linux-trace-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Masami Hiramatsu (Google)" <mhiramat@kernel.org>
To: Alexei Starovoitov <alexei.starovoitov@gmail.com>,
	Steven Rostedt <rostedt@goodmis.org>,
	Florent Revest <revest@chromium.org>
Cc: 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>,
	linux-arch@vger.kernel.org
Subject: [PATCH v14 19/19] fgraph: Skip recording calltime/rettime if it is not nneeded
Date: Fri, 13 Sep 2024 00:11:38 +0900	[thread overview]
Message-ID: <172615389864.133222.14452329708227900626.stgit@devnote2> (raw)
In-Reply-To: <172615368656.133222.2336770908714920670.stgit@devnote2>

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

Skip recording calltime and rettime if the fgraph_ops does not need it.
This is a kind of performance optimization for fprobe. Since the fprobe
user does not use these entries, recording timestamp in fgraph is just
a overhead (e.g. eBPF, ftrace). So introduce the skip_timestamp flag,
and all fgraph_ops sets this flag, skip recording calltime and rettime.

Here is the performance results measured by
 tools/testing/selftests/bpf/benchs/run_bench_trigger.sh

Without this:
kprobe-multi   :    5.700 ± 0.065M/s
kretprobe-multi:    4.239 ± 0.006M/s

With skip-timestamp:
kprobe-multi   :    6.265 ± 0.033M/s	+9.91%
kretprobe-multi:    4.758 ± 0.009M/s	+12.24%

Suggested-by: Jiri Olsa <olsajiri@gmail.com>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
---
 Changes in v11:
  - Simplify it to be symmetric on push and pop. (Thus the timestamp
    getting place is a bit shifted.)
 Changes in v10:
  - Add likely() to skipping timestamp.
 Changes in v9:
  - Newly added.
---
 include/linux/ftrace.h |    2 ++
 kernel/trace/fgraph.c  |   36 +++++++++++++++++++++++++++++++++---
 kernel/trace/fprobe.c  |    1 +
 3 files changed, 36 insertions(+), 3 deletions(-)

diff --git a/include/linux/ftrace.h b/include/linux/ftrace.h
index 63fb91088a23..bab6fabb3fa1 100644
--- a/include/linux/ftrace.h
+++ b/include/linux/ftrace.h
@@ -1160,6 +1160,8 @@ struct fgraph_ops {
 	void				*private;
 	trace_func_graph_ent_t		saved_func;
 	int				idx;
+	/* If skip_timestamp is true, this does not record timestamps. */
+	bool				skip_timestamp;
 };
 
 void *fgraph_reserve_data(int idx, int size_bytes);
diff --git a/kernel/trace/fgraph.c b/kernel/trace/fgraph.c
index 6a3e2db16aa4..c116a92839ae 100644
--- a/kernel/trace/fgraph.c
+++ b/kernel/trace/fgraph.c
@@ -174,6 +174,7 @@ int ftrace_graph_active;
 
 static struct fgraph_ops *fgraph_array[FGRAPH_ARRAY_SIZE];
 static unsigned long fgraph_array_bitmask;
+static bool fgraph_skip_timestamp;
 
 /* LRU index table for fgraph_array */
 static int fgraph_lru_table[FGRAPH_ARRAY_SIZE];
@@ -557,7 +558,11 @@ ftrace_push_return_trace(unsigned long ret, unsigned long func,
 		return -EBUSY;
 	}
 
-	calltime = trace_clock_local();
+	/* This is not really 'likely' but for keeping the least path to be faster. */
+	if (likely(fgraph_skip_timestamp))
+		calltime = 0LL;
+	else
+		calltime = trace_clock_local();
 
 	offset = READ_ONCE(current->curr_ret_stack);
 	ret_stack = RET_STACK(current, offset);
@@ -728,6 +733,12 @@ ftrace_pop_return_trace(struct ftrace_graph_ret *trace, unsigned long *ret,
 	*ret = ret_stack->ret;
 	trace->func = ret_stack->func;
 	trace->calltime = ret_stack->calltime;
+	/* This is not really 'likely' but for keeping the least path to be faster. */
+	if (likely(!trace->calltime))
+		trace->rettime = 0LL;
+	else
+		trace->rettime = trace_clock_local();
+
 	trace->overrun = atomic_read(&current->trace_overrun);
 	trace->depth = current->curr_ret_depth;
 	/*
@@ -788,7 +799,6 @@ __ftrace_return_to_handler(struct ftrace_regs *fregs, unsigned long frame_pointe
 		return (unsigned long)panic;
 	}
 
