powerpc/ftrace: Add support for -mprofile-kernel ftrace ABI

The gcc switch -mprofile-kernel defines a new ABI for calling _mcount()
very early in the function with minimal overhead.

Although mprofile-kernel has been available since GCC 3.4, there were
bugs which were only fixed recently. Currently it is known to work in
GCC 4.9, 5 and 6.

Additionally there are two possible code sequences generated by the
flag, the first uses mflr/std/bl and the second is optimised to omit the
std. Currently only gcc 6 has the optimised sequence. This patch
supports both sequences.

Initial work started by Vojtech Pavlik, used with permission.

Key changes:
 - rework _mcount() to work for both the old and new ABIs.
 - implement new versions of ftrace_caller() and ftrace_graph_caller()
   which deal with the new ABI.
 - updates to __ftrace_make_nop() to recognise the new mcount calling
   sequence.
 - updates to __ftrace_make_call() to recognise the nop'ed sequence.
 - implement ftrace_modify_call().
 - updates to the module loader to surpress the toc save in the module
   stub when calling mcount with the new ABI.

Reviewed-by: Balbir Singh <bsingharora@gmail.com>
Signed-off-by: Torsten Duwe <duwe@suse.de>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>

5 files changed, 324 insertions, 20 deletions

diff --git a/arch/powerpc/include/asm/code-patching.h b/arch/powerpc/include/asm/code-patching.h
index 840a5509b3f1..994c60a857ce 100644
--- a/arch/powerpc/include/asm/code-patching.h
+++ b/arch/powerpc/include/asm/code-patching.h
@@ -99,4 +99,25 @@ static inline unsigned long ppc_global_function_entry(void *func)
 #endif
 }
 
+#ifdef CONFIG_PPC64
+/*
+ * Some instruction encodings commonly used in dynamic ftracing
+ * and function live patching.
+ */
+
+/* This must match the definition of STK_GOT in <asm/ppc_asm.h> */
+#if defined(_CALL_ELF) && _CALL_ELF == 2
+#define R2_STACK_OFFSET         24
+#else
+#define R2_STACK_OFFSET         40
+#endif
+
+#define PPC_INST_LD_TOC		(PPC_INST_LD  | ___PPC_RT(__REG_R2) | \
+				 ___PPC_RA(__REG_R1) | R2_STACK_OFFSET)
+
+/* usually preceded by a mflr r0 */
+#define PPC_INST_STD_LR		(PPC_INST_STD | ___PPC_RS(__REG_R0) | \
+				 ___PPC_RA(__REG_R1) | PPC_LR_STKOFF)
+#endif /* CONFIG_PPC64 */
+
 #endif /* _ASM_POWERPC_CODE_PATCHING_H */
diff --git a/arch/powerpc/include/asm/ftrace.h b/arch/powerpc/include/asm/ftrace.h
index ef89b1465573..50ca7585abe2 100644
--- a/arch/powerpc/include/asm/ftrace.h
+++ b/arch/powerpc/include/asm/ftrace.h
@@ -46,6 +46,8 @@
 extern void _mcount(void);
 
 #ifdef CONFIG_DYNAMIC_FTRACE
+# define FTRACE_ADDR ((unsigned long)ftrace_caller)
+# define FTRACE_REGS_ADDR FTRACE_ADDR
 static inline unsigned long ftrace_call_adjust(unsigned long addr)
 {
        /* reloction of mcount call site is the same as the address */
@@ -58,6 +60,9 @@ struct dyn_arch_ftrace {
 #endif /*  CONFIG_DYNAMIC_FTRACE */
 #endif /* __ASSEMBLY__ */
 
+#ifdef CONFIG_DYNAMIC_FTRACE_WITH_REGS
+#define ARCH_SUPPORTS_FTRACE_OPS 1
+#endif
 #endif
 
