On Thu, Sep 07, 2017 at 09:57:36AM +0200, Vlastimil Babka wrote:
> The save_stack_trace() and save_stack_trace_tsk() wrappers of
> __save_stack_trace() add themselves to the call stack, and thus appear in the
> recorded stacktraces. This is redundant and wasteful when we have limited
> space
> to record the useful part of the backtrace with e.g. page_owner functionality.
>
> Fix this by making sure __save_stack_trace() is noinline (which matches the
> current gcc decision) and bumping the skip in the wrappers. This is similar
> to what was done for arm in 3683f44c42e9 ("ARM: stacktrace: avoid listing
> stacktrace functions in stacktrace") and is pending for arm64.
>
> Signed-off-by: Vlastimil Babka <vba...@suse.cz>
> ---
> arch/x86/kernel/stacktrace.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
>
> diff --git a/arch/x86/kernel/stacktrace.c b/arch/x86/kernel/stacktrace.c
> index 8dabd7bf1673..4b2fd6092739 100644
> --- a/arch/x86/kernel/stacktrace.c
> +++ b/arch/x86/kernel/stacktrace.c
> @@ -30,7 +30,7 @@ static int save_stack_address(struct stack_trace *trace,
> unsigned long addr,
> return 0;
> }
>
> -static void __save_stack_trace(struct stack_trace *trace,
> +static void noinline __save_stack_trace(struct stack_trace *trace,
> struct task_struct *task, struct pt_regs *regs,
> bool nosched)
> {
> @@ -56,6 +56,7 @@ static void __save_stack_trace(struct stack_trace *trace,
> */
> void save_stack_trace(struct stack_trace *trace)
> {
> + trace->skip++;
> __save_stack_trace(trace, current, NULL, false);
> }
> EXPORT_SYMBOL_GPL(save_stack_trace);
> @@ -70,6 +71,7 @@ void save_stack_trace_tsk(struct task_struct *tsk, struct
> stack_trace *trace)
> if (!try_get_task_stack(tsk))
> return;
>
> + trace->skip++;
> __save_stack_trace(trace, tsk, NULL, true);
>
> put_task_stack(tsk);
save_stack_trace_tsk() is usually called for other tasks, in which case
these functions aren't on the stack. So the skip should only be done
when task == current.
Also, I think __save_stack_trace_reliable() should be made
__always_inline (which matches its current GCC behavior) so that it
doesn't accidentally get this problem in the future.
--
Josh
