rcutorture: Comment invocations of tick_dep_set_task()
authorPaul E. McKenney <paulmck@kernel.org>
Sun, 9 Feb 2025 13:09:26 +0000 (05:09 -0800)
committerJoel Fernandes <joelagnelf@nvidia.com>
Fri, 16 May 2025 15:12:26 +0000 (11:12 -0400)
commited0d0db095dd2b3a609f07d49d1342be9fb5c49d
tree90735487df458cde5004367eadd7afcd2b9e84a3
parenta3204f778cf7e37c7344404768398b4f9d43a368
rcutorture: Comment invocations of tick_dep_set_task()

The rcu_torture_reader() and rcu_torture_fwd_prog_cr() functions
run CPU-bound for extended periods of time (tens or even
hundreds of milliseconds), so they invoke tick_dep_set_task() and
tick_dep_clear_task() to ensure that the scheduling-clock tick helps
move grace periods forward.

So why doesn't rcu_torture_fwd_prog_nr() also invoke tick_dep_set_task()
and tick_dep_clear_task()?  Because the point of this function is to test
RCU's ability to (eventually) force grace periods forward even when the
tick has been disabled during long CPU-bound kernel execution.

Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Joel Fernandes <joelagnelf@nvidia.com>
kernel/rcu/rcutorture.c