LoongArch/ftrace: Add basic support

This patch contains basic ftrace support for LoongArch. Specifically,
function tracer (HAVE_FUNCTION_TRACER), function graph tracer (HAVE_
FUNCTION_GRAPH_TRACER) are implemented following the instructions in
Documentation/trace/ftrace-design.txt.

Use `-pg` makes stub like a child function `void _mcount(void *ra)`.
Thus, it can be seen store RA and alloc stack before `call _mcount`.
Find `alloc stack` at first, and then find `store RA`.

Note that the functions in both inst.c and time.c should not be hooked
with the compiler's -pg option: to prevent infinite self-referencing for
the former, and to ignore early setup stuff for the latter.

Co-developed-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Jinyang He <hejinyang@loongson.cn>
Signed-off-by: Qing Zhang <zhangqing@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>

arch/loongarch/Kconfig

@@ -90,6 +90,8 @@ config LOONGARCH
 	select HAVE_EBPF_JIT
 	select HAVE_EXIT_THREAD
 	select HAVE_FAST_GUP
+	select HAVE_FUNCTION_GRAPH_TRACER
+	select HAVE_FUNCTION_TRACER
 	select HAVE_GENERIC_VDSO
 	select HAVE_IOREMAP_PROT
 	select HAVE_IRQ_EXIT_ON_IRQ_STACK

arch/loongarch/include/asm/ftrace.h

@@ -0,0 +1,21 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2022 Loongson Technology Corporation Limited
+ */
+
+#ifndef _ASM_LOONGARCH_FTRACE_H
+#define _ASM_LOONGARCH_FTRACE_H
+
+#ifdef CONFIG_FUNCTION_TRACER
+
+#define MCOUNT_INSN_SIZE 4		/* sizeof mcount call */
+
+#ifndef __ASSEMBLY__
+#define mcount _mcount
+extern void _mcount(void);
+extern void prepare_ftrace_return(unsigned long self_addr, unsigned long callsite_sp, unsigned long old);
+#endif /* __ASSEMBLY__ */
+
+#endif /* CONFIG_FUNCTION_TRACER */
+
+#endif /* _ASM_LOONGARCH_FTRACE_H */

arch/loongarch/kernel/Makefile

@@ -15,6 +15,14 @@ obj-$(CONFIG_EFI) 		+= efi.o
 
 obj-$(CONFIG_CPU_HAS_FPU)	+= fpu.o
 
+ifdef CONFIG_FUNCTION_TRACER
+obj-y += mcount.o ftrace.o
+CFLAGS_REMOVE_ftrace.o = $(CC_FLAGS_FTRACE)
+CFLAGS_REMOVE_inst.o = $(CC_FLAGS_FTRACE)
+CFLAGS_REMOVE_time.o = $(CC_FLAGS_FTRACE)
+CFLAGS_REMOVE_perf_event.o = $(CC_FLAGS_FTRACE)
+endif
+
 obj-$(CONFIG_MODULES)		+= module.o module-sections.o
 obj-$(CONFIG_STACKTRACE)	+= stacktrace.o
 

arch/loongarch/kernel/ftrace.c

@@ -0,0 +1,73 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2022 Loongson Technology Corporation Limited
+ */
+
+#include <linux/init.h>
+#include <linux/ftrace.h>
+#include <linux/syscalls.h>
+#include <linux/uaccess.h>
+
+#include <asm/asm.h>
+#include <asm/asm-offsets.h>
+#include <asm/cacheflush.h>
+#include <asm/inst.h>
+#include <asm/loongarch.h>
+#include <asm/syscall.h>
+
+#include <asm-generic/sections.h>
+
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+
+/*
+ * As `call _mcount` follows LoongArch psABI, ra-saved operation and
+ * stack operation can be found before this insn.
+ */
+
+static int ftrace_get_parent_ra_addr(unsigned long insn_addr, int *ra_off)
+{
+	int limit = 32;
+	union loongarch_instruction *insn;
+
+	insn = (union loongarch_instruction *)insn_addr;
+
+	do {
+		insn--;
+		limit--;
+
+		if (is_ra_save_ins(insn))
+			*ra_off = -((1 << 12) - insn->reg2i12_format.immediate);
+
+	} while (!is_stack_alloc_ins(insn) && limit);
+
+	if (!limit)
+		return -EINVAL;
+
+	return 0;
+}
+
+void prepare_ftrace_return(unsigned long self_addr,
+		unsigned long callsite_sp, unsigned long old)
+{
+	int ra_off;
+	unsigned long return_hooker = (unsigned long)&return_to_handler;
+
+	if (unlikely(ftrace_graph_is_dead()))
+		return;
+
+	if (unlikely(atomic_read(&current->tracing_graph_pause)))
+		return;
+
+	if (ftrace_get_parent_ra_addr(self_addr, &ra_off))
+		goto out;
+
+	if (!function_graph_enter(old, self_addr, 0, NULL))
+		*(unsigned long *)(callsite_sp + ra_off) = return_hooker;
+
+	return;
+
+out:
+	ftrace_graph_stop();
+	WARN_ON(1);
+}
+#endif	/* CONFIG_FUNCTION_GRAPH_TRACER */

