powerpc/papr_scm: Add perf interface support

Performance monitoring support for papr-scm nvdimm devices
via perf interface is added which includes addition of pmu
functions like add/del/read/event_init for nvdimm_pmu struture.

A new parameter 'priv' in added to the pdev_archdata structure to save
nvdimm_pmu device pointer, to handle the unregistering of pmu device.

papr_scm_pmu_register function populates the nvdimm_pmu structure
with name, capabilities, cpumask along with event handling
functions. Finally the populated nvdimm_pmu structure is passed to
register the pmu device. Event handling functions internally uses
hcall to get events and counter data.

Result in power9 machine with 2 nvdimm device:

Ex: List all event by perf list

command:# perf list nmem

  nmem0/cache_rh_cnt/                                [Kernel PMU event]
  nmem0/cache_wh_cnt/                                [Kernel PMU event]
  nmem0/cri_res_util/                                [Kernel PMU event]
  nmem0/ctl_res_cnt/                                 [Kernel PMU event]
  nmem0/ctl_res_tm/                                  [Kernel PMU event]
  nmem0/fast_w_cnt/                                  [Kernel PMU event]
  nmem0/host_l_cnt/                                  [Kernel PMU event]
  nmem0/host_l_dur/                                  [Kernel PMU event]
  nmem0/host_s_cnt/                                  [Kernel PMU event]
  nmem0/host_s_dur/                                  [Kernel PMU event]
  nmem0/med_r_cnt/                                   [Kernel PMU event]
  nmem0/med_r_dur/                                   [Kernel PMU event]
  nmem0/med_w_cnt/                                   [Kernel PMU event]
  nmem0/med_w_dur/                                   [Kernel PMU event]
  nmem0/mem_life/                                    [Kernel PMU event]
  nmem0/poweron_secs/                                [Kernel PMU event]
  ...
  nmem1/mem_life/                                    [Kernel PMU event]
  nmem1/poweron_secs/                                [Kernel PMU event]

Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Nageswara R Sastry <rnsastry@linux.ibm.com>
Signed-off-by: Kajol Jain <kjain@linux.ibm.com>
[Add numa_map_to_online_node function call to get online node id]
Reported-by: Nageswara R Sastry <rnsastry@linux.ibm.com>
Reviewed-by: Madhavan Srinivasan <maddy@in.ibm.com>
Link: https://lore.kernel.org/r/20220225143024.47947-4-kjain@linux.ibm.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>

arch/powerpc/include/asm/device.h

@@ -48,6 +48,11 @@ struct dev_archdata {
 
 struct pdev_archdata {
 	u64 dma_mask;
+	/*
+	 * Pointer to nvdimm_pmu structure, to handle the unregistering
+	 * of pmu device
+	 */
+	void *priv;
 };
 
 #endif /* _ASM_POWERPC_DEVICE_H */

arch/powerpc/platforms/pseries/papr_scm.c

@@ -19,6 +19,7 @@
 #include <asm/papr_pdsm.h>
 #include <asm/mce.h>
 #include <asm/unaligned.h>
+#include <linux/perf_event.h>
 
 #define BIND_ANY_ADDR (~0ul)
 
@@ -68,6 +69,8 @@
 #define PAPR_SCM_PERF_STATS_EYECATCHER __stringify(SCMSTATS)
 #define PAPR_SCM_PERF_STATS_VERSION 0x1
 
+#define to_nvdimm_pmu(_pmu)	container_of(_pmu, struct nvdimm_pmu, pmu)
+
 /* Struct holding a single performance metric */
 struct papr_scm_perf_stat {
 	u8 stat_id[8];
@@ -120,6 +123,9 @@ struct papr_scm_priv {
 
 	/* length of the stat buffer as expected by phyp */
 	size_t stat_buffer_len;
+
+	 /* array to have event_code and stat_id mappings */
+	char **nvdimm_events_map;
 };
 
 static int papr_scm_pmem_flush(struct nd_region *nd_region,
@@ -340,6 +346,218 @@ static ssize_t drc_pmem_query_stats(struct papr_scm_priv *p,
 	return 0;
 }
 
