Skip to contentSkip to conversation

perf, tools: Add support for PMU events in JSON format

PATCH Superseded Sukadev Bhattiprolu 5 participants 70 messages Sep 15, 2016 – Oct 4, 2016 perf events

lore
Share & tools
lore
#0 lore raw

v21 cover perf, tools: Add support for PMU events in JSON format

CPUs support a large number of performance monitoring events (PMU events) and often these events are very specific to an architecture/model of the CPU. To use most of these PMU events with perf, we currently have to identify them by their raw codes:

perf stat -e r100f2 sleep 1

This patchset allows architectures to specify these PMU events in JSON files located in 'tools/perf/pmu-events/arch/' of the mainline tree. The events from the JSON files for the architecture are then built into the perf binary.

At run time, perf identifies the specific set of events for the CPU and creates "event aliases". These aliases allow users to specify events by "name" as:

perf stat -e pm_1plus_ppc_cmpl sleep 1

The file, 'tools/perf/pmu-events/README' in [PATCH 15/19] gives more details.

Note:
- All known events tables for the architecture are included in the perf binary.

- For architectures that don't have any JSON files, an empty mapping table is created and they should continue to build.

Thanks to input from Andi Kleen, Jiri Olsa, Namhyung Kim and Ingo Molnar.

These patches are available from:

https://github.com/sukadev/linux.git

Branch Description
------------------------------------------------------
json-code-v21 Source Code only
json-code+data-v21 Both code and data (for build/test/pull)

NOTE: Only "source code" patches (i.e those in json-code-v21) are being
emailed. Please pull the json-code+data-v21 branch for build/test.

Changelog[v21]
- Rebase to recent perf/core
- Group the PMU events supported by a CPU model into topics and create a separate JSON file for each topic for each CPU (code and input from Jiri Olsa).

Changelog[v20]
- Rebase to recent perf/core
- Add Patch 20/20 to allow perf-stat to work with the period= field

Changelog[v19]
Rebase to recent perf/core; fix couple lines >80 chars.

Changelog[v18]
Rebase to recent perf/core; fix minor merge conflicts.

Changelog[v17]
Rebase to recent perf/core; couple of small fixes to processing Intel JSON files; allow case-insensitive PMU event names.

Changelog[v16]
Rebase to recent perf/core; fix minor merge conflicts; drop 3 patches that were merged into perf/core.

Changelog[v15]
Code changes:
- Fix 'perf list' usage string and update man page.
- Remove a redundant __maybe_unused tag.
- Rebase to recent perf/core branch.

Data files updates: json-files-5 branch
- Rebase to perf/intel-json-files-5 from Andi Kleen
- Add patch from Madhavan Srinivasan for couple more Powerpc models

Changelog[v14]
Comments from Jiri Olsa:
- Change parameter name/type for pmu_add_cpu_aliases (from void *data to list_head *head)
- Use asprintf() in file_name_to_tablename() and simplify/reorg code.
- Use __weak definition from <linux/compile.h>
- Use fopen() with mode "w" and eliminate unlink()
- Remove minor TODO.
- Add error check for return value from strdup() in print_pmu_events().
- Move independent changes from patches 3,11,12 .. to separate patches for easier review/backport.
- Clarify mapfile's "header line support" in patch description.
- Fix build failure with DEBUG=1

Comment from Andi Kleen:
- In tools/perf/pmu-events/Build, check for 'mapfile.csv' rather than 'mapfile*'

Misc:
- Minor changes/clarifications to tools/perf/pmu-events/README.

Changelog[v13]
Version: Individual patches have their own history :-) that I am preserving. Patchset version (v13) is for overall patchset and is somewhat arbitrary.

- Added support for "categories" of events to perf
- Add mapfile, jevents build dependency on pmu-events.c
- Silence jevents when parsing JSON files unless V=1 is specified
- Cleanup error messages
- Fix memory leak with ->cpuid
- Rebase to Arnaldo's tree
- Allow overriding CPUID via environment variable
- Support long descriptions for events
- Handle header line in mapfile.csv
- Cleanup JSON files (trim PublicDescription if identical to/prefix of BriefDescription field)

Andi Kleen (12):
perf, tools: Add jsmn `jasmine' JSON parser
perf, tools, jevents: Program to convert JSON file to C style file perf, tools: Support CPU id matching for x86 v2
perf, tools: Support alias descriptions
perf, tools: Query terminal width and use in perf list perf, tools: Add a --no-desc flag to perf list
perf, tools: Add override support for event list CPUID perf, tools: Add support for event list topics
perf, tools, jevents: Handle header line in mapfile perf, tools: Make alias matching case-insensitive
perf, tools, pmu-events: Fix fixed counters on Intel perf, tools, pmu-events: Add Skylake frontend MSR support

