On 17.03.2021 16:55, Ian Jackson wrote: > Jan Beulich writes ("Re: libxl / xen-pciback interaction [and 1 more > messages]"): >> On 17.03.2021 16:12, Ian Jackson wrote: >>> How does what libxl is doing differ from a setup, immediately followed >>> by a hot-add ? >> >> In the hot-add case libxl drives things through Reconfiguring state. >> I'm not sure this would be an appropriate (and backwards compatible) >> thing to do when initially populating xenstore. > > Ah. Tbanks, that is precisely the answer to my question. > > I think that means, therefore, populating the whole lot in one > transaction. > > (From what you say it doesn't sound like it's possible to write only a > subset, perhaps with state "not ready yet" and then set them all go > "go" at the end.)
Indeed, that's my understanding and a consequence of pciback's watch covering the entire backend subtree, rather than e.g. just the num_devs node. Jan