3a00161a);
 		amdgpu_ring_write(ring, 0x0000002e);
 		break;
-	case CHIP_TOPAZ:
 	case CHIP_CARRIZO:
 		amdgpu_ring_write(ring, 0x00000002);
 		amdgpu_ring_write(ring, 0x00000000);
 		break;
+	case CHIP_TOPAZ:
+		amdgpu_ring_write(ring, adev->gfx.config.num_rbs == 1 ?
+				0x00000000 : 0x00000002);
+		amdgpu_ring_write(ring, 0x00000000);
+		break;
 	case CHIP_STONEY:
 		amdgpu_ring_write(ring, 0x00000000);
 		amdgpu_ring_write(ring, 0x00000000);
diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c b/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c
index 460bc8ad37e6..825ccd63f2dc 100644
--- a/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/iceland_dpm.c
@@ -72,6 +72,11 @@ static int iceland_dpm_sw_init(void *handle)
 
 static int iceland_dpm_sw_fini(void *handle)
 {
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	release_firmware(adev->pm.fw);
+	adev->pm.fw = NULL;
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
index f4c3130d3fdb..b556bd0a8797 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v2_4.c
@@ -105,6 +105,15 @@ static void sdma_v2_4_init_golden_registers(struct amdgpu_device *adev)
 	}
 }
 
+static void sdma_v2_4_free_microcode(struct amdgpu_device *adev)
+{
+	int i;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		release_firmware(adev->sdma.instance[i].fw);
+		adev->sdma.instance[i].fw = NULL;
+	}
+}
+
 /**
  * sdma_v2_4_init_microcode - load ucode images from disk
  *
@@ -461,6 +470,8 @@ static int sdma_v2_4_gfx_resume(struct amdgpu_device *adev)
 		/* Initialize the ring buffer's read and write pointers */
 		WREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i], 0);
 		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_OFFSET + sdma_offsets[i], 0);
 
 		/* set the wb address whether it's enabled or not */
 		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
@@ -489,7 +500,11 @@ static int sdma_v2_4_gfx_resume(struct amdgpu_device *adev)
 		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
 
 		ring->ready = true;
+	}
 
+	sdma_v2_4_enable(adev, true);
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
 		r = amdgpu_ring_test_ring(ring);
 		if (r) {
 			ring->ready = false;
@@ -580,8 +595,8 @@ static int sdma_v2_4_start(struct amdgpu_device *adev)
 			return -EINVAL;
 	}
 
-	/* unhalt the MEs */
-	sdma_v2_4_enable(adev, true);
+	/* halt the engine before programing */
+	sdma_v2_4_enable(adev, false);
 
 	/* start the gfx rings and rlc compute queues */
 	r = sdma_v2_4_gfx_resume(adev);
@@ -1012,6 +1027,7 @@ static int sdma_v2_4_sw_fini(void *handle)
 	for (i = 0; i < adev->sdma.num_instances; i++)
 		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
 
+	sdma_v2_4_free_microcode(adev);
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
index 31d99b0010f7..532ea88da66a 100644
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v3_0.c
@@ -236,6 +236,15 @@ static void sdma_v3_0_init_golden_registers(struct amdgpu_device *adev)
 	}
 }
 
+static void sdma_v3_0_free_microcode(struct amdgpu_device *adev)
+{
+	int i;
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		release_firmware(adev->sdma.instance[i].fw);
+		adev->sdma.instance[i].fw = NULL;
+	}
+}
+
 /**
  * sdma_v3_0_init_microcode - load ucode images from disk
  *
@@ -672,6 +681,8 @@ static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev)
 		/* Initialize the ring buffer's read and write pointers */
 		WREG32(mmSDMA0_GFX_RB_RPTR + sdma_offsets[i], 0);
 		WREG32(mmSDMA0_GFX_RB_WPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_RPTR + sdma_offsets[i], 0);
+		WREG32(mmSDMA0_GFX_IB_OFFSET + sdma_offsets[i], 0);
 
 		/* set the wb address whether it's enabled or not */
 		WREG32(mmSDMA0_GFX_RB_RPTR_ADDR_HI + sdma_offsets[i],
@@ -711,7 +722,15 @@ static int sdma_v3_0_gfx_resume(struct amdgpu_device *adev)
 		WREG32(mmSDMA0_GFX_IB_CNTL + sdma_offsets[i], ib_cntl);
 
 		ring->ready = true;
+	}
+
+	/* unhalt the MEs */
+	sdma_v3_0_enable(adev, true);
+	/* enable sdma ring preemption */
+	sdma_v3_0_ctx_switch_enable(adev, true);
 
+	for (i = 0; i < adev->sdma.num_instances; i++) {
+		ring = &adev->sdma.instance[i].ring;
 		r = amdgpu_ring_test_ring(ring);
 		if (r) {
 			ring->ready = false;
@@ -804,10 +823,9 @@ static int sdma_v3_0_start(struct amdgpu_device *adev)
 		}
 	}
 
