Merge tag 'trace-v5.5' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace

Pull tracing updates from Steven Rostedt:
 "New tracing features:

   - New PERMANENT flag to ftrace_ops when attaching a callback to a
     function.

     As /proc/sys/kernel/ftrace_enabled when set to zero will disable
     all attached callbacks in ftrace, this has a detrimental impact on
     live kernel tracing, as it disables all that it patched. If a
     ftrace_ops is registered to ftrace with the PERMANENT flag set, it
     will prevent ftrace_enabled from being disabled, and if
     ftrace_enabled is already disabled, it will prevent a ftrace_ops
     with PREMANENT flag set from being registered.

   - New register_ftrace_direct().

     As eBPF would like to register its own trampolines to be called by
     the ftrace nop locations directly, without going through the ftrace
     trampoline, this function has been added. This allows for eBPF
     trampolines to live along side of ftrace, perf, kprobe and live
     patching. It also utilizes the ftrace enabled_functions file that
     keeps track of functions that have been modified in the kernel, to
     allow for security auditing.

   - Allow for kernel internal use of ftrace instances.

     Subsystems in the kernel can now create and destroy their own
     tracing instances which allows them to have their own tracing
     buffer, and be able to record events without worrying about other
     users from writing over their data.

   - New seq_buf_hex_dump() that lets users use the hex_dump() in their
     seq_buf usage.

   - Notifications now added to tracing_max_latency to allow user space
     to know when a new max latency is hit by one of the latency
     tracers.

   - Wider spread use of generic compare operations for use of bsearch
     and friends.

   - More synthetic event fields may be defined (32 up from 16)

   - Use of xarray for architectures with sparse system calls, for the
     system call trace events.

  This along with small clean ups and fixes"

* tag 'trace-v5.5' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace: (51 commits)
  tracing: Enable syscall optimization for MIPS
  tracing: Use xarray for syscall trace events
  tracing: Sample module to demonstrate kernel access to Ftrace instances.
  tracing: Adding new functions for kernel access to Ftrace instances
  tracing: Fix Kconfig indentation
  ring-buffer: Fix typos in function ring_buffer_producer
  ftrace: Use BIT() macro
  ftrace: Return ENOTSUPP when DYNAMIC_FTRACE_WITH_DIRECT_CALLS is not configured
  ftrace: Rename ftrace_graph_stub to ftrace_stub_graph
  ftrace: Add a helper function to modify_ftrace_direct() to allow arch optimization
  ftrace: Add helper find_direct_entry() to consolidate code
  ftrace: Add another check for match in register_ftrace_direct()
  ftrace: Fix accounting bug with direct->count in register_ftrace_direct()
  ftrace/selftests: Fix spelling mistake "wakeing" -> "waking"
  tracing: Increase SYNTH_FIELDS_MAX for synthetic_events
  ftrace/samples: Add a sample module that implements modify_ftrace_direct()
  ftrace: Add modify_ftrace_direct()
  tracing: Add missing "inline" in stub function of latency_fsnotify()
  tracing: Remove stray tab in TRACE_EVAL_MAP_FILE's help text
  tracing: Use seq_buf_hex_dump() to dump buffers
  ...

Author: torvalds

Documentation/trace/ftrace-uses.rst

@@ -146,7 +146,7 @@ FTRACE_OPS_FL_RECURSION_SAFE
 	itself or any nested functions that those functions call.
 
 	If this flag is set, it is possible that the callback will also
-	be called with preemption enabled (when CONFIG_PREEMPT is set),
+	be called with preemption enabled (when CONFIG_PREEMPTION is set),
 	but this is not guaranteed.
 
 FTRACE_OPS_FL_IPMODIFY
@@ -170,6 +170,14 @@ FTRACE_OPS_FL_RCU
 	a callback may be executed and RCU synchronization will not protect
 	it.
 
