[PATCH 01/22] drm/amdgpu: fix dereference after null check

[Zhang, Jesse(Jie)]

[AMD Official Use Only - AMD Internal Distribution Only]

Hi Christian,

-----Original Message-----
From: Christian König <ckoenig.leichtzumerken at gmail.com>
Sent: Monday, May 13, 2024 7:41 PM
To: Zhang, Jesse(Jie) <Jesse.Zhang at amd.com>; amd-gfx at lists.freedesktop.org
Cc: Deucher, Alexander <Alexander.Deucher at amd.com>; Koenig, Christian <Christian.Koenig at amd.com>; Huang, Tim <Tim.Huang at amd.com>
Subject: Re: [PATCH 01/22] drm/amdgpu: fix dereference after null check

Am 10.05.24 um 04:50 schrieb Jesse Zhang:
> check the pointer hive before use.
>
> Signed-off-by: Jesse Zhang <Jesse.Zhang at amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> index 77f6fd50002a..00fe3c2d5431 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
> @@ -5725,7 +5725,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
>        * to put adev in the 1st position.
>        */
>       INIT_LIST_HEAD(&device_list);
> -     if (!amdgpu_sriov_vf(adev) && (adev->gmc.xgmi.num_physical_nodes > 1)) {
> +     if (!amdgpu_sriov_vf(adev) && (adev->gmc.xgmi.num_physical_nodes >
> +1) && hive) {

That solution looks not optimal to me.

Checking adev->gmc.xgmi.num_physical_nodes > 1 already makes sure that hive shouldn't be NULL.

If automated checkers complain about that we should probably drop the
adev->gmc.xgmi.num_physical_nodes > 1 check and check for hive instead.

[Zhang, Jesse(Jie)]
gmc.xgmi.num_physical_nodes is obtained by reading register GCMC_VM_XGMI_LFB_CNTL.
But getting hive may fail because of  no memory (NOMEM),
or the kobject of xgmi hive cannot be initialized in the function amdgpu_get_xgmi_hive.
Is (adev->gmc.xgmi.num_physical_nodes > 1) equivalent to (!hive) here?

Regards
Jesse

Regards,
Christian.

>               list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
>                       list_add_tail(&tmp_adev->reset_list, &device_list);
>                       if (adev->shutdown)