Return-Path: Subject: Re: [virtio-comment] [PATCH v1 1/2] virtio-mem: introduce VIRTIO_MEM_F_UNPLUGGED_INACCESSIBLE References: <20210812133150.36146-1-david@redhat.com> <20210812133150.36146-2-david@redhat.com> <87mtpgtpye.fsf@redhat.com> <87h7fotofz.fsf@redhat.com> From: David Hildenbrand Message-ID: <2044101d-2b2e-0fbf-68a7-2c4398084a40@redhat.com> Date: Tue, 17 Aug 2021 11:27:35 +0200 MIME-Version: 1.0 In-Reply-To: <87h7fotofz.fsf@redhat.com> Content-Type: text/plain; charset=utf-8; format=flowed Content-Language: en-US Content-Transfer-Encoding: 7bit To: Cornelia Huck , virtio-comment@lists.oasis-open.org List-ID: On 17.08.21 11:23, Cornelia Huck wrote: > On Tue, Aug 17 2021, David Hildenbrand wrote: > >> On 17.08.21 10:50, Cornelia Huck wrote: >>> On Thu, Aug 12 2021, David Hildenbrand wrote: >>>> +With VIRTIO_MEM_F_UNPLUGGED_INACCESSIBLE, the device MAY allow the CPU to >>> >>> "If VIRTIO_MEM_F_UNPLUGGED_INACCESSIBLE has been negotiated, ..." >>> >> >> ack >> >>>> +read memory of unplugged memory blocks inside \field{usable_region_size}. >>> >>> A compliant driver would not read that memory, would it? >> >> Indeed. The device could decide to allow for reading, but it's pretty >> much undefined behavior. > > Maybe make it "SHOULD NOT"? Less strong than "MUST NOT", but still makes > clear that a driver reading it is pretty much doing the wrong thing. > I was also wondering to just drop it completely, because it's actually undefined behavior. What do you think? -- Thanks, David / dhildenb