13 files changed, 98 insertions, 74 deletions

diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c
index a931b133f3ac..795f865c1366 100644
--- a/tools/perf/builtin-annotate.c
+++ b/tools/perf/builtin-annotate.c
@@ -131,14 +131,14 @@ static int hist_entry__add(struct addr_location *al, u64 count)
         return 0;
 }
 
-static int process_sample_event(event_t *event, struct perf_session *session __used)
+static int process_sample_event(event_t *event, struct perf_session *session)
 {
         struct addr_location al;
 
         dump_printf("(IP, %d): %d: %p\n", event->header.misc,
                     event->ip.pid, (void *)(long)event->ip.ip);
 
-        if (event__preprocess_sample(event, &al, symbol_filter) < 0) {
+        if (event__preprocess_sample(event, session, &al, symbol_filter) < 0) {
                 fprintf(stderr, "problem processing %d event, skipping it.\n",
                         event->header.type);
                 return -1;
@@ -479,7 +479,7 @@ static int __cmd_annotate(void)
         }
 
         if (verbose > 3)
-                threads__fprintf(stdout);
+                perf_session__fprintf(session, stdout);
 
         if (verbose > 2)
                 dsos__fprintf(stdout);
diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c
index 237155fa756b..de194958fe6e 100644
--- a/tools/perf/builtin-kmem.c
+++ b/tools/perf/builtin-kmem.c
@@ -311,7 +311,7 @@ process_raw_event(event_t *raw_event __used, void *data,
         }
 }
 
-static int process_sample_event(event_t *event, struct perf_session *session __used)
+static int process_sample_event(event_t *event, struct perf_session *session)
 {
         struct sample_data data;
         struct thread *thread;
@@ -329,7 +329,7 @@ static int process_sample_event(event_t *event, struct perf_session *session __u
                 (void *)(long)data.ip,
                 (long long)data.period);
 
-        thread = threads__findnew(event->ip.pid);
+        thread = perf_session__findnew(session, event->ip.pid);
         if (thread == NULL) {
                 pr_debug("problem processing %d event, skipping it.\n",
                          event->header.type);
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 26b947860948..efa8147b8991 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -600,7 +600,7 @@ static int validate_chain(struct ip_callchain *chain, event_t *event)
         return 0;
 }
 
-static int process_sample_event(event_t *event, struct perf_session *session __used)
+static int process_sample_event(event_t *event, struct perf_session *session)
 {
         struct sample_data data;
         int cpumode;
@@ -636,7 +636,7 @@ static int process_sample_event(event_t *event, struct perf_session *session __u
                 }
         }
 
-        thread = threads__findnew(data.pid);
+        thread = perf_session__findnew(session, data.pid);
         if (thread == NULL) {
                 pr_debug("problem processing %d event, skipping it.\n",
                         event->header.type);
@@ -679,9 +679,9 @@ static int process_sample_event(event_t *event, struct perf_session *session __u
         return 0;
 }
 
-static int process_comm_event(event_t *event, struct perf_session *session __used)
+static int process_comm_event(event_t *event, struct perf_session *session)
 {
-        struct thread *thread = threads__findnew(event->comm.pid);
+        struct thread *thread = perf_session__findnew(session, event->comm.pid);
 
         dump_printf(": %s:%d\n", event->comm.comm, event->comm.pid);
 
@@ -780,7 +780,7 @@ static int __cmd_report(void)
         }
 
         if (verbose > 3)
-                threads__fprintf(stdout);
+                perf_session__fprintf(session, stdout);
 
         if (verbose > 2)
                 dsos__fprintf(stdout);
diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
index 1e4e508339a8..8d58d9e07a7b 100644
--- a/tools/perf/builtin-sched.c
+++ b/tools/perf/builtin-sched.c
@@ -730,18 +730,21 @@ struct trace_migrate_task_event {
 
 struct trace_sched_handler {
         void (*switch_event)(struct trace_switch_event *,
+                             struct perf_session *,
                              struct event *,
                              int cpu,
                              u64 timestamp,
                              struct thread *thread);
 
         void (*runtime_event)(struct trace_runtime_event *,
+                              struct perf_session *,
                               struct event *,
                               int cpu,
                               u64 timestamp,
                               struct thread *thread);
 
         void (*wakeup_event)(struct trace_wakeup_event *,
+                             struct perf_session *,
                              struct event *,
                              int cpu,
                              u64 timestamp,
@@ -754,6 +757,7 @@ struct trace_sched_handler {
                            struct thread *thread);
 
