Merge tag 'trace-v7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-tracegrafted

Pull tracing fixes from Steven Rostedt:

 - Don't destroy user event fields when removal fails

   User event fields are destroyed before the event is removed from
   visibility. But that can fail leaving the still visible event with no
   fields. Move the destroying of the fields to after the event is
   successfully removed from visibility.

 - Initialize function graph state is fork before calling
   copy_exec_state()

   For non-CLONE_VM forks, copy_exec_state() allocates a new
   task_exec_state. If that allocation fails, ftrace_graph_exit_task()
   will free the tasks ret_stack pointer. Since that pointer is still
   using the parent's ret_stack, it mistakenly frees the parent's
   pointer too.

   Call ftrace_graph_init() on the task first which will NULL out the
   new tasks's ret_stack and if the copy fails, it will not free
   anything.

 - Remove FGRAPH_MAX_INDEX

   The macro FGRAPH_MAX_INDEX was added but never used. Remove it.

 - Save ent_size in function graph printing of nested functions

   The function graph tracer needs to look at the next event to see if
   the next event is the return of the current function entry. If it is,
   it prints a single line:

	ktime_get();

   Otherwise it prints it like a nested function:

	tick_nohz_irq_exit() {
	    ktime_get();
	    kcpustat_irq_exit();
	}

   In order to look at the next event, it must save the current event so
   that it has the information to print from it. It saves the event in
   the iterator descriptor called "ent". What it doesn't save is the
   ent_size of the event which is now used to know if the function graph
   arguments are to be printed. The peek doesn't save the size so the
   size used happens to be that of the size of the last event that was
   seen.

   Save the entry event size in the iterator descriptor so that the
   correct size is used.

 - Fix several errors with freeing data in the histogram code

   The histogram code had a lot of leaked or or incorrect accounting
   when failures happen. Correct them.

 - Fix histogram regression of .percent and .graph modifiers

   Up until 6.3 histogram values could have "percent" or "graph"
   modifiers that changed how they were printed. But a change that added
   restricting histograms values from being strings, stack traces and
   other modifiers inadvertently prevented them from using the percent
   and graph modifiers, which were legal use cases for values.

   Put back the percent and graph modifiers.

 - Fix various typos in the comments

 - Set the trace_clock before initializing a histogram with clock
   argument

   The histogram API allows the user to specific which trace clock to
   use via a "clock=" string. The histogram is set up first before the
   clock is checked. If the passed in clock is not valid, it exits
   without fully fixing up the histogram leaving it on the list and a
   use-after-free can trigger.

   Update the clock argument first and if it fails then exit gracefully
   before the histogram trigger is placed on any lists.

 - Restore :mod: trailer after parsing in ftrace_set_clr_event

   The function ftrace_set_clr_event() modifies the parse string and
   needs to put it back to what was passed in. It searches for ":mod:"
   via a strsep() but fails to put back the first ':' in the string.

   Add back the ':' in the passed in string.

 - Take trace_array reference when opening a tracer options file

   The options files are dynamically created and some tracers add their
   own options. When a tracer adds their own list of options, the
   trace_array holding them has an array to hold the list of options for
   each tracer. This array increases in size via a krealloc(), and the
   new entry gets a newly allocated array to hold the options of the new
   tracer being added.

   The element in each entry of the tracer's option array holds a
   pointer back to the trace_array, a pointer to the tracer it is
   associated to, a pointer to the flags of the option.

   The issue is that these arrays are freed when the trace_array is
   freed when its instance it represents is removed from the instances
   directory. There's a race that an open of one of these options files
   can happen when the instance is being removed.

   Add a new helper function to be called by the open function of the
   options file to iterate all existing trace_arrays under a lock and
   find the one that has the given option element in one of it's tracer
   arrays. If found, then update the associated trace_array's reference
   counter to keep it from being freed. If not found, have the open call
   return -ENODEV.

 - Disable interrupts when acquiring the lock in rb_wake_up_waiters()

   The function rb_wake_up_waiters() assumes it will be called in
   interrupt context and does not disable irqs when taking
   cpu_buffer->reader_lock, which can be called in hard interrupt
   context. The issue is in PREEMPT_RT, this function is called in
   thread context leaving this lock open to a deadlock.

   Take the lock with interrupts disabled.

 - Use rcu_assign_pointer() for tmp_ops filter hash

   The tmp_ops used in update_ftrace_direct_mod() assigns its
   filter_hash field directly, but that field is annotated as __rcu and
   sparse complains. Assign it with rcu_assign_pointer()

 - Fix use-after-free in enable_trigger_private_data_free()

   The trace_event_call is accessed through the event_trigger_data's
   trace_event_file pointer to put the trace_event_call on freeing. The
   issue is that the trace_event_file data may have been freed already
   causing a use-after-free. Add a field to the event_trigger_data that
   points directly to the trace_event_call so that it can decrement its
   reference directly without needing to go through the
   trace_event_file.

