1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
|
/*
* Performance event support for s390x
*
* Copyright IBM Corp. 2012, 2013
* Author(s): Hendrik Brueckner <brueckner@linux.vnet.ibm.com>
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License (version 2 only)
* as published by the Free Software Foundation.
*/
#define KMSG_COMPONENT "perf"
#define pr_fmt(fmt) KMSG_COMPONENT ": " fmt
#include <linux/kernel.h>
#include <linux/perf_event.h>
#include <linux/kvm_host.h>
#include <linux/percpu.h>
#include <linux/export.h>
#include <linux/seq_file.h>
#include <linux/spinlock.h>
#include <linux/sysfs.h>
#include <asm/irq.h>
#include <asm/cpu_mf.h>
#include <asm/lowcore.h>
#include <asm/processor.h>
#include <asm/sysinfo.h>
const char *perf_pmu_name(void)
{
if (cpum_cf_avail() || cpum_sf_avail())
return "CPU-Measurement Facilities (CPU-MF)";
return "pmu";
}
EXPORT_SYMBOL(perf_pmu_name);
int perf_num_counters(void)
{
int num = 0;
if (cpum_cf_avail())
num += PERF_CPUM_CF_MAX_CTR;
if (cpum_sf_avail())
num += PERF_CPUM_SF_MAX_CTR;
return num;
}
EXPORT_SYMBOL(perf_num_counters);
static struct kvm_s390_sie_block *sie_block(struct pt_regs *regs)
{
struct stack_frame *stack = (struct stack_frame *) regs->gprs[15];
if (!stack)
return NULL;
return (struct kvm_s390_sie_block *) stack->empty1[0];
}
static bool is_in_guest(struct pt_regs *regs)
{
if (user_mode(regs))
return false;
#if IS_ENABLED(CONFIG_KVM)
return instruction_pointer(regs) == (unsigned long) &sie_exit;
#else
return false;
#endif
}
static unsigned long guest_is_user_mode(struct pt_regs *regs)
{
return sie_block(regs)->gpsw.mask & PSW_MASK_PSTATE;
}
static unsigned long instruction_pointer_guest(struct pt_regs *regs)
{
return sie_block(regs)->gpsw.addr & PSW_ADDR_INSN;
}
unsigned long perf_instruction_pointer(struct pt_regs *regs)
{
return is_in_guest(regs) ? instruction_pointer_guest(regs)
: instruction_pointer(regs);
}
static unsigned long perf_misc_guest_flags(struct pt_regs *regs)
{
return guest_is_user_mode(regs) ? PERF_RECORD_MISC_GUEST_USER
: PERF_RECORD_MISC_GUEST_KERNEL;
}
static unsigned long perf_misc_flags_sf(struct pt_regs *regs)
{
struct perf_sf_sde_regs *sde_regs;
unsigned long flags;
sde_regs = (struct perf_sf_sde_regs *) ®s->int_parm_long;
if (sde_regs->in_guest)
flags = user_mode(regs) ? PERF_RECORD_MISC_GUEST_USER
: PERF_RECORD_MISC_GUEST_KERNEL;
else
flags = user_mode(regs) ? PERF_RECORD_MISC_USER
: PERF_RECORD_MISC_KERNEL;
return flags;
}
unsigned long perf_misc_flags(struct pt_regs *regs)
{
/* Check if the cpum_sf PMU has created the pt_regs structure.
* In this case, perf misc flags can be easily extracted. Otherwise,
* do regular checks on the pt_regs content.