arch/loongarch/kernel/mcount.S

@@ -0,0 +1,96 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * LoongArch specific _mcount support
+ *
+ * Copyright (C) 2022 Loongson Technology Corporation Limited
+ */
+
+#include <asm/export.h>
+#include <asm/ftrace.h>
+#include <asm/regdef.h>
+#include <asm/stackframe.h>
+
+	.text
+
+#define MCOUNT_S0_OFFSET	(0)
+#define MCOUNT_RA_OFFSET	(SZREG)
+#define MCOUNT_STACK_SIZE	(2 * SZREG)
+
+	.macro MCOUNT_SAVE_REGS
+	PTR_ADDI	sp, sp, -MCOUNT_STACK_SIZE
+	PTR_S		s0, sp, MCOUNT_S0_OFFSET
+	PTR_S		ra, sp, MCOUNT_RA_OFFSET
+	move		s0, a0
+	.endm
+
+	.macro MCOUNT_RESTORE_REGS
+	move		a0, s0
+	PTR_L		ra, sp, MCOUNT_RA_OFFSET
+	PTR_L		s0, sp, MCOUNT_S0_OFFSET
+	PTR_ADDI	sp, sp, MCOUNT_STACK_SIZE
+	.endm
+
+SYM_FUNC_START(_mcount)
+	la.pcrel	t1, ftrace_stub
+	la.pcrel	t2, ftrace_trace_function	/* Prepare t2 for (1) */
+	PTR_L		t2, t2, 0
+	beq		t1, t2, fgraph_trace
+
+	MCOUNT_SAVE_REGS
+
+	move		a0, ra				/* arg0: self return address */
+	move		a1, s0				/* arg1: parent's return address */
+	jirl		ra, t2, 0			/* (1) call *ftrace_trace_function */
+
+	MCOUNT_RESTORE_REGS
+
+fgraph_trace:
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+	la.pcrel	t1, ftrace_stub
+	la.pcrel	t3, ftrace_graph_return
+	PTR_L		t3, t3, 0
+	bne		t1, t3, ftrace_graph_caller
+	la.pcrel	t1, ftrace_graph_entry_stub
+	la.pcrel	t3, ftrace_graph_entry
+	PTR_L		t3, t3, 0
+	bne		t1, t3, ftrace_graph_caller
+#endif
+
+SYM_INNER_LABEL(ftrace_stub, SYM_L_GLOBAL)
+	jr		ra
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+SYM_INNER_LABEL(ftrace_graph_func, SYM_L_GLOBAL)
+	bl		ftrace_stub
+#endif
+SYM_FUNC_END(_mcount)
+EXPORT_SYMBOL(_mcount)
+
+#ifdef CONFIG_FUNCTION_GRAPH_TRACER
+SYM_FUNC_START(ftrace_graph_caller)
+	MCOUNT_SAVE_REGS
+
+	PTR_ADDI	a0, ra, -4			/* arg0: Callsite self return addr */
+	PTR_ADDI	a1, sp, MCOUNT_STACK_SIZE	/* arg1: Callsite sp */
+	move		a2, s0				/* arg2: Callsite parent ra */
+	bl		prepare_ftrace_return
+
+	MCOUNT_RESTORE_REGS
+	jr		ra
+SYM_FUNC_END(ftrace_graph_caller)
+
+SYM_FUNC_START(return_to_handler)
+	PTR_ADDI	sp, sp, -2 * SZREG
+	PTR_S		a0, sp, 0
+	PTR_S		a1, sp, SZREG
+
+	bl		ftrace_return_to_handler
+
+	/* Restore the real parent address: a0 -> ra */
+	move		ra, a0
+
+	PTR_L		a0, sp, 0
+	PTR_L		a1, sp, SZREG
+	PTR_ADDI	sp, sp, 2 * SZREG
+	jr		ra
+SYM_FUNC_END(return_to_handler)
+#endif /* CONFIG_FUNCTION_GRAPH_TRACER */