Received-SPF: pass (google.com: domain of linux-kernel+bounces-68273-linux.lists.archive=gmail.com@vger.kernel.org designates 139.178.88.99 as permitted sender) client-ip=139.178.88.99;
Date: Fri, 16 Feb 2024 17:51:08 +0900
From: Masami Hiramatsu (Google) <mhiramat@kernel.org>
To: Steven Rostedt <rostedt@goodmis.org>
Cc: Alexei Starovoitov <alexei.starovoitov@gmail.com>, Florent Revest
 <revest@chromium.org>, linux-trace-kernel@vger.kernel.org, LKML
 <linux-kernel@vger.kernel.org>, Martin KaFai Lau <martin.lau@linux.dev>,
 bpf <bpf@vger.kernel.org>, Sven Schnelle <svens@linux.ibm.com>, Alexei
 Starovoitov <ast@kernel.org>, Jiri Olsa <jolsa@kernel.org>, Arnaldo
 Carvalho de Melo <acme@kernel.org>, Daniel Borkmann <daniel@iogearbox.net>,
 Alan Maguire <alan.maguire@oracle.com>, Mark Rutland
 <mark.rutland@arm.com>, Peter Zijlstra <peterz@infradead.org>, Thomas
 Gleixner <tglx@linutronix.de>, Guo Ren <guoren@kernel.org>
Subject: Re: [PATCH v7 23/36] function_graph: Add a new exit handler with
 parent_ip and ftrace_regs
Message-Id: <20240216175108.79a256a20c89ed1d672c7e14@kernel.org>
In-Reply-To: <20240215110404.4e8c5a94@gandalf.local.home>
References: <170723204881.502590.11906735097521170661.stgit@devnote2>
	<170723230476.502590.16817817024423790038.stgit@devnote2>
	<20240215110404.4e8c5a94@gandalf.local.home>
Precedence: bulk
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

On Thu, 15 Feb 2024 11:04:04 -0500
Steven Rostedt <rostedt@goodmis.org> wrote:

> On Wed,  7 Feb 2024 00:11:44 +0900
> "Masami Hiramatsu (Google)" <mhiramat@kernel.org> wrote:
> 
> > diff --git a/arch/x86/include/asm/ftrace.h b/arch/x86/include/asm/ftrace.h
> > index c88bf47f46da..a061f8832b20 100644
> > --- a/arch/x86/include/asm/ftrace.h
> > +++ b/arch/x86/include/asm/ftrace.h
> > @@ -72,6 +72,8 @@ arch_ftrace_get_regs(struct ftrace_regs *fregs)
> >  	override_function_with_return(&(fregs)->regs)
> >  #define ftrace_regs_query_register_offset(name) \
> >  	regs_query_register_offset(name)
> > +#define ftrace_regs_get_frame_pointer(fregs) \
> > +	frame_pointer(&(fregs)->regs)
> >  
> 
> Doesn't the above belong in the next patch that implements this for x86?

Yes, thanks for pointing it!

> 
> >  struct ftrace_ops;
> >  #define ftrace_graph_func ftrace_graph_func
> > diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
> > index 65d4d4b68768..da2a23f5a9ed 100644
> > --- a/include/linux/ftrace.h
> > +++ b/include/linux/ftrace.h
> > @@ -43,7 +43,9 @@ struct dyn_ftrace;
> >  
> >  char *arch_ftrace_match_adjust(char *str, const char *search);
> >  
> > -#ifdef CONFIG_HAVE_FUNCTION_GRAPH_RETVAL
> > +#ifdef CONFIG_HAVE_FUNCTION_GRAPH_FREGS
> > +unsigned long ftrace_return_to_handler(struct ftrace_regs *fregs);
> > +#elif defined(CONFIG_HAVE_FUNCTION_GRAPH_RETVAL)
> >  struct fgraph_ret_regs;
> >  unsigned long ftrace_return_to_handler(struct fgraph_ret_regs *ret_regs);
> >  #else
> > @@ -157,6 +159,7 @@ struct ftrace_regs {
> >  #define ftrace_regs_set_instruction_pointer(fregs, ip) do { } while (0)
> >  #endif /* CONFIG_HAVE_DYNAMIC_FTRACE_WITH_ARGS */
> >  
> > +
> 
> spurious newline.

OK, I'll remove.

