[PATCH 2/3] drm/amdgpu: Move xgmi ras initialization from .late_init to .early_init

Chai, Thomas YiPeng.Chai at amd.com
Wed Jan 19 07:22:13 UTC 2022



-----Original Message-----
From: Zhou1, Tao <Tao.Zhou1 at amd.com> 
Sent: Wednesday, January 19, 2022 2:11 PM
To: Chai, Thomas <YiPeng.Chai at amd.com>; amd-gfx at lists.freedesktop.org
Cc: Zhang, Hawking <Hawking.Zhang at amd.com>; Clements, John <John.Clements at amd.com>
Subject: RE: [PATCH 2/3] drm/amdgpu: Move xgmi ras initialization from .late_init to .early_init

[AMD Official Use Only]



> -----Original Message-----
> From: Chai, Thomas <YiPeng.Chai at amd.com>
> Sent: Wednesday, January 19, 2022 10:56 AM
> To: amd-gfx at lists.freedesktop.org
> Cc: Chai, Thomas <YiPeng.Chai at amd.com>; Zhang, Hawking 
> <Hawking.Zhang at amd.com>; Zhou1, Tao <Tao.Zhou1 at amd.com>; Clements, 
> John <John.Clements at amd.com>; Chai, Thomas <YiPeng.Chai at amd.com>
> Subject: [PATCH 2/3] drm/amdgpu: Move xgmi ras initialization from 
> .late_init to .early_init
> 
> Move xgmi ras initialization from .late_init to .early_init, which let 
> xgmi ras be initialized only once.
> 
> Signed-off-by: yipechai <YiPeng.Chai at amd.com>
> ---
>  drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c | 5 ----- 
> drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c  | 1 -
>  drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c   | 9 +++++++++
>  3 files changed, 9 insertions(+), 6 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
> index 3483a82f5734..d83eee1984c8 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c
> @@ -452,11 +452,6 @@ int amdgpu_gmc_ras_late_init(struct amdgpu_device
> *adev)
>  			return r;
>  	}
> 
> -	if (!adev->gmc.xgmi.connected_to_cpu) {
> -		adev->gmc.xgmi.ras = &xgmi_ras;
> -		amdgpu_ras_register_ras_block(adev, &adev->gmc.xgmi.ras-
> >ras_block);
> -	}
> -
>  	if (adev->gmc.xgmi.ras && 
> adev->gmc.xgmi.ras->ras_block.ras_late_init)
> {
>  		r = adev->gmc.xgmi.ras->ras_block.ras_late_init(adev, NULL);
>  		if (r)
> diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
> b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
> index 4f8d356f8432..5f9f82091000 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
> @@ -716,7 +716,6 @@ static void gmc_v10_0_set_gfxhub_funcs(struct 
> amdgpu_device *adev)
>  	}
>  }
> 
> -
>[Tao]: Please don't introduce irrelevant change.

[Thomas] OK

>  static int gmc_v10_0_early_init(void *handle)  {
>  	struct amdgpu_device *adev = (struct amdgpu_device *)handle; diff 
> --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> index c76ffd1a70cd..8d1b11368a7b 100644
> --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
> @@ -1303,6 +1303,14 @@ static void gmc_v9_0_set_hdp_ras_funcs(struct 
> amdgpu_device *adev)
>  	amdgpu_ras_register_ras_block(adev, &adev->hdp.ras->ras_block);  }
> 
> +static void gmc_v9_0_set_xgmi_ras_funcs(struct amdgpu_device *adev) {
> +	if (!adev->gmc.xgmi.connected_to_cpu) {
> +		adev->gmc.xgmi.ras = &xgmi_ras;
> +		amdgpu_ras_register_ras_block(adev, &adev->gmc.xgmi.ras-
> >ras_block);
> +	}
> +}
>[Tao]: Since the initialization of xgmi.ras is common for all versions of IP, I recommend to create a generic ras_early_init func for it.
>BTW, I think we can remove the check for block_obj's existence in register_ras_block now.
 
[Thomas] I will update the patch.

> +
>  static void gmc_v9_0_set_mca_funcs(struct amdgpu_device *adev)  {
>  	/* is UMC the right IP to check for MCA?  Maybe DF? */ @@ -1339,6
> +1347,7 @@ static int gmc_v9_0_early_init(void *handle)
>  	gmc_v9_0_set_gfxhub_funcs(adev);
>  	gmc_v9_0_set_hdp_ras_funcs(adev);
>  	gmc_v9_0_set_mca_funcs(adev);
> +	gmc_v9_0_set_xgmi_ras_funcs(adev);
> 
>  	adev->gmc.shared_aperture_start = 0x2000000000000000ULL;
>  	adev->gmc.shared_aperture_end =
> --
> 2.25.1


More information about the amd-gfx mailing list