Re: [PATCH v2 2/4] arm64: unwind: Prohibit probing on return_address()

[James Morse <james.morse@arm.com>]

Hi,

On 22/07/2019 08:48, Masami Hiramatsu wrote:
> Prohibit probing on return_address() and subroutines which
> is called from return_address(), since the it is invoked from
> trace_hardirqs_off() which is also kprobe blacklisted.

(Nits: "which are called" and "since it is")


> diff --git a/arch/arm64/kernel/return_address.c b/arch/arm64/kernel/return_address.c
> index b21cba90f82d..7f8a143268b0 100644
> --- a/arch/arm64/kernel/return_address.c
> +++ b/arch/arm64/kernel/return_address.c
> @@ -8,6 +8,7 @@
>  
>  #include <linux/export.h>
>  #include <linux/ftrace.h>
> +#include <linux/kprobes.h>
>  
>  #include <asm/stack_pointer.h>
>  #include <asm/stacktrace.h>
> @@ -17,7 +18,7 @@ struct return_address_data {
>  	void *addr;
>  };
>  
> -static int save_return_addr(struct stackframe *frame, void *d)
> +static nokprobe_inline int save_return_addr(struct stackframe *frame, void *d)

This nokprobe_inline ends up as __always_inline if kprobes is enabled.
What do we expect the compiler to do with this? save_return_addr is passed as a
function-pointer to walk_stackframe()... I don't see how the compiler can inline it!

This would be needed for on_accessible_stack().
Should we cover ftrace_graph_get_ret_stack()?, or is that already in hand?


>  {
>  	struct return_address_data *data = d;
>  
> @@ -52,3 +53,4 @@ void *return_address(unsigned int level)
>  		return NULL;
>  }
>  EXPORT_SYMBOL_GPL(return_address);
> +NOKPROBE_SYMBOL(return_address);


Thanks,

James