Re: [PATCH 7/7] arch: Fix STACKTRACE_SUPPORT

From: Mark Rutland <mark.rutland@arm.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: keescook@chromium.org, jannh@google.com,
	linux-kernel@vger.kernel.org, vcaputo@pengaru.com,
	mingo@redhat.com, juri.lelli@redhat.com,
	vincent.guittot@linaro.org, dietmar.eggemann@arm.com,
	rostedt@goodmis.org, bsegall@google.com, mgorman@suse.de,
	bristot@redhat.com, akpm@linux-foundation.org,
	christian.brauner@ubuntu.com, amistry@google.com,
	Kenta.Tada@sony.com, legion@kernel.org,
	michael.weiss@aisec.fraunhofer.de, mhocko@suse.com,
	deller@gmx.de, zhengqi.arch@bytedance.com, me@tobin.cc,
	tycho@tycho.pizza, tglx@linutronix.de, bp@alien8.de,
	hpa@zytor.com, axboe@kernel.dk, metze@samba.org,
	laijs@linux.alibaba.com, luto@kernel.org,
	dave.hansen@linux.intel.com, ebiederm@xmission.com,
	ohoono.kwon@samsung.com, kaleshsingh@google.com,
	yifeifz2@illinois.edu, jpoimboe@redhat.com,
	linux-hardening@vger.kernel.org, linux-arch@vger.kernel.org,
	vgupta@kernel.org, linux@armlinux.org.uk, will@kernel.org,
	guoren@kernel.org, bcain@codeaurora.org, monstr@monstr.eu,
	tsbogend@alpha.franken.de, nickhu@andestech.com,
	jonas@southpole.se, mpe@ellerman.id.au, paul.walmsley@sifive.com,
	hca@linux.ibm.com, ysato@users.sourceforge.jp,
	davem@davemloft.net, chris@zankel.net
Subject: Re: [PATCH 7/7] arch: Fix STACKTRACE_SUPPORT
Date: Fri, 8 Oct 2021 13:52:38 +0100	[thread overview]
Message-ID: <20211008125238.GC976@C02TD0UTHF1T.local> (raw)
In-Reply-To: <20211008111626.455137084@infradead.org>

On Fri, Oct 08, 2021 at 01:15:34PM +0200, Peter Zijlstra wrote:
> A few archs got save_stack_trace_tsk() vs in_sched_functions() wrong.

As mentioned on the last patch, it's not clear to me what the intended
semantic of save_stack_trace_tsk() is w.r.t. sched functions, as the
naive reading is that it should report *everything* a task may return
to.

If it's meant to skip sched functions, I think we need some explicit
documentation/commentary to that effect. In that case, there are other
architectures that need a fixup (e.g. arm64).

TBH, I don't think it *should* skip sched functions, and we should
filter out sched functions as required at a higher level, or deprecate
this interface in favour of arch_stack_walk() where it's easier to have
common filter functions invoked during the walk....

Thanks,
Mark.

> 
> Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> ---
>  arch/csky/kernel/stacktrace.c  |    7 ++++++-
>  arch/mips/kernel/stacktrace.c  |   27 ++++++++++++++++-----------
>  arch/nds32/kernel/stacktrace.c |   21 +++++++++++----------
>  3 files changed, 33 insertions(+), 22 deletions(-)
> 
> --- a/arch/csky/kernel/stacktrace.c
> +++ b/arch/csky/kernel/stacktrace.c
> @@ -122,12 +122,17 @@ static bool save_trace(unsigned long pc,
>  	return __save_trace(pc, arg, false);
>  }
>  
> +static bool save_trace_nosched(unsigned long pc, void *arg)
> +{
> +	return __save_trace(pc, arg, true);
> +}
> +
>  /*
>   * Save stack-backtrace addresses into a stack_trace buffer.
>   */
>  void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
>  {
> -	walk_stackframe(tsk, NULL, save_trace, trace);
> +	walk_stackframe(tsk, NULL, save_trace_nosched, trace);
>  }
>  EXPORT_SYMBOL_GPL(save_stack_trace_tsk);
>  
> --- a/arch/mips/kernel/stacktrace.c
> +++ b/arch/mips/kernel/stacktrace.c
> @@ -66,16 +66,7 @@ static void save_context_stack(struct st
>  #endif
>  }
>  
> -/*
> - * Save stack-backtrace addresses into a stack_trace buffer.
> - */
> -void save_stack_trace(struct stack_trace *trace)
> -{
> -	save_stack_trace_tsk(current, trace);
> -}
> -EXPORT_SYMBOL_GPL(save_stack_trace);
> -
> -void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
> +static void __save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace, bool savesched)
>  {
>  	struct pt_regs dummyregs;
>  	struct pt_regs *regs = &dummyregs;
> @@ -88,6 +79,20 @@ void save_stack_trace_tsk(struct task_st
>  		regs->cp0_epc = tsk->thread.reg31;
>  	} else
>  		prepare_frametrace(regs);
> -	save_context_stack(trace, tsk, regs, tsk == current);
> +	save_context_stack(trace, tsk, regs, savesched);
> +}
> +
> +/*
> + * Save stack-backtrace addresses into a stack_trace buffer.
> + */
> +void save_stack_trace(struct stack_trace *trace)
> +{
> +	__save_stack_trace_tsk(current, trace, true);
> +}
> +EXPORT_SYMBOL_GPL(save_stack_trace);
> +
> +void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
> +{
> +	__save_stack_trace_tsk(tsk, trace, false);
>  }
>  EXPORT_SYMBOL_GPL(save_stack_trace_tsk);
> --- a/arch/nds32/kernel/stacktrace.c
> +++ b/arch/nds32/kernel/stacktrace.c
> @@ -6,25 +6,16 @@
>  #include <linux/stacktrace.h>
>  #include <linux/ftrace.h>
>  
> -void save_stack_trace(struct stack_trace *trace)
> -{
> -	save_stack_trace_tsk(current, trace);
> -}
> -EXPORT_SYMBOL_GPL(save_stack_trace);
> -
> -void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
> +static void __save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace, bool savesched)
>  {
>  	unsigned long *fpn;
>  	int skip = trace->skip;
> -	int savesched;
>  	int graph_idx = 0;
>  
>  	if (tsk == current) {
>  		__asm__ __volatile__("\tori\t%0, $fp, #0\n":"=r"(fpn));
> -		savesched = 1;
>  	} else {
>  		fpn = (unsigned long *)thread_saved_fp(tsk);
> -		savesched = 0;
>  	}
>  
>  	while (!kstack_end(fpn) && !((unsigned long)fpn & 0x3)
> @@ -50,4 +41,14 @@ void save_stack_trace_tsk(struct task_st
>  		fpn = (unsigned long *)fpp;
>  	}
>  }
> +void save_stack_trace(struct stack_trace *trace)
> +{
> +	__save_stack_trace_tsk(current, trace, true);
> +}
> +EXPORT_SYMBOL_GPL(save_stack_trace);
> +
> +void save_stack_trace_tsk(struct task_struct *tsk, struct stack_trace *trace)
> +{
> +	__save_stack_trace_tsk(tsk, trace, false);
> +}
>  EXPORT_SYMBOL_GPL(save_stack_trace_tsk);
> 
> 