 #if defined(CONFIG_FTRACE_SYSCALLS) && defined(CONFIG_PPC64) && !defined(__ASSEMBLY__)
diff --git a/arch/powerpc/kernel/entry_64.S b/arch/powerpc/kernel/entry_64.S
index 0d525ce3717f..ec7f8aada697 100644
--- a/arch/powerpc/kernel/entry_64.S
+++ b/arch/powerpc/kernel/entry_64.S
@@ -1143,8 +1143,12 @@ _GLOBAL(enter_prom)
 #ifdef CONFIG_DYNAMIC_FTRACE
 _GLOBAL(mcount)
 _GLOBAL(_mcount)
-	blr
+	mflr	r12
+	mtctr	r12
+	mtlr	r0
+	bctr
 
+#ifndef CC_USING_MPROFILE_KERNEL
 _GLOBAL_TOC(ftrace_caller)
 	/* Taken from output of objdump from lib64/glibc */
 	mflr	r3
@@ -1166,6 +1170,115 @@ _GLOBAL(ftrace_graph_stub)
 	ld	r0, 128(r1)
 	mtlr	r0
 	addi	r1, r1, 112
+
+#else /* CC_USING_MPROFILE_KERNEL */
+/*
+ *
+ * ftrace_caller() is the function that replaces _mcount() when ftrace is
+ * active.
+ *
+ * We arrive here after a function A calls function B, and we are the trace
+ * function for B. When we enter r1 points to A's stack frame, B has not yet
+ * had a chance to allocate one yet.
+ *
+ * Additionally r2 may point either to the TOC for A, or B, depending on
+ * whether B did a TOC setup sequence before calling us.
+ *
+ * On entry the LR points back to the _mcount() call site, and r0 holds the
+ * saved LR as it was on entry to B, ie. the original return address at the
+ * call site in A.
+ *
+ * Our job is to save the register state into a struct pt_regs (on the stack)
+ * and then arrange for the ftrace function to be called.
+ */
+_GLOBAL(ftrace_caller)
+	/* Save the original return address in A's stack frame */
+	std	r0,LRSAVE(r1)
+
+	/* Create our stack frame + pt_regs */
+	stdu	r1,-SWITCH_FRAME_SIZE(r1)
+
+	/* Save all gprs to pt_regs */
+	SAVE_8GPRS(0,r1)
+	SAVE_8GPRS(8,r1)
+	SAVE_8GPRS(16,r1)
+	SAVE_8GPRS(24,r1)
+
+	/* Load special regs for save below */
+	mfmsr   r8
+	mfctr   r9
+	mfxer   r10
+	mfcr	r11
+
+	/* Get the _mcount() call site out of LR */
+	mflr	r7
+	/* Save it as pt_regs->nip & pt_regs->link */
+	std     r7, _NIP(r1)
+	std     r7, _LINK(r1)
+
+	/* Save callee's TOC in the ABI compliant location */
+	std	r2, 24(r1)
+	ld	r2,PACATOC(r13)	/* get kernel TOC in r2 */
+
+	addis	r3,r2,function_trace_op@toc@ha
+	addi	r3,r3,function_trace_op@toc@l
+	ld	r5,0(r3)
+
+	/* Calculate ip from nip-4 into r3 for call below */
+	subi    r3, r7, MCOUNT_INSN_SIZE
+
+	/* Put the original return address in r4 as parent_ip */
+	mr	r4, r0
+
+	/* Save special regs */
+	std     r8, _MSR(r1)
+	std     r9, _CTR(r1)
+	std     r10, _XER(r1)
+	std     r11, _CCR(r1)
+
+	/* Load &pt_regs in r6 for call below */
+	addi    r6, r1 ,STACK_FRAME_OVERHEAD
+
+	/* ftrace_call(r3, r4, r5, r6) */
+.globl ftrace_call
+ftrace_call:
+	bl	ftrace_stub
+	nop
+
+	/* Load ctr with the possibly modified NIP */
+	ld	r3, _NIP(r1)
+	mtctr	r3
+
+	/* Restore gprs */
+	REST_8GPRS(0,r1)
+	REST_8GPRS(8,r1)
+	REST_8GPRS(16,r1)
+	REST_8GPRS(24,r1)
+
+	/* Restore callee's TOC */
+	ld	r2, 24(r1)
+
+	/* Pop our stack frame */
+	addi r1, r1, SWITCH_FRAME_SIZE
+
+	/* Restore original LR for return to B */
+	ld	r0, LRSAVE(r1)
+	mtlr	r0
+
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+	stdu	r1, -112(r1)
+.globl ftrace_graph_call
+ftrace_graph_call:
+	b	ftrace_graph_stub
+_GLOBAL(ftrace_graph_stub)
+	addi	r1, r1, 112
+#endif
+
+	ld	r0,LRSAVE(r1)	/* restore callee's lr at _mcount site */
+	mtlr	r0
+	bctr			/* jump after _mcount site */
+#endif /* CC_USING_MPROFILE_KERNEL */
+
 _GLOBAL(ftrace_stub)
 	blr
 #else
@@ -1198,6 +1311,7 @@ _GLOBAL(ftrace_stub)
 #endif /* CONFIG_DYNAMIC_FTRACE */
 
