Merge tag 'ftrace-v6.11' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace

Pull ftrace updates from Steven Rostedt:
 "Rewrite of function graph tracer to allow multiple users

  Up until now, the function graph tracer could only have a single user
  attached to it. If another user tried to attach to the function graph
  tracer while one was already attached, it would fail. Allowing
  function graph tracer to have more than one user has been asked for
  since 2009, but it required a rewrite to the logic to pull it off so
  it never happened. Until now!

  There's three systems that trace the return of a function. That is
  kretprobes, function graph tracer, and BPF. kretprobes and function
  graph tracing both do it similarly. The difference is that kretprobes
  uses a shadow stack per callback and function graph tracer creates a
  shadow stack for all tasks. The function graph tracer method makes it
  possible to trace the return of all functions. As kretprobes now needs
  that feature too, allowing it to use function graph tracer was needed.
  BPF also wants to trace the return of many probes and its method
  doesn't scale either. Having it use function graph tracer would
  improve that.

  By allowing function graph tracer to have multiple users allows both
  kretprobes and BPF to use function graph tracer in these cases. This
  will allow kretprobes code to be removed in the future as it's version
  will no longer be needed.

  Note, function graph tracer is only limited to 16 simultaneous users,
  due to shadow stack size and allocated slots"

* tag 'ftrace-v6.11' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace: (49 commits)
  fgraph: Use str_plural() in test_graph_storage_single()
  function_graph: Add READ_ONCE() when accessing fgraph_array[]
  ftrace: Add missing kerneldoc parameters to unregister_ftrace_direct()
  function_graph: Everyone uses HAVE_FUNCTION_GRAPH_RET_ADDR_PTR, remove it
  function_graph: Fix up ftrace_graph_ret_addr()
  function_graph: Make fgraph_update_pid_func() a stub for !DYNAMIC_FTRACE
  function_graph: Rename BYTE_NUMBER to CHAR_NUMBER in selftests
  fgraph: Remove some unused functions
  ftrace: Hide one more entry in stack trace when ftrace_pid is enabled
  function_graph: Do not update pid func if CONFIG_DYNAMIC_FTRACE not enabled
  function_graph: Make fgraph_do_direct static key static
  ftrace: Fix prototypes for ftrace_startup/shutdown_subops()
  ftrace: Assign RCU list variable with rcu_assign_ptr()
  ftrace: Assign ftrace_list_end to ftrace_ops_list type cast to RCU
  ftrace: Declare function_trace_op in header to quiet sparse warning
  ftrace: Add comments to ftrace_hash_move() and friends
  ftrace: Convert "inc" parameter to bool in ftrace_hash_rec_update_modify()
  ftrace: Add comments to ftrace_hash_rec_disable/enable()
  ftrace: Remove "filter_hash" parameter from __ftrace_hash_rec_update()
  ftrace: Rename dup_hash() and comment it
  ...

Documentation/trace/ftrace-design.rst

@@ -217,18 +217,6 @@ along to ftrace_push_return_trace() instead of a stub value of 0.
 
 Similarly, when you call ftrace_return_to_handler(), pass it the frame pointer.
 
-HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
---------------------------------
-
-An arch may pass in a pointer to the return address on the stack.  This
-prevents potential stack unwinding issues where the unwinder gets out of
-sync with ret_stack and the wrong addresses are reported by
-ftrace_graph_ret_addr().
-
-Adding support for it is easy: just define the macro in asm/ftrace.h and
-pass the return address pointer as the 'retp' argument to
-ftrace_push_return_trace().
-
 HAVE_SYSCALL_TRACEPOINTS
 ------------------------
 

arch/arm64/include/asm/ftrace.h

@@ -12,17 +12,6 @@
 
 #define HAVE_FUNCTION_GRAPH_FP_TEST
 
-/*
- * HAVE_FUNCTION_GRAPH_RET_ADDR_PTR means that the architecture can provide a
- * "return address pointer" which can be used to uniquely identify a return
- * address which has been overwritten.
- *
- * On arm64 we use the address of the caller's frame record, which remains the
- * same for the lifetime of the instrumented function, unlike the return
- * address in the LR.
- */
-#define HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
-
 #ifdef CONFIG_DYNAMIC_FTRACE_WITH_ARGS
 #define ARCH_SUPPORTS_FTRACE_OPS 1
 #else

arch/csky/include/asm/ftrace.h

@@ -7,8 +7,6 @@
 
 #define HAVE_FUNCTION_GRAPH_FP_TEST
 
-#define HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
-
 #define ARCH_SUPPORTS_FTRACE_OPS 1
 
 #define MCOUNT_ADDR	((unsigned long)_mcount)

arch/loongarch/include/asm/ftrace.h

@@ -28,7 +28,6 @@ struct dyn_ftrace;
 struct dyn_arch_ftrace { };
 
 #define ARCH_SUPPORTS_FTRACE_OPS 1
-#define HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
 
 #define ftrace_init_nop ftrace_init_nop
 int ftrace_init_nop(struct module *mod, struct dyn_ftrace *rec);

arch/powerpc/include/asm/ftrace.h

@@ -8,8 +8,6 @@
 #define MCOUNT_ADDR		((unsigned long)(_mcount))
 #define MCOUNT_INSN_SIZE	4 /* sizeof mcount call */
 
-#define HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
-
 /* Ignore unused weak functions which will have larger offsets */
 #if defined(CONFIG_MPROFILE_KERNEL) || defined(CONFIG_ARCH_USING_PATCHABLE_FUNCTION_ENTRY)
 #define FTRACE_MCOUNT_MAX_OFFSET	16

arch/riscv/include/asm/ftrace.h

@@ -11,7 +11,6 @@
 #if defined(CONFIG_FUNCTION_GRAPH_TRACER) && defined(CONFIG_FRAME_POINTER)
 #define HAVE_FUNCTION_GRAPH_FP_TEST
 #endif
-#define HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
 
 #define ARCH_SUPPORTS_FTRACE_OPS 1
 #ifndef __ASSEMBLY__

arch/s390/include/asm/ftrace.h

@@ -2,7 +2,6 @@
 #ifndef _ASM_S390_FTRACE_H
 #define _ASM_S390_FTRACE_H
 
-#define HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
 #define ARCH_SUPPORTS_FTRACE_OPS 1
 #define MCOUNT_INSN_SIZE	6
 

arch/x86/include/asm/ftrace.h

@@ -20,8 +20,6 @@
 #define ARCH_SUPPORTS_FTRACE_OPS 1
 #endif
 
-#define HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
-
 #ifndef __ASSEMBLY__
 extern void __fentry__(void);
 

include/linux/ftrace.h

@@ -227,6 +227,7 @@ ftrace_func_t ftrace_ops_get_func(struct ftrace_ops *ops);
  *             ftrace_enabled.
  * DIRECT - Used by the direct ftrace_ops helper for direct functions
  *            (internal ftrace only, should not be used by others)
+ * SUBOP  - Is controlled by another op in field managed.
  */
 enum {
 	FTRACE_OPS_FL_ENABLED			= BIT(0),
@@ -247,6 +248,7 @@ enum {
 	FTRACE_OPS_FL_TRACE_ARRAY		= BIT(15),
 	FTRACE_OPS_FL_PERMANENT                 = BIT(16),
 	FTRACE_OPS_FL_DIRECT			= BIT(17),
+	FTRACE_OPS_FL_SUBOP			= BIT(18),
 };
 
 #ifndef CONFIG_DYNAMIC_FTRACE_WITH_ARGS
@@ -334,7 +336,9 @@ struct ftrace_ops {
 	unsigned long			trampoline;
 	unsigned long			trampoline_size;
 	struct list_head		list;
+	struct list_head		subop_list;
 	ftrace_ops_func_t		ops_func;
+	struct ftrace_ops		*managed;
 #ifdef CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
 	unsigned long			direct_call;
 #endif
@@ -509,6 +513,15 @@ static inline void stack_tracer_disable(void) { }
 static inline void stack_tracer_enable(void) { }
 #endif
 
+enum {
+	FTRACE_UPDATE_CALLS		= (1 << 0),
+	FTRACE_DISABLE_CALLS		= (1 << 1),
+	FTRACE_UPDATE_TRACE_FUNC	= (1 << 2),
+	FTRACE_START_FUNC_RET		= (1 << 3),
+	FTRACE_STOP_FUNC_RET		= (1 << 4),
+	FTRACE_MAY_SLEEP		= (1 << 5),
+};
+
 #ifdef CONFIG_DYNAMIC_FTRACE
 
 void ftrace_arch_code_modify_prepare(void);
@@ -603,15 +616,6 @@ void ftrace_set_global_notrace(unsigned char *buf, int len, int reset);
 void ftrace_free_filter(struct ftrace_ops *ops);
 void ftrace_ops_set_global_filter(struct ftrace_ops *ops);
 
-enum {
-	FTRACE_UPDATE_CALLS		= (1 << 0),
-	FTRACE_DISABLE_CALLS		= (1 << 1),
-	FTRACE_UPDATE_TRACE_FUNC	= (1 << 2),
-	FTRACE_START_FUNC_RET		= (1 << 3),
-	FTRACE_STOP_FUNC_RET		= (1 << 4),
-	FTRACE_MAY_SLEEP		= (1 << 5),
-};
-
 /*
  * The FTRACE_UPDATE_* enum is used to pass information back
  * from the ftrace_update_record() and ftrace_test_record()
@@ -1027,19 +1031,31 @@ struct ftrace_graph_ret {
 	unsigned long long rettime;
 } __packed;
 
+struct fgraph_ops;
+
 /* Type of the callback handlers for tracing function graph*/
-typedef void (*trace_func_graph_ret_t)(struct ftrace_graph_ret *); /* return */
-typedef int (*trace_func_graph_ent_t)(struct ftrace_graph_ent *); /* entry */
+typedef void (*trace_func_graph_ret_t)(struct ftrace_graph_ret *,
+				       struct fgraph_ops *); /* return */
+typedef int (*trace_func_graph_ent_t)(struct ftrace_graph_ent *,
+				      struct fgraph_ops *); /* entry */
 
-extern int ftrace_graph_entry_stub(struct ftrace_graph_ent *trace);
+extern int ftrace_graph_entry_stub(struct ftrace_graph_ent *trace, struct fgraph_ops *gops);
+bool ftrace_pids_enabled(struct ftrace_ops *ops);
 
 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
 
 struct fgraph_ops {
 	trace_func_graph_ent_t		entryfunc;
 	trace_func_graph_ret_t		retfunc;
+	struct ftrace_ops		ops; /* for the hash lists */
+	void				*private;
+	trace_func_graph_ent_t		saved_func;
+	int				idx;
 };
 
+void *fgraph_reserve_data(int idx, int size_bytes);
+void *fgraph_retrieve_data(int idx, int *size_bytes);
+
 /*
  * Stack of return addresses for functions
  * of a thread.
@@ -1055,9 +1071,7 @@ struct ftrace_ret_stack {
 #ifdef HAVE_FUNCTION_GRAPH_FP_TEST
 	unsigned long fp;
 #endif
-#ifdef HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
 	unsigned long *retp;
-#endif
 };
 
 /*
@@ -1072,10 +1086,11 @@ function_graph_enter(unsigned long ret, unsigned long func,
 		     unsigned long frame_pointer, unsigned long *retp);
 
 struct ftrace_ret_stack *
-ftrace_graph_get_ret_stack(struct task_struct *task, int idx);
+ftrace_graph_get_ret_stack(struct task_struct *task, int skip);
 
 unsigned long ftrace_graph_ret_addr(struct task_struct *task, int *idx,
 				    unsigned long ret, unsigned long *retp);
+unsigned long *fgraph_get_task_var(struct fgraph_ops *gops);
 
 /*
  * Sometimes we don't want to trace a function with the function
@@ -1114,6 +1129,9 @@ extern void ftrace_graph_init_task(struct task_struct *t);
 extern void ftrace_graph_exit_task(struct task_struct *t);
 extern void ftrace_graph_init_idle_task(struct task_struct *t, int cpu);
 
+/* Used by assembly, but to quiet sparse warnings */
+extern struct ftrace_ops *function_trace_op;
+
 static inline void pause_graph_tracing(void)
 {
 	atomic_inc(&current->tracing_graph_pause);

include/linux/sched.h

@@ -1413,7 +1413,7 @@ struct task_struct {
 	int				curr_ret_depth;
 
 	/* Stack of return addresses for return function tracing: */
-	struct ftrace_ret_stack		*ret_stack;
+	unsigned long			*ret_stack;
 
 	/* Timestamp for last schedule: */
 	unsigned long long		ftrace_timestamp;

include/linux/trace_recursion.h

@@ -44,35 +44,6 @@ enum {
  */
 	TRACE_IRQ_BIT,
 
-	/* Set if the function is in the set_graph_function file */
-	TRACE_GRAPH_BIT,
-
-	/*
-	 * In the very unlikely case that an interrupt came in
-	 * at a start of graph tracing, and we want to trace
-	 * the function in that interrupt, the depth can be greater
-	 * than zero, because of the preempted start of a previous
-	 * trace. In an even more unlikely case, depth could be 2
-	 * if a softirq interrupted the start of graph tracing,
-	 * followed by an interrupt preempting a start of graph
-	 * tracing in the softirq, and depth can even be 3
-	 * if an NMI came in at the start of an interrupt function
-	 * that preempted a softirq start of a function that
-	 * preempted normal context!!!! Luckily, it can't be
-	 * greater than 3, so the next two bits are a mask
-	 * of what the depth is when we set TRACE_GRAPH_BIT
-	 */
-
-	TRACE_GRAPH_DEPTH_START_BIT,
-	TRACE_GRAPH_DEPTH_END_BIT,
-
-	/*
-	 * To implement set_graph_notrace, if this bit is set, we ignore
-	 * function graph tracing of called functions, until the return
-	 * function is called to clear it.
-	 */
-	TRACE_GRAPH_NOTRACE_BIT,
-
 	/* Used to prevent recursion recording from recursing. */
 	TRACE_RECORD_RECURSION_BIT,
 };
@@ -81,16 +52,6 @@ enum {
 #define trace_recursion_clear(bit)	do { (current)->trace_recursion &= ~(1<<(bit)); } while (0)
 #define trace_recursion_test(bit)	((current)->trace_recursion & (1<<(bit)))
 
-#define trace_recursion_depth() \
-	(((current)->trace_recursion >> TRACE_GRAPH_DEPTH_START_BIT) & 3)
-#define trace_recursion_set_depth(depth) \
-	do {								\
-		current->trace_recursion &=				\
-			~(3 << TRACE_GRAPH_DEPTH_START_BIT);		\
-		current->trace_recursion |=				\
-			((depth) & 3) << TRACE_GRAPH_DEPTH_START_BIT;	\
-	} while (0)
-
 #define TRACE_CONTEXT_BITS	4
 
 #define TRACE_FTRACE_START	TRACE_FTRACE_BIT

kernel/trace/fgraph.c

@@ -7,9 +7,11 @@
  *
  * Highly modified by Steven Rostedt (VMware).
  */
+#include <linux/bits.h>
 #include <linux/jump_label.h>
 #include <linux/suspend.h>
 #include <linux/ftrace.h>
+#include <linux/static_call.h>
 #include <linux/slab.h>
 
 #include <trace/events/sched.h>
@@ -17,17 +19,447 @@
 #include "ftrace_internal.h"
 #include "trace.h"
 
