diff mbox series

[v2,RESEND] drivers/perf: hisi: Schedule perf session according to locality

Message ID 20230815131010.2147-1-yangyicong@huawei.com (mailing list archive)
State New, archived
Headers show
Series [v2,RESEND] drivers/perf: hisi: Schedule perf session according to locality | expand

Commit Message

Yicong Yang Aug. 15, 2023, 1:10 p.m. UTC
From: Yicong Yang <yangyicong@hisilicon.com>

The PCIe PMUs locate on different NUMA node but currently we don't
consider it and likely stack all the sessions on the same CPU:

[root@localhost tmp]# cat /sys/devices/hisi_pcie*/cpumask
0
0
0
0
0
0

This can be optimize a bit to use a local CPU for the PMU.

Signed-off-by: Yicong Yang <yangyicong@hisilicon.com>
---
Change since v2:
- Make interrupt affinity consistent with CPU in online handler
Link: https://lore.kernel.org/all/20230808125147.2080-1-yangyicong@huawei.com/

 drivers/perf/hisilicon/hisi_pcie_pmu.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

Comments

Will Deacon Aug. 16, 2023, 6:25 p.m. UTC | #1
On Tue, 15 Aug 2023 21:10:10 +0800, Yicong Yang wrote:
> From: Yicong Yang <yangyicong@hisilicon.com>
> 
> The PCIe PMUs locate on different NUMA node but currently we don't
> consider it and likely stack all the sessions on the same CPU:
> 
> [root@localhost tmp]# cat /sys/devices/hisi_pcie*/cpumask
> 0
> 0
> 0
> 0
> 0
> 0
> 
> [...]

Applied to will (for-next/perf), thanks!

[1/1] drivers/perf: hisi: Schedule perf session according to locality
      https://git.kernel.org/will/c/83a6d80c2bfd

Cheers,
diff mbox series

Patch

diff --git a/drivers/perf/hisilicon/hisi_pcie_pmu.c b/drivers/perf/hisilicon/hisi_pcie_pmu.c
index e10fc7cb9493..5a00adb2de8c 100644
--- a/drivers/perf/hisilicon/hisi_pcie_pmu.c
+++ b/drivers/perf/hisilicon/hisi_pcie_pmu.c
@@ -665,8 +665,8 @@  static int hisi_pcie_pmu_online_cpu(unsigned int cpu, struct hlist_node *node)
 	struct hisi_pcie_pmu *pcie_pmu = hlist_entry_safe(node, struct hisi_pcie_pmu, node);
 
 	if (pcie_pmu->on_cpu == -1) {
-		pcie_pmu->on_cpu = cpu;
-		WARN_ON(irq_set_affinity(pcie_pmu->irq, cpumask_of(cpu)));
+		pcie_pmu->on_cpu = cpumask_local_spread(0, dev_to_node(&pcie_pmu->pdev->dev));
+		WARN_ON(irq_set_affinity(pcie_pmu->irq, cpumask_of(pcie_pmu->on_cpu)));
 	}
 
 	return 0;
@@ -676,14 +676,23 @@  static int hisi_pcie_pmu_offline_cpu(unsigned int cpu, struct hlist_node *node)
 {
 	struct hisi_pcie_pmu *pcie_pmu = hlist_entry_safe(node, struct hisi_pcie_pmu, node);
 	unsigned int target;
+	cpumask_t mask;
+	int numa_node;
 
 	/* Nothing to do if this CPU doesn't own the PMU */
 	if (pcie_pmu->on_cpu != cpu)
 		return 0;
 
 	pcie_pmu->on_cpu = -1;
-	/* Choose a new CPU from all online cpus. */
-	target = cpumask_any_but(cpu_online_mask, cpu);
+
+	/* Choose a local CPU from all online cpus. */
+	numa_node = dev_to_node(&pcie_pmu->pdev->dev);
+	if (cpumask_and(&mask, cpumask_of_node(numa_node), cpu_online_mask) &&
+	    cpumask_andnot(&mask, &mask, cpumask_of(cpu)))
+		target = cpumask_any(&mask);
+	else
+		target = cpumask_any_but(cpu_online_mask, cpu);
+
 	if (target >= nr_cpu_ids) {
 		pci_err(pcie_pmu->pdev, "There is no CPU to set\n");
 		return 0;