On Tue, Jun 16, 2026 at 06:26:17PM -0400, Joel Fernandes wrote:
> This series adds rcutorture coverage for the RCU priority-boosting "tardy
> deboost" issue. The first 3 patches are Paul's and the next 2 are fixes I had
> to make to the test, so that another patch series (I haven't posted yet) can
> correctly be tested. I am posting just the test patches first in advance,
> which
> can go in independently if we agree on the changes.
OK, I have recently gotten distracted with hazptrtorture, and am likely
to be more distracted with call_rcu*_nolock().
So if you are ready, then you should go first.
> Joel Fernandes (2):
> rcutorture: tighten boost-WARN to exclude any implicit-reader context
> rcutorture: give async deboost mechanisms up to 500us before WARN
You need these, but once I get the immediate-deboosting going, they
should become optional. But that is OK, I can adjust later.
Thanx, Paul
> Paul E. McKenney (3):
> rcutorture: Abstract reader-segment dump into
> rcu_torture_dump_read_segs()
> rcutorture: Check for immediate deboosting at reader end
> rcutorture: Test RCU readers from hardware interrupt handlers
>
> kernel/rcu/rcu.h | 7 ++
> kernel/rcu/rcutorture.c | 257 +++++++++++++++++++++++++++------------
> kernel/rcu/tree_plugin.h | 32 ++++-
> 3 files changed, 218 insertions(+), 78 deletions(-)
>
> --
> 2.34.1
>
