[PATCH] PCI/VGA: Don't assume the only VGA device on a system is `boot_vga`

[PATCH] PCI/VGA: Don't assume the only VGA device on a system is `boot_vga`

[Mario Limonciello (AMD)]

Some systems ship with multiple display class devices but not all
of them are VGA devices. If the "only" VGA device on the system is not
used for displaying the image on the screen marking it as `boot_vga`
because nothing was found is totally wrong.

This behavior actually leads to mistakes of the wrong device being
advertised to userspace and then userspace can make incorrect decisions.

As there is an accurate `boot_display` sysfs file stop lying about
`boot_vga` by assuming if nothing is found it's the right device.

Reported-by: Aaron Erhardt <aer@tuxedocomputers.com>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=220712
Tested-by: Aaron Erhardt <aer@tuxedocomputers.com>
Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
---
 drivers/pci/vgaarb.c | 7 -------
 1 file changed, 7 deletions(-)

diff --git a/drivers/pci/vgaarb.c b/drivers/pci/vgaarb.c
index 436fa7f4c3873..baa242b140993 100644
--- a/drivers/pci/vgaarb.c
+++ b/drivers/pci/vgaarb.c
@@ -652,13 +652,6 @@ static bool vga_is_boot_device(struct vga_device *vgadev)
 		return true;
 	}
 
-	/*
-	 * Vgadev has neither IO nor MEM enabled.  If we haven't found any
-	 * other VGA devices, it is the best candidate so far.
-	 */
-	if (!boot_vga)
-		return true;
-
 	return false;
 }
 
-- 
2.43.0

Re: [PATCH] PCI/VGA: Don't assume the only VGA device on a system is `boot_vga`

[Bjorn Helgaas]

On Wed, Oct 29, 2025 at 01:59:33PM -0500, Mario Limonciello (AMD) wrote:
> Some systems ship with multiple display class devices but not all
> of them are VGA devices. If the "only" VGA device on the system is not
> used for displaying the image on the screen marking it as `boot_vga`
> because nothing was found is totally wrong.
> 
> This behavior actually leads to mistakes of the wrong device being
> advertised to userspace and then userspace can make incorrect decisions.
> 
> As there is an accurate `boot_display` sysfs file stop lying about
> `boot_vga` by assuming if nothing is found it's the right device.
> 
> Reported-by: Aaron Erhardt <aer@tuxedocomputers.com>
> Closes: https://bugzilla.kernel.org/show_bug.cgi?id=220712
> Tested-by: Aaron Erhardt <aer@tuxedocomputers.com>
> Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>

Do we need a Fixes: here?  A stable cc?

The bugzilla suggests this might be a regression and hence v6.18
material?

> ---
>  drivers/pci/vgaarb.c | 7 -------
>  1 file changed, 7 deletions(-)
> 
> diff --git a/drivers/pci/vgaarb.c b/drivers/pci/vgaarb.c
> index 436fa7f4c3873..baa242b140993 100644
> --- a/drivers/pci/vgaarb.c
> +++ b/drivers/pci/vgaarb.c
> @@ -652,13 +652,6 @@ static bool vga_is_boot_device(struct vga_device *vgadev)
>  		return true;
>  	}
>  
> -	/*
> -	 * Vgadev has neither IO nor MEM enabled.  If we haven't found any
> -	 * other VGA devices, it is the best candidate so far.
> -	 */
> -	if (!boot_vga)
> -		return true;
> -
>  	return false;
>  }
>  
> -- 
> 2.43.0
> 

Re: [PATCH] PCI/VGA: Don't assume the only VGA device on a system is `boot_vga`

[Mario Limonciello]

