Hi all,
hope this is the right place to ask this.
>From my understanding of virtio-pmem it can be used to bypass the guest
page cache, so that the host is in control of the guest page cache. I have
successfully run QEMU with a virtio-pmem device managing my drive. However,
while testing (e.g. creating files) I still see the guest page cache
growing, the same as without virtio-pmem. I would have expected to not see
the guest page growing, as virtio-pmem is supposed to bypass it. I thought
that the access through the virtio-pmem driver in the guest would
circumvent the normal page cache mechanism.
Does anyone know if my understanding of what virtio-pmem should do is wrong
or is there maybe some guest kernel specific configuration that I need to
do in order for this to work?
Best Regards,
Marco