Re: [PATCH] drm/amdgpu: Gate VFCT bus fallback on pci=realloc

[Mario Limonciello <mario.limonciello@amd.com>]

On 10/15/26 07:56, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
> - [Low] pci: missing EXPORT_SYMBOL for pci_realloc_enabled_by_user() breaks amdgpu
> --
> 
> commit 9eeda1e99dc9a34bbbe5670a252bc894b1208696
> Author: Mario Limonciello <mario.limonciello@amd.com>
> 
> drm/amdgpu: Gate VFCT bus fallback on pci=realloc
> 
> This patch modifies the soft filter system to gate the VFCT bus fallback
> so it only runs when pci=realloc is explicitly requested by the user. It
> introduces pci_realloc_enabled_by_user() in the PCI core and uses it in
> the AMDGPU driver.
> 
> Closes: https://lore.kernel.org/amd-gfx/20261014080523.186533-1-rezanikzadch@gmail.com/
> 
>> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
>> index f9315e0472fe9..9651260278c18 100644
>> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
>> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c
> [ ... ]
>> @@ -459,7 +457,10 @@ static int amdgpu_acpi_vfct_match(struct amdgpu_device *adev,
>>   	if (vhdr->PCIBus == adev->pdev->bus->number)
>>   		return 0;
>>   
>> -	/* Bus mismatch but device identity matches (PCI renumbering case) */
>> +	/* Allow bus renumbering only when the user requested pci=realloc. */
>> +	if (!pci_realloc_enabled_by_user())
>> +		return -ENODEV;
>> +
> 
> [ ... ]
> 
>> diff --git a/drivers/pci/setup-bus.c b/drivers/pci/setup-bus.c
>> index ed16ef7c26fa7..d9450585082c0 100644
>> --- a/drivers/pci/setup-bus.c
>> +++ b/drivers/pci/setup-bus.c
>> @@ -1799,6 +1799,11 @@ static bool pci_realloc_enabled(enum enable_type enable)
>>   	return enable >= user_enabled;
>>   }
>>   
>> +bool pci_realloc_enabled_by_user(void)
>> +{
>> +	return pci_realloc_enable == user_enabled;
>> +}
>> +
> 
> [Severity: Low]
> Will this cause a module build failure (undefined reference) if the amdgpu
> driver is compiled as a loadable module (CONFIG_DRM_AMDGPU=m)?
> 
> Since setup-bus.c is built into the core kernel and pci_realloc_enabled_by_user()
> lacks an EXPORT_SYMBOL or EXPORT_SYMBOL_GPL macro, the linker won't be able
> to resolve this symbol during modpost for amdgpu.ko.
> 

Yeah; this is a valid issue.  If we're aligned directionally on this 
I'll fix that in a v2.