*/
if (regs->int_code == 0x1407 && regs->int_parm == CPU_MF_INT_SF_PRA)
if (!regs->gprs[15])
return perf_misc_flags_sf(regs);
if (is_in_guest(regs))
return perf_misc_guest_flags(regs);
return user_mode(regs) ? PERF_RECORD_MISC_USER
: PERF_RECORD_MISC_KERNEL;
}
static void print_debug_cf(void)
{
struct cpumf_ctr_info cf_info;
int 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))
return;
pr_info("CPU[%i] CPUM_SF: basic=%i diag=%i min=%lu max=%lu cpu_speed=%u\n",
cpu, si.as, si.ad, si.min_sampl_rate, si.max_sampl_rate,
si.cpu_speed);
if (si.as)
pr_info("CPU[%i] CPUM_SF: Basic-sampling: a=%i e=%i c=%i"
" bsdes=%i tear=%016lx dear=%016lx\n", cpu,
si.as, si.es, si.cs, si.bsdes, si.tear, si.dear);
if (si.ad)
pr_info("CPU[%i] CPUM_SF: Diagnostic-sampling: a=%i e=%i c=%i"
" dsdes=%i tear=%016lx dear=%016lx\n", cpu,
si.ad, si.ed, si.cd, si.dsdes, 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);
}
/* Service level infrastructure */
static void sl_print_counter(struct seq_file *m)
{
struct cpumf_ctr_info ci;
memset(&ci, 0, sizeof(ci));
if (qctri(&ci))
return;
seq_printf(m, "CPU-MF: Counter facility: version=%u.%u "
"authorization=%04x\n", ci.cfvn, ci.csvn, ci.auth_ctl);
}
static void sl_print_sampling(struct seq_file *m)
{
struct hws_qsi_info_block si;
memset(&si, 0, sizeof(si));
if (qsi(&si))
return;
if (!si.as && !si.ad)
return;
seq_printf(m, "CPU-MF: Sampling facility: min_rate=%lu max_rate=%lu"
" cpu_speed=%u\n", si.min_sampl_rate, si.max_sampl_rate,
si.cpu_speed);
if (si.as)
seq_printf(m, "CPU-MF: Sampling facility: mode=basic"
" sample_size=%u\n", si.bsdes);
if (si.ad)
seq_printf(m, "CPU-MF: Sampling facility: mode=diagnostic"
" sample_size=%u\n", si.dsdes);
}
static void service_level_perf_print(struct seq_file *m,
struct service_level *sl)
{
if (cpum_cf_avail())
sl_print_counter(m);
if (cpum_sf_avail())
sl_print_sampling(m);
}
static struct service_level service_level_perf = {
.seq_print = service_level_perf_print,
};
static int __init service_level_perf_register(void)
{
return register_service_level(&service_level_perf);
}
arch_initcall(service_level_perf_register);
/* See also arch/s390/kernel/traps.c */
static unsigned long __store_trace(struct perf_callchain_entry *entry,
unsigned long sp,
unsigned long low, unsigned long high)
{
struct stack_frame *sf;
struct pt_regs *regs;
while (1) {
sp = sp & PSW_ADDR_INSN;
if (sp < low || sp > high - sizeof(*sf))
return sp;
sf = (struct stack_frame *) sp;
perf_callchain_store(entry, sf->gprs[8] & PSW_ADDR_INSN);
/* Follow the backchain. */
while (1) {
low = sp;
sp = sf->back_chain & PSW_ADDR_INSN;
if (!sp)
break;
if (sp <= low || sp > high - sizeof(*sf))
return sp;
sf = (struct stack_frame *) sp;
perf_callchain_store(entry,
sf->gprs[8] & PSW_ADDR_INSN);
}
/* Zero backchain detected, check for interrupt frame. */
sp = (unsigned long) (sf + 1);
if (sp <= low || sp > high - sizeof(*regs))
return sp;
regs = (struct pt_regs *) sp;
perf_callchain_store(entry, sf->gprs[8] & PSW_ADDR_INSN);
low = sp;
sp = regs->gprs[15];
}
}
void perf_callchain_kernel(struct perf_callchain_entry *entry,
struct pt_regs *regs)
{
unsigned long head;
struct stack_frame *head_sf;
if (user_mode(regs))
return;
head = regs->gprs[15];
head_sf = (struct stack_frame *) head;
if (!head_sf || !head_sf->back_chain)
return;
head = head_sf->back_chain;
head = __store_trace(entry, head, S390_lowcore
|