+static int papr_scm_pmu_get_value(struct perf_event *event, struct device *dev, u64 *count)
+{
+	struct papr_scm_perf_stat *stat;
+	struct papr_scm_perf_stats *stats;
+	struct papr_scm_priv *p = (struct papr_scm_priv *)dev->driver_data;
+	int rc, size;
+
+	/* Allocate request buffer enough to hold single performance stat */
+	size = sizeof(struct papr_scm_perf_stats) +
+		sizeof(struct papr_scm_perf_stat);
+
+	if (!p || !p->nvdimm_events_map)
+		return -EINVAL;
+
+	stats = kzalloc(size, GFP_KERNEL);
+	if (!stats)
+		return -ENOMEM;
+
+	stat = &stats->scm_statistic[0];
+	memcpy(&stat->stat_id,
+	       p->nvdimm_events_map[event->attr.config],
+		sizeof(stat->stat_id));
+	stat->stat_val = 0;
+
+	rc = drc_pmem_query_stats(p, stats, 1);
+	if (rc < 0) {
+		kfree(stats);
+		return rc;
+	}
+
+	*count = be64_to_cpu(stat->stat_val);
+	kfree(stats);
+	return 0;
+}
+
+static int papr_scm_pmu_event_init(struct perf_event *event)
+{
+	struct nvdimm_pmu *nd_pmu = to_nvdimm_pmu(event->pmu);
+	struct papr_scm_priv *p;
+
+	if (!nd_pmu)
+		return -EINVAL;
+
+	/* test the event attr type for PMU enumeration */
+	if (event->attr.type != event->pmu->type)
+		return -ENOENT;
+
+	/* it does not support event sampling mode */
+	if (is_sampling_event(event))
+		return -EOPNOTSUPP;
+
+	/* no branch sampling */
+	if (has_branch_stack(event))
+		return -EOPNOTSUPP;
+
+	p = (struct papr_scm_priv *)nd_pmu->dev->driver_data;
+	if (!p)
+		return -EINVAL;
+
+	/* Invalid eventcode */
+	if (event->attr.config == 0 || event->attr.config > 16)
+		return -EINVAL;
+
+	return 0;
+}
+
+static int papr_scm_pmu_add(struct perf_event *event, int flags)
+{
+	u64 count;
+	int rc;
+	struct nvdimm_pmu *nd_pmu = to_nvdimm_pmu(event->pmu);
+
+	if (!nd_pmu)
+		return -EINVAL;
+
+	if (flags & PERF_EF_START) {
+		rc = papr_scm_pmu_get_value(event, nd_pmu->dev, &count);
+		if (rc)
+			return rc;
+
+		local64_set(&event->hw.prev_count, count);
+	}
+
+	return 0;
+}
+
+static void papr_scm_pmu_read(struct perf_event *event)
+{
+	u64 prev, now;
+	int rc;
+	struct nvdimm_pmu *nd_pmu = to_nvdimm_pmu(event->pmu);
+
+	if (!nd_pmu)
+		return;
+
+	rc = papr_scm_pmu_get_value(event, nd_pmu->dev, &now);
+	if (rc)
+		return;
+
+	prev = local64_xchg(&event->hw.prev_count, now);
+	local64_add(now - prev, &event->count);
+}
+
+static void papr_scm_pmu_del(struct perf_event *event, int flags)
+{
+	papr_scm_pmu_read(event);
+}
+
+static int papr_scm_pmu_check_events(struct papr_scm_priv *p, struct nvdimm_pmu *nd_pmu)
+{
+	struct papr_scm_perf_stat *stat;
+	struct papr_scm_perf_stats *stats;
+	char *statid;
+	int index, rc, count;
+	u32 available_events;
+
+	if (!p->stat_buffer_len)
+		return -ENOENT;
+
+	available_events = (p->stat_buffer_len  - sizeof(struct papr_scm_perf_stats))
+			/ sizeof(struct papr_scm_perf_stat);
+
