From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Ingo Molnar <mingo@kernel.org>
Cc: linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org,
Jin Yao <yao.jin@linux.intel.com>,
Alexander Shishkin <alexander.shishkin@linux.intel.com>,
Jiri Olsa <jolsa@kernel.org>, Kan Liang <kan.liang@intel.com>,
Peter Zijlstra <peterz@infradead.org>,
Arnaldo Carvalho de Melo <acme@redhat.com>
Subject: [PATCH 37/40] perf report: Tag branch type/flag on "to" and tag cycles on "from"
Date: Wed, 26 Jul 2017 11:17:27 -0300 [thread overview]
Message-ID: <20170726141730.30750-38-acme@kernel.org> (raw)
In-Reply-To: <20170726141730.30750-1-acme@kernel.org>
From: Jin Yao <yao.jin@linux.intel.com>
Current --branch-history LBR annotation displays confused data. For
example, each cycles report is duplicated on both "from" and "to"
entries.
For example:
perf report --branch-history --no-children --stdio
--2.32%--main div.c:39 (COND_BWD CROSS_2M predicted:49.7% cycles:1)
main div.c:44 (predicted:49.7% cycles:1)
main div.c:42 (RET CROSS_2M cycles:2)
compute_flag div.c:28 (cycles:2)
compute_flag div.c:27 (RET CROSS_2M cycles:1)
rand rand.c:28 (cycles:1)
rand rand.c:28 (RET CROSS_2M cycles:1)
__random random.c:298 (cycles:1)
__random random.c:297 (COND_BWD CROSS_2M cycles:1)
__random random.c:295 (cycles:1)
__random random.c:295 (COND_BWD CROSS_2M cycles:1)
__random random.c:295 (cycles:1)
__random random.c:295 (RET CROSS_2M cycles:9)
The cycles should be tagged only on the "from". It's for the code block
that ends with "from", not for "to".
Another issue is the "predicted:49.7%" is duplicated too (tag on both
"from" and "to").
This patch tags the branch type/flag on "to" and tag the cycles on
"from".
For example:
--2.32%--main div.c:39 (COND_BWD CROSS_2M predicted:49.7%)
main div.c:44 (cycles:1)
main div.c:42 (RET CROSS_2M)
compute_flag div.c:28 (cycles:2)
compute_flag div.c:27 (RET CROSS_2M)
rand rand.c:28 (cycles:1)
rand rand.c:28 (RET CROSS_2M)
__random random.c:298 (cycles:1)
__random random.c:297 (COND_BWD CROSS_2M)
__random random.c:295 (cycles:1)
__random random.c:295 (COND_BWD CROSS_2M)
__random random.c:295 (cycles:1)
__random random.c:295 (RET CROSS_2M)
|
--2.23%--__random_r random_r.c:392 (cycles:9)
In this example, The "main div.c:39 (COND_BWD CROSS_2M predicted:49.7%)"
is "to" of branch and "main div.c:44 (cycles:1)" is "from" of branch.
It should be easier for understanding than before.
Signed-off-by: Yao Jin <yao.jin@linux.intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/1500894547-18411-1-git-send-email-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/util/branch.h | 11 ++--
tools/perf/util/callchain.c | 148 +++++++++++++++++++++++++++++++-------------
2 files changed, 111 insertions(+), 48 deletions(-)
diff --git a/tools/perf/util/branch.h b/tools/perf/util/branch.h
index 686f2b65ba84..1e3c7c5cdc63 100644
--- a/tools/perf/util/branch.h
+++ b/tools/perf/util/branch.h
@@ -5,11 +5,12 @@
#include "../perf.h"
struct branch_type_stat {
- u64 counts[PERF_BR_MAX];
- u64 cond_fwd;
- u64 cond_bwd;
- u64 cross_4k;
- u64 cross_2m;
+ bool branch_to;
+ u64 counts[PERF_BR_MAX];
+ u64 cond_fwd;
+ u64 cond_bwd;
+ u64 cross_4k;
+ u64 cross_2m;
};
struct branch_flags;
diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c
index 1f536418dfb5..f320b0777e0d 100644
--- a/tools/perf/util/callchain.c
+++ b/tools/perf/util/callchain.c
@@ -563,20 +563,33 @@ fill_node(struct callchain_node *node, struct callchain_cursor *cursor)
if (cursor_node->branch) {
call->branch_count = 1;
- if (cursor_node->branch_flags.predicted)
- call->predicted_count = 1;
-
- if (cursor_node->branch_flags.abort)
- call->abort_count = 1;
-
- call->cycles_count = cursor_node->branch_flags.cycles;
- call->iter_count = cursor_node->nr_loop_iter;
- call->samples_count = cursor_node->samples;
-
- branch_type_count(&call->brtype_stat,
- &cursor_node->branch_flags,
- cursor_node->branch_from,
- cursor_node->ip);
+ if (cursor_node->branch_from) {
+ /*
+ * branch_from is set with value somewhere else
+ * to imply it's "to" of a branch.
