tracepoints: add DECLARE_TRACE() and DEFINE_TRACE()

Impact: API *CHANGE*. Must update all tracepoint users.

Add DEFINE_TRACE() to tracepoints to let them declare the tracepoint
structure in a single spot for all the kernel. It helps reducing memory
consumption, especially when declaring a lot of tracepoints, e.g. for
kmalloc tracing.

*API CHANGE WARNING*: now, DECLARE_TRACE() must be used in headers for
tracepoint declarations rather than DEFINE_TRACE(). This is the sane way
to do it. The name previously used was misleading.

Updates scheduler instrumentation to follow this API change.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Ingo Molnar <mingo@elte.hu>

3 files changed, 38 insertions, 22 deletions

diff --git a/include/asm-generic/vmlinux.lds.h b/include/asm-generic/vmlinux.lds.h
index a5e4ed9baec8..3b46ae464933 100644
--- a/include/asm-generic/vmlinux.lds.h
+++ b/include/asm-generic/vmlinux.lds.h
@@ -71,6 +71,7 @@
         VMLINUX_SYMBOL(__start___markers) = .;                          \
         *(__markers)                                                    \
         VMLINUX_SYMBOL(__stop___markers) = .;                           \
+        . = ALIGN(32);                                                  \
         VMLINUX_SYMBOL(__start___tracepoints) = .;                      \
         *(__tracepoints)                                                \
         VMLINUX_SYMBOL(__stop___tracepoints) = .;                       \
diff --git a/include/linux/tracepoint.h b/include/linux/tracepoint.h
index 7e9b42aeae0e..757005458366 100644
--- a/include/linux/tracepoint.h
+++ b/include/linux/tracepoint.h
@@ -24,8 +24,12 @@ struct tracepoint {
         const char *name;               /* Tracepoint name */
         int state;                      /* State. */
         void **funcs;
-} __attribute__((aligned(8)));
-
+} __attribute__((aligned(32)));         /*
+                                         * Aligned on 32 bytes because it is
+                                         * globally visible and gcc happily
+                                         * align these on the structure size.
+                                         * Keep in sync with vmlinux.lds.h.
+                                         */
 
 #define TPPROTO(args...)        args
 #define TPARGS(args...)         args
@@ -55,15 +59,10 @@ struct tracepoint {
  * not add unwanted padding between the beginning of the section and the
  * structure. Force alignment to the same alignment as the section start.
  */
-#define DEFINE_TRACE(name, proto, args)                                 \
+#define DECLARE_TRACE(name, proto, args)                                \
+        extern struct tracepoint __tracepoint_##name;                   \
         static inline void trace_##name(proto)                          \
         {                                                               \
-                static const char __tpstrtab_##name[]                   \
-                __attribute__((section("__tracepoints_strings")))       \
-                = #name;                                                \
-                static struct tracepoint __tracepoint_##name            \
-                __attribute__((section("__tracepoints"), aligned(8))) = \
-                { __tpstrtab_##name, 0, NULL };                         \
                 if (unlikely(__tracepoint_##name.state))                \
                         __DO_TRACE(&__tracepoint_##name,                \
                                 TPPROTO(proto), TPARGS(args));          \
@@ -77,11 +76,23 @@ struct tracepoint {
                 return tracepoint_probe_unregister(#name, (void *)probe);\
         }
 
+#define DEFINE_TRACE(name)                                              \
+        static const char __tpstrtab_##name[]                           \
+        __attribute__((section("__tracepoints_strings"))) = #name;      \
+        struct tracepoint __tracepoint_##name                           \
+        __attribute__((section("__tracepoints"), aligned(32))) =        \
+                { __tpstrtab_##name, 0, NULL }
+
+#define EXPORT_TRACEPOINT_SYMBOL_GPL(name)                              \
+        EXPORT_SYMBOL_GPL(__tracepoint_##name)
+#define EXPORT_TRACEPOINT_SYMBOL(name)                                  \
+        EXPORT_SYMBOL(__tracepoint_##name)
+
 extern void tracepoint_update_probe_range(struct tracepoint *begin,
         struct tracepoint *end);
 
 #else /* !CONFIG_TRACEPOINTS */
-#define DEFINE_TRACE(name, proto, args)                 \
+#define DECLARE_TRACE(name, proto, args)                                \
         static inline void _do_trace_##name(struct tracepoint *tp, proto) \
         { }                                                             \
         static inline void trace_##name(proto)                          \
@@ -95,6 +106,10 @@ extern void tracepoint_update_probe_range(struct tracepoint *begin,
                 return -ENOSYS;                                         \
         }
 
+#define DEFINE_TRACE(name)
+#define EXPORT_TRACEPOINT_SYMBOL_GPL(name)
+#define EXPORT_TRACEPOINT_SYMBOL(name)
+
 static inline void tracepoint_update_probe_range(struct tracepoint *begin,
         struct tracepoint *end)
 { }
diff --git a/include/trace/sched.h b/include/trace/sched.h
index ad47369d01b5..9b2854abf7e2 100644
--- a/include/trace/sched.h
+++ b/include/trace/sched.h
@@ -4,52 +4,52 @@
 #include <linux/sched.h>
 #include <linux/tracepoint.h>
 
-DEFINE_TRACE(sched_kthread_stop,
+DECLARE_TRACE(sched_kthread_stop,
         TPPROTO(struct task_struct *t),
                 TPARGS(t));
 
-DEFINE_TRACE(sched_kthread_stop_ret,
+DECLARE_TRACE(sched_kthread_stop_ret,
         TPPROTO(int ret),
                 TPARGS(ret));
 
-DEFINE_TRACE(sched_wait_task,
+DECLARE_TRACE(sched_wait_task,
         TPPROTO(struct rq *rq, struct task_struct *p),
                 TPARGS(rq, p));
 
-DEFINE_TRACE(sched_wakeup,
+DECLARE_TRACE(sched_wakeup,
         TPPROTO(struct rq *rq, struct task_struct *p),
                 TPARGS(rq, p));
 
-DEFINE_TRACE(sched_wakeup_new,
+DECLARE_TRACE(sched_wakeup_new,
         TPPROTO(struct rq *rq, struct task_struct *p),
                 TPARGS(rq, p));
 
-DEFINE_TRACE(sched_switch,
+DECLARE_TRACE(sched_switch,
         TPPROTO(struct rq *rq, struct task_struct *prev,
                 struct task_struct *next),
                 TPARGS(rq, prev, next));
 
-DEFINE_TRACE(sched_migrate_task,
+DECLARE_TRACE(sched_migrate_task,
         TPPROTO(struct rq *rq, struct task_struct *p, int dest_cpu),
                 TPARGS(rq, p, dest_cpu));
 
-DEFINE_TRACE(sched_process_free,
+DECLARE_TRACE(sched_process_free,
         TPPROTO(struct task_struct *p),
                 TPARGS(p));
 
-DEFINE_TRACE(sched_process_exit,
+DECLARE_TRACE(sched_process_exit,
         TPPROTO(struct task_struct *p),
                 TPARGS(p));
 
-DEFINE_TRACE(sched_process_wait,
+DECLARE_TRACE(sched_process_wait,
         TPPROTO(struct pid *pid),
                 TPARGS(pid));
 
-DEFINE_TRACE(sched_process_fork,
+DECLARE_TRACE(sched_process_fork,
         TPPROTO(struct task_struct *parent, struct task_struct *child),
                 TPARGS(parent, child));
 
-DEFINE_TRACE(sched_signal_send,
+DECLARE_TRACE(sched_signal_send,
         TPPROTO(int sig, struct task_struct *p),
                 TPARGS(sig, p));