+	/* Allocate the buffer for phyp where stats are written */
+	stats = kzalloc(p->stat_buffer_len, GFP_KERNEL);
+	if (!stats) {
+		rc = -ENOMEM;
+		return rc;
+	}
+
+	/* Allocate memory to nvdimm_event_map */
+	p->nvdimm_events_map = kcalloc(available_events, sizeof(char *), GFP_KERNEL);
+	if (!p->nvdimm_events_map) {
+		rc = -ENOMEM;
+		goto out_stats;
+	}
+
+	/* Called to get list of events supported */
+	rc = drc_pmem_query_stats(p, stats, 0);
+	if (rc)
+		goto out_nvdimm_events_map;
+
+	for (index = 0, stat = stats->scm_statistic, count = 0;
+		     index < available_events; index++, ++stat) {
+		statid = kzalloc(strlen(stat->stat_id) + 1, GFP_KERNEL);
+		if (!statid) {
+			rc = -ENOMEM;
+			goto out_nvdimm_events_map;
+		}
+
+		strcpy(statid, stat->stat_id);
+		p->nvdimm_events_map[count] = statid;
+		count++;
+	}
+	p->nvdimm_events_map[count] = NULL;
+	kfree(stats);
+	return 0;
+
+out_nvdimm_events_map:
+	kfree(p->nvdimm_events_map);
+out_stats:
+	kfree(stats);
+	return rc;
+}
+
+static void papr_scm_pmu_register(struct papr_scm_priv *p)
+{
+	struct nvdimm_pmu *nd_pmu;
+	int rc, nodeid;
+
+	nd_pmu = kzalloc(sizeof(*nd_pmu), GFP_KERNEL);
+	if (!nd_pmu) {
+		rc = -ENOMEM;
+		goto pmu_err_print;
+	}
+
+	rc = papr_scm_pmu_check_events(p, nd_pmu);
+	if (rc)
+		goto pmu_check_events_err;
+
+	nd_pmu->pmu.task_ctx_nr = perf_invalid_context;
+	nd_pmu->pmu.name = nvdimm_name(p->nvdimm);
+	nd_pmu->pmu.event_init = papr_scm_pmu_event_init;
+	nd_pmu->pmu.read = papr_scm_pmu_read;
+	nd_pmu->pmu.add = papr_scm_pmu_add;
+	nd_pmu->pmu.del = papr_scm_pmu_del;
+
+	nd_pmu->pmu.capabilities = PERF_PMU_CAP_NO_INTERRUPT |
+				PERF_PMU_CAP_NO_EXCLUDE;
+
+	/*updating the cpumask variable */
+	nodeid = numa_map_to_online_node(dev_to_node(&p->pdev->dev));
+	nd_pmu->arch_cpumask = *cpumask_of_node(nodeid);
+
+	rc = register_nvdimm_pmu(nd_pmu, p->pdev);
+	if (rc)
+		goto pmu_register_err;
+
+	/*
+	 * Set archdata.priv value to nvdimm_pmu structure, to handle the
+	 * unregistering of pmu device.
+	 */
+	p->pdev->archdata.priv = nd_pmu;
+	return;
+
+pmu_register_err:
+	kfree(p->nvdimm_events_map);
+pmu_check_events_err:
+	kfree(nd_pmu);
+pmu_err_print:
+	dev_info(&p->pdev->dev, "nvdimm pmu didn't register rc=%d\n", rc);
+}
+
 /*
  * Issue hcall to retrieve dimm health info and populate papr_scm_priv with the
  * health information.
@@ -1236,6 +1454,7 @@ static int papr_scm_probe(struct platform_device *pdev)
 		goto err2;
 
 	platform_set_drvdata(pdev, p);
+	papr_scm_pmu_register(p);
 
 	return 0;
 
@@ -1254,6 +1473,12 @@ static int papr_scm_remove(struct platform_device *pdev)
 
 	nvdimm_bus_unregister(p->bus);
 	drc_pmem_unbind(p);
+
+	if (pdev->archdata.priv)
+		unregister_nvdimm_pmu(pdev->archdata.priv);
+
+	pdev->archdata.priv = NULL;
+	kfree(p->nvdimm_events_map);
 	kfree(p->bus_desc.provider_name);
 	kfree(p);