         void (*migrate_task_event)(struct trace_migrate_task_event *,
+                           struct perf_session *session,
                            struct event *,
                            int cpu,
                            u64 timestamp,
@@ -763,6 +767,7 @@ struct trace_sched_handler {
 
 static void
 replay_wakeup_event(struct trace_wakeup_event *wakeup_event,
+                    struct perf_session *session __used,
                     struct event *event,
                     int cpu __used,
                     u64 timestamp __used,
@@ -789,6 +794,7 @@ static u64 cpu_last_switched[MAX_CPUS];
 
 static void
 replay_switch_event(struct trace_switch_event *switch_event,
+                    struct perf_session *session __used,
                     struct event *event,
                     int cpu,
                     u64 timestamp,
@@ -1022,6 +1028,7 @@ add_sched_in_event(struct work_atoms *atoms, u64 timestamp)
 
 static void
 latency_switch_event(struct trace_switch_event *switch_event,
+                     struct perf_session *session,
                      struct event *event __used,
                      int cpu,
                      u64 timestamp,
@@ -1045,8 +1052,8 @@ latency_switch_event(struct trace_switch_event *switch_event,
                 die("hm, delta: %Ld < 0 ?\n", delta);
 
 
-        sched_out = threads__findnew(switch_event->prev_pid);
-        sched_in = threads__findnew(switch_event->next_pid);
+        sched_out = perf_session__findnew(session, switch_event->prev_pid);
+        sched_in = perf_session__findnew(session, switch_event->next_pid);
 
         out_events = thread_atoms_search(&atom_root, sched_out, &cmp_pid);
         if (!out_events) {
@@ -1074,12 +1081,13 @@ latency_switch_event(struct trace_switch_event *switch_event,
 
 static void
 latency_runtime_event(struct trace_runtime_event *runtime_event,
+                     struct perf_session *session,
                      struct event *event __used,
                      int cpu,
                      u64 timestamp,
                      struct thread *this_thread __used)
 {
-        struct thread *thread = threads__findnew(runtime_event->pid);
+        struct thread *thread = perf_session__findnew(session, runtime_event->pid);
         struct work_atoms *atoms = thread_atoms_search(&atom_root, thread, &cmp_pid);
 
         BUG_ON(cpu >= MAX_CPUS || cpu < 0);
@@ -1096,6 +1104,7 @@ latency_runtime_event(struct trace_runtime_event *runtime_event,
 
 static void
 latency_wakeup_event(struct trace_wakeup_event *wakeup_event,
+                     struct perf_session *session,
                      struct event *__event __used,
                      int cpu __used,
                      u64 timestamp,
@@ -1109,7 +1118,7 @@ latency_wakeup_event(struct trace_wakeup_event *wakeup_event,
         if (!wakeup_event->success)
                 return;
 
-        wakee = threads__findnew(wakeup_event->pid);
+        wakee = perf_session__findnew(session, wakeup_event->pid);
         atoms = thread_atoms_search(&atom_root, wakee, &cmp_pid);
         if (!atoms) {
                 thread_atoms_insert(wakee);
@@ -1143,6 +1152,7 @@ latency_wakeup_event(struct trace_wakeup_event *wakeup_event,
 
 static void
 latency_migrate_task_event(struct trace_migrate_task_event *migrate_task_event,
+                     struct perf_session *session,
                      struct event *__event __used,
                      int cpu __used,
                      u64 timestamp,
@@ -1158,7 +1168,7 @@ latency_migrate_task_event(struct trace_migrate_task_event *migrate_task_event,
         if (profile_cpu == -1)
                 return;
 
-        migrant = threads__findnew(migrate_task_event->pid);
+        migrant = perf_session__findnew(session, migrate_task_event->pid);
         atoms = thread_atoms_search(&atom_root, migrant, &cmp_pid);
         if (!atoms) {
                 thread_atoms_insert(migrant);
@@ -1353,7 +1363,7 @@ static void sort_lat(void)
 static struct trace_sched_handler *trace_handler;
 
 static void
-process_sched_wakeup_event(void *data,
+process_sched_wakeup_event(void *data, struct perf_session *session,
                            struct event *event,
                            int cpu __used,
                            u64 timestamp __used,
@@ -1370,7 +1380,8 @@ process_sched_wakeup_event(void *data,
         FILL_FIELD(wakeup_event, cpu, event, data);
 
         if (trace_handler->wakeup_event)
-                trace_handler->wakeup_event(&wakeup_event, event, cpu, timestamp, thread);
+                trace_handler->wakeup_event(&wakeup_event, session, event,
+                                            cpu, timestamp, thread);
 }
 
