1 files changed, 100 insertions, 161 deletions

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index 52dea5dd5362..83eab547f1d1 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -354,6 +354,11 @@ static void rb_init_page(struct buffer_data_page *bpage)
 	local_set(&bpage->commit, 0);
 }
 
+static __always_inline unsigned int rb_page_commit(struct buffer_page *bpage)
+{
+	return local_read(&bpage->page->commit);
+}
+
 static void free_buffer_page(struct buffer_page *bpage)
 {
 	free_page((unsigned long)bpage->page);
@@ -639,8 +644,8 @@ static inline bool __rb_time_read(rb_time_t *t, u64 *ret, unsigned long *cnt)
 
 	*cnt = rb_time_cnt(top);
 
-	/* If top and bottom counts don't match, this interrupted a write */
-	if (*cnt != rb_time_cnt(bottom))
+	/* If top, msb or bottom counts don't match, this interrupted a write */
+	if (*cnt != rb_time_cnt(msb) || *cnt != rb_time_cnt(bottom))
 		return false;
 
 	/* The shift to msb will lose its cnt bits */
@@ -692,48 +697,7 @@ static void rb_time_set(rb_time_t *t, u64 val)
 static inline bool
 rb_time_read_cmpxchg(local_t *l, unsigned long expect, unsigned long set)
 {
-	unsigned long ret;
-
-	ret = local_cmpxchg(l, expect, set);
-	return ret == expect;
-}
-
-static bool rb_time_cmpxchg(rb_time_t *t, u64 expect, u64 set)
-{
-	unsigned long cnt, top, bottom, msb;
-	unsigned long cnt2, top2, bottom2, msb2;
-	u64 val;
-
-	/* The cmpxchg always fails if it interrupted an update */
-	 if (!__rb_time_read(t, &val, &cnt2))
-		 return false;
-
-	 if (val != expect)
-		 return false;
-
-	 cnt = local_read(&t->cnt);
-	 if ((cnt & 3) != cnt2)
-		 return false;
-
-	 cnt2 = cnt + 1;
-
-	 rb_time_split(val, &top, &bottom, &msb);
-	 top = rb_time_val_cnt(top, cnt);
-	 bottom = rb_time_val_cnt(bottom, cnt);
-
-	 rb_time_split(set, &top2, &bottom2, &msb2);
-	 top2 = rb_time_val_cnt(top2, cnt2);
-	 bottom2 = rb_time_val_cnt(bottom2, cnt2);
-
-	if (!rb_time_read_cmpxchg(&t->cnt, cnt, cnt2))
-		return false;
-	if (!rb_time_read_cmpxchg(&t->msb, msb, msb2))
-		return false;
-	if (!rb_time_read_cmpxchg(&t->top, top, top2))
-		return false;
-	if (!rb_time_read_cmpxchg(&t->bottom, bottom, bottom2))
-		return false;
-	return true;
+	return local_try_cmpxchg(l, &expect, set);
 }
 
 #else /* 64 bits */
@@ -749,13 +713,6 @@ static void rb_time_set(rb_time_t *t, u64 val)
 {
 	local64_set(&t->time, val);
 }
-
-static bool rb_time_cmpxchg(rb_time_t *t, u64 expect, u64 set)
-{
-	u64 val;
-	val = local64_cmpxchg(&t->time, expect, set);
-	return val == expect;
-}
 #endif
 
 /*
@@ -1137,6 +1094,9 @@ __poll_t ring_buffer_poll_wait(struct trace_buffer *buffer, int cpu,
 	if (full) {
 		poll_wait(filp, &work->full_waiters, poll_table);
 		work->full_waiters_pending = true;
+		if (!cpu_buffer->shortest_full ||
+		    cpu_buffer->shortest_full > full)
+			cpu_buffer->shortest_full = full;
 	} else {
 		poll_wait(filp, &work->waiters, poll_table);
 		work->waiters_pending = true;
@@ -1494,14 +1454,11 @@ static bool rb_head_page_replace(struct buffer_page *old,
 {
 	unsigned long *ptr = (unsigned long *)&old->list.prev->next;
 	unsigned long val;
-	unsigned long ret;
 
 	val = *ptr & ~RB_FLAG_MASK;
 	val |= RB_PAGE_HEAD;
 
-	ret = cmpxchg(ptr, val, (unsigned long)&new->list);
-
-	return ret == val;
+	return try_cmpxchg(ptr, &val, (unsigned long)&new->list);
 }
 