-	trace.rettime = trace_clock_local();
 	if (fregs)
 		ftrace_regs_set_instruction_pointer(fregs, ret);
 
@@ -1248,6 +1258,24 @@ static void ftrace_graph_disable_direct(bool disable_branch)
 	fgraph_direct_gops = &fgraph_stub;
 }
 
+static void update_fgraph_skip_timestamp(void)
+{
+	int i;
+
+	for (i = 0; i < FGRAPH_ARRAY_SIZE; i++) {
+		struct fgraph_ops *gops = fgraph_array[i];
+
+		if (gops == &fgraph_stub)
+			continue;
+
+		if (!gops->skip_timestamp) {
+			fgraph_skip_timestamp = false;
+			return;
+		}
+	}
+	fgraph_skip_timestamp = true;
+}
+
 int register_ftrace_graph(struct fgraph_ops *gops)
 {
 	int command = 0;
@@ -1271,6 +1299,7 @@ int register_ftrace_graph(struct fgraph_ops *gops)
 	gops->idx = i;
 
 	ftrace_graph_active++;
+	update_fgraph_skip_timestamp();
 
 	if (ftrace_graph_active == 2)
 		ftrace_graph_disable_direct(true);
@@ -1303,6 +1332,7 @@ int register_ftrace_graph(struct fgraph_ops *gops)
 		ftrace_graph_active--;
 		gops->saved_func = NULL;
 		fgraph_lru_release_index(i);
+		update_fgraph_skip_timestamp();
 	}
 out:
 	mutex_unlock(&ftrace_lock);
@@ -1326,8 +1356,8 @@ void unregister_ftrace_graph(struct fgraph_ops *gops)
 		goto out;
 
 	fgraph_array[gops->idx] = &fgraph_stub;
-
 	ftrace_graph_active--;
+	update_fgraph_skip_timestamp();
 
 	if (!ftrace_graph_active)
 		command = FTRACE_STOP_FUNC_RET;
diff --git a/kernel/trace/fprobe.c b/kernel/trace/fprobe.c
index 5a0b4ef52fa7..b108d26d7ee5 100644
--- a/kernel/trace/fprobe.c
+++ b/kernel/trace/fprobe.c
@@ -345,6 +345,7 @@ NOKPROBE_SYMBOL(fprobe_return);
 static struct fgraph_ops fprobe_graph_ops = {
 	.entryfunc	= fprobe_entry,
 	.retfunc	= fprobe_return,
+	.skip_timestamp = true,
 };
 static int fprobe_graph_active;
 


  parent reply	other threads:[~2024-09-12 15:11 UTC|newest]