-#ifdef CONFIG_DYNAMIC_FTRACE
-#define ASSIGN_OPS_HASH(opsname, val) \
-	.func_hash		= val, \
-	.local_hash.regex_lock	= __MUTEX_INITIALIZER(opsname.local_hash.regex_lock),
-#else
-#define ASSIGN_OPS_HASH(opsname, val)
-#endif
+/*
+ * FGRAPH_FRAME_SIZE:	Size in bytes of the meta data on the shadow stack
+ * FGRAPH_FRAME_OFFSET:	Size in long words of the meta data frame
+ */
+#define FGRAPH_FRAME_SIZE	sizeof(struct ftrace_ret_stack)
+#define FGRAPH_FRAME_OFFSET	DIV_ROUND_UP(FGRAPH_FRAME_SIZE, sizeof(long))
+
+/*
+ * On entry to a function (via function_graph_enter()), a new fgraph frame
+ * (ftrace_ret_stack) is pushed onto the stack as well as a word that
+ * holds a bitmask and a type (called "bitmap"). The bitmap is defined as:
+ *
+ * bits:  0 -  9	offset in words from the previous ftrace_ret_stack
+ *
+ * bits: 10 - 11	Type of storage
+ *			  0 - reserved
+ *			  1 - bitmap of fgraph_array index
+ *			  2 - reserved data
+ *
+ * For type with "bitmap of fgraph_array index" (FGRAPH_TYPE_BITMAP):
+ *  bits: 12 - 27	The bitmap of fgraph_ops fgraph_array index
+ *			That is, it's a bitmask of 0-15 (16 bits)
+ *			where if a corresponding ops in the fgraph_array[]
+ *			expects a callback from the return of the function
+ *			it's corresponding bit will be set.
+ *
+ *
+ * The top of the ret_stack (when not empty) will always have a reference
+ * word that points to the last fgraph frame that was saved.
+ *
+ * For reserved data:
+ *  bits: 12 - 17	The size in words that is stored
+ *  bits: 18 - 23	The index of fgraph_array, which shows who is stored
+ *
+ * That is, at the end of function_graph_enter, if the first and forth
+ * fgraph_ops on the fgraph_array[] (index 0 and 3) needs their retfunc called
+ * on the return of the function being traced, and the forth fgraph_ops
+ * stored two words of data, this is what will be on the task's shadow
+ * ret_stack: (the stack grows upward)
+ *
+ *  ret_stack[SHADOW_STACK_OFFSET]
+ * | SHADOW_STACK_TASK_VARS(ret_stack)[15]      |
+ * ...
+ * | SHADOW_STACK_TASK_VARS(ret_stack)[0]       |
+ *  ret_stack[SHADOW_STACK_MAX_OFFSET]
+ * ...
+ * |                                            | <- task->curr_ret_stack
+ * +--------------------------------------------+
+ * | (3 << 12) | (3 << 10) | FGRAPH_FRAME_OFFSET|
+ * |         *or put another way*               |
+ * | (3 << FGRAPH_DATA_INDEX_SHIFT)| \          | This is for fgraph_ops[3].
+ * | ((2 - 1) << FGRAPH_DATA_SHIFT)| \          | The data size is 2 words.
+ * | (FGRAPH_TYPE_DATA << FGRAPH_TYPE_SHIFT)| \ |
+ * | (offset2:FGRAPH_FRAME_OFFSET+3)            | <- the offset2 is from here
+ * +--------------------------------------------+ ( It is 4 words from the ret_stack)
+ * |            STORED DATA WORD 2              |
+ * |            STORED DATA WORD 1              |
+ * +--------------------------------------------+
+ * | (9 << 12) | (1 << 10) | FGRAPH_FRAME_OFFSET|
+ * |         *or put another way*               |
+ * | (BIT(3)|BIT(0)) << FGRAPH_INDEX_SHIFT | \  |
+ * | FGRAPH_TYPE_BITMAP << FGRAPH_TYPE_SHIFT| \ |
+ * | (offset1:FGRAPH_FRAME_OFFSET)              | <- the offset1 is from here
+ * +--------------------------------------------+
+ * | struct ftrace_ret_stack                    |
+ * |   (stores the saved ret pointer)           | <- the offset points here
+ * +--------------------------------------------+
+ * |                 (X) | (N)                  | ( N words away from
+ * |                                            |   previous ret_stack)
+ * ...
+ * ret_stack[0]
+ *
+ * If a backtrace is required, and the real return pointer needs to be
+ * fetched, then it looks at the task's curr_ret_stack offset, if it
+ * is greater than zero (reserved, or right before popped), it would mask
+ * the value by FGRAPH_FRAME_OFFSET_MASK to get the offset of the
+ * ftrace_ret_stack structure stored on the shadow stack.
+ */
+
+/*
+ * The following is for the top word on the stack:
+ *
+ *   FGRAPH_FRAME_OFFSET (0-9) holds the offset delta to the fgraph frame
+ *   FGRAPH_TYPE (10-11) holds the type of word this is.
+ *     (RESERVED or BITMAP)
+ */
+#define FGRAPH_FRAME_OFFSET_BITS	10
+#define FGRAPH_FRAME_OFFSET_MASK	GENMASK(FGRAPH_FRAME_OFFSET_BITS - 1, 0)
+
+#define FGRAPH_TYPE_BITS	2
+#define FGRAPH_TYPE_MASK	GENMASK(FGRAPH_TYPE_BITS - 1, 0)
+#define FGRAPH_TYPE_SHIFT	FGRAPH_FRAME_OFFSET_BITS
+
+enum {
+	FGRAPH_TYPE_RESERVED	= 0,
+	FGRAPH_TYPE_BITMAP	= 1,
+	FGRAPH_TYPE_DATA	= 2,
+};
+
+/*
+ * For BITMAP type:
+ *   FGRAPH_INDEX (12-27) bits holding the gops index wanting return callback called
+ */
+#define FGRAPH_INDEX_BITS	16
+#define FGRAPH_INDEX_MASK	GENMASK(FGRAPH_INDEX_BITS - 1, 0)
+#define FGRAPH_INDEX_SHIFT	(FGRAPH_TYPE_SHIFT + FGRAPH_TYPE_BITS)
+
+/*
+ * For DATA type:
+ *  FGRAPH_DATA (12-17) bits hold the size of data (in words)
+ *  FGRAPH_INDEX (18-23) bits hold the index for which gops->idx the data is for
+ *
+ * Note:
+ *  data_size == 0 means 1 word, and 31 (=2^5 - 1) means 32 words.
+ */
+#define FGRAPH_DATA_BITS	5
+#define FGRAPH_DATA_MASK	GENMASK(FGRAPH_DATA_BITS - 1, 0)
+#define FGRAPH_DATA_SHIFT	(FGRAPH_TYPE_SHIFT + FGRAPH_TYPE_BITS)
+#define FGRAPH_MAX_DATA_SIZE (sizeof(long) * (1 << FGRAPH_DATA_BITS))
+
+#define FGRAPH_DATA_INDEX_BITS	4
+#define FGRAPH_DATA_INDEX_MASK	GENMASK(FGRAPH_DATA_INDEX_BITS - 1, 0)
+#define FGRAPH_DATA_INDEX_SHIFT	(FGRAPH_DATA_SHIFT + FGRAPH_DATA_BITS)
+
+#define FGRAPH_MAX_INDEX	\
+	((FGRAPH_INDEX_SIZE << FGRAPH_DATA_BITS) + FGRAPH_RET_INDEX)
+
+#define FGRAPH_ARRAY_SIZE	FGRAPH_INDEX_BITS
+
+/*
+ * SHADOW_STACK_SIZE:	The size in bytes of the entire shadow stack
+ * SHADOW_STACK_OFFSET:	The size in long words of the shadow stack
+ * SHADOW_STACK_MAX_OFFSET: The max offset of the stack for a new frame to be added
+ */
+#define SHADOW_STACK_SIZE	(PAGE_SIZE)
+#define SHADOW_STACK_OFFSET	(SHADOW_STACK_SIZE / sizeof(long))
+/* Leave on a buffer at the end */
+#define SHADOW_STACK_MAX_OFFSET				\
+	(SHADOW_STACK_OFFSET - (FGRAPH_FRAME_OFFSET + 1 + FGRAPH_ARRAY_SIZE))
+
+/* RET_STACK():		Return the frame from a given @offset from task @t */
+#define RET_STACK(t, offset) ((struct ftrace_ret_stack *)(&(t)->ret_stack[offset]))
+
+/*
+ * Each fgraph_ops has a reservered unsigned long at the end (top) of the
+ * ret_stack to store task specific state.
+ */
+#define SHADOW_STACK_TASK_VARS(ret_stack) \
+	((unsigned long *)(&(ret_stack)[SHADOW_STACK_OFFSET - FGRAPH_ARRAY_SIZE]))
 
 DEFINE_STATIC_KEY_FALSE(kill_ftrace_graph);
 int ftrace_graph_active;
 
+static struct fgraph_ops *fgraph_array[FGRAPH_ARRAY_SIZE];
+static unsigned long fgraph_array_bitmask;
+
+/* LRU index table for fgraph_array */
+static int fgraph_lru_table[FGRAPH_ARRAY_SIZE];
+static int fgraph_lru_next;
+static int fgraph_lru_last;
+
+/* Initialize fgraph_lru_table with unused index */
+static void fgraph_lru_init(void)
+{
+	int i;
+
+	for (i = 0; i < FGRAPH_ARRAY_SIZE; i++)
+		fgraph_lru_table[i] = i;
+}
+
+/* Release the used index to the LRU table */
+static int fgraph_lru_release_index(int idx)
+{
+	if (idx < 0 || idx >= FGRAPH_ARRAY_SIZE ||
+	    WARN_ON_ONCE(fgraph_lru_table[fgraph_lru_last] != -1))
+		return -1;
+
+	fgraph_lru_table[fgraph_lru_last] = idx;
+	fgraph_lru_last = (fgraph_lru_last + 1) % FGRAPH_ARRAY_SIZE;
+
+	clear_bit(idx, &fgraph_array_bitmask);
+	return 0;
+}
+
+/* Allocate a new index from LRU table */
+static int fgraph_lru_alloc_index(void)
+{
+	int idx = fgraph_lru_table[fgraph_lru_next];
+
+	/* No id is available */
+	if (idx == -1)
+		return -1;
+
+	fgraph_lru_table[fgraph_lru_next] = -1;
+	fgraph_lru_next = (fgraph_lru_next + 1) % FGRAPH_ARRAY_SIZE;
+
+	set_bit(idx, &fgraph_array_bitmask);
+	return idx;
+}
+
+/* Get the offset to the fgraph frame from a ret_stack value */
+static inline int __get_offset(unsigned long val)
+{
+	return val & FGRAPH_FRAME_OFFSET_MASK;
+}
+
+/* Get the type of word from a ret_stack value */
+static inline int __get_type(unsigned long val)
+{
+	return (val >> FGRAPH_TYPE_SHIFT) & FGRAPH_TYPE_MASK;
+}
+
+/* Get the data_index for a DATA type ret_stack word */
+static inline int __get_data_index(unsigned long val)
+{
+	return (val >> FGRAPH_DATA_INDEX_SHIFT) & FGRAPH_DATA_INDEX_MASK;
+}
+
+/* Get the data_size for a DATA type ret_stack word */
+static inline int __get_data_size(unsigned long val)
+{
+	return ((val >> FGRAPH_DATA_SHIFT) & FGRAPH_DATA_MASK) + 1;
+}
+
+/* Get the word from the ret_stack at @offset */
+static inline unsigned long get_fgraph_entry(struct task_struct *t, int offset)
+{
+	return t->ret_stack[offset];
+}
+
+/* Get the FRAME_OFFSET from the word from the @offset on ret_stack */
+static inline int get_frame_offset(struct task_struct *t, int offset)
+{
+	return __get_offset(t->ret_stack[offset]);
+}
+
+/* For BITMAP type: get the bitmask from the @offset at ret_stack */
+static inline unsigned long
+get_bitmap_bits(struct task_struct *t, int offset)
+{
+	return (t->ret_stack[offset] >> FGRAPH_INDEX_SHIFT) & FGRAPH_INDEX_MASK;
+}
+
+/* Write the bitmap to the ret_stack at @offset (does index, offset and bitmask) */
+static inline void
+set_bitmap(struct task_struct *t, int offset, unsigned long bitmap)
+{
+	t->ret_stack[offset] = (bitmap << FGRAPH_INDEX_SHIFT) |
+		(FGRAPH_TYPE_BITMAP << FGRAPH_TYPE_SHIFT) | FGRAPH_FRAME_OFFSET;
+}
+
+/* For DATA type: get the data saved under the ret_stack word at @offset */
+static inline void *get_data_type_data(struct task_struct *t, int offset)
+{
+	unsigned long val = t->ret_stack[offset];
+
+	if (__get_type(val) != FGRAPH_TYPE_DATA)
+		return NULL;
+	offset -= __get_data_size(val);
+	return (void *)&t->ret_stack[offset];
+}
+
+/* Create the ret_stack word for a DATA type */
+static inline unsigned long make_data_type_val(int idx, int size, int offset)
+{
+	return (idx << FGRAPH_DATA_INDEX_SHIFT) |
+		((size - 1) << FGRAPH_DATA_SHIFT) |
+		(FGRAPH_TYPE_DATA << FGRAPH_TYPE_SHIFT) | offset;
+}
+
+/* ftrace_graph_entry set to this to tell some archs to run function graph */
+static int entry_run(struct ftrace_graph_ent *trace, struct fgraph_ops *ops)
+{
+	return 0;
+}
+
+/* ftrace_graph_return set to this to tell some archs to run function graph */
+static void return_run(struct ftrace_graph_ret *trace, struct fgraph_ops *ops)
+{
+}
+
+static void ret_stack_set_task_var(struct task_struct *t, int idx, long val)
+{
+	unsigned long *gvals = SHADOW_STACK_TASK_VARS(t->ret_stack);
+
+	gvals[idx] = val;
+}
+
+static unsigned long *
+ret_stack_get_task_var(struct task_struct *t, int idx)
+{
+	unsigned long *gvals = SHADOW_STACK_TASK_VARS(t->ret_stack);
+
+	return &gvals[idx];
+}
+
+static void ret_stack_init_task_vars(unsigned long *ret_stack)
+{
+	unsigned long *gvals = SHADOW_STACK_TASK_VARS(ret_stack);
+
+	memset(gvals, 0, sizeof(*gvals) * FGRAPH_ARRAY_SIZE);
+}
+
+/**
+ * fgraph_reserve_data - Reserve storage on the task's ret_stack
+ * @idx:	The index of fgraph_array
+ * @size_bytes: The size in bytes to reserve
+ *
+ * Reserves space of up to FGRAPH_MAX_DATA_SIZE bytes on the
+ * task's ret_stack shadow stack, for a given fgraph_ops during
+ * the entryfunc() call. If entryfunc() returns zero, the storage
+ * is discarded. An entryfunc() can only call this once per iteration.
+ * The fgraph_ops retfunc() can retrieve this stored data with
+ * fgraph_retrieve_data().
+ *
+ * Returns: On success, a pointer to the data on the stack.
+ *   Otherwise, NULL if there's not enough space left on the
+ *   ret_stack for the data, or if fgraph_reserve_data() was called
+ *   more than once for a single entryfunc() call.
+ */
+void *fgraph_reserve_data(int idx, int size_bytes)
+{
+	unsigned long val;
+	void *data;
+	int curr_ret_stack = current->curr_ret_stack;
+	int data_size;
+
+	if (size_bytes > FGRAPH_MAX_DATA_SIZE)
+		return NULL;
+
+	/* Convert the data size to number of longs. */
+	data_size = (size_bytes + sizeof(long) - 1) >> (sizeof(long) == 4 ? 2 : 3);
+
+	val = get_fgraph_entry(current, curr_ret_stack - 1);
+	data = &current->ret_stack[curr_ret_stack];
+
+	curr_ret_stack += data_size + 1;
+	if (unlikely(curr_ret_stack >= SHADOW_STACK_MAX_OFFSET))
+		return NULL;
+
+	val = make_data_type_val(idx, data_size, __get_offset(val) + data_size + 1);
+
+	/* Set the last word to be reserved */
+	current->ret_stack[curr_ret_stack - 1] = val;
+
+	/* Make sure interrupts see this */
+	barrier();
+	current->curr_ret_stack = curr_ret_stack;
+	/* Again sync with interrupts, and reset reserve */
+	current->ret_stack[curr_ret_stack - 1] = val;
+
+	return data;
+}
+
+/**
+ * fgraph_retrieve_data - Retrieve stored data from fgraph_reserve_data()
+ * @idx:	the index of fgraph_array (fgraph_ops::idx)
+ * @size_bytes: pointer to retrieved data size.
+ *
+ * This is to be called by a fgraph_ops retfunc(), to retrieve data that
+ * was stored by the fgraph_ops entryfunc() on the function entry.
+ * That is, this will retrieve the data that was reserved on the
+ * entry of the function that corresponds to the exit of the function
+ * that the fgraph_ops retfunc() is called on.
+ *
+ * Returns: The stored data from fgraph_reserve_data() called by the
+ *    matching entryfunc() for the retfunc() this is called from.
+ *   Or NULL if there was nothing stored.
+ */
+void *fgraph_retrieve_data(int idx, int *size_bytes)
+{
+	int offset = current->curr_ret_stack - 1;
+	unsigned long val;
+
+	val = get_fgraph_entry(current, offset);
+	while (__get_type(val) == FGRAPH_TYPE_DATA) {
+		if (__get_data_index(val) == idx)
+			goto found;
+		offset -= __get_data_size(val) + 1;
+		val = get_fgraph_entry(current, offset);
+	}
+	return NULL;
+found:
+	if (size_bytes)
+		*size_bytes = __get_data_size(val) * sizeof(long);
+	return get_data_type_data(current, offset);
+}
+
+/**
+ * fgraph_get_task_var - retrieve a task specific state variable
+ * @gops: The ftrace_ops that owns the task specific variable
+ *
+ * Every registered fgraph_ops has a task state variable
+ * reserved on the task's ret_stack. This function returns the
+ * address to that variable.
+ *
+ * Returns the address to the fgraph_ops @gops tasks specific
+ * unsigned long variable.
+ */
+unsigned long *fgraph_get_task_var(struct fgraph_ops *gops)
+{
+	return ret_stack_get_task_var(current, gops->idx);
+}
+
+/*
+ * @offset: The offset into @t->ret_stack to find the ret_stack entry
+ * @frame_offset: Where to place the offset into @t->ret_stack of that entry
+ *
+ * Returns a pointer to the previous ret_stack below @offset or NULL
+ *   when it reaches the bottom of the stack.
+ *
+ * Calling this with:
+ *
+ *   offset = task->curr_ret_stack;
+ *   do {
+ *	ret_stack = get_ret_stack(task, offset, &offset);
+ *   } while (ret_stack);
+ *
+ * Will iterate through all the ret_stack entries from curr_ret_stack
+ * down to the first one.
+ */
+static inline struct ftrace_ret_stack *
+get_ret_stack(struct task_struct *t, int offset, int *frame_offset)
+{
+	int offs;
+
+	BUILD_BUG_ON(FGRAPH_FRAME_SIZE % sizeof(long));
+
+	if (unlikely(offset <= 0))
+		return NULL;
+
+	offs = get_frame_offset(t, --offset);
+	if (WARN_ON_ONCE(offs <= 0 || offs > offset))
+		return NULL;
+
+	offset -= offs;
+
+	*frame_offset = offset;
+	return RET_STACK(t, offset);
+}
+
 /* Both enabled by default (can be cleared by function_graph tracer flags */
 static bool fgraph_sleep_time = true;
 
@@ -51,6 +483,27 @@ int __weak ftrace_disable_ftrace_graph_caller(void)
 }
 #endif
 