 - Fix accounting of buffer data remote headers

   trace_buffer_desc_size() and trace_remote_alloc_buffer() undercount
   the number of pages is needed for the asked for size as it doesn't
   take into account the meta data on each page. Add a helper function
   to do the calculation properly and use that in these functions.

 - Catch nr_page_va overflow in ring_buffer_desc sizing

   The number of pages per remote ring buffer is capped by
   ring_buffer_desc::nr_page_va (32 bits). A buffer_size large enough to
   overflow that field would silently allocate a descriptor smaller than
   what was asked for.

 - Do not resize the subbuf order if any per_cpu buffer is disabled

   The mmapping of ring buffers disables resizing the subbuffers, but it
   is done per-cpu whereas the subbuf size change is done for all the
   per_cpu buffers under the buffer->mutex. It could change the size of
   some while the mapping is happening on others. Have the resize of the
   subbuf order check all the per_cpu buffers under the lock to see if
   any of them is disabled before starting and causing an inconsistency
   between buffers that are being mapped.

* tag 'trace-v7.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace: (25 commits)
  ring-buffer: Check resize_disabled before publishing the new subbuf order
  tracing/remotes: Catch nr_page_va overflow in ring_buffer_desc sizing
  tracing/remotes: Account for ring buffer page header in size calculation
  tracing: Don't dereference trace_event_file in deferred trigger free
  ftrace: Use rcu_assign_pointer() for tmp_ops filter hash
  ring-buffer: Acquire the lock with irqsave in rb_wake_up_waiters()
  tracing: Take trace_array reference when opening a tracer options file
  tracing: Fix ring_buffer_read_page_size() kernel-doc
  tracing: Restore :mod: trailer after parsing in ftrace_set_clr_event()
  tracing: Fix memory corruption from a "STACKTRACE" histogram key
  tracing: Fix memory corruption from the histogram stacktrace modifier
  tracing: Undo the registration when enabling the histogram trigger fails
  tracing: Take the reference before publishing the named histogram trigger
  tracing: Set the trace clock before registering the histogram trigger
  tracing: Fix typo "preceeded" in comment
  tracing: Fix typo "availabe" in comment
  tracing: Let histogram values keep the percent and graph modifiers
  tracing: Keep the entry count when the histogram stats allocation fails
  tracing: Free histogram the field rejected for a bad modifier
  tracing: Free histogram the var ref when its initialization fails
  ...

