[RFC PATCH 05/14] arm/thread_info: Introduce TIF_NOTIFY_IPI flag

[K Prateek Nayak]

Add support for TIF_NOTIFY_IPI on ARM. With TIF_NOTIFY_IPI, a sender
sending an IPI to an idle CPU in TIF_POLLING mode will set the
TIF_NOTIFY_IPI flag in the target's idle tasks's thread_info to pull the
CPU out of idle, as opposed to setting TIF_NEED_RESCHED previously. This
avoids spurious calls to schedule_idle() in cases where an IPI does not
necessarily wake up a task on the idle CPU.

IPI throughput measured using a modified version of Anton Blanchard's
ipistorm benchmark [1], configured to measure time taken to perform a
fixed number of smp_call_function_single() (with wait set to 1),
improves significantly with TIF_NOTIFY_IPI on a dual socket Ampere Server
(2 x 64C) with the benchmark time reducing to less than half for
100000 IPIs between two CPUs. (Note: Only WFI idle mode was left enabled
during testing to reduce variance)

cmdline: insmod ipistorm.ko numipi=100000 single=1 offset=8 cpulist=8 wait=1

  ==================================================================
  Test          : ipistorm (modified)
  Units         : Normalized runtime
  Interpretation: Lower is better
  Statistic     : AMean
  ==================================================================
  kernel:				time [pct imp]
  tip:sched/core			1.00 [0.00]
  tip:sched/core + TIF_NOTIFY_IPI	0.41 [59.29]

tip:sched/core was at tag "sched-core-2024-01-08" at the time of
testing.

Cc: Russell King <linux@xxxxxxxxxxxxxxx>
Cc: "Rafael J. Wysocki" <rafael@xxxxxxxxxx>
Cc: Daniel Lezcano <daniel.lezcano@xxxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Juri Lelli <juri.lelli@xxxxxxxxxx>
Cc: Vincent Guittot <vincent.guittot@xxxxxxxxxx>
Cc: Dietmar Eggemann <dietmar.eggemann@xxxxxxx>
Cc: Steven Rostedt <rostedt@xxxxxxxxxxx>
Cc: Ben Segall <bsegall@xxxxxxxxxx>
Cc: Mel Gorman <mgorman@xxxxxxx>
Cc: Daniel Bristot de Oliveira <bristot@xxxxxxxxxx>
Cc: Valentin Schneider <vschneid@xxxxxxxxxx>
Cc: Linus Walleij <linus.walleij@xxxxxxxxxx>
Cc: Ard Biesheuvel <ardb@xxxxxxxxxx>
Cc: linux-arm-kernel@xxxxxxxxxxxxxxxxxxx
Cc: linux-kernel@xxxxxxxxxxxxxxx
Cc: linux-pm@xxxxxxxxxxxxxxx
Link: https://github.com/antonblanchard/ipistorm [1]
Signed-off-by: K Prateek Nayak <kprateek.nayak@xxxxxxx>
---
 arch/arm/include/asm/thread_info.h | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/arch/arm/include/asm/thread_info.h b/arch/arm/include/asm/thread_info.h
index 943ffcf069d2..324248d87c9e 100644
--- a/arch/arm/include/asm/thread_info.h
+++ b/arch/arm/include/asm/thread_info.h
@@ -136,6 +136,7 @@ extern int vfp_restore_user_hwstate(struct user_vfp *,
  * thread information flags:
  *  TIF_USEDFPU		- FPU was used by this task this quantum (SMP)
  *  TIF_POLLING_NRFLAG	- true if poll_idle() is polling TIF_NEED_RESCHED
+ *			  or TIF_NOTIFY_IPI
  *
  * Any bit in the range of 0..15 will cause do_work_pending() to be invoked.
  */
@@ -144,6 +145,7 @@ extern int vfp_restore_user_hwstate(struct user_vfp *,
 #define TIF_NOTIFY_RESUME	2	/* callback before returning to user */
 #define TIF_UPROBE		3	/* breakpointed or singlestepping */
 #define TIF_NOTIFY_SIGNAL	4	/* signal notifications exist */
+#define TIF_NOTIFY_IPI		5	/* pending IPI on TIF_POLLLING idle CPU */
 
 #define TIF_USING_IWMMXT	17
 #define TIF_MEMDIE		18	/* is terminating due to OOM killer */
@@ -164,6 +166,7 @@ extern int vfp_restore_user_hwstate(struct user_vfp *,
 #define _TIF_SECCOMP		(1 << TIF_SECCOMP)
 #define _TIF_NOTIFY_SIGNAL	(1 << TIF_NOTIFY_SIGNAL)
 #define _TIF_USING_IWMMXT	(1 << TIF_USING_IWMMXT)
+#define _TIF_NOTIFY_IPI		(1 << TIF_NOTIFY_IPI)
 
 /* Checks for any syscall work in entry-common.S */
 #define _TIF_SYSCALL_WORK (_TIF_SYSCALL_TRACE | _TIF_SYSCALL_AUDIT | \
-- 
2.34.1