 /*
@@ -1787,6 +1744,8 @@ static void rb_free_cpu_buffer(struct ring_buffer_per_cpu *cpu_buffer)
 		free_buffer_page(bpage);
 	}
 
+	free_page((unsigned long)cpu_buffer->free_page);
+
 	kfree(cpu_buffer);
 }
 
@@ -2011,7 +1970,7 @@ rb_remove_pages(struct ring_buffer_per_cpu *cpu_buffer, unsigned long nr_pages)
 			 * Increment overrun to account for the lost events.
 			 */
 			local_add(page_entries, &cpu_buffer->overrun);
-			local_sub(BUF_PAGE_SIZE, &cpu_buffer->entries_bytes);
+			local_sub(rb_page_commit(to_remove_page), &cpu_buffer->entries_bytes);
 			local_inc(&cpu_buffer->pages_lost);
 		}
 
@@ -2056,7 +2015,7 @@ rb_insert_pages(struct ring_buffer_per_cpu *cpu_buffer)
 	retries = 10;
 	success = false;
 	while (retries--) {
-		struct list_head *head_page, *prev_page, *r;
+		struct list_head *head_page, *prev_page;
 		struct list_head *last_page, *first_page;
 		struct list_head *head_page_with_bit;
 		struct buffer_page *hpage = rb_set_head_page(cpu_buffer);
@@ -2075,9 +2034,9 @@ rb_insert_pages(struct ring_buffer_per_cpu *cpu_buffer)
 		last_page->next = head_page_with_bit;
 		first_page->prev = prev_page;
 
-		r = cmpxchg(&prev_page->next, head_page_with_bit, first_page);
-
-		if (r == head_page_with_bit) {
+		/* caution: head_page_with_bit gets updated on cmpxchg failure */
+		if (try_cmpxchg(&prev_page->next,
+				&head_page_with_bit, first_page)) {
 			/*
 			 * yay, we replaced the page pointer to our new list,
 			 * now, we just have to update to head page's prev
@@ -2206,6 +2165,8 @@ int ring_buffer_resize(struct trace_buffer *buffer, unsigned long size,
 				err = -ENOMEM;
 				goto out_err;
 			}
+
+			cond_resched();
 		}
 
 		cpus_read_lock();
@@ -2373,11 +2334,6 @@ rb_reader_event(struct ring_buffer_per_cpu *cpu_buffer)
 			       cpu_buffer->reader_page->read);
 }
 
-static __always_inline unsigned rb_page_commit(struct buffer_page *bpage)
-{
-	return local_read(&bpage->page->commit);
-}
-
 static struct ring_buffer_event *
 rb_iter_head_event(struct ring_buffer_iter *iter)
 {
@@ -2396,6 +2352,11 @@ rb_iter_head_event(struct ring_buffer_iter *iter)
 	 */
 	commit = rb_page_commit(iter_head_page);
 	smp_rmb();
+
+	/* An event needs to be at least 8 bytes in size */
+	if (iter->head > commit - 8)
+		goto reset;
+
 	event = __rb_page_index(iter_head_page, iter->head);
 	length = rb_event_length(event);
 
@@ -2405,7 +2366,7 @@ rb_iter_head_event(struct ring_buffer_iter *iter)
 	 */
 	barrier();
 
-	if ((iter->head + length) > commit || length > BUF_MAX_DATA_SIZE)
+	if ((iter->head + length) > commit || length > BUF_PAGE_SIZE)
 		/* Writer corrupted the read? */
 		goto reset;
 
@@ -2518,7 +2479,7 @@ rb_handle_head_page(struct ring_buffer_per_cpu *cpu_buffer,
 		 * the counters.
 		 */
 		local_add(entries, &cpu_buffer->overrun);
-		local_sub(BUF_PAGE_SIZE, &cpu_buffer->entries_bytes);
+		local_sub(rb_page_commit(next_page), &cpu_buffer->entries_bytes);
 		local_inc(&cpu_buffer->pages_lost);
 
 		/*
@@ -2661,9 +2622,6 @@ rb_reset_tail(struct ring_buffer_per_cpu *cpu_buffer,
 
 	event = __rb_page_index(tail_page, tail);
 