-	/* unhalt the MEs */
-	sdma_v3_0_enable(adev, true);
-	/* enable sdma ring preemption */
-	sdma_v3_0_ctx_switch_enable(adev, true);
+	/* disble sdma engine before programing it */
+	sdma_v3_0_ctx_switch_enable(adev, false);
+	sdma_v3_0_enable(adev, false);
 
 	/* start the gfx rings and rlc compute queues */
 	r = sdma_v3_0_gfx_resume(adev);
@@ -1247,6 +1265,7 @@ static int sdma_v3_0_sw_fini(void *handle)
 	for (i = 0; i < adev->sdma.num_instances; i++)
 		amdgpu_ring_fini(&adev->sdma.instance[i].ring);
 
+	sdma_v3_0_free_microcode(adev);
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c b/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c
index b7615cefcac4..f06f6f4dc3a8 100644
--- a/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c
+++ b/drivers/gpu/drm/amd/amdgpu/tonga_dpm.c
@@ -71,6 +71,11 @@ static int tonga_dpm_sw_init(void *handle)
 
 static int tonga_dpm_sw_fini(void *handle)
 {
+	struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+
+	release_firmware(adev->pm.fw);
+	adev->pm.fw = NULL;
+
 	return 0;
 }
 
diff --git a/drivers/gpu/drm/amd/include/amd_shared.h b/drivers/gpu/drm/amd/include/amd_shared.h
index 6080951d539d..afce1edbe250 100644
--- a/drivers/gpu/drm/amd/include/amd_shared.h
+++ b/drivers/gpu/drm/amd/include/amd_shared.h
@@ -157,6 +157,7 @@ struct amd_ip_funcs {
 	int (*hw_init)(void *handle);
 	/* tears down the hw state */
 	int (*hw_fini)(void *handle);
+	void (*late_fini)(void *handle);
 	/* handles IP specific hw/sw changes for suspend */
 	int (*suspend)(void *handle);
 	/* handles IP specific hw/sw changes for resume */
diff --git a/drivers/gpu/drm/amd/include/cgs_common.h b/drivers/gpu/drm/amd/include/cgs_common.h
index a461e155a160..7464daf89ca1 100644
--- a/drivers/gpu/drm/amd/include/cgs_common.h
+++ b/drivers/gpu/drm/amd/include/cgs_common.h
@@ -581,6 +581,9 @@ typedef int (*cgs_get_firmware_info)(struct cgs_device *cgs_device,
 				     enum cgs_ucode_id type,
 				     struct cgs_firmware_info *info);
 
+typedef int (*cgs_rel_firmware)(struct cgs_device *cgs_device,
+					 enum cgs_ucode_id type);
+
 typedef int(*cgs_set_powergating_state)(struct cgs_device *cgs_device,
 				  enum amd_ip_block_type block_type,
 				  enum amd_powergating_state state);
@@ -645,6 +648,7 @@ struct cgs_ops {
 	cgs_set_camera_voltages_t set_camera_voltages;
 	/* Firmware Info */
 	cgs_get_firmware_info get_firmware_info;
+	cgs_rel_firmware rel_firmware;
 	/* cg pg interface*/
 	cgs_set_powergating_state set_powergating_state;
 	cgs_set_clockgating_state set_clockgating_state;
@@ -738,6 +742,8 @@ struct cgs_device
 	CGS_CALL(set_camera_voltages,dev,mask,voltages)
 #define cgs_get_firmware_info(dev, type, info)	\
 	CGS_CALL(get_firmware_info, dev, type, info)
+#define cgs_rel_firmware(dev, type)	\
+	CGS_CALL(rel_firmware, dev, type)
 #define cgs_set_powergating_state(dev, block_type, state)	\
 	CGS_CALL(set_powergating_state, dev, block_type, state)
 #define cgs_set_clockgating_state(dev, block_type, state)	\
diff --git a/drivers/gpu/drm/amd/powerplay/amd_powerplay.c b/drivers/gpu/drm/amd/powerplay/amd_powerplay.c
index 8e345bfddb69..e629f8a9fe93 100644
--- a/drivers/gpu/drm/amd/powerplay/amd_powerplay.c
+++ b/drivers/gpu/drm/amd/powerplay/amd_powerplay.c
@@ -73,11 +73,14 @@ static int pp_sw_init(void *handle)
 
 	ret = hwmgr->hwmgr_func->backend_init(hwmgr);
 	if (ret)
-		goto err;
+		goto err1;
 
 	pr_info("amdgpu: powerplay initialized\n");
 
 	return 0;
+err1:
+	if (hwmgr->pptable_func->pptable_fini)
+		hwmgr->pptable_func->pptable_fini(hwmgr);
 err:
 	pr_err("amdgpu: powerplay initialization failed\n");
 	return ret;
@@ -100,6 +103,9 @@ static int pp_sw_fini(void *handle)
 	if (hwmgr->hwmgr_func->backend_fini != NULL)
 		ret = hwmgr->hwmgr_func->backend_fini(hwmgr);