On 10/8/22 06:00, Lin Yang wrote:
With NUMA config:
<devices>
...
<memory model='sgx-epc'>
<source>
<nodemask>0-1</nodemask>
</source>
<target>
<size unit='KiB'>512</size>
<node>0</node>
</target>
</memory>
...
</devices>
Without NUMA config:
<devices>
...
<memory model='sgx-epc'>
<target>
<size unit='KiB'>512</size>
</target>
</memory>
...
</devices>
Patches look good to me. And I'd like to merge them. However, in one of
previous iterations Peter raised a concern whether this should be
exposed as something else than <memory model='sgx-epc'/>. One fact that
would support this is that SGX 'cuts' from guest memory. I mean, memory
backend-epc does not add more RAM into the guest:
# ./qemu-system-x86_64 -S -nographic -nodefaults -m 128 \
-machine pc,sgx-epc.0.memdev=memepc0,sgx-epc.0.node=0 \
-object
'{"qom-type":"memory-backend-epc","id":"memepc0","prealloc":true,"size":67108864,"host-nodes":[0],"policy":"bind"}'
\
-monitor stdio
QEMU 7.1.50 monitor - type 'help' for more information
(qemu) info memory-devices
Memory device [sgx-epc]: ""
memaddr: 0x100000000
size: 67108864
node: 0
memdev: /objects/memepc0
(qemu) info memory_size_summary
base memory: 134217728
plugged memory: 0
(qemu)
Which is also the reason why we need to special case
VIR_DOMAIN_MEMORY_MODEL_SGX_EPC so much.
On the other hand, the fact that QEMU exposes it as a memory backend
means, that it kind of falls into <memory/> category.
So my question then is, Peter, are you okay with leaving this under
<memory/> or do you have other suggestion?
Michal