Hi,

I am wondering why the subtask field is not protected in the function 
with_subtask in xen-api/ocaml/xenops/task_server.ml. It is not clear to me that 
the invariant "only one thread per task" is enforced in xenopsd (where there 
are *a lot* of asynchronous processes) and in the storage layer (ok, there is 
only one call to with_subtask here, and it seems relatively safe as the main 
task is short-lived and not passed around, but how do you ensure it will still 
be enforced in future changes). Is there any good reason to that ?

Thomas
_______________________________________________
Xen-api mailing list
[email protected]
http://lists.xen.org/cgi-bin/mailman/listinfo/xen-api

Reply via email to