Thread overview: 53+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-09-12 15:08 [PATCH v14 00/19] tracing: fprobe: function_graph: Multi-function graph and fprobe on fgraph Masami Hiramatsu (Google)
2024-09-12 15:08 ` [PATCH v14 01/19] tracing: Add a comment about ftrace_regs definition Masami Hiramatsu (Google)
2024-09-12 15:08 ` [PATCH v14 02/19] tracing: Rename ftrace_regs_return_value to ftrace_regs_get_return_value Masami Hiramatsu (Google)
2024-09-12 15:08 ` [PATCH v14 03/19] function_graph: Pass ftrace_regs to entryfunc Masami Hiramatsu (Google)
2024-09-15  8:46   ` Steven Rostedt
2024-09-17  8:26     ` Will Deacon
2024-09-30 18:46       ` Steven Rostedt
2024-10-01  1:57         ` Masami Hiramatsu
2024-09-15  8:50   ` Steven Rostedt
2024-09-15  8:53   ` Steven Rostedt
2024-09-15  8:56   ` Steven Rostedt
2024-09-15  8:58   ` Steven Rostedt
2024-09-12 15:08 ` [PATCH v14 04/19] function_graph: Replace fgraph_ret_regs with ftrace_regs Masami Hiramatsu (Google)
2024-09-15  9:11   ` Steven Rostedt
2024-09-17  9:55     ` Will Deacon
2024-09-30 18:55       ` Steven Rostedt
2024-10-01 23:10         ` Masami Hiramatsu
2024-10-01 23:32           ` Steven Rostedt
2024-10-02 14:31             ` Masami Hiramatsu
2024-09-15  9:13   ` Steven Rostedt
2024-09-15  9:15   ` Steven Rostedt
2024-09-16 12:16     ` Heiko Carstens
2024-09-16 16:29       ` Steven Rostedt
2024-09-16 18:59         ` Heiko Carstens
2024-10-01 12:55           ` Masami Hiramatsu
2024-09-15  9:17   ` Steven Rostedt
2024-09-12 15:09 ` [PATCH v14 05/19] function_graph: Pass ftrace_regs to retfunc Masami Hiramatsu (Google)
2024-09-15  8:49   ` Steven Rostedt
2024-09-17 10:08     ` Will Deacon
2024-09-30 19:03       ` Steven Rostedt
2024-10-01 23:24         ` Masami Hiramatsu
2024-09-15  8:51   ` Steven Rostedt
2024-09-15  8:54   ` Steven Rostedt
2024-09-15  8:57   ` Steven Rostedt
2024-09-15  9:00   ` Steven Rostedt
2024-09-12 15:09 ` [PATCH v14 06/19] fprobe: Use ftrace_regs in fprobe entry handler Masami Hiramatsu (Google)
2024-09-12 15:09 ` [PATCH v14 07/19] fprobe: Use ftrace_regs in fprobe exit handler Masami Hiramatsu (Google)
2024-09-12 15:09 ` [PATCH v14 08/19] tracing: Add ftrace_partial_regs() for converting ftrace_regs to pt_regs Masami Hiramatsu (Google)
2024-09-15  9:22   ` Steven Rostedt
2024-09-17 10:14     ` Will Deacon
2024-10-01 23:26       ` Masami Hiramatsu
2024-09-12 15:09 ` [PATCH v14 09/19] tracing: Add ftrace_fill_perf_regs() for perf event Masami Hiramatsu (Google)
2024-09-12 15:09 ` [PATCH v14 10/19] tracing/fprobe: Enable fprobe events with CONFIG_DYNAMIC_FTRACE_WITH_ARGS Masami Hiramatsu (Google)
2024-09-12 15:10 ` [PATCH v14 11/19] bpf: Enable kprobe_multi feature if CONFIG_FPROBE is enabled Masami Hiramatsu (Google)
2024-09-12 15:10 ` [PATCH v14 12/19] ftrace: Add CONFIG_HAVE_FTRACE_GRAPH_FUNC Masami Hiramatsu (Google)
2024-09-12 15:10 ` [PATCH v14 13/19] fprobe: Rewrite fprobe on function-graph tracer Masami Hiramatsu (Google)
2024-09-12 15:10 ` [PATCH v14 14/19] tracing: Fix function timing profiler to initialize hashtable Masami Hiramatsu (Google)
2024-09-12 15:10 ` [PATCH v14 15/19] tracing/fprobe: Remove nr_maxactive from fprobe Masami Hiramatsu (Google)
2024-09-12 15:11 ` [PATCH v14 16/19] selftests: ftrace: Remove obsolate maxactive syntax check Masami Hiramatsu (Google)
2024-09-12 15:11 ` [PATCH v14 17/19] selftests/ftrace: Add a test case for repeating register/unregister fprobe Masami Hiramatsu (Google)
2024-09-12 15:11 ` [PATCH v14 18/19] Documentation: probes: Update fprobe on function-graph tracer Masami Hiramatsu (Google)
2024-09-12 15:11 ` Masami Hiramatsu (Google) [this message]
2024-09-14 21:53   ` [PATCH v14 19/19] fgraph: Skip recording calltime/rettime if it is not nneeded Steven Rostedt

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=172615389864.133222.14452329708227900626.stgit@devnote2 \
    --to=mhiramat@kernel.org \
    --cc=acme@kernel.org \
    --cc=alan.maguire@oracle.com \
    --cc=alexei.starovoitov@gmail.com \
    --cc=ast@kernel.org \
    --cc=bpf@vger.kernel.org \
    --cc=daniel@iogearbox.net \
    --cc=guoren@kernel.org \
    --cc=jolsa@kernel.org \
    --cc=linux-arch@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-trace-kernel@vger.kernel.org \
    --cc=mark.rutland@arm.com \
    --cc=martin.lau@linux.dev \
    --cc=peterz@infradead.org \
    --cc=revest@chromium.org \
    --cc=rostedt@goodmis.org \
    --cc=svens@linux.ibm.com \
    --cc=tglx@linutronix.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).