s390/perf: add support for the CPU-Measurement Sampling Facility

Introduce a perf PMU, "cpum_sf", to support the CPU-Measurement
Sampling Facility.  You can control the sampling facility through
this perf PMU interfaces.  Perf sampling events are created for
hardware samples.

For details about the CPU-Measurement Sampling Facility, see
"The Load-Program-Parameter and the CPU-Measurement Facilities" (SA23-2260).
Signed-off-by: Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

arch/s390/include/asm/cpu_mf.h

@@ -210,6 +210,20 @@ static inline int lsctl(struct hws_lsctl_request_block *req)
 
 /* Sampling control helper functions */
 
+#include <linux/time.h>
+
+static inline unsigned long freq_to_sample_rate(struct hws_qsi_info_block *qsi,
+						unsigned long freq)
+{
+	return (USEC_PER_SEC / freq) * qsi->cpu_speed;
+}
+
+static inline unsigned long sample_rate_to_freq(struct hws_qsi_info_block *qsi,
+						unsigned long rate)
+{
+	return USEC_PER_SEC * qsi->cpu_speed / rate;
+}
+
 #define SDB_TE_ALERT_REQ_MASK	0x4000000000000000UL
 #define SDB_TE_BUFFER_FULL_MASK 0x8000000000000000UL
 

arch/s390/include/asm/perf_event.h

@@ -15,12 +15,13 @@
 #include <linux/device.h>
 #include <asm/cpu_mf.h>
 
-/* CPU-measurement counter facility */
-#define PERF_CPUM_CF_MAX_CTR		256
-
 /* Per-CPU flags for PMU states */
 #define PMU_F_RESERVED			0x1000
 #define PMU_F_ENABLED			0x2000
+#define PMU_F_IN_USE			0x4000
+#define PMU_F_ERR_IBE			0x0100
+#define PMU_F_ERR_LSDA			0x0200
+#define PMU_F_ERR_MASK			(PMU_F_ERR_IBE|PMU_F_ERR_LSDA)
 
 /* Perf defintions for PMU event attributes in sysfs */
 extern __init const struct attribute_group **cpumf_cf_event_group(void);
@@ -41,5 +42,15 @@ extern unsigned long perf_instruction_pointer(struct pt_regs *regs);
 extern unsigned long perf_misc_flags(struct pt_regs *regs);
 #define perf_misc_flags(regs) perf_misc_flags(regs)
 
+/* Perf PMU definitions for the counter facility */
+#define PERF_CPUM_CF_MAX_CTR		256
+
+/* Perf PMU definitions for the sampling facility */
+#define PERF_CPUM_SF_MAX_CTR		1
+#define PERF_EVENT_CPUM_SF		0xB0000UL	/* Raw event ID */
+
+#define TEAR_REG(hwc)		((hwc)->last_tag)
+#define SAMPL_RATE(hwc)		((hwc)->event_base)
+
 #endif /* CONFIG_64BIT */
 #endif /* _ASM_S390_PERF_EVENT_H */

arch/s390/kernel/Makefile

@@ -60,7 +60,7 @@ obj-$(CONFIG_FTRACE_SYSCALLS)  += ftrace.o
 obj-$(CONFIG_CRASH_DUMP)	+= crash_dump.o
 
 ifdef CONFIG_64BIT
-obj-$(CONFIG_PERF_EVENTS)	+= perf_event.o perf_cpum_cf.o \
+obj-$(CONFIG_PERF_EVENTS)	+= perf_event.o perf_cpum_cf.o perf_cpum_sf.o \
 						perf_cpum_cf_events.o
 obj-y				+= runtime_instr.o cache.o
 endif

arch/s390/kernel/perf_event.c

@@ -16,6 +16,7 @@
 #include <linux/kvm_host.h>
 #include <linux/percpu.h>
 #include <linux/export.h>
+#include <linux/spinlock.h>
 #include <linux/sysfs.h>
 #include <asm/irq.h>
 #include <asm/cpu_mf.h>
@@ -36,6 +37,8 @@ int perf_num_counters(void)
 
 	if (cpum_cf_avail())
 		num += PERF_CPUM_CF_MAX_CTR;
+	if (cpum_sf_avail())
+		num += PERF_CPUM_SF_MAX_CTR;
 
 	return num;
 }
@@ -93,24 +96,45 @@ unsigned long perf_misc_flags(struct pt_regs *regs)
 			       : PERF_RECORD_MISC_KERNEL;
 }
 
-void perf_event_print_debug(void)
+void print_debug_cf(void)
 {
 	struct cpumf_ctr_info cf_info;
-	unsigned long flags;
-	int cpu;
-
-	if (!cpum_cf_avail())
-		return;
-
-	local_irq_save(flags);
+	int cpu = smp_processor_id();
 
-	cpu = smp_processor_id();
 	memset(&cf_info, 0, sizeof(cf_info));
 	if (!qctri(&cf_info))
 		pr_info("CPU[%i] CPUM_CF: ver=%u.%u A=%04x E=%04x C=%04x\n",
 			cpu, cf_info.cfvn, cf_info.csvn,
 			cf_info.auth_ctl, cf_info.enable_ctl, cf_info.act_ctl);
+}
+
+static void print_debug_sf(void)
+{
+	struct hws_qsi_info_block si;
+	int cpu = smp_processor_id();
+
+	memset(&si, 0, sizeof(si));
+	if (qsi(&si)) {
+		pr_err("CPU[%i]: CPM_SF: qsi failed\n");
+		return;
+	}
+
+	pr_info("CPU[%i]: CPM_SF: as=%i es=%i cs=%i bsdes=%i dsdes=%i"
+		" min=%i max=%i cpu_speed=%i tear=%p dear=%p\n",
+		cpu, si.as, si.es, si.cs, si.bsdes, si.dsdes,
+		si.min_sampl_rate, si.max_sampl_rate, si.cpu_speed,
+		si.tear, si.dear);
+}
+
+void perf_event_print_debug(void)
+{
+	unsigned long flags;
 
+	local_irq_save(flags);
+	if (cpum_cf_avail())
+		print_debug_cf();
+	if (cpum_sf_avail())
+		print_debug_sf();
 	local_irq_restore(flags);
 }