On BE32 kernels, the __opcode_to_mem_thumb32() interface is intentionally
not defined, but it is referenced whenever runtime patching is enabled
for the kernel, which may be for ftrace, jump label, kprobes or kgdb:
arch/arm/kernel/patch.c: In function '__patch_text_real':
arch/arm/kernel/patch.c:94:32: error: implicit declaration of function '__opcode_to_mem_thumb32' [-Werror=implicit-function-declaration]
94 | insn = __opcode_to_mem_thumb32(insn);
| ^~~~~~~~~~~~~~~~~~~~~~~
Since BE32 kernels never run Thumb2 code, we never end up using the
result of this call, so providing an extern declaration without
a definition makes it build correctly.
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
#define __opcode_to_mem_thumb16(x) ___opcode_identity16(x)
#define ___asm_opcode_to_mem_arm(x) ___asm_opcode_identity32(x)
#define ___asm_opcode_to_mem_thumb16(x) ___asm_opcode_identity16(x)
-#ifndef CONFIG_CPU_ENDIAN_BE32
+#ifdef CONFIG_CPU_ENDIAN_BE32
+#ifndef __ASSEMBLY__
/*
* On BE32 systems, using 32-bit accesses to store Thumb instructions will not
* work in all cases, due to alignment constraints. For now, a correct
- * version is not provided for BE32.
+ * version is not provided for BE32, but the prototype needs to be there
+ * to compile patch.c.
*/
+extern __u32 __opcode_to_mem_thumb32(__u32);
+#endif
+#else
#define __opcode_to_mem_thumb32(x) ___opcode_swahw32(x)
#define ___asm_opcode_to_mem_thumb32(x) ___asm_opcode_swahw32(x)
#endif