+Thomas

On 10/29/25 2:34 PM, Bjorn Helgaas wrote:
> On Wed, Oct 29, 2025 at 01:59:33PM -0500, Mario Limonciello (AMD) wrote:
>> Some systems ship with multiple display class devices but not all
>> of them are VGA devices. If the "only" VGA device on the system is not
>> used for displaying the image on the screen marking it as `boot_vga`
>> because nothing was found is totally wrong.
>>
>> This behavior actually leads to mistakes of the wrong device being
>> advertised to userspace and then userspace can make incorrect decisions.
>>
>> As there is an accurate `boot_display` sysfs file stop lying about
>> `boot_vga` by assuming if nothing is found it's the right device.
>>
>> Reported-by: Aaron Erhardt <aer@tuxedocomputers.com>
>> Closes: https://bugzilla.kernel.org/show_bug.cgi?id=220712
>> Tested-by: Aaron Erhardt <aer@tuxedocomputers.com>
>> Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
> 
> Do we need a Fixes: here?  A stable cc?
> 
> The bugzilla suggests this might be a regression and hence v6.18
> material?

Yeah I think you're right, we should add these two tags and this should 
go to 6.18 if it's a reasonable change.

Cc: stable@vger.kernel.org
Fixes: ad90860bd10ee ("fbcon: Use screen info to find primary device")

But let's also make sure Thomas Zimmermann agrees with this change.

> 
>> ---
>>   drivers/pci/vgaarb.c | 7 -------
>>   1 file changed, 7 deletions(-)
>>
>> diff --git a/drivers/pci/vgaarb.c b/drivers/pci/vgaarb.c
>> index 436fa7f4c3873..baa242b140993 100644
>> --- a/drivers/pci/vgaarb.c
>> +++ b/drivers/pci/vgaarb.c
>> @@ -652,13 +652,6 @@ static bool vga_is_boot_device(struct vga_device *vgadev)
>>   		return true;
>>   	}
>>   
>> -	/*
>> -	 * Vgadev has neither IO nor MEM enabled.  If we haven't found any
>> -	 * other VGA devices, it is the best candidate so far.
>> -	 */
>> -	if (!boot_vga)
>> -		return true;
>> -
>>   	return false;
>>   }
>>   
>> -- 
>> 2.43.0
>>

Re: [PATCH] PCI/VGA: Don't assume the only VGA device on a system is `boot_vga`

[Mario Limonciello]

On 10/29/25 2:38 PM, Mario Limonciello wrote:
> +Thomas
> 
> On 10/29/25 2:34 PM, Bjorn Helgaas wrote:
>> On Wed, Oct 29, 2025 at 01:59:33PM -0500, Mario Limonciello (AMD) wrote:
>>> Some systems ship with multiple display class devices but not all
>>> of them are VGA devices. If the "only" VGA device on the system is not
>>> used for displaying the image on the screen marking it as `boot_vga`
>>> because nothing was found is totally wrong.
>>>
>>> This behavior actually leads to mistakes of the wrong device being
>>> advertised to userspace and then userspace can make incorrect decisions.
>>>
>>> As there is an accurate `boot_display` sysfs file stop lying about
>>> `boot_vga` by assuming if nothing is found it's the right device.
>>>
>>> Reported-by: Aaron Erhardt <aer@tuxedocomputers.com>
>>> Closes: https://bugzilla.kernel.org/show_bug.cgi?id=220712
>>> Tested-by: Aaron Erhardt <aer@tuxedocomputers.com>
>>> Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
>>
>> Do we need a Fixes: here?  A stable cc?
>>
>> The bugzilla suggests this might be a regression and hence v6.18
>> material?
> 
> Yeah I think you're right, we should add these two tags and this should 
> go to 6.18 if it's a reasonable change.
> 
> Cc: stable@vger.kernel.org
> Fixes: ad90860bd10ee ("fbcon: Use screen info to find primary device")
> 
> But let's also make sure Thomas Zimmermann agrees with this change.
> 

ping?

