[PATCH v2] arm64: ftrace: emit ftrace-mod.o contents through code

[Ard Biesheuvel]

When building the arm64 kernel with both CONFIG_ARM64_MODULE_PLTS and
CONFIG_DYNAMIC_FTRACE enabled, the ftrace-mod.o object file is built
with the kernel and contains a trampoline that is linked into each
module, so that modules can be loaded far away from the kernel and
still reach the ftrace entry point in the core kernel with an ordinary
relative branch, as is emitted by the compiler instrumentation code
dynamic ftrace relies on.

In order to be able to build out of tree modules, this object file
needs to be included into the linux-headers or linux-devel packages,
which is undesirable, as it makes arm64 a special case (although a
precedent does exist for 32-bit PPC).

Given that the trampoline only consists of two instructions, let's
not bother with a source or object file for it, and simply patch in
these instructions whenever the trampoline is being populated.

Signed-off-by: Ard Biesheuvel <ard.biesheuvel at linaro.org>
---
v2: populate the two trampoline instruction slots from the ftrace code so
    there is no longer a need to have a template .o or .S in the first place
    don't use a separate linker scripts for ftrace+PLTs

 arch/arm64/Makefile                   |  3 ---
 arch/arm64/kernel/Makefile            |  3 ---
 arch/arm64/kernel/ftrace-mod.S        | 18 ------------------
 arch/arm64/kernel/ftrace-trampoline.h | 18 ++++++++++++++++++
 arch/arm64/kernel/ftrace.c            | 16 ++++++++++------
 arch/arm64/kernel/module-plts.c       | 14 ++++++++++++++
 arch/arm64/kernel/module.lds          |  1 +
 7 files changed, 43 insertions(+), 30 deletions(-)

diff --git a/arch/arm64/Makefile b/arch/arm64/Makefile
index b35788c909f1..b481b4a7c011 100644
--- a/arch/arm64/Makefile
+++ b/arch/arm64/Makefile
@@ -83,9 +83,6 @@ endif
 
 ifeq ($(CONFIG_ARM64_MODULE_PLTS),y)
 KBUILD_LDFLAGS_MODULE	+= -T $(srctree)/arch/arm64/kernel/module.lds
-ifeq ($(CONFIG_DYNAMIC_FTRACE),y)
-KBUILD_LDFLAGS_MODULE	+= $(objtree)/arch/arm64/kernel/ftrace-mod.o
-endif
 endif
 
 # Default value
diff --git a/arch/arm64/kernel/Makefile b/arch/arm64/kernel/Makefile
index 8265dd790895..067baace74a0 100644
--- a/arch/arm64/kernel/Makefile
+++ b/arch/arm64/kernel/Makefile
@@ -61,6 +61,3 @@ extra-y					+= $(head-y) vmlinux.lds
 ifeq ($(CONFIG_DEBUG_EFI),y)
 AFLAGS_head.o += -DVMLINUX_PATH="\"$(realpath $(objtree)/vmlinux)\""
 endif
-
-# will be included by each individual module but not by the core kernel itself
-extra-$(CONFIG_DYNAMIC_FTRACE) += ftrace-mod.o
diff --git a/arch/arm64/kernel/ftrace-mod.S b/arch/arm64/kernel/ftrace-mod.S
deleted file mode 100644
index 00c4025be4ff..000000000000
--- a/arch/arm64/kernel/ftrace-mod.S
+++ /dev/null
@@ -1,18 +0,0 @@
-/*
- * Copyright (C) 2017 Linaro Ltd <ard.biesheuvel at linaro.org>
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
- */
-
-#include <linux/linkage.h>
-#include <asm/assembler.h>
-
-	.section	".text.ftrace_trampoline", "ax"
-	.align		3
-0:	.quad		0
-__ftrace_trampoline:
-	ldr		x16, 0b
-	br		x16
-ENDPROC(__ftrace_trampoline)
diff --git a/arch/arm64/kernel/ftrace-trampoline.h b/arch/arm64/kernel/ftrace-trampoline.h
new file mode 100644
index 000000000000..caf8434ca52a
--- /dev/null
+++ b/arch/arm64/kernel/ftrace-trampoline.h
@@ -0,0 +1,18 @@
+/*
+ * Copyright (C) 2017 Linaro Ltd. <ard.biesheuvel at linaro.org>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/types.h>
+
+#define FTRACE_TRAMPOLINE_LDR	cpu_to_le32(0x58ffffd0)
+#define FTRACE_TRAMPOLINE_BR	cpu_to_le32(0xd61f0200)
+
+struct ftrace_trampoline {
+	u64	literal;	/* 0:	.quad	0x0		*/
+	__le32	ldr;		/*	ldr	x16, 0b		*/
+	__le32	br;		/*	br	x16		*/
+};
diff --git a/arch/arm64/kernel/ftrace.c b/arch/arm64/kernel/ftrace.c
index c13b1fca0e5b..2198b5a48871 100644
--- a/arch/arm64/kernel/ftrace.c
+++ b/arch/arm64/kernel/ftrace.c
@@ -19,6 +19,8 @@
 #include <asm/ftrace.h>
 #include <asm/insn.h>
 
