perf tools: Refactor cpumap to hold nr and the map

So that later, we can pass the cpu_map instance instead of (nr_cpus, cpu_map)
for things like perf_evsel__open and friends.

Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

5 files changed, 138 insertions, 67 deletions

diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 052de1780f76..220e6e7f0b97 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -39,7 +39,7 @@ static u64			user_interval			= ULLONG_MAX;
 static u64                      default_interval                =      0;
 static u64                      sample_type;
 
-static int                      nr_cpus                         =      0;
+static struct cpu_map           *cpus;
 static unsigned int             page_size;
 static unsigned int             mmap_pages                      =    128;
 static unsigned int             user_freq                       = UINT_MAX;
@@ -670,8 +670,8 @@ static int __cmd_record(int argc, const char **argv)
         if (!system_wide && no_inherit && !cpu_list) {
                 open_counters(-1);
         } else {
-                for (i = 0; i < nr_cpus; i++)
-                        open_counters(cpumap[i]);
+                for (i = 0; i < cpus->nr; i++)
+                        open_counters(cpus->map[i]);
         }
 
         perf_session__set_sample_type(session, sample_type);
@@ -927,14 +927,14 @@ int cmd_record(int argc, const char **argv, const char *prefix __used)
                 thread_num = 1;
         }
 
-        nr_cpus = read_cpu_map(cpu_list);
-        if (nr_cpus < 1) {
-                perror("failed to collect number of CPUs");
+        cpus = cpu_map__new(cpu_list);
+        if (cpus == NULL) {
+                perror("failed to parse CPUs map");
                 return -1;
         }
 
         list_for_each_entry(pos, &evsel_list, node) {
-                if (perf_evsel__alloc_fd(pos, nr_cpus, thread_num) < 0)
+                if (perf_evsel__alloc_fd(pos, cpus->nr, thread_num) < 0)
                         goto out_free_fd;
         }
         event_array = malloc(
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 065e79eb2142..3f4a431fb5a4 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -72,7 +72,7 @@ static struct perf_event_attr default_attrs[] = {
 };
 
 static bool                     system_wide                     =  false;
-static int                      nr_cpus                         =  0;
+static struct cpu_map           *cpus;
 static int                      run_idx                         =  0;
 
 static int                      run_count                       =  1;
@@ -167,7 +167,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel)
                                     PERF_FORMAT_TOTAL_TIME_RUNNING;
 
         if (system_wide)
-                return perf_evsel__open_per_cpu(evsel, nr_cpus, cpumap);
+                return perf_evsel__open_per_cpu(evsel, cpus->nr, cpus->map);
 
         attr->inherit = !no_inherit;
         if (target_pid == -1 && target_tid == -1) {
@@ -200,7 +200,7 @@ static int read_counter_aggr(struct perf_evsel *counter)
         u64 *count = counter->counts->aggr.values;
         int i;
 
-        if (__perf_evsel__read(counter, nr_cpus, thread_num, scale) < 0)
+        if (__perf_evsel__read(counter, cpus->nr, thread_num, scale) < 0)
                 return -1;
 
         for (i = 0; i < 3; i++)
@@ -233,7 +233,7 @@ static int read_counter(struct perf_evsel *counter)
         u64 *count;
         int cpu;
 
-        for (cpu = 0; cpu < nr_cpus; cpu++) {
+        for (cpu = 0; cpu < cpus->nr; cpu++) {
                 if (__perf_evsel__read_on_cpu(counter, cpu, 0, scale) < 0)
                         return -1;
 
@@ -259,9 +259,6 @@ static int run_perf_stat(int argc __used, const char **argv)
         const bool forks = (argc > 0);
         char buf;
 
-        if (!system_wide)
-                nr_cpus = 1;
-
         if (forks && (pipe(child_ready_pipe) < 0 || pipe(go_pipe) < 0)) {
                 perror("failed to create pipes");
                 exit(1);
@@ -351,12 +348,12 @@ static int run_perf_stat(int argc __used, const char **argv)
         if (no_aggr) {
                 list_for_each_entry(counter, &evsel_list, node) {
                         read_counter(counter);
-                        perf_evsel__close_fd(counter, nr_cpus, 1);
+                        perf_evsel__close_fd(counter, cpus->nr, 1);
                 }
         } else {
                 list_for_each_entry(counter, &evsel_list, node) {
                         read_counter_aggr(counter);
-                        perf_evsel__close_fd(counter, nr_cpus, thread_num);
+                        perf_evsel__close_fd(counter, cpus->nr, thread_num);
                 }
         }
 
