Re: [PATCH 2/3] sev/i386: Warn if using -kernel with invalid OVMF hashes table area

[Philippe Mathieu-Daudé]

On 11/2/21 19:38, Dr. David Alan Gilbert wrote:
> * Dov Murik (dovmurik@linux.ibm.com) wrote:

>>> however, maybe it needs to be more thorough before using area->base to
>>> qemu_map_ram_ptr? - I think it'll get unhappy if it's a bad address not
>>> in a ram block. (Or check you're running over the end of a ramblock)
>>>
>>
>> Does address_space_write perform these checks? Or maybe another API for
>> accessing the guest's RAM?
> 
> I'm not sure; for example in address_space_map I don't see an check that
> flatview_translate gives a valid mr.

IIUC the API the MemTxAttrs argument could help you, but I don't think
properly enforced (yet?...).