-	/* account for padding bytes */
-	local_add(BUF_PAGE_SIZE - tail, &cpu_buffer->entries_bytes);
-
 	/*
 	 * Save the original length to the meta data.
 	 * This will be used by the reader to add lost event
@@ -2677,7 +2635,8 @@ rb_reset_tail(struct ring_buffer_per_cpu *cpu_buffer,
 	 * write counter enough to allow another writer to slip
 	 * in on this page.
 	 * We put in a discarded commit instead, to make sure
-	 * that this space is not used again.
+	 * that this space is not used again, and this space will
+	 * not be accounted into 'entries_bytes'.
 	 *
 	 * If we are less than the minimum size, we don't need to
 	 * worry about it.
@@ -2702,6 +2661,9 @@ rb_reset_tail(struct ring_buffer_per_cpu *cpu_buffer,
 	/* time delta must be non zero */
 	event->time_delta = 1;
 
+	/* account for padding bytes */
+	local_add(BUF_PAGE_SIZE - tail, &cpu_buffer->entries_bytes);
+
 	/* Make sure the padding is visible before the tail_page->write update */
 	smp_wmb();
 
@@ -2978,35 +2940,13 @@ static unsigned rb_calculate_event_length(unsigned length)
 	return length;
 }
 
-static u64 rb_time_delta(struct ring_buffer_event *event)
-{
-	switch (event->type_len) {
-	case RINGBUF_TYPE_PADDING:
-		return 0;
-
-	case RINGBUF_TYPE_TIME_EXTEND:
-		return rb_event_time_stamp(event);
-
-	case RINGBUF_TYPE_TIME_STAMP:
-		return 0;
-
-	case RINGBUF_TYPE_DATA:
-		return event->time_delta;
-	default:
-		return 0;
-	}
-}
-
 static inline bool
 rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer,
 		  struct ring_buffer_event *event)
 {
 	unsigned long new_index, old_index;
 	struct buffer_page *bpage;
-	unsigned long index;
 	unsigned long addr;
-	u64 write_stamp;
-	u64 delta;
 
 	new_index = rb_event_index(event);
 	old_index = new_index + rb_event_ts_length(event);
@@ -3015,41 +2955,34 @@ rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer,
 
 	bpage = READ_ONCE(cpu_buffer->tail_page);
 
-	delta = rb_time_delta(event);
-
-	if (!rb_time_read(&cpu_buffer->write_stamp, &write_stamp))
-		return false;
-
-	/* Make sure the write stamp is read before testing the location */
-	barrier();
-
+	/*
+	 * Make sure the tail_page is still the same and
+	 * the next write location is the end of this event
+	 */
 	if (bpage->page == (void *)addr && rb_page_write(bpage) == old_index) {
 		unsigned long write_mask =
 			local_read(&bpage->write) & ~RB_WRITE_MASK;
 		unsigned long event_length = rb_event_length(event);
 
-		/* Something came in, can't discard */
-		if (!rb_time_cmpxchg(&cpu_buffer->write_stamp,
-				       write_stamp, write_stamp - delta))
-			return false;
-
 		/*
-		 * It's possible that the event time delta is zero
-		 * (has the same time stamp as the previous event)
-		 * in which case write_stamp and before_stamp could
-		 * be the same. In such a case, force before_stamp
-		 * to be different than write_stamp. It doesn't
-		 * matter what it is, as long as its different.
+		 * For the before_stamp to be different than the write_stamp
+		 * to make sure that the next event adds an absolute
+		 * value and does not rely on the saved write stamp, which
+		 * is now going to be bogus.
+		 *
+		 * By setting the before_stamp to zero, the next event
+		 * is not going to use the write_stamp and will instead
+		 * create an absolute timestamp. This means there's no
+		 * reason to update the wirte_stamp!
 		 */
-		if (!delta)
-			rb_time_set(&cpu_buffer->before_stamp, 0);
+		rb_time_set(&cpu_buffer->before_stamp, 0);
 
 		/*
 		 * If an event were to come in now, it would see that the
 		 * write_stamp and the before_stamp are different, and assume
 		 * that this event just added itself before updating
 		 * the write stamp. The interrupting event will fix the
-		 * write stamp for us, and use the before stamp as its delta.
+		 * write stamp for us, and use an absolute timestamp.
 		 */
 
