In count_unregister_thread, both adding the counter to total and NULLs
the exiting thread's counterp[] array element happen before publishing
the newly allocated cap. Therefore, count_unregister_thread() can not
result in the outgoing thread's work to be double counted.
Signed-off-by: Alan Huang <mmpgouride@xxxxxxxxx>
---
together/applyrcu.tex | 9 ---------
1 file changed, 9 deletions(-)
diff --git a/together/applyrcu.tex b/together/applyrcu.tex
index 1e69fe3e..09a265f7 100644
--- a/together/applyrcu.tex
+++ b/together/applyrcu.tex
@@ -151,15 +151,6 @@ is invoked by each newly created thread.
So, interestingly enough, when adding a new thread, this
implementation gets the effect of allocating a new structure,
but without actually having to do the allocation.
-
- \begin{fcvref}[ln:count:count_end_rcu:whole:unreg]
- On the other hand, \co{count_unregister_thread()} can result
- in the outgoing thread's work being double counted.
- This can happen when \co{read_count()} is invoked between
- \clnref{add,null}.
- There are efficient ways of avoiding this double-counting, but
- these are left as an exercise for the reader.
- \end{fcvref}
}\QuickQuizEnd
\begin{fcvref}[ln:count:count_end_rcu:whole:unreg]
--
2.34.1
