From: David Sharp <dhsharp@google.com>
To: rostedt@goodmis.org, linux-kernel@vger.kernel.org
Cc: mrubin@google.com, David Sharp <dhsharp@google.com>
Subject: [PATCH 04/15] ftrace: pack event structures.
Date: Fri, 3 Dec 2010 16:13:18 -0800 [thread overview]
Message-ID: <1291421609-14665-5-git-send-email-dhsharp@google.com> (raw)
In-Reply-To: <1291421609-14665-1-git-send-email-dhsharp@google.com>
Ftrace event structures have a 12-byte struct trace_entry at the beginning.
If the structure is aligned, this means that if the first field is 64-bits,
there will be 4 bytes of padding. Ironically, due to the 4-byte ringbuffer
header, this will make 64-bit writes unaligned, if the ring buffer position
is currently 64-bit aligned:
4(rb)+12(ftrace)+4(pad) = 20; 20%8 = 4
Adding __attribute__((packed)) to the event structures removes the extra
space from the trace events, and actually improves alignment of trace
events with a first field that is 64-bits.
About 65 tracepoints have a 4-byte pad at offset 12:
# find events -name format | xargs -n1 awk '
$1=="name:" {name=$2}
$1=="format:"{FS="\t"}
$3=="offset:12;" && $4=="size:4;"{okay=1}
$3=="offset:16;" && !okay {print name}' | wc -l
65
With all 'syscalls' and 'timer' events enabled, this results in a 5%
improvement in a simple 512MB read benchmark with warm caches.
Tested:
setup:
# echo 1 >events/syscalls/enable
# echo 1 >events/timer/enable
# echo 0 > tracing_enabled
off:
# for n in $(seq 10) ; do \
time dd if=/dev/hda3 of=/dev/null bs=1K count=512K ; \
done
on:
# for n in $(seq 10) ; do \
echo 1 >tracing_enabled; \
time dd if=/dev/hda3 of=/dev/null bs=1K count=512K ; \
echo 0 >tracing_enabled; \
echo > trace; \
done
real time mean/median/stdev
w/o patch off: 1.1679/1.164/0.0169
w/o patch on : 1.9432/1.936/0.0274
w/ patch off: 1.1715/1.159/0.0431
w/ patch on : 1.8425/1.836/0.0138
"on" delta: -0.1007 --> -5.2%
Google-Bug-Id: 2895627
Signed-off-by: David Sharp <dhsharp@google.com>
---
include/trace/ftrace.h | 5 +++--
kernel/trace/trace.h | 2 +-
2 files changed, 4 insertions(+), 3 deletions(-)
diff --git a/include/trace/ftrace.h b/include/trace/ftrace.h
index a9377c0..51d1f52 100644
--- a/include/trace/ftrace.h
+++ b/include/trace/ftrace.h
@@ -48,7 +48,8 @@
#define __array(type, item, len) type item[len];
#undef __dynamic_array
-#define __dynamic_array(type, item, len) u32 __data_loc_##item;
+#define __dynamic_array(type, item, len) \
+ u32 __data_loc_##item __attribute__((aligned(4)));
#undef __string
#define __string(item, src) __dynamic_array(char, item, -1)
@@ -62,7 +63,7 @@
struct trace_entry ent; \
tstruct \
char __data[0]; \
- }; \
+ } __attribute__((packed)); \
\
static struct ftrace_event_class event_class_##name;
diff --git a/kernel/trace/trace.h b/kernel/trace/trace.h
index 9021f8c..2e80433 100644
--- a/kernel/trace/trace.h
+++ b/kernel/trace/trace.h
@@ -60,7 +60,7 @@ enum trace_type {
struct struct_name { \
struct trace_entry ent; \
tstruct \
- }
+ } __attribute__((packed))
#undef TP_ARGS
#define TP_ARGS(args...) args
--
1.7.3.1
next prev parent reply other threads:[~2010-12-04 0:13 UTC|newest]
Thread overview: 74+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-12-04 0:13 [Patch 00/15] Reduce tracing payload size David Sharp
2010-12-04 0:13 ` [PATCH 01/15] tracing: Add a 'buffer_overwrite' debugfs file David Sharp
2010-12-04 1:57 ` Steven Rostedt
2010-12-08 20:15 ` David Sharp
2010-12-08 21:46 ` [PATCH] tracing: Add an 'overwrite' trace_option David Sharp
2010-12-14 0:39 ` David Sharp
2011-03-09 0:45 ` David Sharp
2011-03-11 9:45 ` [tip:perf/core] " tip-bot for David Sharp
