[PATCH] arm64: Relocate screen_info.lfb_base on PCI BAR allocation

Alexander Graf agraf at suse.de
Fri Apr 29 14:52:47 PDT 2016



On 29.04.16 23:37, Bjorn Helgaas wrote:
> On Fri, Apr 29, 2016 at 10:51:34PM +0200, Alexander Graf wrote:
>>
>>
>> On 29.04.16 22:25, Ard Biesheuvel wrote:
>>>
>>>> On 29 apr. 2016, at 22:06, Bjorn Helgaas <helgaas at kernel.org> wrote:
>>>>
>>>>> On Fri, Apr 29, 2016 at 03:51:49PM +0200, Ard Biesheuvel wrote:
>>>>>> On 29 April 2016 at 15:41, Bjorn Helgaas <helgaas at kernel.org> wrote:
>>>>>>> On Thu, Apr 28, 2016 at 11:39:35PM +0200, Alexander Graf wrote:
>>>>>>> On 28.04.16 20:06, Bjorn Helgaas wrote:
>>>>
>>>>>>>> If firmware is giving us a bare address of something, that seems like
>>>>>>>> a clue that it might depend on that address staying the same.
>>>>>>>
>>>>>>> Well, I'd look at it from the other side. It gives us a correct address
>>>>>>> on entry with the system configured at exactly the state it's in on
>>>>>>> entry. If Linux changes the system, some guarantees obviously don't work
>>>>>>> anymore.
>>>>>>
>>>>>> Can you point me to the part of the EFI spec that communicates this?
>>>>>> I'm curious what the intent is and whether there's any indication that
>>>>>> EFI expects the OS to preserve some configuration.  I don't think it's
>>>>>> reasonable for the OS to preserve this sort of configuration because
>>>>>> it limits how well we can support hotplug.
>>>>>>
>>>>>> I wonder if we're using this frame buffer address as more than what
>>>>>> EFI intended.  For example, maybe it was intended for use by an early
>>>>>> console driver, but there's some other mechanism we should be using
>>>>>> after that.
>>>>>
>>>>> The UEFI spec describes this as follows (UEFIv2.6 section 11.9)
>>>>>
>>>>> """
>>>>> Graphics output may also be required as part of the startup of an
>>>>> operating system. There are
>>>>> potentially times in modern operating systems prior to the loading of
>>>>> a high performance OS
>>>>> graphics driver where access to graphics output device is required.
>>>>> The Graphics Output Protocol
>>>>> supports this capability by providing the EFI OS loader access to a
>>>>> hardware frame buffer and
>>>>> enough information to allow the OS to draw directly to the graphics
>>>>> output device.
>>>>> """
>>>>>
>>>>> So the intent is to provide minimal framebuffer services until the
>>>>> 'real' driver takes over.
>>>>
>>>> Makes sense.  A 'real' driver for a PCI device would use
>>>> pci_register_driver() and use pci_resource_start() or similar to
>>>> locate the framebuffer, which would avoid the problem because the PCI
>>>> core doesn't change BARs while a driver owns the device.
>>>>
>>>>> The GOP protocol only describes the base and size of the framebuffer,
>>>>> and the pixel format. At boot time, the early UEFI code in the kernel
>>>>> could potentially figure out which PCI device it is related to, if
>>>>> necessary, but i am not sure if this would solve the x86 case as well.
>>>>
>>>> Does drivers/video/fbdev/efifb.c support only a single framebuffer
>>>> device?  If a system has several, how does it decide which to use?  I
>>>> assume UEFI would provide GOP for all the framebuffers?
>>>>
>>>
>>> Yes. The early efi code iterates over all gop instances to figure out which one is connected to the efi console.
> 
> How does it match the EFI console to the GOP instance?  Sorry, I'm
> pretty EFI-illiterate, and "find . -name \*efi\*.c | xargs grep -i gop"
> shows nothing :)
> 
>>>> If we could fix this by making efifb claim a PCI device, I think that
>>>> would be cleaner.  I don't know how to figure out the correct device,
>>>> but that would solve the "BAR changed" problem, and it would have
>>>> cleaner ownership semantics, too.
>>>
>>> If the bus layout is static, the efi init code can record the segment/bus/device and match it with the kernel's view. is that guaranteed to be sufficient for all implementations of pci?
>>
>> I think the only thing we can really take as granted is the physical
>> address we receive. Any efi device topology may be as implementation
>> dependent as your firmware vendor wants to it to be.
>>
>> So couldn't we just try to see whether the efifb is within a pci mmio
>> window? We could then go through all devices on it, search for
>> overlapping BAR allocations and know the device. From there it would
>> just be a matter of officially reserving the region - or setting the
>> resource to FIXED, right?
> 
> You could do something like a quirk that ran on every device after
> reading the BARs.  Then you could see if any of the device's resources
> contain the framebuffer address, and if so, set the FIXED bit for that
> resource.  Or you could save the pci_dev and the BAR number and make
> the framebuffer driver actually claim that device.  That would have
> the advantage of cleaner ownership and allowing the core to reassign
> it if necessary.

Unfortunately the framebuffer driver gets initialized after the BARs got
reassigned, so that wouldn't work :).

Can you point me to the code that does "read[ing] the BARs"? So far my
impression was that we don't even try to read out the current BAR values
from pci config space, but I can't claim I fully grasp the Linux pci
code quite yet.

> It would be a lot nicer if UEFI told us which device was the console.
> We can't figure this out from the ConOut variable or the HCDP or PCDP
> table or something?

I guess you could try to look at the device path for the GOP handle, but
I'm not sure any layout of the device paths is mandated by the spec. For
all I know a valid implementation could just claim the GOP interface
fell from the skies and it'd be legitimate. But I'm happy to stand
corrected if someone points me to the respective part of the spec.

I suppose HCDP and PCDP are ACPI tables? We can't rely on those :).
AArch64 systems can (and some do) boot perfectly fine without any ACPI
tables exposed via uEFI. And the less we have to marry ACPI with uEFI
the happier I am - uEFI seems like a pretty nice path towards
standardized booting while ACPI was quite a nightmare every time I
looked at it so far ;).


Alex



More information about the linux-arm-kernel mailing list