 		/*
@@ -3060,8 +2993,9 @@ rb_try_to_discard(struct ring_buffer_per_cpu *cpu_buffer,
 		 */
 		old_index += write_mask;
 		new_index += write_mask;
-		index = local_cmpxchg(&bpage->write, old_index, new_index);
-		if (index == old_index) {
+
+		/* caution: old_index gets updated on cmpxchg failure */
+		if (local_try_cmpxchg(&bpage->write, &old_index, new_index)) {
 			/* update counters */
 			local_sub(event_length, &cpu_buffer->entries_bytes);
 			return true;
@@ -3485,7 +3419,7 @@ static void check_buffer(struct ring_buffer_per_cpu *cpu_buffer,
 		return;
 
 	/*
-	 * If this interrupted another event, 
+	 * If this interrupted another event,
 	 */
 	if (atomic_inc_return(this_cpu_ptr(&checking)) != 1)
 		goto out;
@@ -3579,7 +3513,10 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
 		 * absolute timestamp.
 		 * Don't bother if this is the start of a new page (w == 0).
 		 */
-		if (unlikely(!a_ok || !b_ok || (info->before != info->after && w))) {
+		if (!w) {
+			/* Use the sub-buffer timestamp */
+			info->delta = 0;
+		} else if (unlikely(!a_ok || !b_ok || info->before != info->after)) {
 			info->add_timestamp |= RB_ADD_STAMP_FORCE | RB_ADD_STAMP_EXTEND;
 			info->length += RB_LEN_TIME_EXTEND;
 		} else {
@@ -3602,26 +3539,19 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
 
 	/* See if we shot pass the end of this buffer page */
 	if (unlikely(write > BUF_PAGE_SIZE)) {
-		/* before and after may now different, fix it up*/
-		b_ok = rb_time_read(&cpu_buffer->before_stamp, &info->before);
-		a_ok = rb_time_read(&cpu_buffer->write_stamp, &info->after);
-		if (a_ok && b_ok && info->before != info->after)
-			(void)rb_time_cmpxchg(&cpu_buffer->before_stamp,
-					      info->before, info->after);
-		if (a_ok && b_ok)
-			check_buffer(cpu_buffer, info, CHECK_FULL_PAGE);
+		check_buffer(cpu_buffer, info, CHECK_FULL_PAGE);
 		return rb_move_tail(cpu_buffer, tail, info);
 	}
 
 	if (likely(tail == w)) {
-		u64 save_before;
-		bool s_ok;
-
 		/* Nothing interrupted us between A and C */
  /*D*/		rb_time_set(&cpu_buffer->write_stamp, info->ts);
-		barrier();
- /*E*/		s_ok = rb_time_read(&cpu_buffer->before_stamp, &save_before);
-		RB_WARN_ON(cpu_buffer, !s_ok);
+		/*
+		 * If something came in between C and D, the write stamp
+		 * may now not be in sync. But that's fine as the before_stamp
+		 * will be different and then next event will just be forced
+		 * to use an absolute timestamp.
+		 */
 		if (likely(!(info->add_timestamp &
 			     (RB_ADD_STAMP_FORCE | RB_ADD_STAMP_ABSOLUTE))))
 			/* This did not interrupt any time update */
@@ -3629,41 +3559,40 @@ __rb_reserve_next(struct ring_buffer_per_cpu *cpu_buffer,
 		else
 			/* Just use full timestamp for interrupting event */
 			info->delta = info->ts;
-		barrier();
 		check_buffer(cpu_buffer, info, tail);
-		if (unlikely(info->ts != save_before)) {
-			/* SLOW PATH - Interrupted between C and E */
-
-			a_ok = rb_time_read(&cpu_buffer->write_stamp, &info->after);
-			RB_WARN_ON(cpu_buffer, !a_ok);
-
-			/* Write stamp must only go forward */
-			if (save_before > info->after) {
-				/*
-				 * We do not care about the result, only that
-				 * it gets updated atomically.
-				 */
-				(void)rb_time_cmpxchg(&cpu_buffer->write_stamp,
-						      info->after, save_before);
-			}
-		}
 	} else {
 		u64 ts;
 		/* SLOW PATH - Interrupted between A and C */
-		a_ok = rb_time_read(&cpu_buffer->write_stamp, &info->after);
-		/* Was interrupted before here, write_stamp must be valid */
+
+		/* Save the old before_stamp */
+		a_ok = rb_time_read(&cpu_buffer->before_stamp, &info->before);
 		RB_WARN_ON(cpu_buffer, !a_ok);
+
+		/*
+		 * Read a new timestamp and update the before_stamp to make
+		 * the next event after this one force using an absolute
+		 * timestamp. This is in case an interrupt were to come in
+		 * between E and F.
+		 */
 		ts = rb_time_stamp(cpu_buffer->buffer);
+		rb_time_set(&cpu_buffer->before_stamp, ts);
+
+		barrier();
+ /*E*/		a_ok = rb_time_read(&cpu_buffer->write_stamp, &info->after);
+		/* Was interrupted before here, write_stamp must be valid */
+		RB_WARN_ON(cpu_buffer, !a_ok);
 		barrier();
- /*E*/		if (write == (local_read(&tail_page->write) & RB_WRITE_MASK) &&
-		    info->after < ts &&
-		    rb_time_cmpxchg(&cpu_buffer->write_stamp,
-				    info->after, ts)) {
-			/* Nothing came after this event between C and E */
+ /*F*/		if (write == (local_read(&tail_page->write) & RB_WRITE_MASK) &&
+		    info->after == info->before && info->after < ts) {
+			/*
+			 * Nothing came after this event between C and F, it is
+			 * safe to use info->after for the delta as it
+			 * matched info->before and is still valid.
+			 */
 			info->delta = ts - info->after;
 		} else {
 			/*
-			 * Interrupted between C and E:
+			 * Interrupted between C and F:
 			 * Lost the previous events time stamp. Just set the
 			 * delta to zero, and this will be the same time as
 			 * the event this event interrupted. And the events that
@@ -3714,6 +3643,12 @@ rb_reserve_next_event(struct trace_buffer *buffer,
 	int nr_loops = 0;
 	int add_ts_default;
 