>>
>>> ---
>>>   drivers/pci/vgaarb.c | 7 -------
>>>   1 file changed, 7 deletions(-)
>>>
>>> diff --git a/drivers/pci/vgaarb.c b/drivers/pci/vgaarb.c
>>> index 436fa7f4c3873..baa242b140993 100644
>>> --- a/drivers/pci/vgaarb.c
>>> +++ b/drivers/pci/vgaarb.c
>>> @@ -652,13 +652,6 @@ static bool vga_is_boot_device(struct vga_device 
>>> *vgadev)
>>>           return true;
>>>       }
>>> -    /*
>>> -     * Vgadev has neither IO nor MEM enabled.  If we haven't found any
>>> -     * other VGA devices, it is the best candidate so far.
>>> -     */
>>> -    if (!boot_vga)
>>> -        return true;
>>> -
>>>       return false;
>>>   }
>>> -- 
>>> 2.43.0
>>>
> 
> 

Re: [PATCH] PCI/VGA: Don't assume the only VGA device on a system is `boot_vga`

[Thomas Zimmermann]

Hi

Am 17.11.25 um 06:20 schrieb Mario Limonciello:
>
>
> On 10/29/25 2:38 PM, Mario Limonciello wrote:
>> +Thomas
>>
>> On 10/29/25 2:34 PM, Bjorn Helgaas wrote:
>>> On Wed, Oct 29, 2025 at 01:59:33PM -0500, Mario Limonciello (AMD) 
>>> wrote:
>>>> Some systems ship with multiple display class devices but not all
>>>> of them are VGA devices. If the "only" VGA device on the system is not
>>>> used for displaying the image on the screen marking it as `boot_vga`
>>>> because nothing was found is totally wrong.
>>>>
>>>> This behavior actually leads to mistakes of the wrong device being
>>>> advertised to userspace and then userspace can make incorrect 
>>>> decisions.
>>>>
>>>> As there is an accurate `boot_display` sysfs file stop lying about
>>>> `boot_vga` by assuming if nothing is found it's the right device.
>>>>
>>>> Reported-by: Aaron Erhardt <aer@tuxedocomputers.com>
>>>> Closes: https://bugzilla.kernel.org/show_bug.cgi?id=220712
>>>> Tested-by: Aaron Erhardt <aer@tuxedocomputers.com>
>>>> Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
>>>
>>> Do we need a Fixes: here?  A stable cc?
>>>
>>> The bugzilla suggests this might be a regression and hence v6.18
>>> material?
>>
>> Yeah I think you're right, we should add these two tags and this 
>> should go to 6.18 if it's a reasonable change.
>>
>> Cc: stable@vger.kernel.org
>> Fixes: ad90860bd10ee ("fbcon: Use screen info to find primary device")
>>
>> But let's also make sure Thomas Zimmermann agrees with this change.
>>
>
> ping?

Sorry, this fell though the cracks. I can see that this backfires, but still

Acked-by: Thomas Zimmermann <tzimmermann@suse.de>

because the overall logic of detecting the boot-up VGA is questionable. 
It picks the first possible candidate instead of looking at all devices 
before making the decision. If we see regressions from the patch here, 
this would be the place to fix it.

Best regards
Thomas

>
>>>
>>>> ---
>>>>   drivers/pci/vgaarb.c | 7 -------
>>>>   1 file changed, 7 deletions(-)
>>>>
>>>> diff --git a/drivers/pci/vgaarb.c b/drivers/pci/vgaarb.c
>>>> index 436fa7f4c3873..baa242b140993 100644
>>>> --- a/drivers/pci/vgaarb.c
>>>> +++ b/drivers/pci/vgaarb.c
>>>> @@ -652,13 +652,6 @@ static bool vga_is_boot_device(struct 
>>>> vga_device *vgadev)
>>>>           return true;
>>>>       }
>>>> -    /*
>>>> -     * Vgadev has neither IO nor MEM enabled.  If we haven't found 
>>>> any
>>>> -     * other VGA devices, it is the best candidate so far.
>>>> -     */
>>>> -    if (!boot_vga)
>>>> -        return true;
>>>> -
>>>>       return false;
>>>>   }
>>>> -- 
>>>> 2.43.0
>>>>
>>
>>
>

-- 
--
Thomas Zimmermann
Graphics Driver Developer
SUSE Software Solutions Germany GmbH
Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)

Re: [PATCH] PCI/VGA: Don't assume the only VGA device on a system is `boot_vga`

[Bjorn Helgaas]

