On 6/13/19 9:45 AM, Roman Stratiienko wrote:

>>
>> Just throw nbd-client in your initramfs.  Every nbd server has it's own
>> handshake protocol, embedding one particular servers handshake protocol into the
>> kernel isn't the answer here.  Thanks,

The handshake protocol is well-specified:
https://github.com/NetworkBlockDevice/nbd/blob/cdb0bc57f3faefd7a5562d57ad57cd990781c415/doc/proto.md

All servers implement various subsets of that document for the handshake.

> Also, as far as I know mainline nbd-server daemon have only 2
> handshake protocols. So called OLD-STYLE and NEW-STYLE. And OLD-STYLE
> is no longer supported. So it should not be a problem, or please fix
> me if I'm wrong.

You are correct that oldstyle is no longer recommended. However, the
current NBD specification states that newstyle has two different
flavors, NBD_OPT_EXPORT_NAME (which you used, but is also old) and
NBD_OPT_GO (which is newer, but is more likely to encounter differences
where not all servers support it).

The NBD specification includes a compatibility baseline:
https://github.com/NetworkBlockDevice/nbd/blob/cdb0bc57f3faefd7a5562d57ad57cd990781c415/doc/proto.md#compatibility-and-interoperability

and right now, NBD_OPT_GO (and _not_ NBD_OPT_EXPORT_NAME) is the
preferred way forward.  As long as your handshake implementation
complies with the baseline documented there, you'll have maximum
portability to the largest number of servers that also support the
baseline - but not all servers are up to that baseline yet.

So, this becomes a question of how much are you reinventing baseline
portability handshake concerns in the kernel, vs. in initramfs.

-- 
Eric Blake, Principal Software Engineer
Red Hat, Inc.           +1-919-301-3226
Virtualization:  qemu.org | libvirt.org