 /*
@@ -1388,6 +1399,7 @@ static char next_shortname2 = '0';
 
 static void
 map_switch_event(struct trace_switch_event *switch_event,
+                 struct perf_session *session,
                  struct event *event __used,
                  int this_cpu,
                  u64 timestamp,
@@ -1415,8 +1427,8 @@ map_switch_event(struct trace_switch_event *switch_event,
                 die("hm, delta: %Ld < 0 ?\n", delta);
 
 
-        sched_out = threads__findnew(switch_event->prev_pid);
-        sched_in = threads__findnew(switch_event->next_pid);
+        sched_out = perf_session__findnew(session, switch_event->prev_pid);
+        sched_in = perf_session__findnew(session, switch_event->next_pid);
 
         curr_thread[this_cpu] = sched_in;
 
@@ -1466,7 +1478,7 @@ map_switch_event(struct trace_switch_event *switch_event,
 
 
 static void
-process_sched_switch_event(void *data,
+process_sched_switch_event(void *data, struct perf_session *session,
                            struct event *event,
                            int this_cpu,
                            u64 timestamp __used,
@@ -1493,13 +1505,14 @@ process_sched_switch_event(void *data,
                         nr_context_switch_bugs++;
         }
         if (trace_handler->switch_event)
-                trace_handler->switch_event(&switch_event, event, this_cpu, timestamp, thread);
+                trace_handler->switch_event(&switch_event, session, event,
+                                            this_cpu, timestamp, thread);
 
         curr_pid[this_cpu] = switch_event.next_pid;
 }
 
 static void
-process_sched_runtime_event(void *data,
+process_sched_runtime_event(void *data, struct perf_session *session,
                            struct event *event,
                            int cpu __used,
                            u64 timestamp __used,
@@ -1513,7 +1526,7 @@ process_sched_runtime_event(void *data,
         FILL_FIELD(runtime_event, vruntime, event, data);
 
         if (trace_handler->runtime_event)
-                trace_handler->runtime_event(&runtime_event, event, cpu, timestamp, thread);
+                trace_handler->runtime_event(&runtime_event, session, event, cpu, timestamp, thread);
 }
 
 static void
@@ -1533,7 +1546,8 @@ process_sched_fork_event(void *data,
         FILL_FIELD(fork_event, child_pid, event, data);
 
         if (trace_handler->fork_event)
-                trace_handler->fork_event(&fork_event, event, cpu, timestamp, thread);
+                trace_handler->fork_event(&fork_event, event,
+                                          cpu, timestamp, thread);
 }
 
 static void
@@ -1547,7 +1561,7 @@ process_sched_exit_event(struct event *event,
 }
 
 static void
-process_sched_migrate_task_event(void *data,
+process_sched_migrate_task_event(void *data, struct perf_session *session,
                            struct event *event,
                            int cpu __used,
                            u64 timestamp __used,
@@ -1563,12 +1577,13 @@ process_sched_migrate_task_event(void *data,
         FILL_FIELD(migrate_task_event, cpu, event, data);
 
         if (trace_handler->migrate_task_event)
-                trace_handler->migrate_task_event(&migrate_task_event, event, cpu, timestamp, thread);
+                trace_handler->migrate_task_event(&migrate_task_event, session,
+                                                 event, cpu, timestamp, thread);
 }
 
 static void
-process_raw_event(event_t *raw_event __used, void *data,
-                  int cpu, u64 timestamp, struct thread *thread)
+process_raw_event(event_t *raw_event __used, struct perf_session *session,
+                  void *data, int cpu, u64 timestamp, struct thread *thread)
 {
         struct event *event;
         int type;
@@ -1578,23 +1593,22 @@ process_raw_event(event_t *raw_event __used, void *data,
         event = trace_find_event(type);
 