+int ftrace_graph_entry_stub(struct ftrace_graph_ent *trace,
+			    struct fgraph_ops *gops)
+{
+	return 0;
+}
+
+static void ftrace_graph_ret_stub(struct ftrace_graph_ret *trace,
+				  struct fgraph_ops *gops)
+{
+}
+
+static struct fgraph_ops fgraph_stub = {
+	.entryfunc = ftrace_graph_entry_stub,
+	.retfunc = ftrace_graph_ret_stub,
+};
+
+static struct fgraph_ops *fgraph_direct_gops = &fgraph_stub;
+DEFINE_STATIC_CALL(fgraph_func, ftrace_graph_entry_stub);
+DEFINE_STATIC_CALL(fgraph_retfunc, ftrace_graph_ret_stub);
+static DEFINE_STATIC_KEY_TRUE(fgraph_do_direct);
+
 /**
  * ftrace_graph_stop - set to permanently disable function graph tracing
  *
@@ -67,10 +520,13 @@ void ftrace_graph_stop(void)
 /* Add a function return address to the trace stack on thread info.*/
 static int
 ftrace_push_return_trace(unsigned long ret, unsigned long func,
-			 unsigned long frame_pointer, unsigned long *retp)
+			 unsigned long frame_pointer, unsigned long *retp,
+			 int fgraph_idx)
 {
+	struct ftrace_ret_stack *ret_stack;
 	unsigned long long calltime;
-	int index;
+	unsigned long val;
+	int offset;
 
 	if (unlikely(ftrace_graph_is_dead()))
 		return -EBUSY;
@@ -78,32 +534,67 @@ ftrace_push_return_trace(unsigned long ret, unsigned long func,
 	if (!current->ret_stack)
 		return -EBUSY;
 
+	BUILD_BUG_ON(SHADOW_STACK_SIZE % sizeof(long));
+
+	/* Set val to "reserved" with the delta to the new fgraph frame */
+	val = (FGRAPH_TYPE_RESERVED << FGRAPH_TYPE_SHIFT) | FGRAPH_FRAME_OFFSET;
+
 	/*
 	 * We must make sure the ret_stack is tested before we read
 	 * anything else.
 	 */
 	smp_rmb();
 
-	/* The return trace stack is full */
-	if (current->curr_ret_stack == FTRACE_RETFUNC_DEPTH - 1) {
+	/*
+	 * Check if there's room on the shadow stack to fit a fraph frame
+	 * and a bitmap word.
+	 */
+	if (current->curr_ret_stack + FGRAPH_FRAME_OFFSET + 1 >= SHADOW_STACK_MAX_OFFSET) {
 		atomic_inc(&current->trace_overrun);
 		return -EBUSY;
 	}
 
 	calltime = trace_clock_local();
 
-	index = ++current->curr_ret_stack;
+	offset = READ_ONCE(current->curr_ret_stack);
+	ret_stack = RET_STACK(current, offset);
+	offset += FGRAPH_FRAME_OFFSET;
+
+	/* ret offset = FGRAPH_FRAME_OFFSET ; type = reserved */
+	current->ret_stack[offset] = val;
+	ret_stack->ret = ret;
+	/*
+	 * The unwinders expect curr_ret_stack to point to either zero
+	 * or an offset where to find the next ret_stack. Even though the
+	 * ret stack might be bogus, we want to write the ret and the
+	 * offset to find the ret_stack before we increment the stack point.
+	 * If an interrupt comes in now before we increment the curr_ret_stack
+	 * it may blow away what we wrote. But that's fine, because the
+	 * offset will still be correct (even though the 'ret' won't be).
+	 * What we worry about is the offset being correct after we increment
+	 * the curr_ret_stack and before we update that offset, as if an
+	 * interrupt comes in and does an unwind stack dump, it will need
+	 * at least a correct offset!
+	 */
+	barrier();
+	WRITE_ONCE(current->curr_ret_stack, offset + 1);
+	/*
+	 * This next barrier is to ensure that an interrupt coming in
+	 * will not corrupt what we are about to write.
+	 */
 	barrier();
-	current->ret_stack[index].ret = ret;
-	current->ret_stack[index].func = func;
-	current->ret_stack[index].calltime = calltime;
+
+	/* Still keep it reserved even if an interrupt came in */
+	current->ret_stack[offset] = val;
+
+	ret_stack->ret = ret;
+	ret_stack->func = func;
+	ret_stack->calltime = calltime;
 #ifdef HAVE_FUNCTION_GRAPH_FP_TEST
-	current->ret_stack[index].fp = frame_pointer;
-#endif
-#ifdef HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
-	current->ret_stack[index].retp = retp;
+	ret_stack->fp = frame_pointer;
 #endif
-	return 0;
+	ret_stack->retp = retp;
+	return offset;
 }
 
 /*
@@ -120,44 +611,85 @@ ftrace_push_return_trace(unsigned long ret, unsigned long func,
 # define MCOUNT_INSN_SIZE 0
 #endif
 
+/* If the caller does not use ftrace, call this function. */
 int function_graph_enter(unsigned long ret, unsigned long func,
 			 unsigned long frame_pointer, unsigned long *retp)
 {
 	struct ftrace_graph_ent trace;
+	unsigned long bitmap = 0;
+	int offset;
+	int i;
 
 	trace.func = func;
 	trace.depth = ++current->curr_ret_depth;
 
-	if (ftrace_push_return_trace(ret, func, frame_pointer, retp))
+	offset = ftrace_push_return_trace(ret, func, frame_pointer, retp, 0);
+	if (offset < 0)
 		goto out;
 
-	/* Only trace if the calling function expects to */
-	if (!ftrace_graph_entry(&trace))
+#ifdef CONFIG_HAVE_STATIC_CALL
+	if (static_branch_likely(&fgraph_do_direct)) {
+		int save_curr_ret_stack = current->curr_ret_stack;
+
+		if (static_call(fgraph_func)(&trace, fgraph_direct_gops))
+			bitmap |= BIT(fgraph_direct_gops->idx);
+		else
+			/* Clear out any saved storage */
+			current->curr_ret_stack = save_curr_ret_stack;
+	} else
+#endif
+	{
+		for_each_set_bit(i, &fgraph_array_bitmask,
+					 sizeof(fgraph_array_bitmask) * BITS_PER_BYTE) {
+			struct fgraph_ops *gops = READ_ONCE(fgraph_array[i]);
+			int save_curr_ret_stack;
+
+			if (gops == &fgraph_stub)
+				continue;
+
+			save_curr_ret_stack = current->curr_ret_stack;
+			if (ftrace_ops_test(&gops->ops, func, NULL) &&
+			    gops->entryfunc(&trace, gops))
+				bitmap |= BIT(i);
+			else
+				/* Clear out any saved storage */
+				current->curr_ret_stack = save_curr_ret_stack;
+		}
+	}
+
+	if (!bitmap)
 		goto out_ret;
 
+	/*
+	 * Since this function uses fgraph_idx = 0 as a tail-call checking
+	 * flag, set that bit always.
+	 */
+	set_bitmap(current, offset, bitmap | BIT(0));
+
 	return 0;
  out_ret:
-	current->curr_ret_stack--;
+	current->curr_ret_stack -= FGRAPH_FRAME_OFFSET + 1;
  out:
 	current->curr_ret_depth--;
 	return -EBUSY;
 }
 
 /* Retrieve a function return address to the trace stack on thread info.*/
-static void
+static struct ftrace_ret_stack *
 ftrace_pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret,
-			unsigned long frame_pointer)
+			unsigned long frame_pointer, int *offset)
 {
-	int index;
+	struct ftrace_ret_stack *ret_stack;
 
-	index = current->curr_ret_stack;
+	ret_stack = get_ret_stack(current, current->curr_ret_stack, offset);
 
-	if (unlikely(index < 0 || index >= FTRACE_RETFUNC_DEPTH)) {
+	if (unlikely(!ret_stack)) {
 		ftrace_graph_stop();
-		WARN_ON(1);
+		WARN(1, "Bad function graph ret_stack pointer: %d",
+		     current->curr_ret_stack);
 		/* Might as well panic, otherwise we have no where to go */
 		*ret = (unsigned long)panic;
-		return;
+		return NULL;
 	}
 
 #ifdef HAVE_FUNCTION_GRAPH_FP_TEST
@@ -175,30 +707,33 @@ ftrace_pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret,
 	 * Note, -mfentry does not use frame pointers, and this test
 	 *  is not needed if CC_USING_FENTRY is set.
 	 */
-	if (unlikely(current->ret_stack[index].fp != frame_pointer)) {
+	if (unlikely(ret_stack->fp != frame_pointer)) {
 		ftrace_graph_stop();
 		WARN(1, "Bad frame pointer: expected %lx, received %lx\n"
 		     "  from func %ps return to %lx\n",
-		     current->ret_stack[index].fp,
+		     ret_stack->fp,
 		     frame_pointer,
-		     (void *)current->ret_stack[index].func,
-		     current->ret_stack[index].ret);
+		     (void *)ret_stack->func,
+		     ret_stack->ret);
 		*ret = (unsigned long)panic;
-		return;
+		return NULL;
 	}
 #endif
 
-	*ret = current->ret_stack[index].ret;
-	trace->func = current->ret_stack[index].func;
-	trace->calltime = current->ret_stack[index].calltime;
+	*offset += FGRAPH_FRAME_OFFSET;
+	*ret = ret_stack->ret;
+	trace->func = ret_stack->func;
+	trace->calltime = ret_stack->calltime;
 	trace->overrun = atomic_read(&current->trace_overrun);
-	trace->depth = current->curr_ret_depth--;
+	trace->depth = current->curr_ret_depth;
 	/*
 	 * We still want to trace interrupts coming in if
 	 * max_depth is set to 1. Make sure the decrement is
 	 * seen before ftrace_graph_return.
 	 */
 	barrier();
+
+	return ret_stack;
 }
 
 /*
@@ -236,30 +771,55 @@ struct fgraph_ret_regs;
 static unsigned long __ftrace_return_to_handler(struct fgraph_ret_regs *ret_regs,
 						unsigned long frame_pointer)
 {
+	struct ftrace_ret_stack *ret_stack;
 	struct ftrace_graph_ret trace;
+	unsigned long bitmap;
 	unsigned long ret;
+	int offset;
+	int i;
+
+	ret_stack = ftrace_pop_return_trace(&trace, &ret, frame_pointer, &offset);
+
+	if (unlikely(!ret_stack)) {
+		ftrace_graph_stop();
+		WARN_ON(1);
+		/* Might as well panic. What else to do? */
+		return (unsigned long)panic;
+	}
 
-	ftrace_pop_return_trace(&trace, &ret, frame_pointer);
+	trace.rettime = trace_clock_local();
 #ifdef CONFIG_FUNCTION_GRAPH_RETVAL
 	trace.retval = fgraph_ret_regs_return_value(ret_regs);
 #endif
-	trace.rettime = trace_clock_local();
-	ftrace_graph_return(&trace);
+
+	bitmap = get_bitmap_bits(current, offset);
+
+#ifdef CONFIG_HAVE_STATIC_CALL
+	if (static_branch_likely(&fgraph_do_direct)) {
+		if (test_bit(fgraph_direct_gops->idx, &bitmap))
+			static_call(fgraph_retfunc)(&trace, fgraph_direct_gops);
+	} else
+#endif
+	{
+		for_each_set_bit(i, &bitmap, sizeof(bitmap) * BITS_PER_BYTE) {
+			struct fgraph_ops *gops = fgraph_array[i];
+
+			if (gops == &fgraph_stub)
+				continue;
+
+			gops->retfunc(&trace, gops);
+		}
+	}
+
 	/*
 	 * The ftrace_graph_return() may still access the current
 	 * ret_stack structure, we need to make sure the update of
 	 * curr_ret_stack is after that.
 	 */
 	barrier();
-	current->curr_ret_stack--;
-
-	if (unlikely(!ret)) {
-		ftrace_graph_stop();
-		WARN_ON(1);
-		/* Might as well panic. What else to do? */
-		ret = (unsigned long)panic;
-	}
+	current->curr_ret_stack = offset - FGRAPH_FRAME_OFFSET;
 
+	current->curr_ret_depth--;
 	return ret;
 }
 
@@ -282,7 +842,7 @@ unsigned long ftrace_return_to_handler(unsigned long frame_pointer)
 
 /**
  * ftrace_graph_get_ret_stack - return the entry of the shadow stack
- * @task: The task to read the shadow stack from
+ * @task: The task to read the shadow stack from.
  * @idx: Index down the shadow stack
  *
  * Return the ret_struct on the shadow stack of the @task at the
@@ -294,104 +854,116 @@ unsigned long ftrace_return_to_handler(unsigned long frame_pointer)
 struct ftrace_ret_stack *
 ftrace_graph_get_ret_stack(struct task_struct *task, int idx)
 {
-	idx = task->curr_ret_stack - idx;
+	struct ftrace_ret_stack *ret_stack = NULL;
+	int offset = task->curr_ret_stack;
 
-	if (idx >= 0 && idx <= task->curr_ret_stack)
-		return &task->ret_stack[idx];
+	if (offset < 0)
+		return NULL;
 
-	return NULL;
+	do {
+		ret_stack = get_ret_stack(task, offset, &offset);
+	} while (ret_stack && --idx >= 0);
+
+	return ret_stack;
 }
 
 /**
- * ftrace_graph_ret_addr - convert a potentially modified stack return address
- *			   to its original value
+ * ftrace_graph_ret_addr - return the original value of the return address
+ * @task: The task the unwinder is being executed on
+ * @idx: An initialized pointer to the next stack index to use
+ * @ret: The current return address (likely pointing to return_handler)
+ * @retp: The address on the stack of the current return location
  *
  * This function can be called by stack unwinding code to convert a found stack
- * return address ('ret') to its original value, in case the function graph
+ * return address (@ret) to its original value, in case the function graph
  * tracer has modified it to be 'return_to_handler'.  If the address hasn't
- * been modified, the unchanged value of 'ret' is returned.
+ * been modified, the unchanged value of @ret is returned.
  *
- * 'idx' is a state variable which should be initialized by the caller to zero
- * before the first call.
+ * @idx holds the last index used to know where to start from. It should be
+ * initialized to zero for the first iteration as that will mean to start
+ * at the top of the shadow stack. If the location is found, this pointer
+ * will be assigned that location so that if called again, it will continue
+ * where it left off.
  *
- * 'retp' is a pointer to the return address on the stack.  It's ignored if
- * the arch doesn't have HAVE_FUNCTION_GRAPH_RET_ADDR_PTR defined.
+ * @retp is a pointer to the return address on the stack.
  */
-#ifdef HAVE_FUNCTION_GRAPH_RET_ADDR_PTR
 unsigned long ftrace_graph_ret_addr(struct task_struct *task, int *idx,
 				    unsigned long ret, unsigned long *retp)
 {
-	int index = task->curr_ret_stack;
-	int i;
+	struct ftrace_ret_stack *ret_stack;
+	unsigned long return_handler = (unsigned long)dereference_kernel_function_descriptor(return_to_handler);
+	int i = task->curr_ret_stack;
 
-	if (ret != (unsigned long)dereference_kernel_function_descriptor(return_to_handler))
+	if (ret != return_handler)
 		return ret;
 
-	if (index < 0)
+	if (!idx)
 		return ret;
 
-	for (i = 0; i <= index; i++)
-		if (task->ret_stack[i].retp == retp)
-			return task->ret_stack[i].ret;
+	i = *idx ? : task->curr_ret_stack;
+	while (i > 0) {
+		ret_stack = get_ret_stack(current, i, &i);
+		if (!ret_stack)
+			break;
+		/*
+		 * For the tail-call, there would be 2 or more ftrace_ret_stacks on
+		 * the ret_stack, which records "return_to_handler" as the return
+		 * address except for the last one.
+		 * But on the real stack, there should be 1 entry because tail-call
+		 * reuses the return address on the stack and jump to the next function.
+		 * Thus we will continue to find real return address.
+		 */
+		if (ret_stack->retp == retp &&
+		    ret_stack->ret != return_handler) {
+			*idx = i;
+			return ret_stack->ret;
+		}
+	}
 
 	return ret;
 }
-#else /* !HAVE_FUNCTION_GRAPH_RET_ADDR_PTR */
-unsigned long ftrace_graph_ret_addr(struct task_struct *task, int *idx,
-				    unsigned long ret, unsigned long *retp)
-{
-	int task_idx;
-
-	if (ret != (unsigned long)dereference_kernel_function_descriptor(return_to_handler))
-		return ret;
-
-	task_idx = task->curr_ret_stack;
-
-	if (!task->ret_stack || task_idx < *idx)
-		return ret;
-
-	task_idx -= *idx;
-	(*idx)++;
-
-	return task->ret_stack[task_idx].ret;
-}
-#endif /* HAVE_FUNCTION_GRAPH_RET_ADDR_PTR */
 
 static struct ftrace_ops graph_ops = {
 	.func			= ftrace_graph_func,
-	.flags			= FTRACE_OPS_FL_INITIALIZED |
-				   FTRACE_OPS_FL_PID |
-				   FTRACE_OPS_GRAPH_STUB,
+	.flags			= FTRACE_OPS_GRAPH_STUB,
 #ifdef FTRACE_GRAPH_TRAMP_ADDR
 	.trampoline		= FTRACE_GRAPH_TRAMP_ADDR,
 	/* trampoline_size is only needed for dynamically allocated tramps */
 #endif
-	ASSIGN_OPS_HASH(graph_ops, &global_ops.local_hash)
 };
 
-void ftrace_graph_sleep_time_control(bool enable)
+void fgraph_init_ops(struct ftrace_ops *dst_ops,
+		     struct ftrace_ops *src_ops)
 {
-	fgraph_sleep_time = enable;
+	dst_ops->flags = FTRACE_OPS_FL_PID | FTRACE_OPS_GRAPH_STUB;
+
+#ifdef CONFIG_DYNAMIC_FTRACE
+	if (src_ops) {
+		dst_ops->func_hash = &src_ops->local_hash;
+		mutex_init(&dst_ops->local_hash.regex_lock);
+		INIT_LIST_HEAD(&dst_ops->subop_list);
+		dst_ops->flags |= FTRACE_OPS_FL_INITIALIZED;
+	}
+#endif
 }
 