@@ -384,7 +381,7 @@ static void nsec_printout(int cpu, struct perf_evsel *evsel, double avg)
         if (no_aggr)
                 sprintf(cpustr, "CPU%*d%s",
                         csv_output ? 0 : -4,
-                        cpumap[cpu], csv_sep);
+                        cpus->map[cpu], csv_sep);
 
         fprintf(stderr, fmt, cpustr, msecs, csv_sep, event_name(evsel));
 
@@ -412,7 +409,7 @@ static void abs_printout(int cpu, struct perf_evsel *evsel, double avg)
         if (no_aggr)
                 sprintf(cpustr, "CPU%*d%s",
                         csv_output ? 0 : -4,
-                        cpumap[cpu], csv_sep);
+                        cpus->map[cpu], csv_sep);
         else
                 cpu = 0;
 
@@ -498,14 +495,14 @@ static void print_counter(struct perf_evsel *counter)
         u64 ena, run, val;
         int cpu;
 
-        for (cpu = 0; cpu < nr_cpus; cpu++) {
+        for (cpu = 0; cpu < cpus->nr; cpu++) {
                 val = counter->counts->cpu[cpu].val;
                 ena = counter->counts->cpu[cpu].ena;
                 run = counter->counts->cpu[cpu].run;
                 if (run == 0 || ena == 0) {
                         fprintf(stderr, "CPU%*d%s%*s%s%-24s",
                                 csv_output ? 0 : -4,
-                                cpumap[cpu], csv_sep,
+                                cpus->map[cpu], csv_sep,
                                 csv_output ? 0 : 18,
                                 "<not counted>", csv_sep,
                                 event_name(counter));
@@ -697,12 +694,15 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used)
         }
 
         if (system_wide)
-                nr_cpus = read_cpu_map(cpu_list);
+                cpus = cpu_map__new(cpu_list);
         else
-                nr_cpus = 1;
+                cpus = cpu_map__dummy_new();
 
-        if (nr_cpus < 1)
+        if (cpus == NULL) {
+                perror("failed to parse CPUs map");
                 usage_with_options(stat_usage, options);
+                return -1;
+        }
 
         if (target_pid != -1) {
                 target_tid = target_pid;
@@ -723,8 +723,8 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used)
 
         list_for_each_entry(pos, &evsel_list, node) {
                 if (perf_evsel__alloc_stat_priv(pos) < 0 ||
-                    perf_evsel__alloc_counts(pos, nr_cpus) < 0 ||
-                    perf_evsel__alloc_fd(pos, nr_cpus, thread_num) < 0)
+                    perf_evsel__alloc_counts(pos, cpus->nr) < 0 ||
+                    perf_evsel__alloc_fd(pos, cpus->nr, thread_num) < 0)
                         goto out_free_fd;
         }
 
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 27b9c14a0a07..0e426665716d 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -71,7 +71,7 @@ static int			target_tid			=     -1;
 static pid_t                    *all_tids                       =      NULL;
 static int                      thread_num                      =      0;
 static bool                     inherit                         =  false;
-static int                      nr_cpus                         =      0;
+static struct cpu_map           *cpus;
 static int                      realtime_prio                   =      0;
 static bool                     group                           =  false;
 static unsigned int             page_size;
@@ -564,12 +564,12 @@ static void print_sym_table(void)
                 printf(" (all");
 
         if (cpu_list)
-                printf(", CPU%s: %s)\n", nr_cpus > 1 ? "s" : "", cpu_list);
+                printf(", CPU%s: %s)\n", cpus->nr > 1 ? "s" : "", cpu_list);
         else {
                 if (target_tid != -1)
                         printf(")\n");
                 else
-                        printf(", %d CPU%s)\n", nr_cpus, nr_cpus > 1 ? "s" : "");
+                        printf(", %d CPU%s)\n", cpus->nr, cpus->nr > 1 ? "s" : "");
         }
 
         printf("%-*.*s\n", win_width, win_width, graph_dotted_line);
@@ -1197,7 +1197,7 @@ static void perf_session__mmap_read(struct perf_session *self)
         struct perf_evsel *counter;
         int i, thread_index;
 
-        for (i = 0; i < nr_cpus; i++) {
+        for (i = 0; i < cpus->nr; i++) {
                 list_for_each_entry(counter, &evsel_list, node) {
                         for (thread_index = 0;
                                 thread_index < thread_num;
@@ -1221,7 +1221,7 @@ static void start_counter(int i, struct perf_evsel *evsel)
         int thread_index;
 
         if (target_tid == -1)
-                cpu = cpumap[i];
+                cpu = cpus->map[i];
 
         attr = &evsel->attr;
 