 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
+#ifndef CC_USING_MPROFILE_KERNEL
 _GLOBAL(ftrace_graph_caller)
 	/* load r4 with local address */
 	ld	r4, 128(r1)
@@ -1222,6 +1336,56 @@ _GLOBAL(ftrace_graph_caller)
 	addi	r1, r1, 112
 	blr
 
+#else /* CC_USING_MPROFILE_KERNEL */
+_GLOBAL(ftrace_graph_caller)
+	/* with -mprofile-kernel, parameter regs are still alive at _mcount */
+	std	r10, 104(r1)
+	std	r9, 96(r1)
+	std	r8, 88(r1)
+	std	r7, 80(r1)
+	std	r6, 72(r1)
+	std	r5, 64(r1)
+	std	r4, 56(r1)
+	std	r3, 48(r1)
+
+	/* Save callee's TOC in the ABI compliant location */
+	std	r2, 24(r1)
+	ld	r2, PACATOC(r13)	/* get kernel TOC in r2 */
+
+	mfctr	r4		/* ftrace_caller has moved local addr here */
+	std	r4, 40(r1)
+	mflr	r3		/* ftrace_caller has restored LR from stack */
+	subi	r4, r4, MCOUNT_INSN_SIZE
+
+	bl	prepare_ftrace_return
+	nop
+
+	/*
+	 * prepare_ftrace_return gives us the address we divert to.
+	 * Change the LR to this.
+	 */
+	mtlr	r3
+
+	ld	r0, 40(r1)
+	mtctr	r0
+	ld	r10, 104(r1)
+	ld	r9, 96(r1)
+	ld	r8, 88(r1)
+	ld	r7, 80(r1)
+	ld	r6, 72(r1)
+	ld	r5, 64(r1)
+	ld	r4, 56(r1)
+	ld	r3, 48(r1)
+
+	/* Restore callee's TOC */
+	ld	r2, 24(r1)
+
+	addi	r1, r1, 112
+	mflr	r0
+	std	r0, LRSAVE(r1)
+	bctr
+#endif /* CC_USING_MPROFILE_KERNEL */
+
 _GLOBAL(return_to_handler)
 	/* need to save return values */
 	std	r4,  -32(r1)
diff --git a/arch/powerpc/kernel/ftrace.c b/arch/powerpc/kernel/ftrace.c
index 62899fbae703..9dac18dabd03 100644
--- a/arch/powerpc/kernel/ftrace.c
+++ b/arch/powerpc/kernel/ftrace.c
@@ -61,8 +61,11 @@ ftrace_modify_code(unsigned long ip, unsigned int old, unsigned int new)
 		return -EFAULT;
 
 	/* Make sure it is what we expect it to be */
-	if (replaced != old)
+	if (replaced != old) {
+		pr_err("%p: replaced (%#x) != old (%#x)",
+		(void *)ip, replaced, old);
 		return -EINVAL;
+	}
 
 	/* replace the text with the new text */
 	if (patch_instruction((unsigned int *)ip, new))
@@ -108,11 +111,13 @@ __ftrace_make_nop(struct module *mod,
 {
 	unsigned long entry, ptr, tramp;
 	unsigned long ip = rec->ip;
-	unsigned int op;
+	unsigned int op, pop;
 
