On Wed, 23 Mar 2016, George Dunlap wrote:
> On 22/03/16 12:52, Roger Pau Monné wrote:
> > On Mon, 21 Mar 2016, George Dunlap wrote:
> >> Signed-off-by: George Dunlap <george.dunlap@citrix.com>
> >> ---
> >> Changes since v1:
> >> - Attempt to make a clear distinction between custom hotplug scripts
> >> and the script called for raw physical devices and files
> >>
> >> CC: Ian Jackson <ian.jackson@citrix.com>
> >> CC: Wei Liu <wei.liu2@citrix.com>
> >> CC: Roger Pau Monne <roger.pau@citrix.com>
> >> ---
> >>  docs/misc/block-scripts.txt | 101 ++++++++++++++++++++++++++++++++++++++++++++
> >>  1 file changed, 101 insertions(+)
> >>
> >> diff --git a/docs/misc/block-scripts.txt b/docs/misc/block-scripts.txt
> >> new file mode 100644
> >> index 0000000..6dd5d48
> >> --- /dev/null
> >> +++ b/docs/misc/block-scripts.txt
> >> @@ -0,0 +1,101 @@
> > [...]
> >> +Inputs
> >> +------
> >> +
> >> +In all cases, the scripts are called with either "add" or "remove" as
> >> +the command.  For custom scripts, the command will be the first
> >> +argument of the script (i.e. $1).
> >> +
> >> +The environment variable XENBUS_PATH will be set to the
> >> +path for the block device to be created.
> > 
> > This is true for Linux, but not for NetBSD. On NetBSD no env variables are 
> > needed, and everything is passed as arguments using the following format:
> > 
> > ./<script> <backend_path> <xenbus state>
> > 
> > Where xenbus state is either 2 or 6.
> > 
> > On FreeBSD I'm aiming of using the same input interface for both block and 
> > network scripts, and it is the following:
> > 
> > ./<script> {add/remove} <backend path>
> > 
> > With no env variables provided at all. So either this section is expanded, 
> > or it is labelled as "Linux Inputs".
> 
> Nothing like consistency across implementations. :-)
> 
> So in the case of NetBSD, "2" means 'add' and "6" means 'remove'?  Or
> how does that work?

Yes, 2 means add and 6 remove.
 
> Presumably there's not much we can do about NetBSD at this point, if
> there are (or may be) out-of-tree scripts that expect the new format.
> But unless there's a good reason, it seems like we should try to
> converge the hotplug script protocol.
> 
> Was there a particular reason you wanted to use an argument instead of
> an environment variable?  If not, it's probably better to just follow
> suit with the Linux protocol.

Don't get me wrong, but the Linux protocol is all but consistent :). I'm 
not sure if those other env variables are used by the block hotplug 
scripts, but we also set:

script=<script_name>
XENBUS_TYPE=<vbd/vif>
XENBUS_PATH=<be_path>
XENBUS_BASE_PATH="backend"

And it's even worse for vifs, where the action parameter that we pass to 
the hotplug script is different depending on whether we are dealing with 
a PV or an emulated interface (PV uses "online/offline", while emulated 
use "add/remove").

I would like to share the same interface, but I think the Linux one is 
simply too broken, and I don't want to put this anywhere close to FreeBSD. 
There at least I have a chance of having something that's simple and 
rational.

> If there is a good reason, we could consider changing the Linux protocol
> as well (leaving the environment variable in for backwards-compatibility
> purposes).
> 
>  -George
>