ring-buffer: Add event descriptor to simplify passing data
authorSteven Rostedt (Red Hat) <rostedt@goodmis.org>
Thu, 28 May 2015 21:13:14 +0000 (17:13 -0400)
committerSteven Rostedt <rostedt@goodmis.org>
Tue, 21 Jul 2015 02:30:46 +0000 (22:30 -0400)
Add rb_event_info descriptor to pass event info to functions a bit
easier than using a bunch of parameters. This will also allow for
changing the code around a bit to find better fast paths.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
kernel/trace/ring_buffer.c

index 6260717c18e3c6fb8eefa4cfcc185ab8322a6929..ba8f25ffcf6f12f4f7757948ba6a4d33e8eed09a 100644 (file)
@@ -399,6 +399,17 @@ struct rb_irq_work {
        bool                            wakeup_full;
 };
 
+/*
+ * Structure to hold event state and handle nested events.
+ */
+struct rb_event_info {
+       u64                     ts;
+       u64                     delta;
+       unsigned long           length;
+       struct buffer_page      *tail_page;
+       int                     add_timestamp;
+};
+
 /*
  * Used for which event context the event is in.
  *  NMI     = 0
@@ -2000,9 +2011,12 @@ rb_add_time_stamp(struct ring_buffer_event *event, u64 delta)
  */
 static void
 rb_update_event(struct ring_buffer_per_cpu *cpu_buffer,
-               struct ring_buffer_event *event, unsigned length,
-               int add_timestamp, u64 delta)
+               struct ring_buffer_event *event,
+               struct rb_event_info *info)
 {
+       unsigned length = info->length;
+       u64 delta = info->delta;
+
        /* Only a commit updates the timestamp */
        if (unlikely(!rb_event_is_commit(cpu_buffer, event)))
                delta = 0;
@@ -2011,7 +2025,7 @@ rb_update_event(struct ring_buffer_per_cpu *cpu_buffer,
         * If we need to add a timestamp, then we
         * add it to the start of the resevered space.
         */
-       if (unlikely(add_timestamp)) {
+       if (unlikely(info->add_timestamp)) {
                event = rb_add_time_stamp(event, delta);
                length -= RB_LEN_TIME_EXTEND;
                delta = 0;
@@ -2203,10 +2217,11 @@ static unsigned rb_calculate_event_length(unsigned length)
 
 static inline void
 rb_reset_tail(struct ring_buffer_per_cpu *cpu_buffer,
-             struct buffer_page *tail_page,
-             unsigned long tail, unsigned long length)
+             unsigned long tail, struct rb_event_info *info)
 {
+       struct buffer_page *tail_page = info->tail_page;
        struct ring_buffer_event *event;
+       unsigned long length = info->length;
 
        /*
         * Only the event that crossed the page boundary
@@ -2276,13 +2291,14 @@ rb_reset_tail(struct ring_buffer_per_cpu *cpu_buffer,
  */
 static noinline struct ring_buffer_event *
 rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer,
-            unsigned long length, unsigned long tail,
-            struct buffer_page *tail_page, u64 ts)
+            unsigned long tail, struct rb_event_info *info)
 {
+       struct buffer_page *tail_page = info->tail_page;
        struct buffer_page *commit_page = cpu_buffer->commit_page;
        struct ring_buffer *buffer = cpu_buffer->buffer;
        struct buffer_page *next_page;
        int ret;
+       u64 ts;
 
        next_page = tail_page;
 
@@ -2368,25 +2384,24 @@ rb_move_tail(struct ring_buffer_per_cpu *cpu_buffer,
 
  out_again:
 
-       rb_reset_tail(cpu_buffer, tail_page, tail, length);
+       rb_reset_tail(cpu_buffer, tail, info);
 
        /* fail and let the caller try again */
        return ERR_PTR(-EAGAIN);
 
  out_reset:
        /* reset write */
-       rb_reset_tail(cpu_buffer, tail_page, tail, length);
+       rb_reset_tail(cpu_buffer, tail, info);
 
        return NULL;
 }
 
 static struct ring_buffer_event *
 __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
-                 unsigned long length, u64 ts,
-                 u64 delta, int add_timestamp)
+                 struct rb_event_info *info)
 {
-       struct buffer_page *tail_page;
        struct ring_buffer_event *event;
+       struct buffer_page *tail_page;
        unsigned long tail, write;
 
        /*
@@ -2394,33 +2409,32 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
         * hold in the time field of the event, then we append a
         * TIME EXTEND event ahead of the data event.
         */
-       if (unlikely(add_timestamp))
-               length += RB_LEN_TIME_EXTEND;
+       if (unlikely(info->add_timestamp))
+               info->length += RB_LEN_TIME_EXTEND;
 