1 files changed, 348 insertions, 0 deletions

diff --git a/Documentation/input/joydev/joystick-api.rst b/Documentation/input/joydev/joystick-api.rst
new file mode 100644
index 000000000..5db6dc6fe
--- /dev/null
+++ b/Documentation/input/joydev/joystick-api.rst
@@ -0,0 +1,348 @@
+.. _joystick-api:
+
+=====================
+Programming Interface
+=====================
+
+:Author: Ragnar Hojland Espinosa <ragnar@macula.net> - 7 Aug 1998
+
+Introduction
+============
+
+.. important::
+   This document describes legacy ``js`` interface. Newer clients are
+   encouraged to switch to the generic event (``evdev``) interface.
+
+The 1.0 driver uses a new, event based approach to the joystick driver.
+Instead of the user program polling for the joystick values, the joystick
+driver now reports only any changes of its state. See joystick-api.txt,
+joystick.h and jstest.c included in the joystick package for more
+information. The joystick device can be used in either blocking or
+nonblocking mode, and supports select() calls.
+
+For backward compatibility the old (v0.x) interface is still included.
+Any call to the joystick driver using the old interface will return values
+that are compatible to the old interface. This interface is still limited
+to 2 axes, and applications using it usually decode only 2 buttons, although
+the driver provides up to 32.
+
+Initialization
+==============
+
+Open the joystick device following the usual semantics (that is, with open).
+Since the driver now reports events instead of polling for changes,
+immediately after the open it will issue a series of synthetic events
+(JS_EVENT_INIT) that you can read to obtain the initial state of the
+joystick.
+
+By default, the device is opened in blocking mode::
+
+	int fd = open ("/dev/input/js0", O_RDONLY);
+
+
+Event Reading
+=============
+
+::
+
+	struct js_event e;
+	read (fd, &e, sizeof(e));
+
+where js_event is defined as::
+
+	struct js_event {
+		__u32 time;     /* event timestamp in milliseconds */
+		__s16 value;    /* value */
+		__u8 type;      /* event type */
+		__u8 number;    /* axis/button number */
+	};
+
+If the read is successful, it will return sizeof(e), unless you wanted to read
+more than one event per read as described in section 3.1.
+
+
+js_event.type
+-------------
+
+The possible values of ``type`` are::
+
+	#define JS_EVENT_BUTTON         0x01    /* button pressed/released */
+	#define JS_EVENT_AXIS           0x02    /* joystick moved */
+	#define JS_EVENT_INIT           0x80    /* initial state of device */
+
+As mentioned above, the driver will issue synthetic JS_EVENT_INIT ORed
+events on open. That is, if it's issuing an INIT BUTTON event, the
+current type value will be::
+
+	int type = JS_EVENT_BUTTON | JS_EVENT_INIT;	/* 0x81 */
+
+If you choose not to differentiate between synthetic or real events
+you can turn off the JS_EVENT_INIT bits::
+
+	type &= ~JS_EVENT_INIT;				/* 0x01 */
+
+
+js_event.number
+---------------
+
+The values of ``number`` correspond to the axis or button that
+generated the event. Note that they carry separate numeration (that
+is, you have both an axis 0 and a button 0). Generally,
+
+        =============== =======
+	Axis		number
+        =============== =======
+	1st Axis X	0
+	1st Axis Y	1
+	2nd Axis X	2
+	2nd Axis Y	3
+	...and so on
+        =============== =======
+
+Hats vary from one joystick type to another. Some can be moved in 8
+directions, some only in 4. The driver, however, always reports a hat as two
+independent axes, even if the hardware doesn't allow independent movement.
+
+
+js_event.value
+--------------
+
+For an axis, ``value`` is a signed integer between -32767 and +32767
+representing the position of the joystick along that axis. If you
+don't read a 0 when the joystick is ``dead``, or if it doesn't span the
+full range, you should recalibrate it (with, for example, jscal).
+
+For a button, ``value`` for a press button event is 1 and for a release
+button event is 0.
+
+Though this::
+
+	if (js_event.type == JS_EVENT_BUTTON) {
+		buttons_state ^= (1 << js_event.number);
+	}
+
+may work well if you handle JS_EVENT_INIT events separately,
+
+::
+
+	if ((js_event.type & ~JS_EVENT_INIT) == JS_EVENT_BUTTON) {
+		if (js_event.value)
+			buttons_state |= (1 << js_event.number);
+		else
+			buttons_state &= ~(1 << js_event.number);
+	}
+
+is much safer since it can't lose sync with the driver. As you would
+have to write a separate handler for JS_EVENT_INIT events in the first
+snippet, this ends up being shorter.
+
+
+js_event.time
+-------------
+
+The time an event was generated is stored in ``js_event.time``. It's a time
+in milliseconds since ... well, since sometime in the past.  This eases the
+task of detecting double clicks, figuring out if movement of axis and button
+presses happened at the same time, and similar.
+
+
+Reading
+=======
+
+If you open the device in blocking mode, a read will block (that is,
+wait) forever until an event is generated and effectively read. There
+are two alternatives if you can't afford to wait forever (which is,
+admittedly, a long time;)
+
+	a) use select to wait until there's data to be read on fd, or
+	   until it timeouts. There's a good example on the select(2)
+	   man page.
+
+	b) open the device in non-blocking mode (O_NONBLOCK)
+
+
+O_NONBLOCK
+----------
+
+If read returns -1 when reading in O_NONBLOCK mode, this isn't
+necessarily a "real" error (check errno(3)); it can just mean there
+are no events pending to be read on the driver queue. You should read
