Re: AMD SME encrpytion and PCI BAR pages to user space

[Tom Lendacky]

On 10/19/20 12:00 PM, Jason Gunthorpe wrote:
> On Mon, Oct 19, 2020 at 11:36:16AM -0500, Tom Lendacky wrote:
> 
>>> Is RDMA missing something? I don't see anything special in VFIO for
>>> instance and the two are very similar - does VFIO work with SME, eg
>>> DPDK or something unrelated to virtualization?
>>
>> If user space is mapping un-encrypted memory, then, yes, it would seem
>> that there is a gap in the support where the pgprot_decrypted() would be
>> needed in order to override the protection map.
> 
> It isn't "memory" it is PCI BAR pages, eg memory mapped IO

Right, I understand that.

> 
>>> Is there a reason not to just add prot_decrypted() to
>>> io_remap_pfn_range()? Is there use cases where a caller actually wants
>>> encrypted io memory?
>>
>> As long as you never have physical memory / ram being mapped in this path,
>> it seems that applying pgprot_decrypted() would be ok.
> 
> I think the word 'io' implies this is the case..

Heh, you would think so, but I found quite a few things that used ioremap
instead of memremap when developing this.

> 
> Let me make a patch for this avenue then, I think it is not OK to add
> pgprot_decrypted to every driver.. We already have the special
> distinction with io and non-io remap, that seems better.

Yup, seems reasonable.

> 
>>> I saw your original patch series edited a few drivers this way, but
>>> not nearly enough. So I feel like I'm missing something.. Does vfio
>>> work with SME? I couldn't find any sign of it calling prot_decrypted()
>>> either?
>>
>> I haven't tested SME with VFIO/DPDK.
> 
> Hum, I assume it is broken also. Actually quite a swath of drivers
> and devices will be broken under this :\

Not sure what you mean by the last statement - in general or when running
under VFIO/DPDK? In general, traditional in kernel drivers work just fine
under SME without any changes.

Thanks,
Tom

> 
> Jason
>