+ */
+ call->brtype_stat.branch_to = true;
+
+ if (cursor_node->branch_flags.predicted)
+ call->predicted_count = 1;
+
+ if (cursor_node->branch_flags.abort)
+ call->abort_count = 1;
+
+ branch_type_count(&call->brtype_stat,
+ &cursor_node->branch_flags,
+ cursor_node->branch_from,
+ cursor_node->ip);
+ } else {
+ /*
+ * It's "from" of a branch
+ */
+ call->brtype_stat.branch_to = false;
+ call->cycles_count =
+ cursor_node->branch_flags.cycles;
+ call->iter_count = cursor_node->nr_loop_iter;
+ call->samples_count = cursor_node->samples;
+ }
}
list_add_tail(&call->list, &node->val);
@@ -685,20 +698,32 @@ static enum match_result match_chain(struct callchain_cursor_node *node,
if (node->branch) {
cnode->branch_count++;
- if (node->branch_flags.predicted)
- cnode->predicted_count++;
-
- if (node->branch_flags.abort)
- cnode->abort_count++;
-
- cnode->cycles_count += node->branch_flags.cycles;
- cnode->iter_count += node->nr_loop_iter;
- cnode->samples_count += node->samples;
-
- branch_type_count(&cnode->brtype_stat,
- &node->branch_flags,
- node->branch_from,
- node->ip);
+ if (node->branch_from) {
+ /*
+ * It's "to" of a branch
+ */
+ cnode->brtype_stat.branch_to = true;
+
+ if (node->branch_flags.predicted)
+ cnode->predicted_count++;
+
+ if (node->branch_flags.abort)
+ cnode->abort_count++;
+
+ branch_type_count(&cnode->brtype_stat,
+ &node->branch_flags,
+ node->branch_from,
+ node->ip);
+ } else {
+ /*
+ * It's "from" of a branch
+ */
+ cnode->brtype_stat.branch_to = false;
+ cnode->cycles_count +=
+ node->branch_flags.cycles;
+ cnode->iter_count += node->nr_loop_iter;
+ cnode->samples_count += node->samples;
+ }
}
return MATCH_EQ;
@@ -1236,27 +1261,26 @@ static int count_pri64_printf(int idx, const char *str, u64 value, char *bf, int
return printed;
}
-static int count_float_printf(int idx, const char *str, float value, char *bf, int bfsize)
+static int count_float_printf(int idx, const char *str, float value,
+ char *bf, int bfsize, float threshold)
{
int printed;
+ if (threshold != 0.0 && value < threshold)
+ return 0;
+
printed = scnprintf(bf, bfsize, "%s%s:%.1f%%", (idx) ? " " : " (", str, value);
return printed;
}
-static int counts_str_build(char *bf, int bfsize,
- u64 branch_count, u64 predicted_count,
- u64 abort_count, u64 cycles_count,
- u64 iter_count, u64 samples_count,
- struct branch_type_stat *brtype_stat)
+static int branch_to_str(char *bf, int bfsize,
+ u64 branch_count, u64 predicted_count,
+ u64 abort_count,
+ struct branch_type_stat *brtype_stat)
{
- u64 cycles;
int printed, i = 0;
- if (branch_count == 0)
- return scnprintf(bf, bfsize, " (calltrace)");
-
printed = branch_type_str(brtype_stat, bf, bfsize);
if (printed)
i++;
@@ -1264,15 +1288,29 @@ static int counts_str_build(char *bf, int bfsize,
if (predicted_count < branch_count) {
printed += count_float_printf(i++, "predicted",
predicted_count * 100.0 / branch_count,
- bf + printed, bfsize - printed);
+ bf + printed, bfsize - printed, 0.0);
}
if (abort_count) {
printed += count_float_printf(i++, "abort",
abort_count * 100.0 / branch_count,
- bf + printed, bfsize - printed);
+ bf + printed, bfsize - printed, 0.1);
}
+ if (i)
+ printed += scnprintf(bf + printed, bfsize - printed, ")");
+
+ return printed;
+}
+
+static int branch_from_str(char *bf, int bfsize,
+ u64 branch_count,
+ u64 cycles_count, u64 iter_count,
+ u64 samples_count)
+{
+ int printed = 0, i = 0;
+ u64 cycles;
+
cycles = cycles_count / branch_count;
if (cycles) {
printed += count_pri64_printf(i++, "cycles",
@@ -1287,10 +1325,34 @@ static int counts_str_build(char *bf, int bfsize,
}
if (i)
- return scnprintf(bf + printed, bfsize - printed, ")");
+ printed += scnprintf(bf + printed, bfsize - printed, ")");
- bf[0] = 0;
- return 0;
+ return printed;
+}
+
+static int counts_str_build(char *bf, int bfsize,
+ u64 branch_count, u64 predicted_count,
+ u64 abort_count, u64 cycles_count,
+ u64 iter_count, u64 samples_count,
+ struct branch_type_stat *brtype_stat)
+{
+ int printed;
+
+ if (branch_count == 0)
+ return scnprintf(bf, bfsize, " (calltrace)");
+
+ if (brtype_stat->branch_to) {
+ printed = branch_to_str(bf, bfsize, branch_count,
+ predicted_count, abort_count, brtype_stat);
+ } else {
+ printed = branch_from_str(bf, bfsize, branch_count,
+ cycles_count, iter_count, samples_count);
+ }
+
+ if (!printed)
+ bf[0] = 0;
+
+ return printed;
}
static int callchain_counts_printf(FILE *fp, char *bf, int bfsize,
--
2.9.4
next prev parent reply other threads:[~2017-07-26 14:21 UTC|newest]
Thread overview: 42+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-07-26 14:16 [GIT PULL 00/40] perf/core improvements and fixes Arnaldo Carvalho de Melo
2017-07-26 14:16 ` [PATCH 01/40] perf intel-pt: Set no_aux_samples for the tracking event Arnaldo Carvalho de Melo
2017-07-26 14:16 ` [PATCH 02/40] perf intel-pt: Always set no branch for dummy event Arnaldo Carvalho de Melo