         if (!strcmp(event->name, "sched_switch"))
-                process_sched_switch_event(data, event, cpu, timestamp, thread);
+                process_sched_switch_event(data, session, event, cpu, timestamp, thread);
         if (!strcmp(event->name, "sched_stat_runtime"))
-                process_sched_runtime_event(data, event, cpu, timestamp, thread);
+                process_sched_runtime_event(data, session, event, cpu, timestamp, thread);
         if (!strcmp(event->name, "sched_wakeup"))
-                process_sched_wakeup_event(data, event, cpu, timestamp, thread);
+                process_sched_wakeup_event(data, session, event, cpu, timestamp, thread);
         if (!strcmp(event->name, "sched_wakeup_new"))
-                process_sched_wakeup_event(data, event, cpu, timestamp, thread);
+                process_sched_wakeup_event(data, session, event, cpu, timestamp, thread);
         if (!strcmp(event->name, "sched_process_fork"))
                 process_sched_fork_event(data, event, cpu, timestamp, thread);
         if (!strcmp(event->name, "sched_process_exit"))
                 process_sched_exit_event(event, cpu, timestamp, thread);
         if (!strcmp(event->name, "sched_migrate_task"))
-                process_sched_migrate_task_event(data, event, cpu, timestamp, thread);
+                process_sched_migrate_task_event(data, session, event, cpu, timestamp, thread);
 }
 
-static int process_sample_event(event_t *event,
-                                struct perf_session *session __used)
+static int process_sample_event(event_t *event, struct perf_session *session)
 {
         struct sample_data data;
         struct thread *thread;
@@ -1615,7 +1629,7 @@ static int process_sample_event(event_t *event,
                 (void *)(long)data.ip,
                 (long long)data.period);
 
-        thread = threads__findnew(data.pid);
+        thread = perf_session__findnew(session, data.pid);
         if (thread == NULL) {
                 pr_debug("problem processing %d event, skipping it.\n",
                          event->header.type);
@@ -1627,7 +1641,7 @@ static int process_sample_event(event_t *event,
         if (profile_cpu != -1 && profile_cpu != (int)data.cpu)
                 return 0;
 
-        process_raw_event(event, data.raw_data, data.cpu, data.time, thread);
+        process_raw_event(event, session, data.raw_data, data.cpu, data.time, thread);
 
         return 0;
 }
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index b13f42625549..0f7a4da2924c 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -20,8 +20,9 @@
 
 #include "perf.h"
 
-#include "util/symbol.h"
 #include "util/color.h"
+#include "util/session.h"
+#include "util/symbol.h"
 #include "util/thread.h"
 #include "util/util.h"
 #include <linux/rbtree.h>
@@ -926,7 +927,8 @@ static int symbol_filter(struct map *map, struct symbol *sym)
         return 0;
 }
 
-static void event__process_sample(const event_t *self, int counter)
+static void event__process_sample(const event_t *self,
+                                 struct perf_session *session, int counter)
 {
         u64 ip = self->ip.ip;
         struct sym_entry *syme;
@@ -946,7 +948,7 @@ static void event__process_sample(const event_t *self, int counter)
                 return;
         }
 
-        if (event__preprocess_sample(self, &al, symbol_filter) < 0 ||
+        if (event__preprocess_sample(self, session, &al, symbol_filter) < 0 ||
             al.sym == NULL)
                 return;
 
@@ -1053,7 +1055,7 @@ static void perf_session__mmap_read_counter(struct perf_session *self,
                 }
 
                 if (event->header.type == PERF_RECORD_SAMPLE)
-                        event__process_sample(event, md->counter);
+                        event__process_sample(event, self, md->counter);
                 else
                         event__process(event, self);
                 old += size;
@@ -1157,10 +1159,13 @@ static int __cmd_top(void)
         int i, counter;
         int ret;
         /*
-         * XXX perf_session__new should allow passing a O_MMAP, so that all this
-         * mmap reading, etc is encapsulated in it.
+         * FIXME: perf_session__new should allow passing a O_MMAP, so that all this
+         * mmap reading, etc is encapsulated in it. Use O_WRONLY for now.
          */
-        struct perf_session *session = NULL;
+        struct perf_session *session = perf_session__new(NULL, O_WRONLY, false);
+
+        if (session == NULL)
+                return -ENOMEM;
 
         if (target_pid != -1)
                 event__synthesize_thread(target_pid, event__process, session);
diff --git a/tools/perf/builtin-trace.c b/tools/perf/builtin-trace.c
index b7eb3fcc224e..d76532375054 100644
--- a/tools/perf/builtin-trace.c
+++ b/tools/perf/builtin-trace.c
@@ -63,7 +63,7 @@ static char const		*input_name = "perf.data";
 
 static u64                      sample_type;
 
-static int process_sample_event(event_t *event, struct perf_session *session __used)
+static int process_sample_event(event_t *event, struct perf_session *session)
 {
         struct sample_data data;
         struct thread *thread;
@@ -81,7 +81,7 @@ static int process_sample_event(event_t *event, struct perf_session *session __u
                 (void *)(long)data.ip,
                 (long long)data.period);
 