> 
> >  static __always_inline struct pt_regs *ftrace_get_regs(struct ftrace_regs *fregs)
> >  {
> >  	if (!fregs)
> > @@ -1067,6 +1070,10 @@ typedef int (*trace_func_graph_regs_ent_t)(unsigned long func,
> >  					   unsigned long parent_ip,
> >  					   struct ftrace_regs *fregs,
> >  					   struct fgraph_ops *); /* entry w/ regs */
> > +typedef void (*trace_func_graph_regs_ret_t)(unsigned long func,
> > +					    unsigned long parent_ip,
> > +					    struct ftrace_regs *,
> > +					    struct fgraph_ops *); /* return w/ regs */
> >  
> >  extern int ftrace_graph_entry_stub(struct ftrace_graph_ent *trace, struct fgraph_ops *gops);
> >  
> > @@ -1076,6 +1083,7 @@ struct fgraph_ops {
> >  	trace_func_graph_ent_t		entryfunc;
> >  	trace_func_graph_ret_t		retfunc;
> >  	trace_func_graph_regs_ent_t	entryregfunc;
> > +	trace_func_graph_regs_ret_t	retregfunc;
> >  	struct ftrace_ops		ops; /* for the hash lists */
> >  	void				*private;
> >  	int				idx;
> > diff --git a/kernel/trace/Kconfig b/kernel/trace/Kconfig
> > index 61c541c36596..308b3bec01b1 100644
> > --- a/kernel/trace/Kconfig
> > +++ b/kernel/trace/Kconfig
> > @@ -34,6 +34,9 @@ config HAVE_FUNCTION_GRAPH_TRACER
> >  config HAVE_FUNCTION_GRAPH_RETVAL
> >  	bool
> >  
> > +config HAVE_FUNCTION_GRAPH_FREGS
> > +	bool
> > +
> >  config HAVE_DYNAMIC_FTRACE
> >  	bool
> >  	help
> > @@ -232,7 +235,7 @@ config FUNCTION_GRAPH_TRACER
> >  
> >  config FUNCTION_GRAPH_RETVAL
> >  	bool "Kernel Function Graph Return Value"
> > -	depends on HAVE_FUNCTION_GRAPH_RETVAL
> > +	depends on HAVE_FUNCTION_GRAPH_RETVAL || HAVE_FUNCTION_GRAPH_FREGS
> >  	depends on FUNCTION_GRAPH_TRACER
> >  	default n
> >  	help
> > diff --git a/kernel/trace/fgraph.c b/kernel/trace/fgraph.c
> > index 459912ca72e0..12e5f108e242 100644
> > --- a/kernel/trace/fgraph.c
> > +++ b/kernel/trace/fgraph.c
> > @@ -752,8 +752,8 @@ int function_graph_enter_ops(unsigned long ret, unsigned long func,
> >  
> >  /* Retrieve a function return address to the trace stack on thread info.*/
> >  static struct ftrace_ret_stack *
> > -ftrace_pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret,
> > -			unsigned long frame_pointer, int *index)
> > +ftrace_pop_return_trace(unsigned long *ret, unsigned long frame_pointer,
> > +			int *index)
> >  {
> >  	struct ftrace_ret_stack *ret_stack;
> >  
> > @@ -798,10 +798,6 @@ ftrace_pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret,
> >  
> >  	*index += FGRAPH_RET_INDEX;
> >  	*ret = ret_stack->ret;
> > -	trace->func = ret_stack->func;
> > -	trace->calltime = ret_stack->calltime;
> > -	trace->overrun = atomic_read(&current->trace_overrun);
> > -	trace->depth = current->curr_ret_depth;
> 
> There's a lot of information stored in the trace structure. Why not pass
> that to the new retregfunc?
> 
> Then you don't need to separate this code out.

Sorry, I couldn't catch what you meant, Would you mean to call
ftrace_pop_return_trace() before calling retregfunc()?? because some of the
information are found from ret_stack, which is poped from shadow stack.