2017-07-26 14:16 ` [PATCH 03/40] perf trace: Add missing ' = ' in the default formatting of syscall returns Arnaldo Carvalho de Melo
2017-07-26 14:16 ` [PATCH 04/40] perf trace beauty mmap: Ignore 'fd' and 'offset' args for MAP_ANONYMOUS Arnaldo Carvalho de Melo
2017-07-26 14:16 ` [PATCH 05/40] perf trace: Allow allocating sc->arg_fmt even without the syscall tracepoint Arnaldo Carvalho de Melo
2017-07-26 14:16 ` [PATCH 06/40] perf trace: Use the syscall_fmt formatters without a tracepoint Arnaldo Carvalho de Melo
2017-07-26 14:16 ` [PATCH 07/40] perf trace: Ditch __syscall__arg_val() variant, not needed anymore Arnaldo Carvalho de Melo
2017-07-26 14:16 ` [PATCH 08/40] perf trace: Allow specifying number of syscall args for tracepointless syscalls Arnaldo Carvalho de Melo
2017-07-26 14:16 ` [PATCH 09/40] perf trace: Allow specifying names to syscall arguments formatters Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 10/40] tools include uapi: Grab a copy of linux/sched.h Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 11/40] perf trace beauty clone: Beautify syscall arguments Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 12/40] perf trace beauty clone: Suppress unused args according to 'flags' arg Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 13/40] perf trace: Introduce filter_loop_pids() Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 14/40] perf trace: Filter out 'sshd' in the tracer ancestry in syswide tracing Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 15/40] tools include: Adopt strstarts() from the kernel Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 16/40] tools lib: Update copy of strtobool from the kernel sources Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 17/40] perf annotate: Introduce struct sym_hist_entry Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 18/40] perf annotate: Rename 'sum' to 'nr_samples' in struct sym_hist Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 19/40] perf hists: Pass perf_sample to __symbol__inc_addr_samples() Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 20/40] perf annotate: Store the sample period in each histogram bucket Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 21/40] perf annotate: Do not overwrite sample->period Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 22/40] perf annotate stdio: Fix --show-total-period Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 23/40] perf report: Fix kernel symbol adjustment for s390x Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 24/40] perf cgroup: Fix refcount usage Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 25/40] perf tools: Add EXCLUDE_EXTLIBS and EXTRA_PERFLIBS to makefile Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 26/40] perf annotate: Process tracing data in pipe mode Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 27/40] perf jvmti: Fix linker error when libelf config is disabled Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 28/40] perf evsel: Add verbose output for sys_perf_event_open fallback Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 29/40] perf top: Support lookup of symbols in other mount namespaces Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 30/40] perf script: Remove some bogus error handling Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 31/40] perf script python: Allocate memory only if handler exists Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 32/40] perf script python: Refactor creation of perf sample dict Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 33/40] perf script python: Add sample_read to dict Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 34/40] perf script python: Add perf_sample dict to tracepoint handlers Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 35/40] perf script python: Generate hooks with additional argument Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 36/40] perf report: Make --branch-history work without callgraphs(-g) option in perf record Arnaldo Carvalho de Melo
2017-07-26 14:17 ` Arnaldo Carvalho de Melo [this message]
2017-07-26 14:17 ` [PATCH 38/40] perf jevents: Make build fail on JSON parse error Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 39/40] perf annotate stdio: Fix column header when using --show-total-period Arnaldo Carvalho de Melo
2017-07-26 14:17 ` [PATCH 40/40] perf tools: Add tools/include/uapi/asm-generic/fcntl.h to the MANIFEST Arnaldo Carvalho de Melo
2017-07-26 17:10 ` [GIT PULL 00/40] perf/core improvements and fixes Ingo Molnar
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=20170726141730.30750-38-acme@kernel.org \
--to=acme@kernel.org \
--cc=acme@redhat.com \
--cc=alexander.shishkin@linux.intel.com \
--cc=jolsa@kernel.org \
--cc=kan.liang@intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=peterz@infradead.org \
--cc=yao.jin@linux.intel.com \
/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
Powered by JetHome