+#include "ftrace-trampoline.h"
+
 #ifdef CONFIG_DYNAMIC_FTRACE
 /*
  * Replace a single instruction, which may be a branch or NOP.
@@ -76,7 +78,7 @@ int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
 
 	if (offset < -SZ_128M || offset >= SZ_128M) {
 #ifdef CONFIG_ARM64_MODULE_PLTS
-		unsigned long *trampoline;
+		struct ftrace_trampoline *trampoline;
 		struct module *mod;
 
 		/*
@@ -104,22 +106,24 @@ int ftrace_make_call(struct dyn_ftrace *rec, unsigned long addr)
 		 * is added in the future, but for now, the pr_err() below
 		 * deals with a theoretical issue only.
 		 */
-		trampoline = (unsigned long *)mod->arch.ftrace_trampoline;
-		if (trampoline[0] != addr) {
-			if (trampoline[0] != 0) {
+		trampoline = (void *)mod->arch.ftrace_trampoline;
+		if (trampoline->literal != addr) {
+			if (trampoline->literal != 0) {
 				pr_err("ftrace: far branches to multiple entry points unsupported inside a single module\n");
 				return -EINVAL;
 			}
 
 			/* point the trampoline to our ftrace entry point */
 			module_disable_ro(mod);
-			trampoline[0] = addr;
+			trampoline->literal = addr;
+			trampoline->ldr = FTRACE_TRAMPOLINE_LDR;
+			trampoline->br = FTRACE_TRAMPOLINE_BR;
 			module_enable_ro(mod, true);
 
 			/* update trampoline before patching in the branch */
 			smp_wmb();
 		}
-		addr = (unsigned long)&trampoline[1];
+		addr = (unsigned long)&trampoline->ldr;
 #else /* CONFIG_ARM64_MODULE_PLTS */
 		return -EINVAL;
 #endif /* CONFIG_ARM64_MODULE_PLTS */
diff --git a/arch/arm64/kernel/module-plts.c b/arch/arm64/kernel/module-plts.c
index d05dbe658409..c8fe9c98abc4 100644
--- a/arch/arm64/kernel/module-plts.c
+++ b/arch/arm64/kernel/module-plts.c
@@ -11,6 +11,8 @@
 #include <linux/module.h>
 #include <linux/sort.h>
 
+#include "ftrace-trampoline.h"
+
 struct plt_entry {
 	/*
 	 * A program that conforms to the AArch64 Procedure Call Standard
@@ -154,6 +156,7 @@ int module_frob_arch_sections(Elf_Ehdr *ehdr, Elf_Shdr *sechdrs,
 	unsigned long core_plts = 0;
 	unsigned long init_plts = 0;
 	Elf64_Sym *syms = NULL;
+	Elf_Shdr *tramp = NULL;
 	int i;
 
 	/*
@@ -165,6 +168,10 @@ int module_frob_arch_sections(Elf_Ehdr *ehdr, Elf_Shdr *sechdrs,
 			mod->arch.core.plt = sechdrs + i;
 		else if (!strcmp(secstrings + sechdrs[i].sh_name, ".init.plt"))
 			mod->arch.init.plt = sechdrs + i;
+		else if (IS_ENABLED(CONFIG_DYNAMIC_FTRACE) &&
+			 !strcmp(secstrings + sechdrs[i].sh_name,
+				 ".text.ftrace_trampoline"))
+			tramp = sechdrs + i;
 		else if (sechdrs[i].sh_type == SHT_SYMTAB)
 			syms = (Elf64_Sym *)sechdrs[i].sh_addr;
 	}
@@ -215,5 +222,12 @@ int module_frob_arch_sections(Elf_Ehdr *ehdr, Elf_Shdr *sechdrs,
 	mod->arch.init.plt_num_entries = 0;
 	mod->arch.init.plt_max_entries = init_plts;
 
+	if (tramp) {
+		tramp->sh_type = SHT_NOBITS;
+		tramp->sh_flags = SHF_EXECINSTR | SHF_ALLOC;
+		tramp->sh_addralign = __alignof__(struct ftrace_trampoline);
+		tramp->sh_size = sizeof(struct ftrace_trampoline);
+	}
+
 	return 0;
 }
diff --git a/arch/arm64/kernel/module.lds b/arch/arm64/kernel/module.lds
index f7c9781a9d48..22e36a21c113 100644
--- a/arch/arm64/kernel/module.lds
+++ b/arch/arm64/kernel/module.lds
@@ -1,4 +1,5 @@
 SECTIONS {
 	.plt (NOLOAD) : { BYTE(0) }
 	.init.plt (NOLOAD) : { BYTE(0) }
+	.text.ftrace_trampoline (NOLOAD) : { BYTE(0) }
 }
-- 
2.11.0