Re: [PATCH v9 0/5] KVM statistics data fd-based binary interface

On Tue, Jun 15, 2021 at 09:06:43AM +0200, Paolo Bonzini wrote:

On 15/06/21 07:25, Leon Romanovsky wrote:

quoted

Sorry for my naive questions, but how does telemetry get statistics
for hypervisors? Why is KVM different from hypervisors or NIC's statistics
or any other high speed devices (RDMA) that generate tons of data?

Right now, the only way is debugfs but it's slow, and it's disabled when
using lockdown mode; this series is a way to fix this.

I sense that there is another question in there; are you wondering if
another mechanism should be used, for example netlink?  The main issue there
is how to identify a VM, since KVM file descriptors don't have a name.
Using a pid works (sort of) for debugfs, but pids are not appropriate for a
stable API.  Using a file descriptor as in this series requires
collaboration from the userspace program; howver, once the file descriptor
has been transmitted via SCM_RIGHTS, telemetry can read it forever without
further IPC, and there is proper privilege separation.

Yeah, sorry for mixing different questions into one.

So the answer to the question "why KVM is different" is that it doesn't
have any stable identification except file descriptor. While hypervisors
have stable names, NICs and RDMA devices have interface indexes e.t.c.
Did I get it right?

And this was second part of my question, the first part was my attempt to
get on answer why current statistics like process info (/proc/xxx/*), NICs
(netlink) and RDMA (sysfs) are not using binary format.

Thanks

Paolo