 	/* read where this goes */
-	if (probe_kernel_read(&op, (void *)ip, sizeof(int)))
+	if (probe_kernel_read(&op, (void *)ip, sizeof(int))) {
+		pr_err("Fetching opcode failed.\n");
 		return -EFAULT;
+	}
 
 	/* Make sure that that this is still a 24bit jump */
 	if (!is_bl_op(op)) {
@@ -152,10 +157,42 @@ __ftrace_make_nop(struct module *mod,
 	 *
 	 * Use a b +8 to jump over the load.
 	 */
-	op = 0x48000008;	/* b +8 */
 
-	if (patch_instruction((unsigned int *)ip, op))
+	pop = PPC_INST_BRANCH | 8;	/* b +8 */
+
+	/*
+	 * Check what is in the next instruction. We can see ld r2,40(r1), but
+	 * on first pass after boot we will see mflr r0.
+	 */
+	if (probe_kernel_read(&op, (void *)(ip+4), MCOUNT_INSN_SIZE)) {
+		pr_err("Fetching op failed.\n");
+		return -EFAULT;
+	}
+
+	if (op != PPC_INST_LD_TOC) {
+		unsigned int inst;
+
+		if (probe_kernel_read(&inst, (void *)(ip - 4), 4)) {
+			pr_err("Fetching instruction at %lx failed.\n", ip - 4);
+			return -EFAULT;
+		}
+
+		/* We expect either a mlfr r0, or a std r0, LRSAVE(r1) */
+		if (inst != PPC_INST_MFLR && inst != PPC_INST_STD_LR) {
+			pr_err("Unexpected instructions around bl _mcount\n"
+			       "when enabling dynamic ftrace!\t"
+			       "(%08x,bl,%08x)\n", inst, op);
+			return -EINVAL;
+		}
+
+		/* When using -mkernel_profile there is no load to jump over */
+		pop = PPC_INST_NOP;
+	}
+
+	if (patch_instruction((unsigned int *)ip, pop)) {
+		pr_err("Patching NOP failed.\n");
 		return -EPERM;
+	}
 
 	return 0;
 }
@@ -281,16 +318,15 @@ int ftrace_make_nop(struct module *mod,
 
 #ifdef CONFIG_MODULES
 #ifdef CONFIG_PPC64
+/*
+ * Examine the existing instructions for __ftrace_make_call.
+ * They should effectively be a NOP, and follow formal constraints,
+ * depending on the ABI. Return false if they don't.
+ */
+#ifndef CC_USING_MPROFILE_KERNEL
 static int
-__ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
+expected_nop_sequence(void *ip, unsigned int op0, unsigned int op1)
 {
-	unsigned int op[2];
-	void *ip = (void *)rec->ip;
-
-	/* read where this goes */
-	if (probe_kernel_read(op, ip, sizeof(op)))
-		return -EFAULT;
-
 	/*
 	 * We expect to see:
 	 *
@@ -300,8 +336,34 @@ __ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
 	 * The load offset is different depending on the ABI. For simplicity
 	 * just mask it out when doing the compare.
 	 */
-	if ((op[0] != 0x48000008) || ((op[1] & 0xffff0000) != 0xe8410000)) {
-		pr_err("Unexpected call sequence: %x %x\n", op[0], op[1]);
+	if ((op0 != 0x48000008) || ((op1 & 0xffff0000) != 0xe8410000))
+		return 0;
+	return 1;
+}
+#else
+static int
+expected_nop_sequence(void *ip, unsigned int op0, unsigned int op1)
+{
+	/* look for patched "NOP" on ppc64 with -mprofile-kernel */
+	if (op0 != PPC_INST_NOP)
+		return 0;
+	return 1;
+}
+#endif
+
+static int
+__ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
+{
+	unsigned int op[2];
+	void *ip = (void *)rec->ip;
+
+	/* read where this goes */
+	if (probe_kernel_read(op, ip, sizeof(op)))
+		return -EFAULT;
+
+	if (!expected_nop_sequence(ip, op[0], op[1])) {
+		pr_err("Unexpected call sequence at %p: %x %x\n",
+		ip, op[0], op[1]);
 		return -EINVAL;
 	}
 
