Re: [PATCH v2] PCI: Avoid putting some root ports into D3 on some Ryzen chips

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

On 12/17/2024 05:58, Werner Sembach wrote:
> 
> Am 17.12.24 um 11:10 schrieb Richard Hughes:
>> On Monday, 16 December 2024 at 23:37, Werner Sembach 
>> <wse@tuxedocomputers.com> wrote:
>>> - AfuEfix64.efi <bios>.bin /p /r /capsule -> overwrites nothing
>>> I tried to explain fwupd and the requirements to our contact at the 
>>> ODM, but
>>> just got the unhelpful reply to use the command above.
>> Can you name the ODM? I think essentially all the ODMs are uploading 
>> [valid] capsules to the LVFS now. If it helps, it's the same capsule 
>> needed for the LVFS as for the Microsoft WU (Windows Update) process 
>> and all ODMs should be intimately familiar with those requirements.
> 
> In this case it was a Tonfang/Uniwill barebone and I talked to a 
> Tongfang representative.
> 
> Want to point out again that I could determine that /k did do nothing in 
> the /capsule mode while /p and /r did effect the flash (iirc /p was 
> required for /capsule or the flash didn't start). I could not determine 
> if /b /n and /l did something (and probably can't without proper 
> documentation, which I don't have access to). I guess /x is irrelevant 
> as long as the flash works.
> 
> Do you have a Microsoft help page for OEMs about BIOS and EC upgrades 
> via Windows Update? Having some Windows requirements to point to often 
> helps when talking with ODMs in my experience.

https://learn.microsoft.com/en-us/windows-hardware/drivers/bringup/windows-uefi-firmware-update-platform

> 
>>> Do you know how these AfuEfix64.efi flags are passed over to the 
>>> capsule flash
>>> process? Then it might be possible to implement them in fwupd too.
>> The capsule, as expected by LVFS and WU, is actually a single *signed* 
>> binary that contains the flasher binary and the payload all wrapped up 
>> into one. The only time I've seen AfuEfix64.efi in use is for the 
>> system preload, as done in the factory.
> Yeah but since at least the /r flag influences the flash process there 
> must be some way the efi shell can pass on options to the flasher 
> included in the capsule ... EFI variables? Some bits appended to the 
> capsule?
>>
>> Richard.
>>