Merge tag 'trace-v6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace

Pull tracing updates from Steven Rostedt:
 "Major changes:

   - Changed location of tracing repo from personal git repo to:
     git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace.git

   - Added Masami Hiramatsu as co-maintainer

   - Updated MAINTAINERS file to separate out FTRACE as it is more than
     just TRACING.

  Minor changes:

   - Added Mark Rutland as FTRACE reviewer

   - Updated user_events to make it on its way to remove the BROKEN tag.
     The changes should now be acceptable but will run it through a
     cycle and hopefully we can remove the BROKEN tag next release.

   - Added filtering to eprobes

   - Added a delta time to the benchmark trace event

   - Have the histogram and filter callbacks called via a switch
     statement instead of indirect functions. This speeds it up to avoid
     retpolines.

   - Add a way to wake up ring buffer waiters waiting for the ring
     buffer to fill up to its watermark.

   - New ioctl() on the trace_pipe_raw file to wake up ring buffer
     waiters.

   - Wake up waiters when the ring buffer is disabled. A reader may
     block when the ring buffer is disabled, but if it was blocked when
     the ring buffer is disabled it should then wake up.

  Fixes:

   - Allow splice to read partially read ring buffer pages. This fixes
     splice never moving forward.

   - Fix inverted compare that made the "shortest" ring buffer wait
     queue actually the longest.

   - Fix a race in the ring buffer between resetting a page when a
     writer goes to another page, and the reader.

   - Fix ftrace accounting bug when function hooks are added at boot up
     before the weak functions are set to "disabled".

   - Fix bug that freed a user allocated snapshot buffer when enabling a
     tracer.

   - Fix possible recursive locks in osnoise tracer

   - Fix recursive locking direct functions

   - Other minor clean ups and fixes"

* tag 'trace-v6.1' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace: (44 commits)
  ftrace: Create separate entry in MAINTAINERS for function hooks
  tracing: Update MAINTAINERS to reflect new tracing git repo
  tracing: Do not free snapshot if tracer is on cmdline
  ftrace: Still disable enabled records marked as disabled
  tracing/user_events: Move pages/locks into groups to prepare for namespaces
  tracing: Add Masami Hiramatsu as co-maintainer
  tracing: Remove unused variable 'dups'
  MAINTAINERS: add myself as a tracing reviewer
  ring-buffer: Fix race between reset page and reading page
  tracing/user_events: Update ABI documentation to align to bits vs bytes
  tracing/user_events: Use bits vs bytes for enabled status page data
  tracing/user_events: Use refcount instead of atomic for ref tracking
  tracing/user_events: Ensure user provided strings are safely formatted
  tracing/user_events: Use WRITE instead of READ for io vector import
  tracing/user_events: Use NULL for strstr checks
  tracing: Fix spelling mistake "preapre" -> "prepare"
  tracing: Wake up waiters when tracing is disabled
  tracing: Add ioctl() to force ring buffer waiters to wake up
  tracing: Wake up ring buffer waiters on closing of the file
  ring-buffer: Add ring_buffer_wake_waiters()
  ...

1 files changed, 84 insertions, 3 deletions

diff --git a/kernel/trace/ring_buffer.c b/kernel/trace/ring_buffer.c
index d59b6a328b7f..c3f354cfc5ba 100644
--- a/kernel/trace/ring_buffer.c
+++ b/kernel/trace/ring_buffer.c
@@ -413,6 +413,7 @@ struct rb_irq_work {
 	struct irq_work			work;
 	wait_queue_head_t		waiters;
 	wait_queue_head_t		full_waiters;
+	long				wait_index;
 	bool				waiters_pending;
 	bool				full_waiters_pending;
 	bool				wakeup_full;
@@ -917,13 +918,45 @@ static void rb_wake_up_waiters(struct irq_work *work)
 	struct rb_irq_work *rbwork = container_of(work, struct rb_irq_work, work);
 
 	wake_up_all(&rbwork->waiters);
-	if (rbwork->wakeup_full) {
+	if (rbwork->full_waiters_pending || rbwork->wakeup_full) {
 		rbwork->wakeup_full = false;
+		rbwork->full_waiters_pending = false;
 		wake_up_all(&rbwork->full_waiters);
 	}
 }
 