+	/* ring buffer does cmpxchg, make sure it is safe in NMI context */
+	if (!IS_ENABLED(CONFIG_ARCH_HAVE_NMI_SAFE_CMPXCHG) &&
+	    (unlikely(in_nmi()))) {
+		return NULL;
+	}
+
 	rb_start_commit(cpu_buffer);
 	/* The commit page can not change after this */
 
@@ -3737,6 +3672,8 @@ rb_reserve_next_event(struct trace_buffer *buffer,
 	if (ring_buffer_time_stamp_abs(cpu_buffer->buffer)) {
 		add_ts_default = RB_ADD_STAMP_ABSOLUTE;
 		info.length += RB_LEN_TIME_EXTEND;
+		if (info.length > BUF_MAX_DATA_SIZE)
+			goto out_fail;
 	} else {
 		add_ts_default = RB_ADD_STAMP_NONE;
 	}
@@ -4216,7 +4153,7 @@ u64 ring_buffer_oldest_event_ts(struct trace_buffer *buffer, int cpu)
 EXPORT_SYMBOL_GPL(ring_buffer_oldest_event_ts);
 
 /**
- * ring_buffer_bytes_cpu - get the number of bytes consumed in a cpu buffer
+ * ring_buffer_bytes_cpu - get the number of bytes unconsumed in a cpu buffer
  * @buffer: The ring buffer
  * @cpu: The per CPU buffer to read from.
  */
@@ -4724,6 +4661,7 @@ static void rb_advance_reader(struct ring_buffer_per_cpu *cpu_buffer)
 
 	length = rb_event_length(event);
 	cpu_buffer->reader_page->read += length;
+	cpu_buffer->read_bytes += length;
 }
 
 static void rb_advance_iter(struct ring_buffer_iter *iter)
@@ -5117,7 +5055,8 @@ ring_buffer_read_prepare(struct trace_buffer *buffer, int cpu, gfp_t flags)
 	if (!iter)
 		return NULL;
 
-	iter->event = kmalloc(BUF_MAX_DATA_SIZE, flags);
+	/* Holds the entire event: data and meta data */
+	iter->event = kmalloc(BUF_PAGE_SIZE, flags);
 	if (!iter->event) {
 		kfree(iter);
 		return NULL;
@@ -5817,7 +5756,7 @@ int ring_buffer_read_page(struct trace_buffer *buffer,
 	} else {
 		/* update the entry counter */
 		cpu_buffer->read += rb_page_entries(reader);
-		cpu_buffer->read_bytes += BUF_PAGE_SIZE;
+		cpu_buffer->read_bytes += rb_page_commit(reader);
 
 		/* swap the pages */
 		rb_init_page(bpage);