@@ -1310,7 +1310,7 @@ static int __cmd_top(void)
         else
                 event__synthesize_threads(event__process, session);
 
-        for (i = 0; i < nr_cpus; i++) {
+        for (i = 0; i < cpus->nr; i++) {
                 group_fd = -1;
                 list_for_each_entry(counter, &evsel_list, node)
                         start_counter(i, counter);
@@ -1460,16 +1460,16 @@ int cmd_top(int argc, const char **argv, const char *prefix __used)
         }
 
         if (target_tid != -1)
-                nr_cpus = 1;
+                cpus = cpu_map__dummy_new();
         else
-                nr_cpus = read_cpu_map(cpu_list);
+                cpus = cpu_map__new(cpu_list);
 
-        if (nr_cpus < 1)
+        if (cpus == NULL)
                 usage_with_options(top_usage, options);
 
         list_for_each_entry(pos, &evsel_list, node) {
-                if (perf_evsel__alloc_mmap_per_thread(pos, nr_cpus, thread_num) < 0 ||
-                    perf_evsel__alloc_fd(pos, nr_cpus, thread_num) < 0)
+                if (perf_evsel__alloc_mmap_per_thread(pos, cpus->nr, thread_num) < 0 ||
+                    perf_evsel__alloc_fd(pos, cpus->nr, thread_num) < 0)
                         goto out_free_fd;
                 /*
                  * Fill in the ones not specifically initialized via -c:
diff --git a/tools/perf/util/cpumap.c b/tools/perf/util/cpumap.c
index 0f9b8d7a7d7e..3ccaa1043383 100644
--- a/tools/perf/util/cpumap.c
+++ b/tools/perf/util/cpumap.c
@@ -4,32 +4,53 @@
 #include <assert.h>
 #include <stdio.h>
 
-int cpumap[MAX_NR_CPUS];
-
-static int default_cpu_map(void)
+static struct cpu_map *cpu_map__default_new(void)
 {
-        int nr_cpus, i;
+        struct cpu_map *cpus;
+        int nr_cpus;
 
         nr_cpus = sysconf(_SC_NPROCESSORS_ONLN);
-        assert(nr_cpus <= MAX_NR_CPUS);
-        assert((int)nr_cpus >= 0);
+        if (nr_cpus < 0)
+                return NULL;
+
+        cpus = malloc(sizeof(*cpus) + nr_cpus * sizeof(int));
+        if (cpus != NULL) {
+                int i;
+                for (i = 0; i < nr_cpus; ++i)
+                        cpus->map[i] = i;
 
-        for (i = 0; i < nr_cpus; ++i)
-                cpumap[i] = i;
+                cpus->nr = nr_cpus;
+        }
 
-        return nr_cpus;
+        return cpus;
 }
 
-static int read_all_cpu_map(void)
+static struct cpu_map *cpu_map__trim_new(int nr_cpus, int *tmp_cpus)
 {
+        size_t payload_size = nr_cpus * sizeof(int);
+        struct cpu_map *cpus = malloc(sizeof(*cpus) + payload_size);
+
+        if (cpus != NULL) {
+                cpus->nr = nr_cpus;
+                memcpy(cpus->map, tmp_cpus, payload_size);
+        }
+
+        return cpus;
+}
+
+static struct cpu_map *cpu_map__read_all_cpu_map(void)
+{
+        struct cpu_map *cpus = NULL;
         FILE *onlnf;
         int nr_cpus = 0;
+        int *tmp_cpus = NULL, *tmp;
+        int max_entries = 0;
         int n, cpu, prev;
         char sep;
 
         onlnf = fopen("/sys/devices/system/cpu/online", "r");
         if (!onlnf)
-                return default_cpu_map();
+                return cpu_map__default_new();
 
