[PATCH] drm/amdgpu: defer cmd/fence/fw buffers destroy on hw_init failure

[Quan, Evan]

Ping..

> -----Original Message-----
> From: Evan Quan <evan.quan at amd.com>
> Sent: 2019年3月18日 10:42
> To: amd-gfx at lists.freedesktop.org
> Cc: Quan, Evan <Evan.Quan at amd.com>
> Subject: [PATCH] drm/amdgpu: defer cmd/fence/fw buffers destroy on
> hw_init failure
> 
> As the cleanup jobs performed in pre_fini may still need these buffers. NULL
> pointer dereference will be triggered without them.
> 
> Change-Id: I398efd652584e022debf237950207199a4ea78fc
> Signed-off-by: Evan Quan <evan.quan at amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c | 31 ++++++++++--------------
> -
>  1 file changed, 12 insertions(+), 19 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
> index 5888e24219d9..2206bb4b0903 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
> @@ -890,59 +890,52 @@ static int psp_load_fw(struct amdgpu_device
> *adev)
>  					&psp->fence_buf_mc_addr,
>  					&psp->fence_buf);
>  	if (ret)
> -		goto failed_mem2;
> +		goto failed;
> 
>  	ret = amdgpu_bo_create_kernel(adev, PSP_CMD_BUFFER_SIZE,
> PAGE_SIZE,
>  				      AMDGPU_GEM_DOMAIN_VRAM,
>  				      &psp->cmd_buf_bo, &psp-
> >cmd_buf_mc_addr,
>  				      (void **)&psp->cmd_buf_mem);
>  	if (ret)
> -		goto failed_mem1;
> +		goto failed;
> 
>  	memset(psp->fence_buf, 0, PSP_FENCE_BUFFER_SIZE);
> 
>  	ret = psp_ring_init(psp, PSP_RING_TYPE__KM);
>  	if (ret) {
>  		DRM_ERROR("PSP ring init failed!\n");
> -		goto failed_mem;
> +		goto failed;
>  	}
> 
>  	ret = psp_tmr_init(psp);
>  	if (ret) {
>  		DRM_ERROR("PSP tmr init failed!\n");
> -		goto failed_mem;
> +		goto failed;
>  	}
> 
>  	ret = psp_asd_init(psp);
>  	if (ret) {
>  		DRM_ERROR("PSP asd init failed!\n");
> -		goto failed_mem;
> +		goto failed;
>  	}
> 
>  skip_memalloc:
>  	ret = psp_hw_start(psp);
>  	if (ret)
> -		goto failed_mem;
> +		goto failed;
> 
>  	ret = psp_np_fw_load(psp);
>  	if (ret)
> -		goto failed_mem;
> +		goto failed;
> 
>  	return 0;
> 
> -failed_mem:
> -	amdgpu_bo_free_kernel(&psp->cmd_buf_bo,
> -			      &psp->cmd_buf_mc_addr,
> -			      (void **)&psp->cmd_buf_mem);
> -failed_mem1:
> -	amdgpu_bo_free_kernel(&psp->fence_buf_bo,
> -			      &psp->fence_buf_mc_addr, &psp->fence_buf);
> -failed_mem2:
> -	amdgpu_bo_free_kernel(&psp->fw_pri_bo,
> -			      &psp->fw_pri_mc_addr, &psp->fw_pri_buf);
>  failed:
> -	kfree(psp->cmd);
> -	psp->cmd = NULL;
> +	/*
> +	 * all cleanup jobs (xgmi terminate, ras terminate,
> +	 * ring destroy, cmd/fence/fw buffers destory,
> +	 * psp->cmd destory) are delayed to psp_hw_fini
> +	 */
>  	return ret;
>  }
> 
> --
> 2.21.0