[Linux-kernel-mentees] [PATCH] doc: listRCU: Add some more listRCU patterns in the kernel
corbet at lwn.net
Tue Dec 3 13:41:32 UTC 2019
On Tue, 3 Dec 2019 12:09:43 +0530
Amol Grover <frextrite at gmail.com> wrote:
> - Add more information about listRCU patterns taking examples
> from audit subsystem in the linux kernel.
> - The initially written audit examples are kept, even though they are
> slightly different in the kernel.
> - Modify inline text for better passage quality.
> - Fix typo in code-blocks and improve code comments.
> - Add text formatting (italics, bold and code) for better emphasis.
Thanks for improving the documentation! I'll leave the RCU stuff to the
experts, but I do have one request...
> +When a process exits, ``release_task()`` calls ``list_del_rcu(&p->tasks)`` under
> +``tasklist_lock`` writer lock protection, to remove the task from the list of
> +all tasks. The ``tasklist_lock`` prevents concurrent list additions/removals
> +from corrupting the list. Readers using ``for_each_process()`` are not protected
> +with the ``tasklist_lock``. To prevent readers from noticing changes in the list
> +pointers, the ``task_struct`` object is freed only after one or more grace
> +periods elapse (with the help of ``call_rcu()``). This deferring of destruction
> +ensures that any readers traversing the list will see valid ``p->tasks.next``
> +pointers and deletion/freeing can happen in parallel with traversal of the list.
> +This pattern is also called an **existence lock**, since RCU pins the object in
> +memory until all existing readers finish.
Please don't put function names as literal text. If you just say
call_rcu(), it will be formatted correctly and cross-linked to the
appropriate kerneldoc entry. Saying ``call_rcu()`` defeats that and
clutters the plain-text reading experience.
More information about the Linux-kernel-mentees