search for: 17296k

...increasing the loop iterations should have no effect on the max memory usage. Replacing the h.set_request_structured_replies(False) line with h.add_meta_context(nbd.CONTEXT_BASE_ALLOCATION) (so that the block status request succeeds instead) proves the point; after that change, I'm seeing 17296k with count=1 and 17284k with count=10000 (that is, the memory usage is higher and somewhat less deterministic because there is now server interactions, but definitely no longer something that scales up as count increases). And in proving that, I found several _other_ bugs, now fixed: Python.ml...

I noticed while reading the code that we have a documentation hole that may cause memory leaks for clients that are unaware, in relation to completion callbacks. The situation arises as follows: for all commands with a completion callback, I checked that the code has clean semantics: either nbd_aio_FOO() returns -1 and we never call the callback cleanup, or nbd_aio_FOO() returns a cookie and