[PATCH V3 1/3] drm/amdgpu: Adjust removal control flow for smu v13_0_2

YiPeng Chai YiPeng.Chai at amd.com
Thu Sep 8 08:42:34 UTC 2022


Adjust removal control flow for smu v13_0_2:
   During amdgpu uninstallation, when removing the first
device, the kernel needs to first send a mode1reset message
to all gpu devices. Otherwise, smu initialization will fail
the next time amdgpu is installed.

V2:
1. Update commit comments.
2. Remove the global variable amdgpu_device_remove_cnt
   and add a variable to the structure amdgpu_hive_info.
3. Use hive to detect the first removed device instead of
   a global variable.

V3:
 1. Update commit comments.
 2. Split a patch into multiple patches.
 3. The current patch does:
    a. Add a work mode of AMDGPU_RESET_FOR_DEVICE_REMOVE into
       the existing gpu recover path, which make all devices
       in hive list only have HW reset but no resume (except
       the base IP).
    b. Call AMDGPU_RESET_FOR_DEVICE_REMOVE and
       AMDGPU_NEED_FULL_RESET mode of amdgpu_device_gpu_recover
       in amdgpu_pci_remove when removing the first device in
       hive list.
    c. When removing the first device, the IP blocks keyword
       function call sequence is as follows:
.suspend->mode1reset->.resume(basic ip)->.hw_fini->.early_fini->.sw_fini.
   ^                           |
   |-<----------<---------<----|
	The first three sequences are because of a call to
        amdgpu_device_gpu_recover. The three sequences will be
        executed in a loop until all devices in the hive list
        are iterated.
        The sequences starting from .hw_fini only apply to the
        first device. Since .suspend has been called before,
        except the resumed phase1 basic ip blocks, all other ip
        blocks .hw_fini of current device will do nothing.
     d. When removing other devices, the calling sequences is the
        same as legacy:
	   .hw_fini -> .early_fini -> .sw_fini.
	Since .suspend has been called when removing the first device,
        except the resumed phase1 basic ip blocks, all of other ip
        blocks .hw_fini of current device will do nothing.

Signed-off-by: YiPeng Chai <YiPeng.Chai at amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 32 ++++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c    | 29 ++++++++++++++++++++
 drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h  |  1 +
 drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h   |  1 +
 4 files changed, 63 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 62b26f0e37b0..a2614b881128 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -2999,6 +2999,7 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev)
 			DRM_ERROR("suspend of IP block <%s> failed %d\n",
 				  adev->ip_blocks[i].version->funcs->name, r);
 		}
+
 		adev->ip_blocks[i].status.hw = false;
 		/* handle putting the SMC in the appropriate state */
 		if(!amdgpu_sriov_vf(adev)){
@@ -4739,6 +4740,7 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle,
 	struct amdgpu_device *tmp_adev = NULL;
 	bool need_full_reset, skip_hw_reset, vram_lost = false;
 	int r = 0;
+	bool gpu_reset_for_dev_remove = 0;
 
 	/* Try reset handler method first */
 	tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device,
@@ -4758,6 +4760,10 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle,
 		test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
 	skip_hw_reset = test_bit(AMDGPU_SKIP_HW_RESET, &reset_context->flags);
 
+	gpu_reset_for_dev_remove =
+		test_bit(AMDGPU_RESET_FOR_DEVICE_REMOVE, &reset_context->flags) &&
+			test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
+
 	/*
 	 * ASIC reset has to be done on all XGMI hive nodes ASAP
 	 * to allow proper links negotiation in FW (within 1 sec)
@@ -4802,6 +4808,19 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle,
 		amdgpu_ras_intr_cleared();
 	}
 
