[PATCH RFC 0/3] Fix cpuidle_idle_call() RCU usage

From: Paul E. McKenney
Date: Thu Feb 02 2012 - 20:13:36 EST


Hello!

This patch series is a second attempt to fix the idle-loop uses of RCU,
see https://lkml.org/lkml/2012/2/1/741 for v1. Where the first series
attempted to drive rcu_idle_enter() and rcu_idle_exit() further down
into the Linux kernels multitude of idle loops, this patch instead
marks specific idle-loop operations containing RCU read-side critical
sections, as suggested by Nicolas Pitre and Steven Rostedt. The possibility
of code shared between idle and non-idle also requires the ability to nest
rcu_idle_enter() calls. The individual patches are as follows:

1. Allow nesting of rcu_idle_enter() and rcu_idle_exit().
2. Add an RCU_NONIDLE() macro to enclose idle-loop code that
contains RCU read-side critical sections.
3. Use RCU_NONIDLE() to protect cpuidle_idle_call()'s tracepoints.

This patchset has the distinct advantage of avoiding touching any
architecture-specific code. ;-)

Thanx, Paul

------------------------------------------------------------------------

b/drivers/cpuidle/cpuidle.c | 12 ++++++++----
b/include/linux/rcupdate.h | 27 +++++++++++++++++++++++++++
b/kernel/rcu.h | 18 +++++++++++++++++-
b/kernel/rcutiny.c | 16 ++++++++++++----
b/kernel/rcutree.c | 19 +++++++++++++------
kernel/rcutiny.c | 2 ++
kernel/rcutree.c | 2 ++
7 files changed, 81 insertions(+), 15 deletions(-)

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/