-        thread = threads__findnew(event->ip.pid);
+        thread = perf_session__findnew(session, event->ip.pid);
         if (thread == NULL) {
                 pr_debug("problem processing %d event, skipping it.\n",
                          event->header.type);
diff --git a/tools/perf/util/data_map.c b/tools/perf/util/data_map.c
index ba2eb2ce018a..44dea211cc65 100644
--- a/tools/perf/util/data_map.c
+++ b/tools/perf/util/data_map.c
@@ -125,9 +125,9 @@ out:
         return err;
 }
 
-static struct thread *perf_session__register_idle_thread(struct perf_session *self __used)
+static struct thread *perf_session__register_idle_thread(struct perf_session *self)
 {
-        struct thread *thread = threads__findnew(0);
+        struct thread *thread = perf_session__findnew(self, 0);
 
         if (!thread || thread__set_comm(thread, "swapper")) {
                 pr_err("problem inserting idle task.\n");
diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c
index 40d8d842a21f..2d09c29b3a6c 100644
--- a/tools/perf/util/event.c
+++ b/tools/perf/util/event.c
@@ -189,9 +189,9 @@ void event__synthesize_threads(int (*process)(event_t *event,
 
 struct events_stats event__stats;
 
-int event__process_comm(event_t *self, struct perf_session *session __used)
+int event__process_comm(event_t *self, struct perf_session *session)
 {
-        struct thread *thread = threads__findnew(self->comm.pid);
+        struct thread *thread = perf_session__findnew(session, self->comm.pid);
 
         dump_printf(": %s:%d\n", self->comm.comm, self->comm.pid);
 
@@ -212,7 +212,7 @@ int event__process_lost(event_t *self, struct perf_session *session __used)
 
 int event__process_mmap(event_t *self, struct perf_session *session)
 {
-        struct thread *thread = threads__findnew(self->mmap.pid);
+        struct thread *thread = perf_session__findnew(session, self->mmap.pid);
         struct map *map = map__new(&self->mmap, MAP__FUNCTION,
                                    session->cwd, session->cwdlen);
 
@@ -231,10 +231,10 @@ int event__process_mmap(event_t *self, struct perf_session *session)
         return 0;
 }
 
-int event__process_task(event_t *self, struct perf_session *session __used)
+int event__process_task(event_t *self, struct perf_session *session)
 {
-        struct thread *thread = threads__findnew(self->fork.pid);
-        struct thread *parent = threads__findnew(self->fork.ppid);
+        struct thread *thread = perf_session__findnew(session, self->fork.pid);
+        struct thread *parent = perf_session__findnew(session, self->fork.ppid);
 
         dump_printf("(%d:%d):(%d:%d)\n", self->fork.pid, self->fork.tid,
                     self->fork.ppid, self->fork.ptid);
@@ -300,11 +300,11 @@ try_again:
         }
 }
 
-int event__preprocess_sample(const event_t *self, struct addr_location *al,
-                             symbol_filter_t filter)
+int event__preprocess_sample(const event_t *self, struct perf_session *session,
+                             struct addr_location *al, symbol_filter_t filter)
 {
         u8 cpumode = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
-        struct thread *thread = threads__findnew(self->ip.pid);
+        struct thread *thread = perf_session__findnew(session, self->ip.pid);
 
         if (thread == NULL)
                 return -1;
diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h
index 6b6429b63da3..bb090257570e 100644
--- a/tools/perf/util/event.h
+++ b/tools/perf/util/event.h
@@ -177,8 +177,8 @@ int event__process_mmap(event_t *self, struct perf_session *session);
 int event__process_task(event_t *self, struct perf_session *session);
 
 struct addr_location;
-int event__preprocess_sample(const event_t *self, struct addr_location *al,
-                             symbol_filter_t filter);
+int event__preprocess_sample(const event_t *self, struct perf_session *session,
+                             struct addr_location *al, symbol_filter_t filter);
 int event__parse_sample(event_t *event, u64 type, struct sample_data *data);
 