         sep = 0;
         prev = -1;
@@ -38,12 +59,28 @@ static int read_all_cpu_map(void)
                 if (n <= 0)
                         break;
                 if (prev >= 0) {
-                        assert(nr_cpus + cpu - prev - 1 < MAX_NR_CPUS);
+                        int new_max = nr_cpus + cpu - prev - 1;
+
+                        if (new_max >= max_entries) {
+                                max_entries = new_max + MAX_NR_CPUS / 2;
+                                tmp = realloc(tmp_cpus, max_entries * sizeof(int));
+                                if (tmp == NULL)
+                                        goto out_free_tmp;
+                                tmp_cpus = tmp;
+                        }
+
                         while (++prev < cpu)
-                                cpumap[nr_cpus++] = prev;
+                                tmp_cpus[nr_cpus++] = prev;
+                }
+                if (nr_cpus == max_entries) {
+                        max_entries += MAX_NR_CPUS;
+                        tmp = realloc(tmp_cpus, max_entries * sizeof(int));
+                        if (tmp == NULL)
+                                goto out_free_tmp;
+                        tmp_cpus = tmp;
                 }
-                assert (nr_cpus < MAX_NR_CPUS);
-                cpumap[nr_cpus++] = cpu;
+
+                tmp_cpus[nr_cpus++] = cpu;
                 if (n == 2 && sep == '-')
                         prev = cpu;
                 else
@@ -51,24 +88,31 @@ static int read_all_cpu_map(void)
                 if (n == 1 || sep == '\n')
                         break;
         }
-        fclose(onlnf);
-        if (nr_cpus > 0)
-                return nr_cpus;
 
-        return default_cpu_map();
+        if (nr_cpus > 0)
+                cpus = cpu_map__trim_new(nr_cpus, tmp_cpus);
+        else
+                cpus = cpu_map__default_new();
+out_free_tmp:
+        free(tmp_cpus);
+        fclose(onlnf);
+        return cpus;
 }
 
-int read_cpu_map(const char *cpu_list)
+struct cpu_map *cpu_map__new(const char *cpu_list)
 {
+        struct cpu_map *cpus = NULL;
         unsigned long start_cpu, end_cpu = 0;
         char *p = NULL;
         int i, nr_cpus = 0;
+        int *tmp_cpus = NULL, *tmp;
+        int max_entries = 0;
 
         if (!cpu_list)
-                return read_all_cpu_map();
+                return cpu_map__read_all_cpu_map();
 
         if (!isdigit(*cpu_list))
-                goto invalid;
+                goto out;
 
         while (isdigit(*cpu_list)) {
                 p = NULL;
@@ -94,21 +138,42 @@ int read_cpu_map(const char *cpu_list)
                 for (; start_cpu <= end_cpu; start_cpu++) {
                         /* check for duplicates */
                         for (i = 0; i < nr_cpus; i++)
-                                if (cpumap[i] == (int)start_cpu)
+                                if (tmp_cpus[i] == (int)start_cpu)
                                         goto invalid;
 
-                        assert(nr_cpus < MAX_NR_CPUS);
-                        cpumap[nr_cpus++] = (int)start_cpu;
+                        if (nr_cpus == max_entries) {
+                                max_entries += MAX_NR_CPUS;
+                                tmp = realloc(tmp_cpus, max_entries * sizeof(int));
+                                if (tmp == NULL)
+                                        goto invalid;
+                                tmp_cpus = tmp;
+                        }
+                        tmp_cpus[nr_cpus++] = (int)start_cpu;
                 }
                 if (*p)
                         ++p;
 
                 cpu_list = p;
         }
-        if (nr_cpus > 0)
-                return nr_cpus;
 
-        return default_cpu_map();
+        if (nr_cpus > 0)
+                cpus = cpu_map__trim_new(nr_cpus, tmp_cpus);
+        else
+                cpus = cpu_map__default_new();
 invalid:
-        return -1;
+        free(tmp_cpus);
+out:
+        return cpus;
+}
+
+struct cpu_map *cpu_map__dummy_new(void)
+{
+        struct cpu_map *cpus = malloc(sizeof(*cpus) + sizeof(int));
+
+        if (cpus != NULL) {
+                cpus->nr = 1;
+                cpus->map[0] = -1;
+        }
+
+        return cpus;
 }
diff --git a/tools/perf/util/cpumap.h b/tools/perf/util/cpumap.h
index 3e60f56e490e..f7a4f42f6307 100644
--- a/tools/perf/util/cpumap.h
+++ b/tools/perf/util/cpumap.h
@@ -1,7 +1,13 @@
 #ifndef __PERF_CPUMAP_H
 #define __PERF_CPUMAP_H
 
-extern int read_cpu_map(const char *cpu_list);
-extern int cpumap[];
+struct cpu_map {
+        int nr;
+        int map[];
+};
+
+struct cpu_map *cpu_map__new(const char *cpu_list);
+struct cpu_map *cpu_map__dummy_new(void);
+void *cpu_map__delete(struct cpu_map *map);
 
 #endif /* __PERF_CPUMAP_H */