-int ftrace_graph_entry_stub(struct ftrace_graph_ent *trace)
+void ftrace_graph_sleep_time_control(bool enable)
 {
-	return 0;
+	fgraph_sleep_time = enable;
 }
 
 /*
  * Simply points to ftrace_stub, but with the proper protocol.
  * Defined by the linker script in linux/vmlinux.lds.h
  */
-extern void ftrace_stub_graph(struct ftrace_graph_ret *);
+void ftrace_stub_graph(struct ftrace_graph_ret *trace, struct fgraph_ops *gops);
 
 /* The callbacks that hook a function */
 trace_func_graph_ret_t ftrace_graph_return = ftrace_stub_graph;
 trace_func_graph_ent_t ftrace_graph_entry = ftrace_graph_entry_stub;
-static trace_func_graph_ent_t __ftrace_graph_entry = ftrace_graph_entry_stub;
 
 /* Try to assign a return stack array on FTRACE_RETSTACK_ALLOC_SIZE tasks. */
-static int alloc_retstack_tasklist(struct ftrace_ret_stack **ret_stack_list)
+static int alloc_retstack_tasklist(unsigned long **ret_stack_list)
 {
 	int i;
 	int ret = 0;
@@ -399,10 +971,7 @@ static int alloc_retstack_tasklist(struct ftrace_ret_stack **ret_stack_list)
 	struct task_struct *g, *t;
 
 	for (i = 0; i < FTRACE_RETSTACK_ALLOC_SIZE; i++) {
-		ret_stack_list[i] =
-			kmalloc_array(FTRACE_RETFUNC_DEPTH,
-				      sizeof(struct ftrace_ret_stack),
-				      GFP_KERNEL);
+		ret_stack_list[i] = kmalloc(SHADOW_STACK_SIZE, GFP_KERNEL);
 		if (!ret_stack_list[i]) {
 			start = 0;
 			end = i;
@@ -420,9 +989,10 @@ static int alloc_retstack_tasklist(struct ftrace_ret_stack **ret_stack_list)
 
 		if (t->ret_stack == NULL) {
 			atomic_set(&t->trace_overrun, 0);
-			t->curr_ret_stack = -1;
+			ret_stack_init_task_vars(ret_stack_list[start]);
+			t->curr_ret_stack = 0;
 			t->curr_ret_depth = -1;
-			/* Make sure the tasks see the -1 first: */
+			/* Make sure the tasks see the 0 first: */
 			smp_wmb();
 			t->ret_stack = ret_stack_list[start++];
 		}
@@ -442,8 +1012,9 @@ ftrace_graph_probe_sched_switch(void *ignore, bool preempt,
 				struct task_struct *next,
 				unsigned int prev_state)
 {
+	struct ftrace_ret_stack *ret_stack;
 	unsigned long long timestamp;
-	int index;
+	int offset;
 
 	/*
 	 * Does the user want to count the time a function was asleep.
@@ -466,57 +1037,23 @@ ftrace_graph_probe_sched_switch(void *ignore, bool preempt,
 	 */
 	timestamp -= next->ftrace_timestamp;
 
-	for (index = next->curr_ret_stack; index >= 0; index--)
-		next->ret_stack[index].calltime += timestamp;
-}
-
-static int ftrace_graph_entry_test(struct ftrace_graph_ent *trace)
-{
-	if (!ftrace_ops_test(&global_ops, trace->func, NULL))
-		return 0;
-	return __ftrace_graph_entry(trace);
-}
-
-/*
- * The function graph tracer should only trace the functions defined
- * by set_ftrace_filter and set_ftrace_notrace. If another function
- * tracer ops is registered, the graph tracer requires testing the
- * function against the global ops, and not just trace any function
- * that any ftrace_ops registered.
- */
-void update_function_graph_func(void)
-{
-	struct ftrace_ops *op;
-	bool do_test = false;
-
-	/*
-	 * The graph and global ops share the same set of functions
-	 * to test. If any other ops is on the list, then
-	 * the graph tracing needs to test if its the function
-	 * it should call.
-	 */
-	do_for_each_ftrace_op(op, ftrace_ops_list) {
-		if (op != &global_ops && op != &graph_ops &&
-		    op != &ftrace_list_end) {
-			do_test = true;
-			/* in double loop, break out with goto */
-			goto out;
-		}
-	} while_for_each_ftrace_op(op);
- out:
-	if (do_test)
-		ftrace_graph_entry = ftrace_graph_entry_test;
-	else
-		ftrace_graph_entry = __ftrace_graph_entry;
+	for (offset = next->curr_ret_stack; offset > 0; ) {
+		ret_stack = get_ret_stack(next, offset, &offset);
+		if (ret_stack)
+			ret_stack->calltime += timestamp;
+	}
 }
 
-static DEFINE_PER_CPU(struct ftrace_ret_stack *, idle_ret_stack);
+static DEFINE_PER_CPU(unsigned long *, idle_ret_stack);
 
 static void
-graph_init_task(struct task_struct *t, struct ftrace_ret_stack *ret_stack)
+graph_init_task(struct task_struct *t, unsigned long *ret_stack)
 {
 	atomic_set(&t->trace_overrun, 0);
+	ret_stack_init_task_vars(ret_stack);
 	t->ftrace_timestamp = 0;
+	t->curr_ret_stack = 0;
+	t->curr_ret_depth = -1;
 	/* make curr_ret_stack visible before we add the ret_stack */
 	smp_wmb();
 	t->ret_stack = ret_stack;
@@ -528,7 +1065,7 @@ graph_init_task(struct task_struct *t, struct ftrace_ret_stack *ret_stack)
  */
 void ftrace_graph_init_idle_task(struct task_struct *t, int cpu)
 {
-	t->curr_ret_stack = -1;
+	t->curr_ret_stack = 0;
 	t->curr_ret_depth = -1;
 	/*
 	 * The idle task has no parent, it either has its own
@@ -538,14 +1075,11 @@ void ftrace_graph_init_idle_task(struct task_struct *t, int cpu)
 		WARN_ON(t->ret_stack != per_cpu(idle_ret_stack, cpu));
 
 	if (ftrace_graph_active) {
-		struct ftrace_ret_stack *ret_stack;
+		unsigned long *ret_stack;
 
 		ret_stack = per_cpu(idle_ret_stack, cpu);
 		if (!ret_stack) {
-			ret_stack =
-				kmalloc_array(FTRACE_RETFUNC_DEPTH,
-					      sizeof(struct ftrace_ret_stack),
-					      GFP_KERNEL);
+			ret_stack = kmalloc(SHADOW_STACK_SIZE, GFP_KERNEL);
 			if (!ret_stack)
 				return;
 			per_cpu(idle_ret_stack, cpu) = ret_stack;
@@ -559,15 +1093,13 @@ void ftrace_graph_init_task(struct task_struct *t)
 {
 	/* Make sure we do not use the parent ret_stack */
 	t->ret_stack = NULL;
-	t->curr_ret_stack = -1;
+	t->curr_ret_stack = 0;
 	t->curr_ret_depth = -1;
 
 	if (ftrace_graph_active) {
-		struct ftrace_ret_stack *ret_stack;
+		unsigned long *ret_stack;
 
-		ret_stack = kmalloc_array(FTRACE_RETFUNC_DEPTH,
-					  sizeof(struct ftrace_ret_stack),
-					  GFP_KERNEL);
+		ret_stack = kmalloc(SHADOW_STACK_SIZE, GFP_KERNEL);
 		if (!ret_stack)
 			return;
 		graph_init_task(t, ret_stack);
@@ -576,7 +1108,7 @@ void ftrace_graph_init_task(struct task_struct *t)
 
 void ftrace_graph_exit_task(struct task_struct *t)
 {
-	struct ftrace_ret_stack	*ret_stack = t->ret_stack;
+	unsigned long *ret_stack = t->ret_stack;
 
 	t->ret_stack = NULL;
 	/* NULL must become visible to IRQs before we free it: */
@@ -585,15 +1117,52 @@ void ftrace_graph_exit_task(struct task_struct *t)
 	kfree(ret_stack);
 }
 
+#ifdef CONFIG_DYNAMIC_FTRACE
+static int fgraph_pid_func(struct ftrace_graph_ent *trace,
+			   struct fgraph_ops *gops)
+{
+	struct trace_array *tr = gops->ops.private;
+	int pid;
+
+	if (tr) {
+		pid = this_cpu_read(tr->array_buffer.data->ftrace_ignore_pid);
+		if (pid == FTRACE_PID_IGNORE)
+			return 0;
+		if (pid != FTRACE_PID_TRACE &&
+		    pid != current->pid)
+			return 0;
+	}
+
+	return gops->saved_func(trace, gops);
+}
+
+void fgraph_update_pid_func(void)
+{
+	struct fgraph_ops *gops;
+	struct ftrace_ops *op;
+
+	if (!(graph_ops.flags & FTRACE_OPS_FL_INITIALIZED))
+		return;
+
+	list_for_each_entry(op, &graph_ops.subop_list, list) {
+		if (op->flags & FTRACE_OPS_FL_PID) {
+			gops = container_of(op, struct fgraph_ops, ops);
+			gops->entryfunc = ftrace_pids_enabled(op) ?
+				fgraph_pid_func : gops->saved_func;
+			if (ftrace_graph_active == 1)
+				static_call_update(fgraph_func, gops->entryfunc);
+		}
+	}
+}
+#endif
+
 /* Allocate a return stack for each task */
 static int start_graph_tracing(void)
 {
-	struct ftrace_ret_stack **ret_stack_list;
+	unsigned long **ret_stack_list;
 	int ret, cpu;
 
-	ret_stack_list = kmalloc_array(FTRACE_RETSTACK_ALLOC_SIZE,
-				       sizeof(struct ftrace_ret_stack *),
-				       GFP_KERNEL);
+	ret_stack_list = kmalloc(SHADOW_STACK_SIZE, GFP_KERNEL);
 
 	if (!ret_stack_list)
 		return -ENOMEM;
@@ -619,40 +1188,111 @@ static int start_graph_tracing(void)
 	return ret;
 }
 
+static void init_task_vars(int idx)
+{
+	struct task_struct *g, *t;
+	int cpu;
+
+	for_each_online_cpu(cpu) {
+		if (idle_task(cpu)->ret_stack)
+			ret_stack_set_task_var(idle_task(cpu), idx, 0);
+	}
+
+	read_lock(&tasklist_lock);
+	for_each_process_thread(g, t) {
+		if (t->ret_stack)
+			ret_stack_set_task_var(t, idx, 0);
+	}
+	read_unlock(&tasklist_lock);
+}
+
+static void ftrace_graph_enable_direct(bool enable_branch)
+{
+	trace_func_graph_ent_t func = NULL;
+	trace_func_graph_ret_t retfunc = NULL;
+	int i;
+
+	for_each_set_bit(i, &fgraph_array_bitmask,
+			 sizeof(fgraph_array_bitmask) * BITS_PER_BYTE) {
+		func = fgraph_array[i]->entryfunc;
+		retfunc = fgraph_array[i]->retfunc;
+		fgraph_direct_gops = fgraph_array[i];
+	 }
+	if (WARN_ON_ONCE(!func))
+		return;
+
+	static_call_update(fgraph_func, func);
+	static_call_update(fgraph_retfunc, retfunc);
+	if (enable_branch)
+		static_branch_disable(&fgraph_do_direct);
+}
+
+static void ftrace_graph_disable_direct(bool disable_branch)
+{
+	if (disable_branch)
+		static_branch_disable(&fgraph_do_direct);
+	static_call_update(fgraph_func, ftrace_graph_entry_stub);
+	static_call_update(fgraph_retfunc, ftrace_graph_ret_stub);
+	fgraph_direct_gops = &fgraph_stub;
+}
+
 int register_ftrace_graph(struct fgraph_ops *gops)
 {
+	int command = 0;
 	int ret = 0;
+	int i = -1;
 
 	mutex_lock(&ftrace_lock);
 
-	/* we currently allow only one tracer registered at a time */
-	if (ftrace_graph_active) {
-		ret = -EBUSY;
+	if (!fgraph_array[0]) {
+		/* The array must always have real data on it */
+		for (i = 0; i < FGRAPH_ARRAY_SIZE; i++)
+			fgraph_array[i] = &fgraph_stub;
+		fgraph_lru_init();
+	}
+
+	i = fgraph_lru_alloc_index();
+	if (i < 0 || WARN_ON_ONCE(fgraph_array[i] != &fgraph_stub)) {
+		ret = -ENOSPC;
 		goto out;
 	}
 
-	register_pm_notifier(&ftrace_suspend_notifier);
+	fgraph_array[i] = gops;
+	gops->idx = i;
 
 	ftrace_graph_active++;
-	ret = start_graph_tracing();
-	if (ret) {
-		ftrace_graph_active--;
-		goto out;
-	}
 
-	ftrace_graph_return = gops->retfunc;
+	if (ftrace_graph_active == 2)
+		ftrace_graph_disable_direct(true);
 
-	/*
-	 * Update the indirect function to the entryfunc, and the
-	 * function that gets called to the entry_test first. Then
-	 * call the update fgraph entry function to determine if
-	 * the entryfunc should be called directly or not.
-	 */
-	__ftrace_graph_entry = gops->entryfunc;
-	ftrace_graph_entry = ftrace_graph_entry_test;
-	update_function_graph_func();
+	if (ftrace_graph_active == 1) {
+		ftrace_graph_enable_direct(false);
+		register_pm_notifier(&ftrace_suspend_notifier);
+		ret = start_graph_tracing();
+		if (ret)
+			goto error;
+		/*
+		 * Some archs just test to see if these are not
+		 * the default function
+		 */
+		ftrace_graph_return = return_run;
+		ftrace_graph_entry = entry_run;
+		command = FTRACE_START_FUNC_RET;
+	} else {
+		init_task_vars(gops->idx);
+	}
+
+	/* Always save the function, and reset at unregistering */
+	gops->saved_func = gops->entryfunc;
 
-	ret = ftrace_startup(&graph_ops, FTRACE_START_FUNC_RET);
+	ret = ftrace_startup_subops(&graph_ops, &gops->ops, command);
+error:
+	if (ret) {
+		fgraph_array[i] = &fgraph_stub;
+		ftrace_graph_active--;
+		gops->saved_func = NULL;
+		fgraph_lru_release_index(i);
+	}
 out:
 	mutex_unlock(&ftrace_lock);
 	return ret;
@@ -660,19 +1300,41 @@ int register_ftrace_graph(struct fgraph_ops *gops)
 
 void unregister_ftrace_graph(struct fgraph_ops *gops)
 {
+	int command = 0;
+
 	mutex_lock(&ftrace_lock);
 
 	if (unlikely(!ftrace_graph_active))
 		goto out;
 
+	if (unlikely(gops->idx < 0 || gops->idx >= FGRAPH_ARRAY_SIZE ||
+		     fgraph_array[gops->idx] != gops))
+		goto out;
+
+	if (fgraph_lru_release_index(gops->idx) < 0)
+		goto out;
+
+	fgraph_array[gops->idx] = &fgraph_stub;
+
 	ftrace_graph_active--;
-	ftrace_graph_return = ftrace_stub_graph;
-	ftrace_graph_entry = ftrace_graph_entry_stub;
-	__ftrace_graph_entry = ftrace_graph_entry_stub;
-	ftrace_shutdown(&graph_ops, FTRACE_STOP_FUNC_RET);
-	unregister_pm_notifier(&ftrace_suspend_notifier);
-	unregister_trace_sched_switch(ftrace_graph_probe_sched_switch, NULL);
 
+	if (!ftrace_graph_active)
+		command = FTRACE_STOP_FUNC_RET;
+
+	ftrace_shutdown_subops(&graph_ops, &gops->ops, command);
+
+	if (ftrace_graph_active == 1)
+		ftrace_graph_enable_direct(true);
+	else if (!ftrace_graph_active)
+		ftrace_graph_disable_direct(false);
+
+	if (!ftrace_graph_active) {
+		ftrace_graph_return = ftrace_stub_graph;
+		ftrace_graph_entry = ftrace_graph_entry_stub;
+		unregister_pm_notifier(&ftrace_suspend_notifier);
+		unregister_trace_sched_switch(ftrace_graph_probe_sched_switch, NULL);
+	}
  out:
+	gops->saved_func = NULL;
 	mutex_unlock(&ftrace_lock);
 }

kernel/trace/ftrace.c

@@ -74,7 +74,8 @@
 #ifdef CONFIG_DYNAMIC_FTRACE
 #define INIT_OPS_HASH(opsname)	\
 	.func_hash		= &opsname.local_hash,			\
-	.local_hash.regex_lock	= __MUTEX_INITIALIZER(opsname.local_hash.regex_lock),
+	.local_hash.regex_lock	= __MUTEX_INITIALIZER(opsname.local_hash.regex_lock), \
+	.subop_list		= LIST_HEAD_INIT(opsname.subop_list),
 #else
 #define INIT_OPS_HASH(opsname)
 #endif
@@ -99,7 +100,7 @@ struct ftrace_ops *function_trace_op __read_mostly = &ftrace_list_end;
 /* What to set function_trace_op to */
 static struct ftrace_ops *set_function_trace_op;
 
-static bool ftrace_pids_enabled(struct ftrace_ops *ops)
+bool ftrace_pids_enabled(struct ftrace_ops *ops)
 {
 	struct trace_array *tr;
 
@@ -121,7 +122,7 @@ static int ftrace_disabled __read_mostly;
 
 DEFINE_MUTEX(ftrace_lock);
 
