[v5] arm64: psci: Reduce the waiting time for cpu_psci_cpu_kill()
diff mbox series

Message ID 710429cc-4d88-b7c3-b068-5459cf8133b5@huawei.com
State New
Headers show
Series
  • [v5] arm64: psci: Reduce the waiting time for cpu_psci_cpu_kill()
Related show

Commit Message

Yunfeng Ye Oct. 21, 2019, 10:52 a.m. UTC
In cases like suspend-to-disk and suspend-to-ram, a large number of CPU
cores need to be shut down. At present, the CPU hotplug operation is
serialised, and the CPU cores can only be shut down one by one. In this
process, if PSCI affinity_info() does not return LEVEL_OFF quickly,
cpu_psci_cpu_kill() needs to wait for 10ms. If hundreds of CPU cores
need to be shut down, it will take a long time.

Normally, there is no need to wait 10ms in cpu_psci_cpu_kill(). So
change the wait interval from 10 ms to max 1 ms and use usleep_range()
instead of msleep() for more accurate timer.

In addition, reducing the time interval will increase the messages
output, so remove the "Retry ..." message, instead, track time and
output to the the sucessful message.

Signed-off-by: Yunfeng Ye <yeyunfeng@huawei.com>
---
v4 -> v5:
 - track time instead of loop counter

v3 -> v4:
 - using time_before(jiffies, timeout) to check
 - update the comment as review suggest

v2 -> v3:
 - update the comment
 - remove the busy-wait logic, modify the loop logic and output message

v1 -> v2:
 - use usleep_range() instead of udelay() after waiting for a while

 arch/arm64/kernel/psci.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

Comments

Sudeep Holla Oct. 21, 2019, 11:21 a.m. UTC | #1
On Mon, Oct 21, 2019 at 06:52:16PM +0800, Yunfeng Ye wrote:
> In cases like suspend-to-disk and suspend-to-ram, a large number of CPU
> cores need to be shut down. At present, the CPU hotplug operation is
> serialised, and the CPU cores can only be shut down one by one. In this
> process, if PSCI affinity_info() does not return LEVEL_OFF quickly,
> cpu_psci_cpu_kill() needs to wait for 10ms. If hundreds of CPU cores
> need to be shut down, it will take a long time.
> 
> Normally, there is no need to wait 10ms in cpu_psci_cpu_kill(). So
> change the wait interval from 10 ms to max 1 ms and use usleep_range()
> instead of msleep() for more accurate timer.
> 
> In addition, reducing the time interval will increase the messages
> output, so remove the "Retry ..." message, instead, track time and
> output to the the successful message.

Reviewed-by: Sudeep Holla <sudeep.holla@arm.com>

--
Regards,
Sudeep

Patch
diff mbox series

diff --git a/arch/arm64/kernel/psci.c b/arch/arm64/kernel/psci.c
index c9f72b2665f1..43ae4e0c968f 100644
--- a/arch/arm64/kernel/psci.c
+++ b/arch/arm64/kernel/psci.c
@@ -81,7 +81,8 @@  static void cpu_psci_cpu_die(unsigned int cpu)

 static int cpu_psci_cpu_kill(unsigned int cpu)
 {
-	int err, i;
+	int err;
+	unsigned long start, end;

 	if (!psci_ops.affinity_info)
 		return 0;
@@ -91,16 +92,18 @@  static int cpu_psci_cpu_kill(unsigned int cpu)
 	 * while it is dying. So, try again a few times.
 	 */

-	for (i = 0; i < 10; i++) {
+	start = jiffies;
+	end = start + msecs_to_jiffies(100);
+	do {
 		err = psci_ops.affinity_info(cpu_logical_map(cpu), 0);
 		if (err == PSCI_0_2_AFFINITY_LEVEL_OFF) {
-			pr_info("CPU%d killed.\n", cpu);
+			pr_info("CPU%d killed (polled %d ms)\n", cpu,
+				jiffies_to_msecs(jiffies - start));
 			return 0;
 		}

-		msleep(10);
-		pr_info("Retrying again to check for CPU kill\n");
-	}
+		usleep_range(100, 1000);
+	} while (time_before(jiffies, end));

 	pr_warn("CPU%d may not have shut down cleanly (AFFINITY_INFO reports %d)\n",
 			cpu, err);