 #endif /* __PERF_RECORD_H */
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index 534a8770ee7f..09836a537fc5 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -51,7 +51,7 @@ out_close:
 struct perf_session *perf_session__new(const char *filename, int mode,
                                        bool force)
 {
-        size_t len = strlen(filename) + 1;
+        size_t len = filename ? strlen(filename) + 1 : 0;
         struct perf_session *self = zalloc(sizeof(*self) + len);
 
         if (self == NULL)
@@ -61,6 +61,8 @@ struct perf_session *perf_session__new(const char *filename, int mode,
                 goto out_delete;
 
         memcpy(self->filename, filename, len);
+        self->threads = RB_ROOT;
+        self->last_match = NULL;
         self->mmap_window = 32;
         self->cwd = NULL;
         self->cwdlen = 0;
diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h
index 1e0da9ca31aa..1dbef7cdd489 100644
--- a/tools/perf/util/session.h
+++ b/tools/perf/util/session.h
@@ -3,11 +3,16 @@
 
 #include "event.h"
 #include "header.h"
+#include <linux/rbtree.h>
+
+struct thread;
 
 struct perf_session {
         struct perf_header      header;
         unsigned long           size;
         unsigned long           mmap_window;
+        struct rb_root          threads;
+        struct thread           *last_match;
         int                     fd;
         int                     cwdlen;
         char                    *cwd;
diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c
index 5c0ab14f3dba..634b7f7140d5 100644
--- a/tools/perf/util/thread.c
+++ b/tools/perf/util/thread.c
@@ -2,13 +2,11 @@
 #include <stdlib.h>
 #include <stdio.h>
 #include <string.h>
+#include "session.h"
 #include "thread.h"
 #include "util.h"
 #include "debug.h"
 
-static struct rb_root threads;
-static struct thread *last_match;
-
 void map_groups__init(struct map_groups *self)
 {
         int i;
@@ -122,9 +120,9 @@ static size_t thread__fprintf(struct thread *self, FILE *fp)
                map_groups__fprintf(&self->mg, fp);
 }
 
-struct thread *threads__findnew(pid_t pid)
+struct thread *perf_session__findnew(struct perf_session *self, pid_t pid)
 {
-        struct rb_node **p = &threads.rb_node;
+        struct rb_node **p = &self->threads.rb_node;
         struct rb_node *parent = NULL;
         struct thread *th;
 
@@ -133,15 +131,15 @@ struct thread *threads__findnew(pid_t pid)
          * so most of the time we dont have to look up
          * the full rbtree:
          */
-        if (last_match && last_match->pid == pid)
-                return last_match;
+        if (self->last_match && self->last_match->pid == pid)
+                return self->last_match;
 
         while (*p != NULL) {
                 parent = *p;
                 th = rb_entry(parent, struct thread, rb_node);
 
                 if (th->pid == pid) {
-                        last_match = th;
+                        self->last_match = th;
                         return th;
                 }
 
@@ -154,8 +152,8 @@ struct thread *threads__findnew(pid_t pid)
         th = thread__new(pid);
         if (th != NULL) {
                 rb_link_node(&th->rb_node, parent, p);
-                rb_insert_color(&th->rb_node, &threads);
-                last_match = th;
+                rb_insert_color(&th->rb_node, &self->threads);
+                self->last_match = th;
         }
 
         return th;
@@ -269,12 +267,12 @@ int thread__fork(struct thread *self, struct thread *parent)
         return 0;
 }
 
-size_t threads__fprintf(FILE *fp)
+size_t perf_session__fprintf(struct perf_session *self, FILE *fp)
 {
         size_t ret = 0;
         struct rb_node *nd;
 
-        for (nd = rb_first(&threads); nd; nd = rb_next(nd)) {
+        for (nd = rb_first(&self->threads); nd; nd = rb_next(nd)) {
                 struct thread *pos = rb_entry(nd, struct thread, rb_node);
 
                 ret += thread__fprintf(pos, fp);
diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h
index 2e35e1f6bb43..e93abf2d9cb6 100644
--- a/tools/perf/util/thread.h
+++ b/tools/perf/util/thread.h
@@ -23,11 +23,11 @@ struct thread {
 void map_groups__init(struct map_groups *self);
 int thread__set_comm(struct thread *self, const char *comm);
 int thread__comm_len(struct thread *self);
-struct thread *threads__findnew(pid_t pid);
+struct thread *perf_session__findnew(struct perf_session *self, pid_t pid);
 void thread__insert_map(struct thread *self, struct map *map);
 int thread__fork(struct thread *self, struct thread *parent);
 size_t map_groups__fprintf_maps(struct map_groups *self, FILE *fp);
-size_t threads__fprintf(FILE *fp);
+size_t perf_session__fprintf(struct perf_session *self, FILE *fp);
 
 void maps__insert(struct rb_root *maps, struct map *map);
 struct map *maps__find(struct rb_root *maps, u64 addr);