Commit
41918ec82eb6 ("ARM: ftrace: enable the graph tracer with the EABI
unwinder") removed the dummy version of return_address() that was
provided for the CONFIG_ARM_UNWIND=y case, on the assumption that the
removal of the kernel_text_address() call from unwind_frame() in the
preceding patch made it safe to do so.
However, this turns out not to be the case: Corentin reports warnings
about suspicious RCU usage and other strange behavior that seems to
originate in the stack unwinding that occurs in return_address().
Given that the function graph tracer (which is what these changes were
enabling for CONFIG_ARM_UNWIND=y builds) does not appear to care about
this distinction, let's revert return_address() to the old state.
Cc: Corentin Labbe <clabbe.montjoie@gmail.com>
Fixes: 41918ec82eb6 ("ARM: ftrace: enable the graph tracer with the EABI unwinder")
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reported-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Tested-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
#ifndef __ASSEMBLY__
+#if defined(CONFIG_FRAME_POINTER) && !defined(CONFIG_ARM_UNWIND)
+/*
+ * return_address uses walk_stackframe to do it's work. If both
+ * CONFIG_FRAME_POINTER=y and CONFIG_ARM_UNWIND=y walk_stackframe uses unwind
+ * information. For this to work in the function tracer many functions would
+ * have to be marked with __notrace. So for now just depend on
+ * !CONFIG_ARM_UNWIND.
+ */
+
void *return_address(unsigned int);
+#else
+
+static inline void *return_address(unsigned int level)
+{
+ return NULL;
+}
+
+#endif
+
#define ftrace_return_address(n) return_address(n)
#define ARCH_HAS_SYSCALL_MATCH_SYM_NAME
KASAN_SANITIZE_stacktrace.o := n
KASAN_SANITIZE_traps.o := n
-obj-y += return_address.o
+ifneq ($(CONFIG_ARM_UNWIND),y)
+obj-$(CONFIG_FRAME_POINTER) += return_address.o
+endif
+
obj-$(CONFIG_ATAGS) += atags_parse.o
obj-$(CONFIG_ATAGS_PROC) += atags_proc.o
obj-$(CONFIG_DEPRECATED_PARAM_STRUCT) += atags_compat.o