-struct ftrace_ops __rcu *ftrace_ops_list __read_mostly = &ftrace_list_end;
+struct ftrace_ops __rcu *ftrace_ops_list __read_mostly = (struct ftrace_ops __rcu *)&ftrace_list_end;
 ftrace_func_t ftrace_trace_function __read_mostly = ftrace_stub;
 struct ftrace_ops global_ops;
 
@@ -161,12 +162,14 @@ static inline void ftrace_ops_init(struct ftrace_ops *ops)
 #ifdef CONFIG_DYNAMIC_FTRACE
 	if (!(ops->flags & FTRACE_OPS_FL_INITIALIZED)) {
 		mutex_init(&ops->local_hash.regex_lock);
+		INIT_LIST_HEAD(&ops->subop_list);
 		ops->func_hash = &ops->local_hash;
 		ops->flags |= FTRACE_OPS_FL_INITIALIZED;
 	}
 #endif
 }
 
+/* Call this function for when a callback filters on set_ftrace_pid */
 static void ftrace_pid_func(unsigned long ip, unsigned long parent_ip,
 			    struct ftrace_ops *op, struct ftrace_regs *fregs)
 {
@@ -235,8 +238,6 @@ static void update_ftrace_function(void)
 		func = ftrace_ops_list_func;
 	}
 
-	update_function_graph_func();
-
 	/* If there's no change, then do nothing more here */
 	if (ftrace_trace_function == func)
 		return;
@@ -310,7 +311,7 @@ static int remove_ftrace_ops(struct ftrace_ops __rcu **list,
 			lockdep_is_held(&ftrace_lock)) == ops &&
 	    rcu_dereference_protected(ops->next,
 			lockdep_is_held(&ftrace_lock)) == &ftrace_list_end) {
-		*list = &ftrace_list_end;
+		rcu_assign_pointer(*list, &ftrace_list_end);
 		return 0;
 	}
 
@@ -406,6 +407,8 @@ static void ftrace_update_pid_func(void)
 		}
 	} while_for_each_ftrace_op(op);
 
+	fgraph_update_pid_func();
+
 	update_ftrace_function();
 }
 
@@ -817,7 +820,8 @@ void ftrace_graph_graph_time_control(bool enable)
 	fgraph_graph_time = enable;
 }
 
-static int profile_graph_entry(struct ftrace_graph_ent *trace)
+static int profile_graph_entry(struct ftrace_graph_ent *trace,
+			       struct fgraph_ops *gops)
 {
 	struct ftrace_ret_stack *ret_stack;
 
@@ -834,7 +838,8 @@ static int profile_graph_entry(struct ftrace_graph_ent *trace)
 	return 1;
 }
 
-static void profile_graph_return(struct ftrace_graph_ret *trace)
+static void profile_graph_return(struct ftrace_graph_ret *trace,
+				 struct fgraph_ops *gops)
 {
 	struct ftrace_ret_stack *ret_stack;
 	struct ftrace_profile_stat *stat;
@@ -1314,7 +1319,7 @@ static struct ftrace_hash *alloc_ftrace_hash(int size_bits)
 	return hash;
 }
 
-
+/* Used to save filters on functions for modules not loaded yet */
 static int ftrace_add_mod(struct trace_array *tr,
 			  const char *func, const char *module,
 			  int enable)
@@ -1380,15 +1385,17 @@ alloc_and_copy_ftrace_hash(int size_bits, struct ftrace_hash *hash)
 	return NULL;
 }
 
-static void
-ftrace_hash_rec_disable_modify(struct ftrace_ops *ops, int filter_hash);
-static void
-ftrace_hash_rec_enable_modify(struct ftrace_ops *ops, int filter_hash);
+static void ftrace_hash_rec_disable_modify(struct ftrace_ops *ops);
+static void ftrace_hash_rec_enable_modify(struct ftrace_ops *ops);
 
 static int ftrace_hash_ipmodify_update(struct ftrace_ops *ops,
 				       struct ftrace_hash *new_hash);
 
-static struct ftrace_hash *dup_hash(struct ftrace_hash *src, int size)
+/*
+ * Allocate a new hash and remove entries from @src and move them to the new hash.
+ * On success, the @src hash will be empty and should be freed.
+ */
+static struct ftrace_hash *__move_hash(struct ftrace_hash *src, int size)
 {
 	struct ftrace_func_entry *entry;
 	struct ftrace_hash *new_hash;
@@ -1424,6 +1431,7 @@ static struct ftrace_hash *dup_hash(struct ftrace_hash *src, int size)
 	return new_hash;
 }
 
+/* Move the @src entries to a newly allocated hash */
 static struct ftrace_hash *
 __ftrace_hash_move(struct ftrace_hash *src)
 {
@@ -1435,9 +1443,29 @@ __ftrace_hash_move(struct ftrace_hash *src)
 	if (ftrace_hash_empty(src))
 		return EMPTY_HASH;
 
-	return dup_hash(src, size);
+	return __move_hash(src, size);
 }
 
+/**
+ * ftrace_hash_move - move a new hash to a filter and do updates
+ * @ops: The ops with the hash that @dst points to
+ * @enable: True if for the filter hash, false for the notrace hash
+ * @dst: Points to the @ops hash that should be updated
+ * @src: The hash to update @dst with
+ *
+ * This is called when an ftrace_ops hash is being updated and the
+ * the kernel needs to reflect this. Note, this only updates the kernel
+ * function callbacks if the @ops is enabled (not to be confused with
+ * @enable above). If the @ops is enabled, its hash determines what
+ * callbacks get called. This function gets called when the @ops hash
+ * is updated and it requires new callbacks.
+ *
+ * On success the elements of @src is moved to @dst, and @dst is updated
+ * properly, as well as the functions determined by the @ops hashes
+ * are now calling the @ops callback function.
+ *
+ * Regardless of return type, @src should be freed with free_ftrace_hash().
+ */
 static int
 ftrace_hash_move(struct ftrace_ops *ops, int enable,
 		 struct ftrace_hash **dst, struct ftrace_hash *src)
@@ -1467,11 +1495,11 @@ ftrace_hash_move(struct ftrace_ops *ops, int enable,
 	 * Remove the current set, update the hash and add
 	 * them back.
 	 */
-	ftrace_hash_rec_disable_modify(ops, enable);
+	ftrace_hash_rec_disable_modify(ops);
 
 	rcu_assign_pointer(*dst, new_hash);
 
-	ftrace_hash_rec_enable_modify(ops, enable);
+	ftrace_hash_rec_enable_modify(ops);
 
 	return 0;
 }
@@ -1694,12 +1722,21 @@ static bool skip_record(struct dyn_ftrace *rec)
 		!(rec->flags & FTRACE_FL_ENABLED);
 }
 
+/*
+ * This is the main engine to the ftrace updates to the dyn_ftrace records.
+ *
+ * It will iterate through all the available ftrace functions
+ * (the ones that ftrace can have callbacks to) and set the flags
+ * in the associated dyn_ftrace records.
+ *
+ * @inc: If true, the functions associated to @ops are added to
+ *       the dyn_ftrace records, otherwise they are removed.
+ */
 static bool __ftrace_hash_rec_update(struct ftrace_ops *ops,
-				     int filter_hash,
 				     bool inc)
 {
 	struct ftrace_hash *hash;
-	struct ftrace_hash *other_hash;
+	struct ftrace_hash *notrace_hash;
 	struct ftrace_page *pg;
 	struct dyn_ftrace *rec;
 	bool update = false;
@@ -1711,35 +1748,16 @@ static bool __ftrace_hash_rec_update(struct ftrace_ops *ops,
 		return false;
 
 	/*
-	 * In the filter_hash case:
 	 *   If the count is zero, we update all records.
 	 *   Otherwise we just update the items in the hash.
-	 *
-	 * In the notrace_hash case:
-	 *   We enable the update in the hash.
-	 *   As disabling notrace means enabling the tracing,
-	 *   and enabling notrace means disabling, the inc variable
-	 *   gets inversed.
 	 */
-	if (filter_hash) {
-		hash = ops->func_hash->filter_hash;
-		other_hash = ops->func_hash->notrace_hash;
-		if (ftrace_hash_empty(hash))
-			all = true;
-	} else {
-		inc = !inc;
-		hash = ops->func_hash->notrace_hash;
-		other_hash = ops->func_hash->filter_hash;
-		/*
-		 * If the notrace hash has no items,
-		 * then there's nothing to do.
-		 */
-		if (ftrace_hash_empty(hash))
-			return false;
-	}
+	hash = ops->func_hash->filter_hash;
+	notrace_hash = ops->func_hash->notrace_hash;
+	if (ftrace_hash_empty(hash))
+		all = true;
 
 	do_for_each_ftrace_rec(pg, rec) {
-		int in_other_hash = 0;
+		int in_notrace_hash = 0;
 		int in_hash = 0;
 		int match = 0;
 
@@ -1751,26 +1769,17 @@ static bool __ftrace_hash_rec_update(struct ftrace_ops *ops,
 			 * Only the filter_hash affects all records.
 			 * Update if the record is not in the notrace hash.
 			 */
-			if (!other_hash || !ftrace_lookup_ip(other_hash, rec->ip))
+			if (!notrace_hash || !ftrace_lookup_ip(notrace_hash, rec->ip))
 				match = 1;
 		} else {
 			in_hash = !!ftrace_lookup_ip(hash, rec->ip);
-			in_other_hash = !!ftrace_lookup_ip(other_hash, rec->ip);
+			in_notrace_hash = !!ftrace_lookup_ip(notrace_hash, rec->ip);
 
 			/*
-			 * If filter_hash is set, we want to match all functions
-			 * that are in the hash but not in the other hash.
-			 *
-			 * If filter_hash is not set, then we are decrementing.
-			 * That means we match anything that is in the hash
-			 * and also in the other_hash. That is, we need to turn
-			 * off functions in the other hash because they are disabled
-			 * by this hash.
+			 * We want to match all functions that are in the hash but
+			 * not in the other hash.
 			 */
-			if (filter_hash && in_hash && !in_other_hash)
-				match = 1;
-			else if (!filter_hash && in_hash &&
-				 (in_other_hash || ftrace_hash_empty(other_hash)))
+			if (in_hash && !in_notrace_hash)
 				match = 1;
 		}
 		if (!match)
@@ -1876,24 +1885,48 @@ static bool __ftrace_hash_rec_update(struct ftrace_ops *ops,
 	return update;
 }
 
-static bool ftrace_hash_rec_disable(struct ftrace_ops *ops,
-				    int filter_hash)
+/*
+ * This is called when an ops is removed from tracing. It will decrement
+ * the counters of the dyn_ftrace records for all the functions that
+ * the @ops attached to.
+ */
+static bool ftrace_hash_rec_disable(struct ftrace_ops *ops)
 {
-	return __ftrace_hash_rec_update(ops, filter_hash, 0);
+	return __ftrace_hash_rec_update(ops, false);
 }
 
-static bool ftrace_hash_rec_enable(struct ftrace_ops *ops,
-				   int filter_hash)
+/*
+ * This is called when an ops is added to tracing. It will increment
+ * the counters of the dyn_ftrace records for all the functions that
+ * the @ops attached to.
+ */
+static bool ftrace_hash_rec_enable(struct ftrace_ops *ops)
 {
-	return __ftrace_hash_rec_update(ops, filter_hash, 1);
+	return __ftrace_hash_rec_update(ops, true);
 }
 
-static void ftrace_hash_rec_update_modify(struct ftrace_ops *ops,
-					  int filter_hash, int inc)
+/*
+ * This function will update what functions @ops traces when its filter
+ * changes.
+ *
+ * The @inc states if the @ops callbacks are going to be added or removed.
+ * When one of the @ops hashes are updated to a "new_hash" the dyn_ftrace
+ * records are update via:
+ *
+ * ftrace_hash_rec_disable_modify(ops);
+ * ops->hash = new_hash
+ * ftrace_hash_rec_enable_modify(ops);
+ *
+ * Where the @ops is removed from all the records it is tracing using
+ * its old hash. The @ops hash is updated to the new hash, and then
+ * the @ops is added back to the records so that it is tracing all
+ * the new functions.
+ */
+static void ftrace_hash_rec_update_modify(struct ftrace_ops *ops, bool inc)
 {
 	struct ftrace_ops *op;
 
-	__ftrace_hash_rec_update(ops, filter_hash, inc);
+	__ftrace_hash_rec_update(ops, inc);
 
 	if (ops->func_hash != &global_ops.local_hash)
 		return;
@@ -1907,20 +1940,18 @@ static void ftrace_hash_rec_update_modify(struct ftrace_ops *ops,
 		if (op == ops)
 			continue;
 		if (op->func_hash == &global_ops.local_hash)
-			__ftrace_hash_rec_update(op, filter_hash, inc);
+			__ftrace_hash_rec_update(op, inc);
 	} while_for_each_ftrace_op(op);
 }
 
-static void ftrace_hash_rec_disable_modify(struct ftrace_ops *ops,
-					   int filter_hash)
+static void ftrace_hash_rec_disable_modify(struct ftrace_ops *ops)
 {
-	ftrace_hash_rec_update_modify(ops, filter_hash, 0);
+	ftrace_hash_rec_update_modify(ops, false);
 }
 
-static void ftrace_hash_rec_enable_modify(struct ftrace_ops *ops,
-					  int filter_hash)
+static void ftrace_hash_rec_enable_modify(struct ftrace_ops *ops)
 {
-	ftrace_hash_rec_update_modify(ops, filter_hash, 1);
+	ftrace_hash_rec_update_modify(ops, true);
 }
 
 /*
@@ -3043,7 +3074,7 @@ int ftrace_startup(struct ftrace_ops *ops, int command)
 		return ret;
 	}
 
-	if (ftrace_hash_rec_enable(ops, 1))
+	if (ftrace_hash_rec_enable(ops))
 		command |= FTRACE_UPDATE_CALLS;
 
 	ftrace_startup_enable(command);
@@ -3085,7 +3116,7 @@ int ftrace_shutdown(struct ftrace_ops *ops, int command)
 	/* Disabling ipmodify never fails */
 	ftrace_hash_ipmodify_disable(ops);
 
-	if (ftrace_hash_rec_disable(ops, 1))
+	if (ftrace_hash_rec_disable(ops))
 		command |= FTRACE_UPDATE_CALLS;
 
 	ops->flags &= ~FTRACE_OPS_FL_ENABLED;
@@ -3164,6 +3195,474 @@ int ftrace_shutdown(struct ftrace_ops *ops, int command)
 	return 0;
 }
 