+FTRACE_OPS_FL_PERMANENT
+        If this is set on any ftrace ops, then the tracing cannot disabled by
+        writing 0 to the proc sysctl ftrace_enabled. Equally, a callback with
+        the flag set cannot be registered if ftrace_enabled is 0.
+
+        Livepatch uses it not to lose the function redirection, so the system
+        stays protected.
+
 
 Filtering which functions to trace
 ==================================

Documentation/trace/ftrace.rst

@@ -2976,7 +2976,9 @@ Note, the proc sysctl ftrace_enable is a big on/off switch for the
 function tracer. By default it is enabled (when function tracing is
 enabled in the kernel). If it is disabled, all function tracing is
 disabled. This includes not only the function tracers for ftrace, but
-also for any other uses (perf, kprobes, stack tracing, profiling, etc).
+also for any other uses (perf, kprobes, stack tracing, profiling, etc). It
+cannot be disabled if there is a callback with FTRACE_OPS_FL_PERMANENT set
+registered.
 
 Please disable this with care.
 

arch/Kconfig

@@ -939,6 +939,14 @@ config RELR
 config ARCH_HAS_MEM_ENCRYPT
 	bool
 
+config HAVE_SPARSE_SYSCALL_NR
+       bool
+       help
+          An architecture should select this if its syscall numbering is sparse
+	  to save space. For example, MIPS architecture has a syscall array with
+	  entries at 4000, 5000 and 6000 locations. This option turns on syscall
+	  related optimizations for a given architecture.
+
 source "kernel/gcov/Kconfig"
 
 source "scripts/gcc-plugins/Kconfig"

arch/mips/Kconfig

@@ -74,6 +74,7 @@ config MIPS
 	select HAVE_PERF_EVENTS
 	select HAVE_REGS_AND_STACK_ACCESS_API
 	select HAVE_RSEQ
+	select HAVE_SPARSE_SYSCALL_NR
 	select HAVE_STACKPROTECTOR
 	select HAVE_SYSCALL_TRACEPOINTS
 	select HAVE_VIRT_CPU_ACCOUNTING_GEN if 64BIT || !SMP

arch/sh/boot/compressed/misc.c

@@ -111,6 +111,11 @@ void __stack_chk_fail(void)
 	error("stack-protector: Kernel stack is corrupted\n");
 }
 
+/* Needed because vmlinux.lds.h references this */
+void ftrace_stub(void)
+{
+}
+
 #ifdef CONFIG_SUPERH64
 #define stackalign	8
 #else

arch/x86/Kconfig

@@ -157,6 +157,7 @@ config X86
 	select HAVE_DMA_CONTIGUOUS
 	select HAVE_DYNAMIC_FTRACE
 	select HAVE_DYNAMIC_FTRACE_WITH_REGS
+	select HAVE_DYNAMIC_FTRACE_WITH_DIRECT_CALLS
 	select HAVE_EBPF_JIT
 	select HAVE_EFFICIENT_UNALIGNED_ACCESS
 	select HAVE_EISA

arch/x86/include/asm/ftrace.h

@@ -28,6 +28,19 @@ static inline unsigned long ftrace_call_adjust(unsigned long addr)
 	return addr;
 }
 
+/*
+ * When a ftrace registered caller is tracing a function that is
+ * also set by a register_ftrace_direct() call, it needs to be
+ * differentiated in the ftrace_caller trampoline. To do this, we
+ * place the direct caller in the ORIG_AX part of pt_regs. This
+ * tells the ftrace_caller that there's a direct caller.
+ */
+static inline void arch_ftrace_set_direct_caller(struct pt_regs *regs, unsigned long addr)
+{
+	/* Emulate a call */
+	regs->orig_ax = addr;
+}
+
 #ifdef CONFIG_DYNAMIC_FTRACE
 
 struct dyn_arch_ftrace {

arch/x86/include/asm/unwind_hints.h

@@ -86,6 +86,14 @@
 	UNWIND_HINT sp_offset=\sp_offset
 .endm
 
+.macro UNWIND_HINT_SAVE
+	UNWIND_HINT type=UNWIND_HINT_TYPE_SAVE
+.endm
+
+.macro UNWIND_HINT_RESTORE
+	UNWIND_HINT type=UNWIND_HINT_TYPE_RESTORE
+.endm
+
 #else /* !__ASSEMBLY__ */
 
