[Xen-tools] b0893b876c8c4c4eb507d48fc1c4af4268ddecde and watch behaviour

[Rusty Russell <rusty@xxxxxxxxxxxxxxx>]

Hi Christian!

        I was looking through b0893b876c8c4c4eb507d48fc1c4af4268ddecde, and it
took me a while to figure out what the code was doing.  I think there
are two issues here:

1) Directories implicitly created by mkdir/write don't fire watches (eg.
mkdir /a/sub/dir only fires a watch event for /a/sub/dir, even if it
also created /a/sub), and
2) Rm only fires a single event for a watch, as an optimization.

The first one is a bug, I think, and I'm testing a patch as I type this.
Fixing the second would actually simplify the xenstored code, but will
mean flooding xenbus_probe with events, so I think we would also want to
introduce a "max_depth" arg to xs_watch.

What do you think?  I can prepare a patch to see how it looks, but my
gut feel is that it'll be a net simplification (xenbus_probe.c and
xenstored_watch.c).

Thoughts?
Rusty.
-- 
A bad analogy is like a leaky screwdriver -- Richard Braakman


_______________________________________________
Xen-tools mailing list
Xen-tools@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-tools