+/* Simply make a copy of @src and return it */
+static struct ftrace_hash *copy_hash(struct ftrace_hash *src)
+{
+	if (ftrace_hash_empty(src))
+		return EMPTY_HASH;
+
+	return alloc_and_copy_ftrace_hash(src->size_bits, src);
+}
+
+/*
+ * Append @new_hash entries to @hash:
+ *
+ *  If @hash is the EMPTY_HASH then it traces all functions and nothing
+ *  needs to be done.
+ *
+ *  If @new_hash is the EMPTY_HASH, then make *hash the EMPTY_HASH so
+ *  that it traces everything.
+ *
+ *  Otherwise, go through all of @new_hash and add anything that @hash
+ *  doesn't already have, to @hash.
+ *
+ *  The filter_hash updates uses just the append_hash() function
+ *  and the notrace_hash does not.
+ */
+static int append_hash(struct ftrace_hash **hash, struct ftrace_hash *new_hash)
+{
+	struct ftrace_func_entry *entry;
+	int size;
+	int i;
+
+	/* An empty hash does everything */
+	if (ftrace_hash_empty(*hash))
+		return 0;
+
+	/* If new_hash has everything make hash have everything */
+	if (ftrace_hash_empty(new_hash)) {
+		free_ftrace_hash(*hash);
+		*hash = EMPTY_HASH;
+		return 0;
+	}
+
+	size = 1 << new_hash->size_bits;
+	for (i = 0; i < size; i++) {
+		hlist_for_each_entry(entry, &new_hash->buckets[i], hlist) {
+			/* Only add if not already in hash */
+			if (!__ftrace_lookup_ip(*hash, entry->ip) &&
+			    add_hash_entry(*hash, entry->ip) == NULL)
+				return -ENOMEM;
+		}
+	}
+	return 0;
+}
+
+/*
+ * Add to @hash only those that are in both @new_hash1 and @new_hash2
+ *
+ * The notrace_hash updates uses just the intersect_hash() function
+ * and the filter_hash does not.
+ */
+static int intersect_hash(struct ftrace_hash **hash, struct ftrace_hash *new_hash1,
+			  struct ftrace_hash *new_hash2)
+{
+	struct ftrace_func_entry *entry;
+	int size;
+	int i;
+
+	/*
+	 * If new_hash1 or new_hash2 is the EMPTY_HASH then make the hash
+	 * empty as well as empty for notrace means none are notraced.
+	 */
+	if (ftrace_hash_empty(new_hash1) || ftrace_hash_empty(new_hash2)) {
+		free_ftrace_hash(*hash);
+		*hash = EMPTY_HASH;
+		return 0;
+	}
+
+	size = 1 << new_hash1->size_bits;
+	for (i = 0; i < size; i++) {
+		hlist_for_each_entry(entry, &new_hash1->buckets[i], hlist) {
+			/* Only add if in both @new_hash1 and @new_hash2 */
+			if (__ftrace_lookup_ip(new_hash2, entry->ip) &&
+			    add_hash_entry(*hash, entry->ip) == NULL)
+				return -ENOMEM;
+		}
+	}
+	/* If nothing intersects, make it the empty set */
+	if (ftrace_hash_empty(*hash)) {
+		free_ftrace_hash(*hash);
+		*hash = EMPTY_HASH;
+	}
+	return 0;
+}
+
+/* Return a new hash that has a union of all @ops->filter_hash entries */
+static struct ftrace_hash *append_hashes(struct ftrace_ops *ops)
+{
+	struct ftrace_hash *new_hash;
+	struct ftrace_ops *subops;
+	int ret;
+
+	new_hash = alloc_ftrace_hash(ops->func_hash->filter_hash->size_bits);
+	if (!new_hash)
+		return NULL;
+
+	list_for_each_entry(subops, &ops->subop_list, list) {
+		ret = append_hash(&new_hash, subops->func_hash->filter_hash);
+		if (ret < 0) {
+			free_ftrace_hash(new_hash);
+			return NULL;
+		}
+		/* Nothing more to do if new_hash is empty */
+		if (ftrace_hash_empty(new_hash))
+			break;
+	}
+	return new_hash;
+}
+
+/* Make @ops trace evenything except what all its subops do not trace */
+static struct ftrace_hash *intersect_hashes(struct ftrace_ops *ops)
+{
+	struct ftrace_hash *new_hash = NULL;
+	struct ftrace_ops *subops;
+	int size_bits;
+	int ret;
+
+	list_for_each_entry(subops, &ops->subop_list, list) {
+		struct ftrace_hash *next_hash;
+
+		if (!new_hash) {
+			size_bits = subops->func_hash->notrace_hash->size_bits;
+			new_hash = alloc_and_copy_ftrace_hash(size_bits, ops->func_hash->notrace_hash);
+			if (!new_hash)
+				return NULL;
+			continue;
+		}
+		size_bits = new_hash->size_bits;
+		next_hash = new_hash;
+		new_hash = alloc_ftrace_hash(size_bits);
+		ret = intersect_hash(&new_hash, next_hash, subops->func_hash->notrace_hash);
+		free_ftrace_hash(next_hash);
+		if (ret < 0) {
+			free_ftrace_hash(new_hash);
+			return NULL;
+		}
+		/* Nothing more to do if new_hash is empty */
+		if (ftrace_hash_empty(new_hash))
+			break;
+	}
+	return new_hash;
+}
+
+static bool ops_equal(struct ftrace_hash *A, struct ftrace_hash *B)
+{
+	struct ftrace_func_entry *entry;
+	int size;
+	int i;
+
+	if (ftrace_hash_empty(A))
+		return ftrace_hash_empty(B);
+
+	if (ftrace_hash_empty(B))
+		return ftrace_hash_empty(A);
+
+	if (A->count != B->count)
+		return false;
+
+	size = 1 << A->size_bits;
+	for (i = 0; i < size; i++) {
+		hlist_for_each_entry(entry, &A->buckets[i], hlist) {
+			if (!__ftrace_lookup_ip(B, entry->ip))
+				return false;
+		}
+	}
+
+	return true;
+}
+
+static void ftrace_ops_update_code(struct ftrace_ops *ops,
+				   struct ftrace_ops_hash *old_hash);
+
+static int __ftrace_hash_move_and_update_ops(struct ftrace_ops *ops,
+					     struct ftrace_hash **orig_hash,
+					     struct ftrace_hash *hash,
+					     int enable)
+{
+	struct ftrace_ops_hash old_hash_ops;
+	struct ftrace_hash *old_hash;
+	int ret;
+
+	old_hash = *orig_hash;
+	old_hash_ops.filter_hash = ops->func_hash->filter_hash;
+	old_hash_ops.notrace_hash = ops->func_hash->notrace_hash;
+	ret = ftrace_hash_move(ops, enable, orig_hash, hash);
+	if (!ret) {
+		ftrace_ops_update_code(ops, &old_hash_ops);
+		free_ftrace_hash_rcu(old_hash);
+	}
+	return ret;
+}
+
+static int ftrace_update_ops(struct ftrace_ops *ops, struct ftrace_hash *filter_hash,
+			     struct ftrace_hash *notrace_hash)
+{
+	int ret;
+
+	if (!ops_equal(filter_hash, ops->func_hash->filter_hash)) {
+		ret = __ftrace_hash_move_and_update_ops(ops, &ops->func_hash->filter_hash,
+							filter_hash, 1);
+		if (ret < 0)
+			return ret;
+	}
+
+	if (!ops_equal(notrace_hash, ops->func_hash->notrace_hash)) {
+		ret = __ftrace_hash_move_and_update_ops(ops, &ops->func_hash->notrace_hash,
+							notrace_hash, 0);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+/**
+ * ftrace_startup_subops - enable tracing for subops of an ops
+ * @ops: Manager ops (used to pick all the functions of its subops)
+ * @subops: A new ops to add to @ops
+ * @command: Extra commands to use to enable tracing
+ *
+ * The @ops is a manager @ops that has the filter that includes all the functions
+ * that its list of subops are tracing. Adding a new @subops will add the
+ * functions of @subops to @ops.
+ */
+int ftrace_startup_subops(struct ftrace_ops *ops, struct ftrace_ops *subops, int command)
+{
+	struct ftrace_hash *filter_hash;
+	struct ftrace_hash *notrace_hash;
+	struct ftrace_hash *save_filter_hash;
+	struct ftrace_hash *save_notrace_hash;
+	int size_bits;
+	int ret;
+
+	if (unlikely(ftrace_disabled))
+		return -ENODEV;
+
+	ftrace_ops_init(ops);
+	ftrace_ops_init(subops);
+
+	if (WARN_ON_ONCE(subops->flags & FTRACE_OPS_FL_ENABLED))
+		return -EBUSY;
+
+	/* Make everything canonical (Just in case!) */
+	if (!ops->func_hash->filter_hash)
+		ops->func_hash->filter_hash = EMPTY_HASH;
+	if (!ops->func_hash->notrace_hash)
+		ops->func_hash->notrace_hash = EMPTY_HASH;
+	if (!subops->func_hash->filter_hash)
+		subops->func_hash->filter_hash = EMPTY_HASH;
+	if (!subops->func_hash->notrace_hash)
+		subops->func_hash->notrace_hash = EMPTY_HASH;
+
+	/* For the first subops to ops just enable it normally */
+	if (list_empty(&ops->subop_list)) {
+		/* Just use the subops hashes */
+		filter_hash = copy_hash(subops->func_hash->filter_hash);
+		notrace_hash = copy_hash(subops->func_hash->notrace_hash);
+		if (!filter_hash || !notrace_hash) {
+			free_ftrace_hash(filter_hash);
+			free_ftrace_hash(notrace_hash);
+			return -ENOMEM;
+		}
+
+		save_filter_hash = ops->func_hash->filter_hash;
+		save_notrace_hash = ops->func_hash->notrace_hash;
+
+		ops->func_hash->filter_hash = filter_hash;
+		ops->func_hash->notrace_hash = notrace_hash;
+		list_add(&subops->list, &ops->subop_list);
+		ret = ftrace_startup(ops, command);
+		if (ret < 0) {
+			list_del(&subops->list);
+			ops->func_hash->filter_hash = save_filter_hash;
+			ops->func_hash->notrace_hash = save_notrace_hash;
+			free_ftrace_hash(filter_hash);
+			free_ftrace_hash(notrace_hash);
+		} else {
+			free_ftrace_hash(save_filter_hash);
+			free_ftrace_hash(save_notrace_hash);
+			subops->flags |= FTRACE_OPS_FL_ENABLED | FTRACE_OPS_FL_SUBOP;
+			subops->managed = ops;
+		}
+		return ret;
+	}
+
+	/*
+	 * Here there's already something attached. Here are the rules:
+	 *   o If either filter_hash is empty then the final stays empty
+	 *      o Otherwise, the final is a superset of both hashes
+	 *   o If either notrace_hash is empty then the final stays empty
+	 *      o Otherwise, the final is an intersection between the hashes
+	 */
+	if (ftrace_hash_empty(ops->func_hash->filter_hash) ||
+	    ftrace_hash_empty(subops->func_hash->filter_hash)) {
+		filter_hash = EMPTY_HASH;
+	} else {
+		size_bits = max(ops->func_hash->filter_hash->size_bits,
+				subops->func_hash->filter_hash->size_bits);
+		filter_hash = alloc_and_copy_ftrace_hash(size_bits, ops->func_hash->filter_hash);
+		if (!filter_hash)
+			return -ENOMEM;
+		ret = append_hash(&filter_hash, subops->func_hash->filter_hash);
+		if (ret < 0) {
+			free_ftrace_hash(filter_hash);
+			return ret;
+		}
+	}
+
+	if (ftrace_hash_empty(ops->func_hash->notrace_hash) ||
+	    ftrace_hash_empty(subops->func_hash->notrace_hash)) {
+		notrace_hash = EMPTY_HASH;
+	} else {
+		size_bits = max(ops->func_hash->filter_hash->size_bits,
+				subops->func_hash->filter_hash->size_bits);
+		notrace_hash = alloc_ftrace_hash(size_bits);
+		if (!notrace_hash) {
+			free_ftrace_hash(filter_hash);
+			return -ENOMEM;
+		}
+
+		ret = intersect_hash(&notrace_hash, ops->func_hash->filter_hash,
+				     subops->func_hash->filter_hash);
+		if (ret < 0) {
+			free_ftrace_hash(filter_hash);
+			free_ftrace_hash(notrace_hash);
+			return ret;
+		}
+	}
+
+	list_add(&subops->list, &ops->subop_list);
+
+	ret = ftrace_update_ops(ops, filter_hash, notrace_hash);
+	free_ftrace_hash(filter_hash);
+	free_ftrace_hash(notrace_hash);
+	if (ret < 0) {
+		list_del(&subops->list);
+	} else {
+		subops->flags |= FTRACE_OPS_FL_ENABLED | FTRACE_OPS_FL_SUBOP;
+		subops->managed = ops;
+	}
+	return ret;
+}
+
+/**
+ * ftrace_shutdown_subops - Remove a subops from a manager ops
+ * @ops: A manager ops to remove @subops from
+ * @subops: The subops to remove from @ops
+ * @command: Any extra command flags to add to modifying the text
+ *
+ * Removes the functions being traced by the @subops from @ops. Note, it
+ * will not affect functions that are being traced by other subops that
+ * still exist in @ops.
+ *
+ * If the last subops is removed from @ops, then @ops is shutdown normally.
+ */
+int ftrace_shutdown_subops(struct ftrace_ops *ops, struct ftrace_ops *subops, int command)
+{
+	struct ftrace_hash *filter_hash;
+	struct ftrace_hash *notrace_hash;
+	int ret;
+
+	if (unlikely(ftrace_disabled))
+		return -ENODEV;
+
+	if (WARN_ON_ONCE(!(subops->flags & FTRACE_OPS_FL_ENABLED)))
+		return -EINVAL;
+
+	list_del(&subops->list);
+
+	if (list_empty(&ops->subop_list)) {
+		/* Last one, just disable the current ops */
+
+		ret = ftrace_shutdown(ops, command);
+		if (ret < 0) {
+			list_add(&subops->list, &ops->subop_list);
+			return ret;
+		}
+
+		subops->flags &= ~FTRACE_OPS_FL_ENABLED;
+
+		free_ftrace_hash(ops->func_hash->filter_hash);
+		free_ftrace_hash(ops->func_hash->notrace_hash);
+		ops->func_hash->filter_hash = EMPTY_HASH;
+		ops->func_hash->notrace_hash = EMPTY_HASH;
+		subops->flags &= ~(FTRACE_OPS_FL_ENABLED | FTRACE_OPS_FL_SUBOP);
+		subops->managed = NULL;
+
+		return 0;
+	}
+
+	/* Rebuild the hashes without subops */
+	filter_hash = append_hashes(ops);
+	notrace_hash = intersect_hashes(ops);
+	if (!filter_hash || !notrace_hash) {
+		free_ftrace_hash(filter_hash);
+		free_ftrace_hash(notrace_hash);
+		list_add(&subops->list, &ops->subop_list);
+		return -ENOMEM;
+	}
+
+	ret = ftrace_update_ops(ops, filter_hash, notrace_hash);
+	if (ret < 0) {
+		list_add(&subops->list, &ops->subop_list);
+	} else {
+		subops->flags &= ~(FTRACE_OPS_FL_ENABLED | FTRACE_OPS_FL_SUBOP);
+		subops->managed = NULL;
+	}
+	free_ftrace_hash(filter_hash);
+	free_ftrace_hash(notrace_hash);
+	return ret;
+}
+
+static int ftrace_hash_move_and_update_subops(struct ftrace_ops *subops,
+					      struct ftrace_hash **orig_subhash,
+					      struct ftrace_hash *hash,
+					      int enable)
+{
+	struct ftrace_ops *ops = subops->managed;
+	struct ftrace_hash **orig_hash;
+	struct ftrace_hash *save_hash;
+	struct ftrace_hash *new_hash;
+	int ret;
+
+	/* Manager ops can not be subops (yet) */
+	if (WARN_ON_ONCE(!ops || ops->flags & FTRACE_OPS_FL_SUBOP))
+		return -EINVAL;
+
+	/* Move the new hash over to the subops hash */
+	save_hash = *orig_subhash;
+	*orig_subhash = __ftrace_hash_move(hash);
+	if (!*orig_subhash) {
+		*orig_subhash = save_hash;
+		return -ENOMEM;
+	}
+
+	/* Create a new_hash to hold the ops new functions */
+	if (enable) {
+		orig_hash = &ops->func_hash->filter_hash;
+		new_hash = append_hashes(ops);
+	} else {
+		orig_hash = &ops->func_hash->notrace_hash;
+		new_hash = intersect_hashes(ops);
+	}
+
+	/* Move the hash over to the new hash */
+	ret = __ftrace_hash_move_and_update_ops(ops, orig_hash, new_hash, enable);
+
+	free_ftrace_hash(new_hash);
+
+	if (ret) {
+		/* Put back the original hash */
+		free_ftrace_hash_rcu(*orig_subhash);
+		*orig_subhash = save_hash;
+	} else {
+		free_ftrace_hash_rcu(save_hash);
+	}
+	return ret;
+}
+
+
 static u64		ftrace_update_time;
 unsigned long		ftrace_update_tot_cnt;
 unsigned long		ftrace_number_of_pages;
@@ -4380,19 +4879,33 @@ static int ftrace_hash_move_and_update_ops(struct ftrace_ops *ops,
 					   struct ftrace_hash *hash,
 					   int enable)
 {
-	struct ftrace_ops_hash old_hash_ops;
-	struct ftrace_hash *old_hash;
-	int ret;
+	if (ops->flags & FTRACE_OPS_FL_SUBOP)
+		return ftrace_hash_move_and_update_subops(ops, orig_hash, hash, enable);
 
-	old_hash = *orig_hash;
-	old_hash_ops.filter_hash = ops->func_hash->filter_hash;
-	old_hash_ops.notrace_hash = ops->func_hash->notrace_hash;
-	ret = ftrace_hash_move(ops, enable, orig_hash, hash);
-	if (!ret) {
-		ftrace_ops_update_code(ops, &old_hash_ops);
-		free_ftrace_hash_rcu(old_hash);
+	/*
+	 * If this ops is not enabled, it could be sharing its filters
+	 * with a subop. If that's the case, update the subop instead of
+	 * this ops. Shared filters are only allowed to have one ops set
+	 * at a time, and if we update the ops that is not enabled,
+	 * it will not affect subops that share it.
+	 */
+	if (!(ops->flags & FTRACE_OPS_FL_ENABLED)) {
+		struct ftrace_ops *op;
+
+		/* Check if any other manager subops maps to this hash */
+		do_for_each_ftrace_op(op, ftrace_ops_list) {
+			struct ftrace_ops *subops;
+
+			list_for_each_entry(subops, &op->subop_list, list) {
+				if ((subops->flags & FTRACE_OPS_FL_ENABLED) &&
+				     subops->func_hash == ops->func_hash) {
+					return ftrace_hash_move_and_update_subops(subops, orig_hash, hash, enable);
+				}
+			}
+		} while_for_each_ftrace_op(op);
 	}
-	return ret;
+
+	return __ftrace_hash_move_and_update_ops(ops, orig_hash, hash, enable);
 }
 
 static bool module_exists(const char *module)
@@ -5475,6 +5988,8 @@ EXPORT_SYMBOL_GPL(register_ftrace_direct);
  * unregister_ftrace_direct - Remove calls to custom trampoline
  * previously registered by register_ftrace_direct for @ops object.
  * @ops: The address of the struct ftrace_ops object
+ * @addr: The address of the direct function that is called by the @ops functions
+ * @free_filters: Set to true to remove all filters for the ftrace_ops, false otherwise
  *
  * This is used to remove a direct calls to @addr from the nop locations
  * of the functions registered in @ops (with by ftrace_set_filter_ip
@@ -7324,6 +7839,7 @@ __init void ftrace_init_global_array_ops(struct trace_array *tr)
 	tr->ops = &global_ops;
 	tr->ops->private = tr;
 	ftrace_init_trace_array(tr);
+	init_array_fgraph_ops(tr, tr->ops);
 }
 
 void ftrace_init_array_ops(struct trace_array *tr, ftrace_func_t func)

kernel/trace/ftrace_internal.h

@@ -15,6 +15,8 @@ extern struct ftrace_ops global_ops;
 int ftrace_startup(struct ftrace_ops *ops, int command);
 int ftrace_shutdown(struct ftrace_ops *ops, int command);
 int ftrace_ops_test(struct ftrace_ops *ops, unsigned long ip, void *regs);
+int ftrace_startup_subops(struct ftrace_ops *ops, struct ftrace_ops *subops, int command);
+int ftrace_shutdown_subops(struct ftrace_ops *ops, struct ftrace_ops *subops, int command);
 