@@ -324,7 +386,16 @@ __ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
 
 	return 0;
 }
-#else
+
+#ifdef CONFIG_DYNAMIC_FTRACE_WITH_REGS
+int ftrace_modify_call(struct dyn_ftrace *rec, unsigned long old_addr,
+			unsigned long addr)
+{
+	return ftrace_make_call(rec, addr);
+}
+#endif
+
+#else  /* !CONFIG_PPC64: */
 static int
 __ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
 {
diff --git a/arch/powerpc/kernel/module_64.c b/arch/powerpc/kernel/module_64.c
index 76c0963572f5..848b47499a27 100644
--- a/arch/powerpc/kernel/module_64.c
+++ b/arch/powerpc/kernel/module_64.c
@@ -42,7 +42,6 @@
    --RR.  */
 
 #if defined(_CALL_ELF) && _CALL_ELF == 2
-#define R2_STACK_OFFSET 24
 
 /* An address is simply the address of the function. */
 typedef unsigned long func_desc_t;
@@ -74,7 +73,6 @@ static unsigned int local_entry_offset(const Elf64_Sym *sym)
 	return PPC64_LOCAL_ENTRY_OFFSET(sym->st_other);
 }
 #else
-#define R2_STACK_OFFSET 40
 
 /* An address is address of the OPD entry, which contains address of fn. */
 typedef struct ppc64_opd_entry func_desc_t;
@@ -451,17 +449,60 @@ static unsigned long stub_for_addr(const Elf64_Shdr *sechdrs,
 	return (unsigned long)&stubs[i];
 }
 
+#ifdef CC_USING_MPROFILE_KERNEL
+static bool is_early_mcount_callsite(u32 *instruction)
+{
+	/*
+	 * Check if this is one of the -mprofile-kernel sequences.
+	 */
+	if (instruction[-1] == PPC_INST_STD_LR &&
+	    instruction[-2] == PPC_INST_MFLR)
+		return true;
+
+	if (instruction[-1] == PPC_INST_MFLR)
+		return true;
+
+	return false;
+}
+
+/*
+ * In case of _mcount calls, do not save the current callee's TOC (in r2) into
+ * the original caller's stack frame. If we did we would clobber the saved TOC
+ * value of the original caller.
+ */
+static void squash_toc_save_inst(const char *name, unsigned long addr)
+{
+	struct ppc64_stub_entry *stub = (struct ppc64_stub_entry *)addr;
+
+	/* Only for calls to _mcount */
+	if (strcmp("_mcount", name) != 0)
+		return;
+
+	stub->jump[2] = PPC_INST_NOP;
+}
+#else
+static void squash_toc_save_inst(const char *name, unsigned long addr) { }
+
+/* without -mprofile-kernel, mcount calls are never early */
+static bool is_early_mcount_callsite(u32 *instruction)
+{
+	return false;
+}
+#endif
+
 /* We expect a noop next: if it is, replace it with instruction to
    restore r2. */
 static int restore_r2(u32 *instruction, struct module *me)
 {
 	if (*instruction != PPC_INST_NOP) {
+		if (is_early_mcount_callsite(instruction - 1))
+			return 1;
 		pr_err("%s: Expect noop after relocate, got %08x\n",
 		       me->name, *instruction);
 		return 0;
 	}
 	/* ld r2,R2_STACK_OFFSET(r1) */
-	*instruction = 0xe8410000 | R2_STACK_OFFSET;
+	*instruction = PPC_INST_LD_TOC;
 	return 1;
 }
 
@@ -586,6 +627,8 @@ int apply_relocate_add(Elf64_Shdr *sechdrs,
 					return -ENOENT;
 				if (!restore_r2((u32 *)location + 1, me))
 					return -ENOEXEC;
+
+				squash_toc_save_inst(strtab + sym->st_name, value);
 			} else
 				value += local_entry_offset(sym);