On Mon, Nov 17, 2025 at 08:25:05AM +0100, Thomas Zimmermann wrote:
> Am 17.11.25 um 06:20 schrieb Mario Limonciello:
> > On 10/29/25 2:38 PM, Mario Limonciello wrote:
> > > On 10/29/25 2:34 PM, Bjorn Helgaas wrote:
> > > > On Wed, Oct 29, 2025 at 01:59:33PM -0500, Mario Limonciello
> > > > (AMD) wrote:
> > > > > Some systems ship with multiple display class devices but not all
> > > > > of them are VGA devices. If the "only" VGA device on the system is not
> > > > > used for displaying the image on the screen marking it as `boot_vga`
> > > > > because nothing was found is totally wrong.
> > > > > 
> > > > > This behavior actually leads to mistakes of the wrong device being
> > > > > advertised to userspace and then userspace can make
> > > > > incorrect decisions.
> > > > > 
> > > > > As there is an accurate `boot_display` sysfs file stop lying about
> > > > > `boot_vga` by assuming if nothing is found it's the right device.
> > > > > 
> > > > > Reported-by: Aaron Erhardt <aer@tuxedocomputers.com>
> > > > > Closes: https://bugzilla.kernel.org/show_bug.cgi?id=220712
> > > > > Tested-by: Aaron Erhardt <aer@tuxedocomputers.com>
> > > > > Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
> > > > 
> > > > Do we need a Fixes: here?  A stable cc?
> > > > 
> > > > The bugzilla suggests this might be a regression and hence v6.18
> > > > material?
> > > 
> > > Yeah I think you're right, we should add these two tags and this
> > > should go to 6.18 if it's a reasonable change.
> > > 
> > > Cc: stable@vger.kernel.org
> > > Fixes: ad90860bd10ee ("fbcon: Use screen info to find primary device")
> > > 
> > > But let's also make sure Thomas Zimmermann agrees with this change.
> > 
> > ping?
> 
> Sorry, this fell though the cracks. I can see that this backfires, but still
> 
> Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
> 
> because the overall logic of detecting the boot-up VGA is questionable. It
> picks the first possible candidate instead of looking at all devices before
> making the decision. If we see regressions from the patch here, this would
> be the place to fix it.

One problem is that it's hard to know when we've looked at all the
devices.  I don't think we currently have a special hook for VGA after
boot-time enumeration.

> > > > > ---
> > > > >   drivers/pci/vgaarb.c | 7 -------
> > > > >   1 file changed, 7 deletions(-)
> > > > > 
> > > > > diff --git a/drivers/pci/vgaarb.c b/drivers/pci/vgaarb.c
> > > > > index 436fa7f4c3873..baa242b140993 100644
> > > > > --- a/drivers/pci/vgaarb.c
> > > > > +++ b/drivers/pci/vgaarb.c
> > > > > @@ -652,13 +652,6 @@ static bool vga_is_boot_device(struct
> > > > > vga_device *vgadev)
> > > > >           return true;
> > > > >       }
> > > > > -    /*
> > > > > -     * Vgadev has neither IO nor MEM enabled.  If we
> > > > > haven't found any
> > > > > -     * other VGA devices, it is the best candidate so far.
> > > > > -     */
> > > > > -    if (!boot_vga)
> > > > > -        return true;
> > > > > -
> > > > >       return false;
> > > > >   }
> > > > > -- 
> > > > > 2.43.0
> > > > > 
> > > 
> > > 
> > 
> 
> -- 
> --
> Thomas Zimmermann
> Graphics Driver Developer
> SUSE Software Solutions Germany GmbH
> Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
> GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)
> 
> 

Re: [PATCH] PCI/VGA: Don't assume the only VGA device on a system is `boot_vga`

[Mario Limonciello (AMD) (kernel.org)]