+
+	/* Since the mode1 reset affects base ip blocks, the
+	 * phase1 ip blocks need to be resumed. Otherwise there
+	 * will be a BIOS signature error and the psp bootloader
+	 * can't load kdb on the next amdgpu install.
+	 */
+	if (gpu_reset_for_dev_remove) {
+		list_for_each_entry(tmp_adev, device_list_handle, reset_list)
+			amdgpu_device_ip_resume_phase1(tmp_adev);
+
+		goto end;
+	}
+
 	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
 		if (need_full_reset) {
 			/* post card */
@@ -5124,6 +5143,11 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 	bool need_emergency_restart = false;
 	bool audio_suspended = false;
 	int tmp_vram_lost_counter;
+	bool gpu_reset_for_dev_remove = false;
+
+	gpu_reset_for_dev_remove =
+			test_bit(AMDGPU_RESET_FOR_DEVICE_REMOVE, &reset_context->flags) &&
+				test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
 
 	/*
 	 * Special case: RAS triggered and full reset isn't supported
@@ -5243,6 +5267,10 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 
 retry:	/* Rest of adevs pre asic reset from XGMI hive. */
 	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+		if (gpu_reset_for_dev_remove) {
+			/* Workaroud for ASICs need to disable SMC first */
+			amdgpu_device_smu_fini_early(tmp_adev);
+		}
 		r = amdgpu_device_pre_asic_reset(tmp_adev, reset_context);
 		/*TODO Should we stop ?*/
 		if (r) {
@@ -5276,6 +5304,9 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 			adev->asic_reset_res = 0;
 			goto retry;
 		}
+
+		if (!r && gpu_reset_for_dev_remove)
+			goto recover_end;
 	}
 
 skip_hw_reset:
@@ -5349,6 +5380,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 		amdgpu_device_unset_mp1_state(tmp_adev);
 	}
 
+recover_end:
 	tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device,
 					    reset_list);
 	amdgpu_device_unlock_reset_domain(tmp_adev->reset_domain);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 728a0933ea6f..8933f8f63d13 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -2186,6 +2186,35 @@ amdgpu_pci_remove(struct pci_dev *pdev)
 		pm_runtime_forbid(dev->dev);
 	}
 
+	if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2)) {
+		bool need_to_reset_gpu = false;
+
+		if (adev->gmc.xgmi.num_physical_nodes > 1) {
+			struct amdgpu_hive_info *hive;
+
+			hive = amdgpu_get_xgmi_hive(adev);
+			if (hive->device_remove_count == 0)
+				need_to_reset_gpu = true;
+			hive->device_remove_count++;
+			amdgpu_put_xgmi_hive(hive);
+		} else
+			need_to_reset_gpu = true;
+
+		/* Workaround for ASICs need to reset SMU.
+		 * Called only when the first device is removed.
+		 */
+		if (need_to_reset_gpu) {
+			struct amdgpu_reset_context reset_context;
+
+			memset(&reset_context, 0, sizeof(reset_context));
+			reset_context.method = AMD_RESET_METHOD_NONE;
+			reset_context.reset_req_dev = adev;
+			set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+			set_bit(AMDGPU_RESET_FOR_DEVICE_REMOVE, &reset_context.flags);
+			amdgpu_device_gpu_recover(adev, NULL, &reset_context);
+		}
+	}
+
 	amdgpu_driver_unload_kms(dev);
 
 	drm_dev_unplug(dev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
index f71b83c42590..dc43fcb93eac 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
@@ -31,6 +31,7 @@ enum AMDGPU_RESET_FLAGS {
 	AMDGPU_NEED_FULL_RESET = 0,
 	AMDGPU_SKIP_HW_RESET = 1,
 	AMDGPU_SKIP_MODE2_RESET = 2,
+	AMDGPU_RESET_FOR_DEVICE_REMOVE = 3,
 };
 
 struct amdgpu_reset_context {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
index 552e6fb55aa8..30dcc1681b4e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
@@ -43,6 +43,7 @@ struct amdgpu_hive_info {
 	} pstate;
 
 	struct amdgpu_reset_domain *reset_domain;
+	uint32_t device_remove_count;
 };
 
 struct amdgpu_pcs_ras_field {
-- 
2.25.1



More information about the amd-gfx mailing list