 #else /* !CONFIG_DYNAMIC_FTRACE */
 
@@ -38,14 +40,26 @@ ftrace_ops_test(struct ftrace_ops *ops, unsigned long ip, void *regs)
 {
 	return 1;
 }
+static inline int ftrace_startup_subops(struct ftrace_ops *ops, struct ftrace_ops *subops, int command)
+{
+	return -EINVAL;
+}
+static inline int ftrace_shutdown_subops(struct ftrace_ops *ops, struct ftrace_ops *subops, int command)
+{
+	return -EINVAL;
+}
 #endif /* CONFIG_DYNAMIC_FTRACE */
 
 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
 extern int ftrace_graph_active;
-void update_function_graph_func(void);
+# ifdef CONFIG_DYNAMIC_FTRACE
+extern void fgraph_update_pid_func(void);
+# else
+static inline void fgraph_update_pid_func(void) {}
+# endif
 #else /* !CONFIG_FUNCTION_GRAPH_TRACER */
 # define ftrace_graph_active 0
-static inline void update_function_graph_func(void) { }
+static inline void fgraph_update_pid_func(void) {}
 #endif /* CONFIG_FUNCTION_GRAPH_TRACER */
 
 #else /* !CONFIG_FUNCTION_TRACER */

kernel/trace/trace.h

@@ -397,6 +397,9 @@ struct trace_array {
 	struct ftrace_ops	*ops;
 	struct trace_pid_list	__rcu *function_pids;
 	struct trace_pid_list	__rcu *function_no_pids;
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+	struct fgraph_ops	*gops;
+#endif
 #ifdef CONFIG_DYNAMIC_FTRACE
 	/* All of these are protected by the ftrace_lock */
 	struct list_head	func_probes;
@@ -679,9 +682,8 @@ void trace_latency_header(struct seq_file *m);
 void trace_default_header(struct seq_file *m);
 void print_trace_header(struct seq_file *m, struct trace_iterator *iter);
 
-void trace_graph_return(struct ftrace_graph_ret *trace);
-int trace_graph_entry(struct ftrace_graph_ent *trace);
-void set_graph_array(struct trace_array *tr);
+void trace_graph_return(struct ftrace_graph_ret *trace, struct fgraph_ops *gops);
+int trace_graph_entry(struct ftrace_graph_ent *trace, struct fgraph_ops *gops);
 
 void tracing_start_cmdline_record(void);
 void tracing_stop_cmdline_record(void);
@@ -892,12 +894,59 @@ extern int __trace_graph_entry(struct trace_array *tr,
 extern void __trace_graph_return(struct trace_array *tr,
 				 struct ftrace_graph_ret *trace,
 				 unsigned int trace_ctx);
+extern void init_array_fgraph_ops(struct trace_array *tr, struct ftrace_ops *ops);
+extern int allocate_fgraph_ops(struct trace_array *tr, struct ftrace_ops *ops);
+extern void free_fgraph_ops(struct trace_array *tr);
+
+enum {
+	TRACE_GRAPH_FL		= 1,
+
+	/*
+	 * In the very unlikely case that an interrupt came in
+	 * at a start of graph tracing, and we want to trace
+	 * the function in that interrupt, the depth can be greater
+	 * than zero, because of the preempted start of a previous
+	 * trace. In an even more unlikely case, depth could be 2
+	 * if a softirq interrupted the start of graph tracing,
+	 * followed by an interrupt preempting a start of graph
+	 * tracing in the softirq, and depth can even be 3
+	 * if an NMI came in at the start of an interrupt function
+	 * that preempted a softirq start of a function that
+	 * preempted normal context!!!! Luckily, it can't be
+	 * greater than 3, so the next two bits are a mask
+	 * of what the depth is when we set TRACE_GRAPH_FL
+	 */
+
+	TRACE_GRAPH_DEPTH_START_BIT,
+	TRACE_GRAPH_DEPTH_END_BIT,
+
+	/*
+	 * To implement set_graph_notrace, if this bit is set, we ignore
+	 * function graph tracing of called functions, until the return
+	 * function is called to clear it.
+	 */
+	TRACE_GRAPH_NOTRACE_BIT,
+};
+
+#define TRACE_GRAPH_NOTRACE		(1 << TRACE_GRAPH_NOTRACE_BIT)
+
+static inline unsigned long ftrace_graph_depth(unsigned long *task_var)
+{
+	return (*task_var >> TRACE_GRAPH_DEPTH_START_BIT) & 3;
+}
+
+static inline void ftrace_graph_set_depth(unsigned long *task_var, int depth)
+{
+	*task_var &= ~(3 << TRACE_GRAPH_DEPTH_START_BIT);
+	*task_var |= (depth & 3) << TRACE_GRAPH_DEPTH_START_BIT;
+}
 
 #ifdef CONFIG_DYNAMIC_FTRACE
 extern struct ftrace_hash __rcu *ftrace_graph_hash;
 extern struct ftrace_hash __rcu *ftrace_graph_notrace_hash;
 
-static inline int ftrace_graph_addr(struct ftrace_graph_ent *trace)
+static inline int
+ftrace_graph_addr(unsigned long *task_var, struct ftrace_graph_ent *trace)
 {
 	unsigned long addr = trace->func;
 	int ret = 0;
@@ -919,13 +968,12 @@ static inline int ftrace_graph_addr(struct ftrace_graph_ent *trace)
 	}
 
 	if (ftrace_lookup_ip(hash, addr)) {
-
 		/*
 		 * This needs to be cleared on the return functions
 		 * when the depth is zero.
 		 */
-		trace_recursion_set(TRACE_GRAPH_BIT);
-		trace_recursion_set_depth(trace->depth);
+		*task_var |= TRACE_GRAPH_FL;
+		ftrace_graph_set_depth(task_var, trace->depth);
 
 		/*
 		 * If no irqs are to be traced, but a set_graph_function
@@ -944,11 +992,14 @@ static inline int ftrace_graph_addr(struct ftrace_graph_ent *trace)
 	return ret;
 }
 
-static inline void ftrace_graph_addr_finish(struct ftrace_graph_ret *trace)
+static inline void
+ftrace_graph_addr_finish(struct fgraph_ops *gops, struct ftrace_graph_ret *trace)
 {
-	if (trace_recursion_test(TRACE_GRAPH_BIT) &&
-	    trace->depth == trace_recursion_depth())
-		trace_recursion_clear(TRACE_GRAPH_BIT);
+	unsigned long *task_var = fgraph_get_task_var(gops);
+
+	if ((*task_var & TRACE_GRAPH_FL) &&
+	    trace->depth == ftrace_graph_depth(task_var))
+		*task_var &= ~TRACE_GRAPH_FL;
 }
 
 static inline int ftrace_graph_notrace_addr(unsigned long addr)
@@ -974,7 +1025,7 @@ static inline int ftrace_graph_notrace_addr(unsigned long addr)
 	return ret;
 }
 #else
-static inline int ftrace_graph_addr(struct ftrace_graph_ent *trace)
+static inline int ftrace_graph_addr(unsigned long *task_var, struct ftrace_graph_ent *trace)
 {
 	return 1;
 }
@@ -983,27 +1034,37 @@ static inline int ftrace_graph_notrace_addr(unsigned long addr)
 {
 	return 0;
 }
-static inline void ftrace_graph_addr_finish(struct ftrace_graph_ret *trace)
+static inline void ftrace_graph_addr_finish(struct fgraph_ops *gops, struct ftrace_graph_ret *trace)
 { }
 #endif /* CONFIG_DYNAMIC_FTRACE */
 
 extern unsigned int fgraph_max_depth;
 
-static inline bool ftrace_graph_ignore_func(struct ftrace_graph_ent *trace)
+static inline bool
+ftrace_graph_ignore_func(struct fgraph_ops *gops, struct ftrace_graph_ent *trace)
 {
+	unsigned long *task_var = fgraph_get_task_var(gops);
+
 	/* trace it when it is-nested-in or is a function enabled. */
-	return !(trace_recursion_test(TRACE_GRAPH_BIT) ||
-		 ftrace_graph_addr(trace)) ||
+	return !((*task_var & TRACE_GRAPH_FL) ||
+		 ftrace_graph_addr(task_var, trace)) ||
 		(trace->depth < 0) ||
 		(fgraph_max_depth && trace->depth >= fgraph_max_depth);
 }
 
+void fgraph_init_ops(struct ftrace_ops *dst_ops,
+		     struct ftrace_ops *src_ops);
+
 #else /* CONFIG_FUNCTION_GRAPH_TRACER */
 static inline enum print_line_t
 print_graph_function_flags(struct trace_iterator *iter, u32 flags)
 {
 	return TRACE_TYPE_UNHANDLED;
 }
+static inline void free_fgraph_ops(struct trace_array *tr) { }
+/* ftrace_ops may not be defined */
+#define init_array_fgraph_ops(tr, ops) do { } while (0)
+#define allocate_fgraph_ops(tr, ops) ({ 0; })
 #endif /* CONFIG_FUNCTION_GRAPH_TRACER */
 
 extern struct list_head ftrace_pids;

kernel/trace/trace_functions.c

@@ -80,6 +80,7 @@ void ftrace_free_ftrace_ops(struct trace_array *tr)
 int ftrace_create_function_files(struct trace_array *tr,
 				 struct dentry *parent)
 {
+	int ret;
 	/*
 	 * The top level array uses the "global_ops", and the files are
 	 * created on boot up.
@@ -90,6 +91,12 @@ int ftrace_create_function_files(struct trace_array *tr,
 	if (!tr->ops)
 		return -EINVAL;
 
+	ret = allocate_fgraph_ops(tr, tr->ops);
+	if (ret) {
+		kfree(tr->ops);
+		return ret;
+	}
+
 	ftrace_create_filter_files(tr->ops, parent);
 
 	return 0;
@@ -99,6 +106,7 @@ void ftrace_destroy_function_files(struct trace_array *tr)
 {
 	ftrace_destroy_filter_files(tr->ops);
 	ftrace_free_ftrace_ops(tr);
+	free_fgraph_ops(tr);
 }
 
 static ftrace_func_t select_trace_function(u32 flags_val)
@@ -223,6 +231,7 @@ function_stack_trace_call(unsigned long ip, unsigned long parent_ip,
 	long disabled;
 	int cpu;
 	unsigned int trace_ctx;
+	int skip = STACK_SKIP;
 
 	if (unlikely(!tr->function_enabled))
 		return;
@@ -239,7 +248,11 @@ function_stack_trace_call(unsigned long ip, unsigned long parent_ip,
 	if (likely(disabled == 1)) {
 		trace_ctx = tracing_gen_ctx_flags(flags);
 		trace_function(tr, ip, parent_ip, trace_ctx);
-		__trace_stack(tr, trace_ctx, STACK_SKIP);
+#ifdef CONFIG_UNWINDER_FRAME_POINTER
+		if (ftrace_pids_enabled(op))
+			skip++;
+#endif
+		__trace_stack(tr, trace_ctx, skip);
 	}
 
 	atomic_dec(&data->disabled);

kernel/trace/trace_functions_graph.c

@@ -83,8 +83,6 @@ static struct tracer_flags tracer_flags = {
 	.opts = trace_opts
 };
 
-static struct trace_array *graph_array;
-
 /*
  * DURATION column is being also used to display IRQ signs,
  * following values are used by print_graph_irq and others
@@ -129,9 +127,11 @@ static inline int ftrace_graph_ignore_irqs(void)
 	return in_hardirq();
 }
 
-int trace_graph_entry(struct ftrace_graph_ent *trace)
+int trace_graph_entry(struct ftrace_graph_ent *trace,
+		      struct fgraph_ops *gops)
 {
-	struct trace_array *tr = graph_array;
+	unsigned long *task_var = fgraph_get_task_var(gops);
+	struct trace_array *tr = gops->private;
 	struct trace_array_cpu *data;
 	unsigned long flags;
 	unsigned int trace_ctx;
@@ -139,7 +139,7 @@ int trace_graph_entry(struct ftrace_graph_ent *trace)
 	int ret;
 	int cpu;
 
-	if (trace_recursion_test(TRACE_GRAPH_NOTRACE_BIT))
+	if (*task_var & TRACE_GRAPH_NOTRACE)
 		return 0;
 
 	/*
@@ -150,7 +150,7 @@ int trace_graph_entry(struct ftrace_graph_ent *trace)
 	 * returning from the function.
 	 */
 	if (ftrace_graph_notrace_addr(trace->func)) {
-		trace_recursion_set(TRACE_GRAPH_NOTRACE_BIT);
+		*task_var |= TRACE_GRAPH_NOTRACE_BIT;
 		/*
 		 * Need to return 1 to have the return called
 		 * that will clear the NOTRACE bit.
@@ -161,7 +161,7 @@ int trace_graph_entry(struct ftrace_graph_ent *trace)
 	if (!ftrace_trace_task(tr))
 		return 0;
 
-	if (ftrace_graph_ignore_func(trace))
+	if (ftrace_graph_ignore_func(gops, trace))
 		return 0;
 
 	if (ftrace_graph_ignore_irqs())
@@ -238,19 +238,21 @@ void __trace_graph_return(struct trace_array *tr,
 		trace_buffer_unlock_commit_nostack(buffer, event);
 }
 
-void trace_graph_return(struct ftrace_graph_ret *trace)
+void trace_graph_return(struct ftrace_graph_ret *trace,
+			struct fgraph_ops *gops)
 {
-	struct trace_array *tr = graph_array;
+	unsigned long *task_var = fgraph_get_task_var(gops);
+	struct trace_array *tr = gops->private;
 	struct trace_array_cpu *data;
 	unsigned long flags;
 	unsigned int trace_ctx;
 	long disabled;
 	int cpu;
 
-	ftrace_graph_addr_finish(trace);
+	ftrace_graph_addr_finish(gops, trace);
 
-	if (trace_recursion_test(TRACE_GRAPH_NOTRACE_BIT)) {
-		trace_recursion_clear(TRACE_GRAPH_NOTRACE_BIT);
+	if (*task_var & TRACE_GRAPH_NOTRACE) {
+		*task_var &= ~TRACE_GRAPH_NOTRACE;
 		return;
 	}
 
@@ -266,18 +268,10 @@ void trace_graph_return(struct ftrace_graph_ret *trace)
 	local_irq_restore(flags);
 }
 
-void set_graph_array(struct trace_array *tr)
-{
-	graph_array = tr;
-
-	/* Make graph_array visible before we start tracing */
-
-	smp_mb();
-}
-
-static void trace_graph_thresh_return(struct ftrace_graph_ret *trace)
+static void trace_graph_thresh_return(struct ftrace_graph_ret *trace,
+				      struct fgraph_ops *gops)
 {
-	ftrace_graph_addr_finish(trace);
+	ftrace_graph_addr_finish(gops, trace);
 
 	if (trace_recursion_test(TRACE_GRAPH_NOTRACE_BIT)) {
 		trace_recursion_clear(TRACE_GRAPH_NOTRACE_BIT);
@@ -288,28 +282,60 @@ static void trace_graph_thresh_return(struct ftrace_graph_ret *trace)
 	    (trace->rettime - trace->calltime < tracing_thresh))
 		return;
 	else
-		trace_graph_return(trace);
+		trace_graph_return(trace, gops);
 }
 
-static struct fgraph_ops funcgraph_thresh_ops = {
-	.entryfunc = &trace_graph_entry,
-	.retfunc = &trace_graph_thresh_return,
-};
-
 static struct fgraph_ops funcgraph_ops = {
 	.entryfunc = &trace_graph_entry,
 	.retfunc = &trace_graph_return,
 };
 
+int allocate_fgraph_ops(struct trace_array *tr, struct ftrace_ops *ops)
+{
+	struct fgraph_ops *gops;
+
+	gops = kzalloc(sizeof(*gops), GFP_KERNEL);
+	if (!gops)
+		return -ENOMEM;
+
+	gops->entryfunc = &trace_graph_entry;
+	gops->retfunc = &trace_graph_return;
+
+	tr->gops = gops;
+	gops->private = tr;
+
+	fgraph_init_ops(&gops->ops, ops);
+
+	return 0;
+}
+
+void free_fgraph_ops(struct trace_array *tr)
+{
+	kfree(tr->gops);
+}
+
+__init void init_array_fgraph_ops(struct trace_array *tr, struct ftrace_ops *ops)
+{
+	tr->gops = &funcgraph_ops;
+	funcgraph_ops.private = tr;
+	fgraph_init_ops(&tr->gops->ops, ops);
+}
+
 static int graph_trace_init(struct trace_array *tr)
 {
 	int ret;
 
-	set_graph_array(tr);
+	tr->gops->entryfunc = trace_graph_entry;
+
 	if (tracing_thresh)
-		ret = register_ftrace_graph(&funcgraph_thresh_ops);
+		tr->gops->retfunc = trace_graph_thresh_return;
 	else
-		ret = register_ftrace_graph(&funcgraph_ops);
+		tr->gops->retfunc = trace_graph_return;
+
+	/* Make gops functions are visible before we start tracing */
+	smp_mb();
+
+	ret = register_ftrace_graph(tr->gops);
 	if (ret)
 		return ret;
 	tracing_start_cmdline_record();
@@ -320,10 +346,7 @@ static int graph_trace_init(struct trace_array *tr)
 static void graph_trace_reset(struct trace_array *tr)
 {
 	tracing_stop_cmdline_record();
-	if (tracing_thresh)
-		unregister_ftrace_graph(&funcgraph_thresh_ops);
-	else
-		unregister_ftrace_graph(&funcgraph_ops);
+	unregister_ftrace_graph(tr->gops);
 }
 
 static int graph_trace_update_thresh(struct trace_array *tr)
@@ -1362,6 +1385,7 @@ static struct tracer graph_trace __tracer_data = {
 	.print_header	= print_graph_headers,
 	.flags		= &tracer_flags,
 	.set_flag	= func_graph_set_flag,
+	.allow_instances = true,
 #ifdef CONFIG_FTRACE_SELFTEST
 	.selftest	= trace_selftest_startup_function_graph,
 #endif

kernel/trace/trace_irqsoff.c

@@ -175,7 +175,8 @@ static int irqsoff_display_graph(struct trace_array *tr, int set)
 	return start_irqsoff_tracer(irqsoff_trace, set);
 }
 
