Re: [PATCH] drm/amdgpu: correct single device PCIe reset flow for DPC status

["Lazar, Lijo" <lijo.lazar@amd.com>]

On 08-Apr-26 11:11 AM, Ce Sun wrote:
> For triggering the dpc event with a single device, we still need
> to set the in_link_reset flag and the dpc status.
> 
> Signed-off-by: Ce Sun <cesun102@amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 8 +++++++-
>   1 file changed, 7 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index deb41c095b59..0bdb54ab9a53 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -6291,6 +6291,9 @@ pci_ers_result_t amdgpu_pci_error_detected(struct pci_dev *pdev, pci_channel_sta
>   			amdgpu_reset_set_dpc_status(adev, true);
>   
>   			mutex_lock(&hive->hive_lock);
> +		} else {
> +			if (amdgpu_device_bus_status_check(adev))
> +				amdgpu_reset_set_dpc_status(adev, true);
>   		}
>   		memset(&reset_context, 0, sizeof(reset_context));
>   		INIT_LIST_HEAD(&device_list);
> @@ -6411,6 +6414,7 @@ pci_ers_result_t amdgpu_pci_slot_reset(struct pci_dev *pdev)
>   		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head)
>   			tmp_adev->pcie_reset_ctx.in_link_reset = true;
>   	} else {
> +		adev->pcie_reset_ctx.in_link_reset = true;

For the device which gets into DPC, this status is set in detect stage 
itself within amdgpu_device_recovery_prepare() -

https://github.com/torvalds/linux/blob/master/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c#L6245

You may keep the same for single device case also.

Thanks,
Lijo

>   		set_bit(AMDGPU_SKIP_HW_RESET, &reset_context.flags);
>   	}
>   
> @@ -6467,8 +6471,10 @@ void amdgpu_pci_resume(struct pci_dev *pdev)
>   			tmp_adev->pcie_reset_ctx.in_link_reset = false;
>   			list_add_tail(&tmp_adev->reset_list, &device_list);
>   		}
> -	} else
> +	} else {
> +		adev->pcie_reset_ctx.in_link_reset = false;
>   		list_add_tail(&adev->reset_list, &device_list);
> +	}
>   
>   	amdgpu_device_sched_resume(&device_list, NULL, NULL);
>   	amdgpu_device_gpu_resume(adev, &device_list, false);