On 11/17/2025 9:44 AM, Bjorn Helgaas wrote:
> On Mon, Nov 17, 2025 at 08:25:05AM +0100, Thomas Zimmermann wrote:
>> Am 17.11.25 um 06:20 schrieb Mario Limonciello:
>>> On 10/29/25 2:38 PM, Mario Limonciello wrote:
>>>> On 10/29/25 2:34 PM, Bjorn Helgaas wrote:
>>>>> On Wed, Oct 29, 2025 at 01:59:33PM -0500, Mario Limonciello
>>>>> (AMD) wrote:
>>>>>> Some systems ship with multiple display class devices but not all
>>>>>> of them are VGA devices. If the "only" VGA device on the system is not
>>>>>> used for displaying the image on the screen marking it as `boot_vga`
>>>>>> because nothing was found is totally wrong.
>>>>>>
>>>>>> This behavior actually leads to mistakes of the wrong device being
>>>>>> advertised to userspace and then userspace can make
>>>>>> incorrect decisions.
>>>>>>
>>>>>> As there is an accurate `boot_display` sysfs file stop lying about
>>>>>> `boot_vga` by assuming if nothing is found it's the right device.
>>>>>>
>>>>>> Reported-by: Aaron Erhardt <aer@tuxedocomputers.com>
>>>>>> Closes: https://bugzilla.kernel.org/show_bug.cgi?id=220712
>>>>>> Tested-by: Aaron Erhardt <aer@tuxedocomputers.com>
>>>>>> Signed-off-by: Mario Limonciello (AMD) <superm1@kernel.org>
>>>>>
>>>>> Do we need a Fixes: here?  A stable cc?
>>>>>
>>>>> The bugzilla suggests this might be a regression and hence v6.18
>>>>> material?
>>>>
>>>> Yeah I think you're right, we should add these two tags and this
>>>> should go to 6.18 if it's a reasonable change.
>>>>
>>>> Cc: stable@vger.kernel.org
>>>> Fixes: ad90860bd10ee ("fbcon: Use screen info to find primary device")
>>>>
>>>> But let's also make sure Thomas Zimmermann agrees with this change.
>>>
>>> ping?
>>
>> Sorry, this fell though the cracks. I can see that this backfires, but still
>>
>> Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
>>
>> because the overall logic of detecting the boot-up VGA is questionable. It
>> picks the first possible candidate instead of looking at all devices before
>> making the decision. If we see regressions from the patch here, this would
>> be the place to fix it.
> 
> One problem is that it's hard to know when we've looked at all the
> devices.  I don't think we currently have a special hook for VGA after
> boot-time enumeration.

I fail to understand why this fallback was even there in the first 
place.  How can we have a VGA device that doesn't have IO or MEM enabled?

It seemed like it was specifically (37114e4d1547e) for hot added 
devices, but that sounds like faulty logic to me.  The whole point of it 
is the device that displayed the boot image - which shouldn't be hot added.
> 
>>>>>> ---
>>>>>>    drivers/pci/vgaarb.c | 7 -------
>>>>>>    1 file changed, 7 deletions(-)
>>>>>>
>>>>>> diff --git a/drivers/pci/vgaarb.c b/drivers/pci/vgaarb.c
>>>>>> index 436fa7f4c3873..baa242b140993 100644
>>>>>> --- a/drivers/pci/vgaarb.c
>>>>>> +++ b/drivers/pci/vgaarb.c
>>>>>> @@ -652,13 +652,6 @@ static bool vga_is_boot_device(struct
>>>>>> vga_device *vgadev)
>>>>>>            return true;
>>>>>>        }
>>>>>> -    /*
>>>>>> -     * Vgadev has neither IO nor MEM enabled.  If we
>>>>>> haven't found any
>>>>>> -     * other VGA devices, it is the best candidate so far.
>>>>>> -     */
>>>>>> -    if (!boot_vga)
>>>>>> -        return true;
>>>>>> -
>>>>>>        return false;
>>>>>>    }
>>>>>> -- 
>>>>>> 2.43.0
>>>>>>
>>>>
>>>>
>>>
>>
>> -- 
>> --
>> Thomas Zimmermann
>> Graphics Driver Developer
>> SUSE Software Solutions Germany GmbH
>> Frankenstr. 146, 90461 Nürnberg, Germany, www.suse.com
>> GF: Jochen Jaser, Andrew McDonald, Werner Knoblich, (HRB 36809, AG Nürnberg)
>>
>>