2010-12-04 0:13 ` [PATCH 02/15] ring_buffer.c: Remove unused #include <linux/trace_irq.h> David Sharp
2011-03-11 9:46 ` [tip:perf/core] ring-buffer: " tip-bot for David Sharp
2010-12-04 0:13 ` [PATCH 03/15] ring_buffer: Align buffer_page struct allocations only to fit the flags David Sharp
2010-12-04 1:43 ` Steven Rostedt
2010-12-07 22:44 ` David Sharp
2010-12-04 0:13 ` David Sharp [this message]
2011-03-08 23:30 ` [PATCH 04/15] ftrace: pack event structures Steven Rostedt
2011-03-09 1:09 ` Steven Rostedt
2011-03-09 6:39 ` David Miller
2011-03-09 15:18 ` Steven Rostedt
2011-03-10 23:21 ` David Sharp
2011-03-11 3:37 ` Steven Rostedt
2011-03-09 13:01 ` Steven Rostedt
2010-12-04 0:13 ` [PATCH 05/15] ftrace: fix event alignment: ftrace:context_switch and ftrace:wakeup David Sharp
2011-03-11 9:48 ` [tip:perf/core] tracing: Fix " tip-bot for David Sharp
2010-12-04 0:13 ` [PATCH 06/15] ftrace: fix event alignment: module:module_request David Sharp
2010-12-04 1:47 ` Steven Rostedt
2010-12-06 1:28 ` Li Zefan
2011-03-11 9:48 ` [tip:perf/core] tracing: Fix " tip-bot for David Sharp
2010-12-04 0:13 ` [PATCH 07/15] ftrace: fix event alignment: kvm:kvm_hv_hypercall David Sharp
2010-12-04 1:49 ` Steven Rostedt
2010-12-04 8:11 ` Avi Kivity
2010-12-06 20:38 ` David Sharp
2010-12-07 9:22 ` Avi Kivity
2010-12-07 21:16 ` David Sharp
2010-12-08 9:18 ` Avi Kivity
2011-03-08 23:55 ` Steven Rostedt
2011-03-09 8:50 ` Avi Kivity
2011-03-09 12:54 ` Steven Rostedt
2011-03-09 13:01 ` Avi Kivity
2011-03-11 9:49 ` [tip:perf/core] tracing: Fix " tip-bot for David Sharp
2010-12-04 0:13 ` [PATCH 08/15] ftrace: fix event alignment: mce:mce_record David Sharp
2010-12-04 1:50 ` Steven Rostedt
2010-12-09 13:33 ` Frederic Weisbecker
2011-03-11 9:49 ` [tip:perf/core] tracing: Fix " tip-bot for David Sharp
2010-12-04 0:13 ` [PATCH 09/15] ftrace: fix event alignment: skb:kfree_skb David Sharp
2010-12-04 1:52 ` Steven Rostedt
2010-12-04 13:38 ` Neil Horman
2011-03-11 9:50 ` [tip:perf/core] tracing: Fix " tip-bot for David Sharp
2010-12-04 0:13 ` [PATCH 10/15] ftrace: fix event alignment: jbd2:* David Sharp
2010-12-04 1:52 ` Steven Rostedt
2011-03-09 0:03 ` Steven Rostedt
2011-03-09 0:31 ` Ted Ts'o
2011-03-09 0:41 ` Steven Rostedt
2010-12-04 0:13 ` [PATCH 11/15] ftrace: fix event alignment: ext4:* David Sharp
2010-12-04 1:53 ` Steven Rostedt
2011-03-09 0:04 ` Steven Rostedt
2010-12-04 0:13 ` [PATCH 12/15] trace_output.c: adjust conditional expression formatting David Sharp
2011-03-11 9:50 ` [tip:perf/core] tracing: Adjust conditional expression latency formatting tip-bot for David Sharp
2010-12-04 0:13 ` [PATCH 13/15] small_traces: Add config option to shrink trace events David Sharp
2010-12-04 1:56 ` Steven Rostedt
2010-12-04 2:33 ` David Sharp
2010-12-04 2:54 ` Steven Rostedt
2010-12-09 14:55 ` Frederic Weisbecker
2010-12-09 15:08 ` Steven Rostedt
2010-12-09 15:28 ` Frederic Weisbecker
2010-12-09 16:16 ` Mathieu Desnoyers
2011-03-09 0:23 ` Steven Rostedt
2010-12-04 0:13 ` [PATCH 14/15] small_traces: Remove trace output of large fields David Sharp
2010-12-04 0:13 ` [PATCH 15/15] small_traces: Remove 8 bytes from trace_entry David Sharp
2010-12-06 13:22 ` [Patch 00/15] Reduce tracing payload size Andi Kleen
2010-12-06 13:56 ` Ted Ts'o
2010-12-06 14:58 ` Andi Kleen
2010-12-06 16:17 ` Steven Rostedt
2010-12-06 16:31 ` Miguel Ojeda
2010-12-06 16:41 ` Andi Kleen
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=1291421609-14665-5-git-send-email-dhsharp@google.com \
--to=dhsharp@google.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mrubin@google.com \
--cc=rostedt@goodmis.org \
/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