-static int irqsoff_graph_entry(struct ftrace_graph_ent *trace)
+static int irqsoff_graph_entry(struct ftrace_graph_ent *trace,
+			       struct fgraph_ops *gops)
 {
 	struct trace_array *tr = irqsoff_trace;
 	struct trace_array_cpu *data;
@@ -183,7 +184,7 @@ static int irqsoff_graph_entry(struct ftrace_graph_ent *trace)
 	unsigned int trace_ctx;
 	int ret;
 
-	if (ftrace_graph_ignore_func(trace))
+	if (ftrace_graph_ignore_func(gops, trace))
 		return 0;
 	/*
 	 * Do not trace a function if it's filtered by set_graph_notrace.
@@ -205,14 +206,15 @@ static int irqsoff_graph_entry(struct ftrace_graph_ent *trace)
 	return ret;
 }
 
-static void irqsoff_graph_return(struct ftrace_graph_ret *trace)
+static void irqsoff_graph_return(struct ftrace_graph_ret *trace,
+				 struct fgraph_ops *gops)
 {
 	struct trace_array *tr = irqsoff_trace;
 	struct trace_array_cpu *data;
 	unsigned long flags;
 	unsigned int trace_ctx;
 
-	ftrace_graph_addr_finish(trace);
+	ftrace_graph_addr_finish(gops, trace);
 
 	if (!func_prolog_dec(tr, &data, &flags))
 		return;

kernel/trace/trace_sched_wakeup.c

@@ -112,14 +112,15 @@ static int wakeup_display_graph(struct trace_array *tr, int set)
 	return start_func_tracer(tr, set);
 }
 
-static int wakeup_graph_entry(struct ftrace_graph_ent *trace)
+static int wakeup_graph_entry(struct ftrace_graph_ent *trace,
+			      struct fgraph_ops *gops)
 {
 	struct trace_array *tr = wakeup_trace;
 	struct trace_array_cpu *data;
 	unsigned int trace_ctx;
 	int ret = 0;
 
-	if (ftrace_graph_ignore_func(trace))
+	if (ftrace_graph_ignore_func(gops, trace))
 		return 0;
 	/*
 	 * Do not trace a function if it's filtered by set_graph_notrace.
@@ -141,13 +142,14 @@ static int wakeup_graph_entry(struct ftrace_graph_ent *trace)
 	return ret;
 }
 
-static void wakeup_graph_return(struct ftrace_graph_ret *trace)
+static void wakeup_graph_return(struct ftrace_graph_ret *trace,
+				struct fgraph_ops *gops)
 {
 	struct trace_array *tr = wakeup_trace;
 	struct trace_array_cpu *data;
 	unsigned int trace_ctx;
 
-	ftrace_graph_addr_finish(trace);
+	ftrace_graph_addr_finish(gops, trace);
 
 	if (!func_prolog_preempt_disable(tr, &data, &trace_ctx))
 		return;

kernel/trace/trace_selftest.c

@@ -756,13 +756,262 @@ trace_selftest_startup_function(struct tracer *trace, struct trace_array *tr)
 
 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
 
+#ifdef CONFIG_DYNAMIC_FTRACE
+
+#define CHAR_NUMBER 123
+#define SHORT_NUMBER 12345
+#define WORD_NUMBER 1234567890
+#define LONG_NUMBER 1234567890123456789LL
+#define ERRSTR_BUFLEN 128
+
+struct fgraph_fixture {
+	struct fgraph_ops gops;
+	int store_size;
+	const char *store_type_name;
+	char error_str_buf[ERRSTR_BUFLEN];
+	char *error_str;
+};
+
+static __init int store_entry(struct ftrace_graph_ent *trace,
+			      struct fgraph_ops *gops)
+{
+	struct fgraph_fixture *fixture = container_of(gops, struct fgraph_fixture, gops);
+	const char *type = fixture->store_type_name;
+	int size = fixture->store_size;
+	void *p;
+
+	p = fgraph_reserve_data(gops->idx, size);
+	if (!p) {
+		snprintf(fixture->error_str_buf, ERRSTR_BUFLEN,
+			 "Failed to reserve %s\n", type);
+		return 0;
+	}
+
+	switch (size) {
+	case 1:
+		*(char *)p = CHAR_NUMBER;
+		break;
+	case 2:
+		*(short *)p = SHORT_NUMBER;
+		break;
+	case 4:
+		*(int *)p = WORD_NUMBER;
+		break;
+	case 8:
+		*(long long *)p = LONG_NUMBER;
+		break;
+	}
+
+	return 1;
+}
+
+static __init void store_return(struct ftrace_graph_ret *trace,
+				struct fgraph_ops *gops)
+{
+	struct fgraph_fixture *fixture = container_of(gops, struct fgraph_fixture, gops);
+	const char *type = fixture->store_type_name;
+	long long expect = 0;
+	long long found = -1;
+	int size;
+	char *p;
+
+	p = fgraph_retrieve_data(gops->idx, &size);
+	if (!p) {
+		snprintf(fixture->error_str_buf, ERRSTR_BUFLEN,
+			 "Failed to retrieve %s\n", type);
+		return;
+	}
+	if (fixture->store_size > size) {
+		snprintf(fixture->error_str_buf, ERRSTR_BUFLEN,
+			 "Retrieved size %d is smaller than expected %d\n",
+			 size, (int)fixture->store_size);
+		return;
+	}
+
+	switch (fixture->store_size) {
+	case 1:
+		expect = CHAR_NUMBER;
+		found = *(char *)p;
+		break;
+	case 2:
+		expect = SHORT_NUMBER;
+		found = *(short *)p;
+		break;
+	case 4:
+		expect = WORD_NUMBER;
+		found = *(int *)p;
+		break;
+	case 8:
+		expect = LONG_NUMBER;
+		found = *(long long *)p;
+		break;
+	}
+
+	if (found != expect) {
+		snprintf(fixture->error_str_buf, ERRSTR_BUFLEN,
+			 "%s returned not %lld but %lld\n", type, expect, found);
+		return;
+	}
+	fixture->error_str = NULL;
+}
+
+static int __init init_fgraph_fixture(struct fgraph_fixture *fixture)
+{
+	char *func_name;
+	int len;
+
+	snprintf(fixture->error_str_buf, ERRSTR_BUFLEN,
+		 "Failed to execute storage %s\n", fixture->store_type_name);
+	fixture->error_str = fixture->error_str_buf;
+
+	func_name = "*" __stringify(DYN_FTRACE_TEST_NAME);
+	len = strlen(func_name);
+
+	return ftrace_set_filter(&fixture->gops.ops, func_name, len, 1);
+}
+
+/* Test fgraph storage for each size */
+static int __init test_graph_storage_single(struct fgraph_fixture *fixture)
+{
+	int size = fixture->store_size;
+	int ret;
+
+	pr_cont("PASSED\n");
+	pr_info("Testing fgraph storage of %d byte%s: ", size, str_plural(size));
+
+	ret = init_fgraph_fixture(fixture);
+	if (ret && ret != -ENODEV) {
+		pr_cont("*Could not set filter* ");
+		return -1;
+	}
+
+	ret = register_ftrace_graph(&fixture->gops);
+	if (ret) {
+		pr_warn("Failed to init store_bytes fgraph tracing\n");
+		return -1;
+	}
+
+	DYN_FTRACE_TEST_NAME();
+
+	unregister_ftrace_graph(&fixture->gops);
+
+	if (fixture->error_str) {
+		pr_cont("*** %s ***", fixture->error_str);
+		return -1;
+	}
+
+	return 0;
+}
+
+static struct fgraph_fixture store_bytes[4] __initdata = {
+	[0] = {
+		.gops = {
+			.entryfunc		= store_entry,
+			.retfunc		= store_return,
+		},
+		.store_size = 1,
+		.store_type_name = "byte",
+	},
+	[1] = {
+		.gops = {
+			.entryfunc		= store_entry,
+			.retfunc		= store_return,
+		},
+		.store_size = 2,
+		.store_type_name = "short",
+	},
+	[2] = {
+		.gops = {
+			.entryfunc		= store_entry,
+			.retfunc		= store_return,
+		},
+		.store_size = 4,
+		.store_type_name = "word",
+	},
+	[3] = {
+		.gops = {
+			.entryfunc		= store_entry,
+			.retfunc		= store_return,
+		},
+		.store_size = 8,
+		.store_type_name = "long long",
+	},
+};
+
+static __init int test_graph_storage_multi(void)
+{
+	struct fgraph_fixture *fixture;
+	bool printed = false;
+	int i, ret;
+
+	pr_cont("PASSED\n");
+	pr_info("Testing multiple fgraph storage on a function: ");
+
+	for (i = 0; i < ARRAY_SIZE(store_bytes); i++) {
+		fixture = &store_bytes[i];
+		ret = init_fgraph_fixture(fixture);
+		if (ret && ret != -ENODEV) {
+			pr_cont("*Could not set filter* ");
+			printed = true;
+			goto out;
+		}
+
+		ret = register_ftrace_graph(&fixture->gops);
+		if (ret) {
+			pr_warn("Failed to init store_bytes fgraph tracing\n");
+			printed = true;
+			goto out;
+		}
+	}
+
+	DYN_FTRACE_TEST_NAME();
+out:
+	while (--i >= 0) {
+		fixture = &store_bytes[i];
+		unregister_ftrace_graph(&fixture->gops);
+
+		if (fixture->error_str && !printed) {
+			pr_cont("*** %s ***", fixture->error_str);
+			printed = true;
+		}
+	}
+	return printed ? -1 : 0;
+}
+
+/* Test the storage passed across function_graph entry and return */
+static __init int test_graph_storage(void)
+{
+	int ret;
+
+	ret = test_graph_storage_single(&store_bytes[0]);
+	if (ret)
+		return ret;
+	ret = test_graph_storage_single(&store_bytes[1]);
+	if (ret)
+		return ret;
+	ret = test_graph_storage_single(&store_bytes[2]);
+	if (ret)
+		return ret;
+	ret = test_graph_storage_single(&store_bytes[3]);
+	if (ret)
+		return ret;
+	ret = test_graph_storage_multi();
+	if (ret)
+		return ret;
+	return 0;
+}
+#else
+static inline int test_graph_storage(void) { return 0; }
+#endif /* CONFIG_DYNAMIC_FTRACE */
+
 /* Maximum number of functions to trace before diagnosing a hang */
 #define GRAPH_MAX_FUNC_TEST	100000000
 
 static unsigned int graph_hang_thresh;
 
 /* Wrap the real function entry probe to avoid possible hanging */
-static int trace_graph_entry_watchdog(struct ftrace_graph_ent *trace)
+static int trace_graph_entry_watchdog(struct ftrace_graph_ent *trace,
+				      struct fgraph_ops *gops)
 {
 	/* This is harmlessly racy, we want to approximately detect a hang */
 	if (unlikely(++graph_hang_thresh > GRAPH_MAX_FUNC_TEST)) {
@@ -776,7 +1025,7 @@ static int trace_graph_entry_watchdog(struct ftrace_graph_ent *trace)
 		return 0;
 	}
 
-	return trace_graph_entry(trace);
+	return trace_graph_entry(trace, gops);
 }
 
 static struct fgraph_ops fgraph_ops __initdata  = {
@@ -812,7 +1061,7 @@ trace_selftest_startup_function_graph(struct tracer *trace,
 	 * to detect and recover from possible hangs
 	 */
 	tracing_reset_online_cpus(&tr->array_buffer);
-	set_graph_array(tr);
+	fgraph_ops.private = tr;
 	ret = register_ftrace_graph(&fgraph_ops);
 	if (ret) {
 		warn_failed_init_tracer(trace, ret);
@@ -855,7 +1104,7 @@ trace_selftest_startup_function_graph(struct tracer *trace,
 	cond_resched();
 
 	tracing_reset_online_cpus(&tr->array_buffer);
-	set_graph_array(tr);
+	fgraph_ops.private = tr;
 
 	/*
 	 * Some archs *cough*PowerPC*cough* add characters to the
@@ -912,6 +1161,8 @@ trace_selftest_startup_function_graph(struct tracer *trace,
 	ftrace_set_global_filter(NULL, 0, 1);
 #endif
 
+	ret = test_graph_storage();
+
 	/* Don't test dynamic tracing, the function tracer already did */
 out:
 	/* Stop it if we failed */

tools/testing/selftests/ftrace/test.d/ftrace/fgraph-multi.tc

+#!/bin/sh
+# SPDX-License-Identifier: GPL-2.0
+# description: ftrace - function graph filters
+# requires: set_ftrace_filter function_graph:tracer
+
+# Make sure that function graph filtering works
+
+INSTANCE1="instances/test1_$$"
+INSTANCE2="instances/test2_$$"
+INSTANCE3="instances/test3_$$"
+
+WD=`pwd`
+
+do_reset() {
+    cd $WD
+    if [ -d $INSTANCE1 ]; then
+	echo nop > $INSTANCE1/current_tracer
+	rmdir $INSTANCE1
+    fi
+    if [ -d $INSTANCE2 ]; then
+	echo nop > $INSTANCE2/current_tracer
+	rmdir $INSTANCE2
+    fi
+    if [ -d $INSTANCE3 ]; then
+	echo nop > $INSTANCE3/current_tracer
+	rmdir $INSTANCE3
+    fi
+}
+
+mkdir $INSTANCE1
+if ! grep -q function_graph $INSTANCE1/available_tracers; then
+    echo "function_graph not allowed with instances"
+    rmdir $INSTANCE1
+    exit_unsupported
+fi
+
+mkdir $INSTANCE2
+mkdir $INSTANCE3
+
+fail() { # msg
+    do_reset
+    echo $1
+    exit_fail
+}
+
+disable_tracing
+clear_trace
+
+do_test() {
+    REGEX=$1
+    TEST=$2
+
+    # filter something, schedule is always good
+    if ! echo "$REGEX" > set_ftrace_filter; then
+	fail "can not enable filter $REGEX"
+    fi
+
+    echo > trace
+    echo function_graph > current_tracer
+    enable_tracing
+    sleep 1
+    # search for functions (has "{" or ";" on the line)
+    echo 0 > tracing_on
+    count=`cat trace | grep -v '^#' | grep -e '{' -e ';' | grep -v "$TEST" | wc -l`
+    echo 1 > tracing_on
+    if [ $count -ne 0 ]; then
+	fail "Graph filtering not working by itself against $TEST?"
+    fi
+
+    # Make sure we did find something
+    echo 0 > tracing_on
+    count=`cat trace | grep -v '^#' | grep -e '{' -e ';' | grep "$TEST" | wc -l`
+    echo 1 > tracing_on
+    if [ $count -eq 0 ]; then
+	fail "No traces found with $TEST?"
+    fi
+}
+
+do_test '*sched*' 'sched'
+cd $INSTANCE1
+do_test '*lock*' 'lock'
+cd $WD
+cd $INSTANCE2
+do_test '*rcu*' 'rcu'
+cd $WD
+cd $INSTANCE3
+echo function_graph > current_tracer
+
+sleep 1
+count=`cat trace | grep -v '^#' | grep -e '{' -e ';' | grep "$TEST" | wc -l`
+if [ $count -eq 0 ]; then
+    fail "No traces found with all tracing?"
+fi
+
+cd $WD
+echo nop > current_tracer
+echo nop > $INSTANCE1/current_tracer
+echo nop > $INSTANCE2/current_tracer
+echo nop > $INSTANCE3/current_tracer
+
+do_reset
+
+exit 0

tools/testing/selftests/ftrace/test.d/ftrace/func-filter-pid.tc

@@ -8,12 +8,18 @@
 # Also test it on an instance directory
 
 do_function_fork=1
+do_funcgraph_proc=1
 
 if [ ! -f options/function-fork ]; then
     do_function_fork=0
     echo "no option for function-fork found. Option will not be tested."
 fi
 
+if [ ! -f options/funcgraph-proc ]; then
+    do_funcgraph_proc=0
+    echo "no option for function-fork found. Option will not be tested."
+fi
+
 read PID _ < /proc/self/stat
 
 if [ $do_function_fork -eq 1 ]; then
@@ -21,12 +27,19 @@ if [ $do_function_fork -eq 1 ]; then
     orig_value=`grep function-fork trace_options`
 fi
 
+if [ $do_funcgraph_proc -eq 1 ]; then
+    orig_value2=`cat options/funcgraph-proc`
+    echo 1 > options/funcgraph-proc
+fi
+
 do_reset() {
-    if [ $do_function_fork -eq 0 ]; then
-	return
+    if [ $do_function_fork -eq 1 ]; then
+	echo $orig_value > trace_options
     fi
 
-    echo $orig_value > trace_options
+    if [ $do_funcgraph_proc -eq 1 ]; then
+	echo $orig_value2 > options/funcgraph-proc
+    fi
 }
 
 fail() { # msg
@@ -36,13 +49,15 @@ fail() { # msg
 }
 
 do_test() {
+    TRACER=$1
+
     disable_tracing
 
     echo do_execve* > set_ftrace_filter
     echo $FUNCTION_FORK >> set_ftrace_filter
 
     echo $PID > set_ftrace_pid
-    echo function > current_tracer
+    echo $TRACER > current_tracer
 
     if [ $do_function_fork -eq 1 ]; then
 	# don't allow children to be traced
@@ -82,7 +97,11 @@ do_test() {
     fi
 }
 
-do_test
+do_test function
+if grep -s function_graph available_tracers; then
+    do_test function_graph
+fi
+
 do_reset
 
 exit 0