[PATCH] amdgpu: prevent NULL pointer dereference if ATIF is not supported

[Mario Limonciello]

On 10/30/2024 16:06, Antonio Quartulli wrote:
> Hi Mario,
> 
> On 30/10/2024 02:41, Mario Limonciello wrote:
>> On 10/29/2024 18:32, Antonio Quartulli wrote:
>>> acpi_evaluate_object() may return AE_NOT_FOUND (failure), which
>>> would result in dereferencing buffer.pointer (obj) while being NULL.
>>>
>>> Bail out also when status is AE_NOT_FOUND with a proper error message.
>>>
>>> This fixes 1 FORWARD_NULL issue reported by Coverity
>>> Report: CID 1600951:  Null pointer dereferences  (FORWARD_NULL)
>>>
>>> Signed-off-by: Antonio Quartulli <antonio at mandelbit.com>
>>
>> I'm not really sure how realistic this failure is.  Can you share the 
>> full call trace that Coverity identified?
> 
> I just checked Coverity Scan and it only says:
> 
>      5. Condition status, taking true branch.
>      6. Condition status != 5U /* (acpi_status)(5 | 0) */, taking false 
> branch.
> 
> The above points are related to:
> 
>      if (ACPI_FAILURE(status) && status != AE_NOT_FOUND)
> 
> It doesn't show how acpi_evaluate_object() is expected to return 
> AE_NOT_FOUND.
> 
> This said, if you think this case is unrealistic, why do you check for 
> "status != AE_NOT_FOUND" at all?
> 
> At this point maybe it would make more sense to simply drop this check 
> and always bail out with the current error message.
> 
> Basically a patch with the following only:
> 
> -       /* Fail if calling the method fails and ATIF is supported */
> -       if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
> +       /* Fail if calling the method fails */
> +       if (ACPI_FAILURE(status)) {
> 
> This way we don't make a fuzz for a possibly unrealistic case, while 
> still protecting against bugs and null-dereferences.

Yeah I think that's a good idea.  Can you respin it as a v2?

> 
> 
> Regards,
> 
>>
>> amdgpu_atif_pci_probe_handle() will check whether the handle is 
>> available in the first place.  We'll never this this far if that failed.
>>
>> Because of that I don't follow how this could return AE_NOT_FOUND.
>>> ---
>>>   drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c | 11 +++++++----
>>>   1 file changed, 7 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/ 
>>> drm/amd/amdgpu/amdgpu_acpi.c
>>> index cce85389427f..f10c3261a4ab 100644
>>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
>>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
>>> @@ -172,10 +172,13 @@ static union acpi_object 
>>> *amdgpu_atif_call(struct amdgpu_atif *atif,
>>>                         &buffer);
>>>       obj = (union acpi_object *)buffer.pointer;
>>> -    /* Fail if calling the method fails and ATIF is supported */
>>> -    if (ACPI_FAILURE(status) && status != AE_NOT_FOUND) {
>>> -        DRM_DEBUG_DRIVER("failed to evaluate ATIF got %s\n",
>>> -                 acpi_format_exception(status));
>>> +    /* Fail if calling the method fails */
>>> +    if (ACPI_FAILURE(status)) {
>>> +        if (status != AE_NOT_FOUND)
>>> +            DRM_DEBUG_DRIVER("failed to evaluate ATIF got %s\n",
>>> +                     acpi_format_exception(status));
>>> +        else
>>> +            DRM_DEBUG_DRIVER("ATIF not supported\n");
>>>           kfree(obj);
>>>           return NULL;
>>>       }
>>
>