RE: [RFC PATCH 10/15] virt: tdx-guest: Set up and accept private MMIO ranges

[Duan, Zhenzhong]

>-----Original Message-----
>From: Li, Xiaoyao <xiaoyao.li@xxxxxxxxx>
>Subject: Re: [RFC PATCH 10/15] virt: tdx-guest: Set up and accept private
>MMIO ranges
>
>On 9/24/2026 12:10 PM, Zhenzhong Duan wrote:
>> Build private MMIO descriptors from the TDI report using the PCI TSM
>> MMIO helpers and accept each private range through TDG.MMIO.ACCEPT.
>>
>> Keep the descriptors while the device remains locked and tear them
>> down on acceptance failure and during unlock.
>
>1. The changelog talks what it does but not why the patch is needed and
>what is
>the design consideration. E.g.,
>
> - why choose to setup MMIO in the .run callback instead of .lock?
>
> - is tdx_tdi_mmio_teardown() safe for MMIO partial acceptance? when
>current
>"run" fails with partial MMIO acceptance, can the second "run" work well?

OK, will rewrite this commit log

>
>2. So far tdx_devsec_run() always return -EOPNOTSUPP, and MMIO descriptors
>are
>never retained.

This is intentionally, as we haven't finish all the work in .run() yet.