 /**
+ * ring_buffer_wake_waiters - wake up any waiters on this ring buffer
+ * @buffer: The ring buffer to wake waiters on
+ *
+ * In the case of a file that represents a ring buffer is closing,
+ * it is prudent to wake up any waiters that are on this.
+ */
+void ring_buffer_wake_waiters(struct trace_buffer *buffer, int cpu)
+{
+	struct ring_buffer_per_cpu *cpu_buffer;
+	struct rb_irq_work *rbwork;
+
+	if (cpu == RING_BUFFER_ALL_CPUS) {
+
+		/* Wake up individual ones too. One level recursion */
+		for_each_buffer_cpu(buffer, cpu)
+			ring_buffer_wake_waiters(buffer, cpu);
+
+		rbwork = &buffer->irq_work;
+	} else {
+		cpu_buffer = buffer->buffers[cpu];
+		rbwork = &cpu_buffer->irq_work;
+	}
+
+	rbwork->wait_index++;
+	/* make sure the waiters see the new index */
+	smp_wmb();
+
+	rb_wake_up_waiters(&rbwork->work);
+}
+
+/**
  * ring_buffer_wait - wait for input to the ring buffer
  * @buffer: buffer to wait on
  * @cpu: the cpu buffer to wait on
@@ -938,6 +971,7 @@ int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full)
 	struct ring_buffer_per_cpu *cpu_buffer;
 	DEFINE_WAIT(wait);
 	struct rb_irq_work *work;
+	long wait_index;
 	int ret = 0;
 
 	/*
@@ -956,6 +990,7 @@ int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full)
 		work = &cpu_buffer->irq_work;
 	}
 
+	wait_index = READ_ONCE(work->wait_index);
 
 	while (true) {
 		if (full)
@@ -1011,7 +1046,7 @@ int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full)
 			nr_pages = cpu_buffer->nr_pages;
 			dirty = ring_buffer_nr_dirty_pages(buffer, cpu);
 			if (!cpu_buffer->shortest_full ||
-			    cpu_buffer->shortest_full < full)
+			    cpu_buffer->shortest_full > full)
 				cpu_buffer->shortest_full = full;
 			raw_spin_unlock_irqrestore(&cpu_buffer->reader_lock, flags);
 			if (!pagebusy &&
@@ -1020,6 +1055,11 @@ int ring_buffer_wait(struct trace_buffer *buffer, int cpu, int full)
 		}
 
 		schedule();
+
+		/* Make sure to see the new wait index */
+		smp_rmb();
+		if (wait_index != work->wait_index)
+			break;
 	}
 
 	if (full)
@@ -2608,6 +2648,9 @@ rb_reset_tail(struct ring_buffer_per_cpu *cpu_buffer,
 		/* Mark the rest of the page with padding */
 		rb_event_set_padding(event);
 
+		/* Make sure the padding is visible before the write update */
+		smp_wmb();
+
 		/* Set the write back to the previous setting */
 		local_sub(length, &tail_page->write);
 		return;
@@ -2619,6 +2662,9 @@ rb_reset_tail(struct ring_buffer_per_cpu *cpu_buffer,
 	/* time delta must be non zero */
 	event->time_delta = 1;
 
+	/* Make sure the padding is visible before the tail_page->write update */
+	smp_wmb();
+
 	/* Set write to end of buffer */
 	length = (tail + length) - BUF_PAGE_SIZE;
 	local_sub(length, &tail_page->write);
@@ -4587,6 +4633,33 @@ rb_get_reader_page(struct ring_buffer_per_cpu *cpu_buffer)
 	arch_spin_unlock(&cpu_buffer->lock);
 	local_irq_restore(flags);
 
+	/*
+	 * The writer has preempt disable, wait for it. But not forever
+	 * Although, 1 second is pretty much "forever"
+	 */
+#define USECS_WAIT	1000000
+        for (nr_loops = 0; nr_loops < USECS_WAIT; nr_loops++) {
+		/* If the write is past the end of page, a writer is still updating it */
+		if (likely(!reader || rb_page_write(reader) <= BUF_PAGE_SIZE))
+			break;
+
+		udelay(1);
+
+		/* Get the latest version of the reader write value */
+		smp_rmb();
+	}
+
+	/* The writer is not moving forward? Something is wrong */
+	if (RB_WARN_ON(cpu_buffer, nr_loops == USECS_WAIT))
+		reader = NULL;
+
+	/*
+	 * Make sure we see any padding after the write update
+	 * (see rb_reset_tail())
+	 */
+	smp_rmb();
+
+
 	return reader;
 }
 
@@ -5616,7 +5689,15 @@ int ring_buffer_read_page(struct trace_buffer *buffer,
 		unsigned int pos = 0;
 		unsigned int size;
 
-		if (full)
+		/*
+		 * If a full page is expected, this can still be returned
+		 * if there's been a previous partial read and the
+		 * rest of the page can be read and the commit page is off
+		 * the reader page.
+		 */
+		if (full &&
+		    (!read || (len < (commit - read)) ||
+		     cpu_buffer->reader_page == cpu_buffer->commit_page))
 			goto out_unlock;
 
 		if (len > (commit - read))