 #define UNWIND_HINT(sp_reg, sp_offset, type, end)		\

arch/x86/kernel/ftrace.c

@@ -1042,6 +1042,20 @@ void prepare_ftrace_return(unsigned long self_addr, unsigned long *parent,
 	if (unlikely(atomic_read(&current->tracing_graph_pause)))
 		return;
 
+	/*
+	 * If the return location is actually pointing directly to
+	 * the start of a direct trampoline (if we trace the trampoline
+	 * it will still be offset by MCOUNT_INSN_SIZE), then the
+	 * return address is actually off by one word, and we
+	 * need to adjust for that.
+	 */
+	if (ftrace_direct_func_count) {
+		if (ftrace_find_direct_func(self_addr + MCOUNT_INSN_SIZE)) {
+			self_addr = *parent;
+			parent++;
+		}
+	}
+
 	/*
 	 * Protect against fault, even if it shouldn't
 	 * happen. This tool is too much intrusive to

arch/x86/kernel/ftrace_64.S

@@ -85,6 +85,7 @@
 	movq %rdi, RDI(%rsp)
 	movq %r8, R8(%rsp)
 	movq %r9, R9(%rsp)
+	movq $0, ORIG_RAX(%rsp)
 	/*
 	 * Save the original RBP. Even though the mcount ABI does not
 	 * require this, it helps out callers.
@@ -111,7 +112,11 @@
 	subq $MCOUNT_INSN_SIZE, %rdi
 	.endm
 
-.macro restore_mcount_regs
+.macro restore_mcount_regs save=0
+
+	/* ftrace_regs_caller or frame pointers require this */
+	movq RBP(%rsp), %rbp
+
 	movq R9(%rsp), %r9
 	movq R8(%rsp), %r8
 	movq RDI(%rsp), %rdi
@@ -120,10 +125,7 @@
 	movq RCX(%rsp), %rcx
 	movq RAX(%rsp), %rax
 
-	/* ftrace_regs_caller can modify %rbp */
-	movq RBP(%rsp), %rbp
-
-	addq $MCOUNT_REG_SIZE, %rsp
+	addq $MCOUNT_REG_SIZE-\save, %rsp
 
 	.endm
 
@@ -174,6 +176,8 @@ SYM_FUNC_START(ftrace_regs_caller)
 	/* Save the current flags before any operations that can change them */
 	pushfq
 
+	UNWIND_HINT_SAVE
+
 	/* added 8 bytes to save flags */
 	save_mcount_regs 8
 	/* save_mcount_regs fills in first two parameters */
@@ -226,7 +230,33 @@ SYM_INNER_LABEL(ftrace_regs_call, SYM_L_GLOBAL)
 	movq R10(%rsp), %r10
 	movq RBX(%rsp), %rbx
 
-	restore_mcount_regs
+	movq ORIG_RAX(%rsp), %rax
+	movq %rax, MCOUNT_REG_SIZE-8(%rsp)
+
+	/* If ORIG_RAX is anything but zero, make this a call to that */
+	movq ORIG_RAX(%rsp), %rax
+	cmpq	$0, %rax
+	je	1f
+
+	/* Swap the flags with orig_rax */
+	movq MCOUNT_REG_SIZE(%rsp), %rdi
+	movq %rdi, MCOUNT_REG_SIZE-8(%rsp)
+	movq %rax, MCOUNT_REG_SIZE(%rsp)
+
+	restore_mcount_regs 8
+
+	jmp	2f
+
+1:	restore_mcount_regs
+
+
+2:
+	/*
+	 * The stack layout is nondetermistic here, depending on which path was
+	 * taken.  This confuses objtool and ORC, rightfully so.  For now,
+	 * pretend the stack always looks like the non-direct case.
+	 */
+	UNWIND_HINT_RESTORE
 
 	/* Restore flags */
 	popfq