+all events on the queue (that is, until you get a -1).
+
+For example,
+
+::
+
+	while (1) {
+		while (read (fd, &e, sizeof(e)) > 0) {
+			process_event (e);
+		}
+		/* EAGAIN is returned when the queue is empty */
+		if (errno != EAGAIN) {
+			/* error */
+		}
+		/* do something interesting with processed events */
+	}
+
+One reason for emptying the queue is that if it gets full you'll start
+missing events since the queue is finite, and older events will get
+overwritten.
+
+The other reason is that you want to know all that happened, and not
+delay the processing till later.
+
+Why can the queue get full? Because you don't empty the queue as
+mentioned, or because too much time elapses from one read to another
+and too many events to store in the queue get generated. Note that
+high system load may contribute to space those reads even more.
+
+If time between reads is enough to fill the queue and lose an event,
+the driver will switch to startup mode and next time you read it,
+synthetic events (JS_EVENT_INIT) will be generated to inform you of
+the actual state of the joystick.
+
+
+.. note::
+
+ As of version 1.2.8, the queue is circular and able to hold 64
+ events. You can increment this size bumping up JS_BUFF_SIZE in
+ joystick.h and recompiling the driver.
+
+
+In the above code, you might as well want to read more than one event
+at a time using the typical read(2) functionality. For that, you would
+replace the read above with something like::
+
+	struct js_event mybuffer[0xff];
+	int i = read (fd, mybuffer, sizeof(mybuffer));
+
+In this case, read would return -1 if the queue was empty, or some
+other value in which the number of events read would be i /
+sizeof(js_event)  Again, if the buffer was full, it's a good idea to
+process the events and keep reading it until you empty the driver queue.
+
+
+IOCTLs
+======
+
+The joystick driver defines the following ioctl(2) operations::
+
+				/* function			3rd arg  */
+	#define JSIOCGAXES	/* get number of axes		char	 */
+	#define JSIOCGBUTTONS	/* get number of buttons	char	 */
+	#define JSIOCGVERSION	/* get driver version		int	 */
+	#define JSIOCGNAME(len) /* get identifier string	char	 */
+	#define JSIOCSCORR	/* set correction values	&js_corr */
+	#define JSIOCGCORR	/* get correction values	&js_corr */
+
+For example, to read the number of axes::
+
+	char number_of_axes;
+	ioctl (fd, JSIOCGAXES, &number_of_axes);
+
+
+JSIOGCVERSION
+-------------
+
+JSIOGCVERSION is a good way to check in run-time whether the running
+driver is 1.0+ and supports the event interface. If it is not, the
+IOCTL will fail. For a compile-time decision, you can test the
+JS_VERSION symbol::
+
+	#ifdef JS_VERSION
+	#if JS_VERSION > 0xsomething
+
+
+JSIOCGNAME
+----------
+
+JSIOCGNAME(len) allows you to get the name string of the joystick - the same
+as is being printed at boot time. The 'len' argument is the length of the
+buffer provided by the application asking for the name. It is used to avoid
+possible overrun should the name be too long::
+
+	char name[128];
+	if (ioctl(fd, JSIOCGNAME(sizeof(name)), name) < 0)
+		strscpy(name, "Unknown", sizeof(name));
+	printf("Name: %s\n", name);
+
+
+JSIOC[SG]CORR
+-------------
+
+For usage on JSIOC[SG]CORR I suggest you to look into jscal.c  They are
+not needed in a normal program, only in joystick calibration software
+such as jscal or kcmjoy. These IOCTLs and data types aren't considered
+to be in the stable part of the API, and therefore may change without
+warning in following releases of the driver.
+
+Both JSIOCSCORR and JSIOCGCORR expect &js_corr to be able to hold
+information for all axes. That is, struct js_corr corr[MAX_AXIS];
+
+struct js_corr is defined as::
+
+	struct js_corr {
+		__s32 coef[8];
+		__u16 prec;
+		__u16 type;
+	};
+
+and ``type``::
+
+	#define JS_CORR_NONE            0x00    /* returns raw values */
+	#define JS_CORR_BROKEN          0x01    /* broken line */
+
+
+Backward compatibility
+======================
+
+The 0.x joystick driver API is quite limited and its usage is deprecated.
+The driver offers backward compatibility, though. Here's a quick summary::
+
+	struct JS_DATA_TYPE js;
+	while (1) {
+		if (read (fd, &js, JS_RETURN) != JS_RETURN) {
+			/* error */
+		}
+		usleep (1000);
+	}
+
+As you can figure out from the example, the read returns immediately,
+with the actual state of the joystick::
+
+	struct JS_DATA_TYPE {
+		int buttons;    /* immediate button state */
+		int x;          /* immediate x axis value */
+		int y;          /* immediate y axis value */
+	};
+
+and JS_RETURN is defined as::
+
+	#define JS_RETURN       sizeof(struct JS_DATA_TYPE)
+
+To test the state of the buttons,
+
+::
+
+	first_button_state  = js.buttons & 1;
+	second_button_state = js.buttons & 2;
+
+The axis values do not have a defined range in the original 0.x driver,
+except that the values are non-negative. The 1.2.8+ drivers use a
+fixed range for reporting the values, 1 being the minimum, 128 the
+center, and 255 maximum value.
+
+The v0.8.0.2 driver also had an interface for 'digital joysticks', (now
+called Multisystem joysticks in this driver), under /dev/djsX. This driver
+doesn't try to be compatible with that interface.
+
+
+Final Notes
+===========
+
+::
+
+  ____/|	Comments, additions, and specially corrections are welcome.
+  \ o.O|	Documentation valid for at least version 1.2.8 of the joystick
+   =(_)=	driver and as usual, the ultimate source for documentation is
+     U		to "Use The Source Luke" or, at your convenience, Vojtech ;)