> 
> >  	/*
> >  	 * We still want to trace interrupts coming in if
> >  	 * max_depth is set to 1. Make sure the decrement is
> > @@ -840,21 +836,42 @@ static struct notifier_block ftrace_suspend_notifier = {
> >  /* fgraph_ret_regs is not defined without CONFIG_FUNCTION_GRAPH_RETVAL */
> >  struct fgraph_ret_regs;
> >  
> > +static void fgraph_call_retfunc(struct ftrace_regs *fregs,
> > +				struct fgraph_ret_regs *ret_regs,
> > +				struct ftrace_ret_stack *ret_stack,
> > +				struct fgraph_ops *gops)
> > +{
> > +	struct ftrace_graph_ret trace;
> > +
> > +	trace.func = ret_stack->func;
> > +	trace.calltime = ret_stack->calltime;
> > +	trace.overrun = atomic_read(&current->trace_overrun);
> > +	trace.depth = current->curr_ret_depth;
> > +	trace.rettime = trace_clock_local();
> > +#ifdef CONFIG_FUNCTION_GRAPH_RETVAL
> > +	if (fregs)
> > +		trace.retval = ftrace_regs_get_return_value(fregs);
> > +	else
> > +		trace.retval = fgraph_ret_regs_return_value(ret_regs);
> > +#endif
> > +	gops->retfunc(&trace, gops);
> > +}
> > +
> >  /*
> >   * Send the trace to the ring-buffer.
> >   * @return the original return address.
> >   */
> > -static unsigned long __ftrace_return_to_handler(struct fgraph_ret_regs *ret_regs,
> > +static unsigned long __ftrace_return_to_handler(struct ftrace_regs *fregs,
> > +						struct fgraph_ret_regs *ret_regs,
> >  						unsigned long frame_pointer)
> >  {
> >  	struct ftrace_ret_stack *ret_stack;
> > -	struct ftrace_graph_ret trace;
> >  	unsigned long bitmap;
> >  	unsigned long ret;
> >  	int index;
> >  	int i;
> >  
> > -	ret_stack = ftrace_pop_return_trace(&trace, &ret, frame_pointer, &index);
> > +	ret_stack = ftrace_pop_return_trace(&ret, frame_pointer, &index);
> >  
> >  	if (unlikely(!ret_stack)) {
> >  		ftrace_graph_stop();
> > @@ -863,10 +880,8 @@ static unsigned long __ftrace_return_to_handler(struct fgraph_ret_regs *ret_regs
> >  		return (unsigned long)panic;
> >  	}
> >  
> > -	trace.rettime = trace_clock_local();
> > -#ifdef CONFIG_FUNCTION_GRAPH_RETVAL
> > -	trace.retval = fgraph_ret_regs_return_value(ret_regs);
> > -#endif
> > +	if (fregs)
> > +		ftrace_regs_set_instruction_pointer(fregs, ret);
> >  
> >  	bitmap = get_fgraph_index_bitmap(current, index);
> >  	for (i = 0; i < FGRAPH_ARRAY_SIZE; i++) {
> > @@ -877,7 +892,10 @@ static unsigned long __ftrace_return_to_handler(struct fgraph_ret_regs *ret_regs
> >  		if (gops == &fgraph_stub)
> >  			continue;
> >  
> > -		gops->retfunc(&trace, gops);
> > +		if (gops->retregfunc)
> > +			gops->retregfunc(ret_stack->func, ret, fregs, gops);
> > +		else
> > +			fgraph_call_retfunc(fregs, ret_regs, ret_stack, gops);
> >  	}
> >  
> >  	/*
> > @@ -892,20 +910,22 @@ static unsigned long __ftrace_return_to_handler(struct fgraph_ret_regs *ret_regs
> >  	return ret;
> >  }
> >  
> > -/*
> > - * After all architecures have selected HAVE_FUNCTION_GRAPH_RETVAL, we can
> > - * leave only ftrace_return_to_handler(ret_regs).
> > - */
> > -#ifdef CONFIG_HAVE_FUNCTION_GRAPH_RETVAL
> > +#ifdef CONFIG_HAVE_FUNCTION_GRAPH_FREGS
> > +unsigned long ftrace_return_to_handler(struct ftrace_regs *fregs)
> > +{
> > +	return __ftrace_return_to_handler(fregs, NULL,
> > +				ftrace_regs_get_frame_pointer(fregs));
> > +}
> > +#elif defined(CONFIG_HAVE_FUNCTION_GRAPH_RETVAL)
> >  unsigned long ftrace_return_to_handler(struct fgraph_ret_regs *ret_regs)
> >  {
> > -	return __ftrace_return_to_handler(ret_regs,
> > +	return __ftrace_return_to_handler(NULL, ret_regs,
> >  				fgraph_ret_regs_frame_pointer(ret_regs));
> >  }
> >  #else
> >  unsigned long ftrace_return_to_handler(unsigned long frame_pointer)
> >  {
> > -	return __ftrace_return_to_handler(NULL, frame_pointer);
> > +	return __ftrace_return_to_handler(NULL, NULL, frame_pointer);
> >  }
> >  #endif
> >  
> > @@ -1262,9 +1282,15 @@ int register_ftrace_graph(struct fgraph_ops *gops)
> >  	int ret = 0;
> >  	int i;
> >  
> > -	if (gops->entryfunc && gops->entryregfunc)
> > +	if ((gops->entryfunc && gops->entryregfunc) ||
> > +	    (gops->retfunc && gops->retregfunc))
> >  		return -EINVAL;
> 
> With a union, you don't need this.

Indeed.

> 
> Now, is it possible to have a entryregfunc and a retfunc, or a entryfunc
> and a retregfunc?

It is possisble, but may not give any benefit.

Thank you,

> 
> -- Steve
> 
> >  
> > +#ifndef CONFIG_HAVE_FUNCTION_GRAPH_FREGS
> > +	if (gops->retregfunc)
> > +		return -EOPNOTSUPP;
> > +#endif
> > +
> >  	mutex_lock(&ftrace_lock);
> >  
> >  	if (!gops->ops.func) {
> 


-- 
Masami Hiramatsu (Google) <mhiramat@kernel.org>