x86/ftrace: Fix ORC unwinding from ftrace handlers

Steven Rostedt discovered that the ftrace stack tracer is broken when
it's used with the ORC unwinder.  The problem is that objtool is
instructed by the Makefile to ignore the ftrace_64.S code, so it doesn't
generate any ORC data for it.

Fix it by making the asm code objtool-friendly:

- Objtool doesn't like the fact that save_mcount_regs pushes RBP at the
  beginning, but it's never restored (directly, at least).  So just skip
  the original RBP push, which is only needed for frame pointers anyway.

- Annotate some functions as normal callable functions with
  ENTRY/ENDPROC.

- Add an empty unwind hint to return_to_handler().  The return address
  isn't on the stack, so there's nothing ORC can do there.  It will just
  punt in the unlikely case it tries to unwind from that code.

With all that fixed, remove the OBJECT_FILES_NON_STANDARD Makefile
annotation so objtool can read the file.

Link: http://lkml.kernel.org/r/20180123040746.ih4ep3tk4pbjvg7c@trebleReported-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>

arch/x86/kernel/Makefile

@@ -29,10 +29,13 @@ KASAN_SANITIZE_stacktrace.o				:= n
 KASAN_SANITIZE_paravirt.o				:= n
 
 OBJECT_FILES_NON_STANDARD_relocate_kernel_$(BITS).o	:= y
-OBJECT_FILES_NON_STANDARD_ftrace_$(BITS).o		:= y
 OBJECT_FILES_NON_STANDARD_test_nx.o			:= y
 OBJECT_FILES_NON_STANDARD_paravirt_patch_$(BITS).o	:= y
 
+ifdef CONFIG_FRAME_POINTER
+OBJECT_FILES_NON_STANDARD_ftrace_$(BITS).o		:= y
+endif
+
 # If instrumentation of this dir is enabled, boot hangs during first second.
 # Probably could be more selective here, but note that files related to irqs,
 # boot, dumpstack/stacktrace, etc are either non-interesting or can lead to

arch/x86/kernel/ftrace_64.S

@@ -8,6 +8,7 @@
 #include <asm/ftrace.h>
 #include <asm/export.h>
 #include <asm/nospec-branch.h>
+#include <asm/unwind_hints.h>
 
 	.code64
 	.section .entry.text, "ax"
@@ -20,7 +21,6 @@ EXPORT_SYMBOL(__fentry__)
 EXPORT_SYMBOL(mcount)
 #endif
 
-/* All cases save the original rbp (8 bytes) */
 #ifdef CONFIG_FRAME_POINTER
 # ifdef CC_USING_FENTRY
 /* Save parent and function stack frames (rip and rbp) */
@@ -31,7 +31,7 @@ EXPORT_SYMBOL(mcount)
 # endif
 #else
 /* No need to save a stack frame */
-# define MCOUNT_FRAME_SIZE	8
+# define MCOUNT_FRAME_SIZE	0
 #endif /* CONFIG_FRAME_POINTER */
 
 /* Size of stack used to save mcount regs in save_mcount_regs */
@@ -64,10 +64,10 @@ EXPORT_SYMBOL(mcount)
  */
 .macro save_mcount_regs added=0
 
-	/* Always save the original rbp */
+#ifdef CONFIG_FRAME_POINTER
+	/* Save the original rbp */
 	pushq %rbp
 
-#ifdef CONFIG_FRAME_POINTER
 	/*
 	 * Stack traces will stop at the ftrace trampoline if the frame pointer
 	 * is not set up properly. If fentry is used, we need to save a frame
@@ -105,7 +105,11 @@ EXPORT_SYMBOL(mcount)
 	 * Save the original RBP. Even though the mcount ABI does not
 	 * require this, it helps out callers.
 	 */
+#ifdef CONFIG_FRAME_POINTER
 	movq MCOUNT_REG_SIZE-8(%rsp), %rdx
+#else
+	movq %rbp, %rdx
+#endif
 	movq %rdx, RBP(%rsp)
 
 	/* Copy the parent address into %rsi (second parameter) */
@@ -148,7 +152,7 @@ EXPORT_SYMBOL(mcount)
 
 ENTRY(function_hook)
 	retq
-END(function_hook)
+ENDPROC(function_hook)
 
 ENTRY(ftrace_caller)
 	/* save_mcount_regs fills in first two parameters */
@@ -184,7 +188,7 @@ GLOBAL(ftrace_graph_call)
 /* This is weak to keep gas from relaxing the jumps */
 WEAK(ftrace_stub)
 	retq
-END(ftrace_caller)
+ENDPROC(ftrace_caller)
 
 ENTRY(ftrace_regs_caller)
 	/* Save the current flags before any operations that can change them */
@@ -255,7 +259,7 @@ GLOBAL(ftrace_regs_caller_end)
 
 	jmp ftrace_epilogue
 
-END(ftrace_regs_caller)
+ENDPROC(ftrace_regs_caller)
 
 
 #else /* ! CONFIG_DYNAMIC_FTRACE */
@@ -313,9 +317,10 @@ ENTRY(ftrace_graph_caller)
 	restore_mcount_regs
 
 	retq
-END(ftrace_graph_caller)
+ENDPROC(ftrace_graph_caller)
 
-GLOBAL(return_to_handler)
+ENTRY(return_to_handler)
+	UNWIND_HINT_EMPTY
 	subq  $24, %rsp
 
 	/* Save the return values */
@@ -330,4 +335,5 @@ GLOBAL(return_to_handler)
 	movq (%rsp), %rax
 	addq $24, %rsp
 	JMP_NOSPEC %rdi
+END(return_to_handler)
 #endif