-       tail_page = cpu_buffer->tail_page;
-       write = local_add_return(length, &tail_page->write);
+       tail_page = info->tail_page = cpu_buffer->tail_page;
+       write = local_add_return(info->length, &tail_page->write);
 
        /* set write to only the index of the write */
        write &= RB_WRITE_MASK;
-       tail = write - length;
+       tail = write - info->length;
 
        /*
         * If this is the first commit on the page, then it has the same
         * timestamp as the page itself.
         */
        if (!tail)
-               delta = 0;
+               info->delta = 0;
 
        /* See if we shot pass the end of this buffer page */
        if (unlikely(write > BUF_PAGE_SIZE))
-               return rb_move_tail(cpu_buffer, length, tail,
-                                   tail_page, ts);
+               return rb_move_tail(cpu_buffer, tail, info);
 
        /* We reserved something on the buffer */
 
        event = __rb_page_index(tail_page, tail);
        kmemcheck_annotate_bitfield(event, bitfield);
-       rb_update_event(cpu_buffer, event, length, add_timestamp, delta);
+       rb_update_event(cpu_buffer, event, info);
 
        local_inc(&tail_page->entries);
 
@@ -2429,10 +2443,10 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
         * its timestamp.
         */
        if (!tail)
-               tail_page->page->time_stamp = ts;
+               tail_page->page->time_stamp = info->ts;
 
        /* account for these added bytes */
-       local_add(length, &cpu_buffer->entries_bytes);
+       local_add(info->length, &cpu_buffer->entries_bytes);
 
        return event;
 }
@@ -2521,9 +2535,8 @@ rb_reserve_next_event(struct ring_buffer *buffer,
                      unsigned long length)
 {
        struct ring_buffer_event *event;
-       u64 ts, delta;
+       struct rb_event_info info;
        int nr_loops = 0;
-       int add_timestamp;
        u64 diff;
 
        rb_start_commit(cpu_buffer);
@@ -2543,10 +2556,10 @@ rb_reserve_next_event(struct ring_buffer *buffer,
        }
 #endif
 
-       length = rb_calculate_event_length(length);
+       info.length = rb_calculate_event_length(length);
  again:
-       add_timestamp = 0;
-       delta = 0;
+       info.add_timestamp = 0;
+       info.delta = 0;
 
        /*
         * We allow for interrupts to reenter here and do a trace.
@@ -2560,35 +2573,35 @@ rb_reserve_next_event(struct ring_buffer *buffer,
        if (RB_WARN_ON(cpu_buffer, ++nr_loops > 1000))
                goto out_fail;
 
-       ts = rb_time_stamp(cpu_buffer->buffer);
-       diff = ts - cpu_buffer->write_stamp;
+       info.ts = rb_time_stamp(cpu_buffer->buffer);
+       diff = info.ts - cpu_buffer->write_stamp;
 
        /* make sure this diff is calculated here */
        barrier();
 
        /* Did the write stamp get updated already? */
-       if (likely(ts >= cpu_buffer->write_stamp)) {
-               delta = diff;
-               if (unlikely(test_time_stamp(delta))) {
+       if (likely(info.ts >= cpu_buffer->write_stamp)) {
+               info.delta = diff;
+               if (unlikely(test_time_stamp(info.delta))) {
                        int local_clock_stable = 1;
 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
                        local_clock_stable = sched_clock_stable();
 #endif
-                       WARN_ONCE(delta > (1ULL << 59),
+                       WARN_ONCE(info.delta > (1ULL << 59),
                                  KERN_WARNING "Delta way too big! %llu ts=%llu write stamp = %llu\n%s",
-                                 (unsigned long long)delta,
-                                 (unsigned long long)ts,
+                                 (unsigned long long)info.delta,
+                                 (unsigned long long)info.ts,
                                  (unsigned long long)cpu_buffer->write_stamp,
                                  local_clock_stable ? "" :
                                  "If you just came from a suspend/resume,\n"
                                  "please switch to the trace global clock:\n"
                                  "  echo global > /sys/kernel/debug/tracing/trace_clock\n");
-                       add_timestamp = 1;
+                       info.add_timestamp = 1;
                }
        }
 
-       event = __rb_reserve_next(cpu_buffer, length, ts,
-                                 delta, add_timestamp);
+       event = __rb_reserve_next(cpu_buffer, &info);
+
        if (unlikely(PTR_ERR(event) == -EAGAIN))
                goto again;