Sukadev Bhattiprolu (7):
perf, tools: Use pmu_events table to create aliases perf, tools: Support CPU ID matching for Powerpc perf, tools, jevents: Add support for long descriptions perf, tools: Add alias support for long descriptions perf, tools: Support long descriptions with perf list perf, tools: Add README for info on parsing JSON/map files Allow period= in perf stat CPU event descriptions.

 tools/lib/subcmd/pager.c               |  16 +
 tools/lib/subcmd/pager.h               |   1 +
 tools/perf/Documentation/perf-list.txt |  12 +-
 tools/perf/Makefile.perf               |  28 +-
 tools/perf/arch/powerpc/util/header.c  |  11 +
 tools/perf/arch/x86/util/header.c      |  24 +-
 tools/perf/builtin-list.c              |  20 +-
 tools/perf/pmu-events/Build            |  11 +
 tools/perf/pmu-events/README           | 148 ++++++
 tools/perf/pmu-events/jevents.c        | 811 +++++++++++++++++++++++++++++++++
 tools/perf/pmu-events/jevents.h        |  18 +
 tools/perf/pmu-events/jsmn.c           | 313 +++++++++++++
 tools/perf/pmu-events/jsmn.h           |  67 +++
 tools/perf/pmu-events/json.c           | 162 +++++++
 tools/perf/pmu-events/json.h           |  42 ++
 tools/perf/pmu-events/pmu-events.h     |  37 ++
 tools/perf/util/header.h               |   1 +
 tools/perf/util/parse-events.c         |   8 +-
 tools/perf/util/parse-events.h         |   3 +-
 tools/perf/util/pmu.c                  | 177 ++++++-
 tools/perf/util/pmu.h                  |   6 +-
 21 files changed, 1880 insertions(+), 36 deletions(-)
 create mode 100644 tools/perf/pmu-events/Build
 create mode 100644 tools/perf/pmu-events/README
 create mode 100644 tools/perf/pmu-events/jevents.c
 create mode 100644 tools/perf/pmu-events/jevents.h
 create mode 100644 tools/perf/pmu-events/jsmn.c
 create mode 100644 tools/perf/pmu-events/jsmn.h
 create mode 100644 tools/perf/pmu-events/json.c
 create mode 100644 tools/perf/pmu-events/json.h
 create mode 100644 tools/perf/pmu-events/pmu-events.h
Signature · 1 line
1.8.3.1
#1 lore raw

PATCH v21 1/19 perf, tools: Add jsmn `jasmine' JSON parser

From: Andi Kleen <[email protected]>

I need a JSON parser. This adds the simplest JSON parser I could find -- Serge Zaitsev's jsmn `jasmine' -- to the perf library. I merely converted it to (mostly) Linux style and added support for non 0 terminated input.

The parser is quite straight forward and does not copy any data, just returns tokens with offsets into the input buffer. So it's relatively efficient and simple to use.

The code is not fully checkpatch clean, but I didn't want to completely fork the upstream code.

Original source: http://zserge.bitbucket.org/jsmn.html

In addition I added a simple wrapper that mmaps a json file and provides some straight forward access functions.

Used in follow-on patches to parse event files.

---
v2: Address review feedback.
v3: Minor checkpatch fixes.
v4 (by Sukadev Bhattiprolu)
- Rebase to 4.0 and fix minor conflicts in tools/perf/Makefile.perf
- Report error if specified events file is invalid.
v5 (Sukadev Bhattiprolu)
- Move files to tools/perf/pmu-events/ since parsing of JSON file now occurs when _building_ rather than running perf.

---
 tools/perf/pmu-events/jsmn.c | 313 +++++++++++++++++++++++++++++++++++++++++++
 tools/perf/pmu-events/jsmn.h |  67 +++++++++
 tools/perf/pmu-events/json.c | 162 ++++++++++++++++++++++
 tools/perf/pmu-events/json.h |  36 +++++
 4 files changed, 578 insertions(+)
 create mode 100644 tools/perf/pmu-events/jsmn.c
 create mode 100644 tools/perf/pmu-events/jsmn.h
 create mode 100644 tools/perf/pmu-events/json.c
 create mode 100644 tools/perf/pmu-events/json.h
tools/perf/pmu-events/jsmn.c new file +313 -0
new file mode 100644
index 0000000..11d1fa1
@@ -0,0 +1,313 @@
+/*
+ * Copyright (c) 2010 Serge A. Zaitsev
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ *
+ * Slightly modified by AK to not assume 0 terminated input.
+ */
+
+#include <stdlib.h>
+#include "jsmn.h"
+
+/*
+ * Allocates a fresh unused token from the token pool.
+ */
+static jsmntok_t *jsmn_alloc_token(jsmn_parser *parser,
+				   jsmntok_t *tokens, size_t num_tokens)
+{
+	jsmntok_t *tok;
+
+	if ((unsigned)parser->toknext >= num_tokens)
+		return NULL;
+	tok = &tokens[parser->toknext++];
+	tok->start = tok->end = -1;
+	tok->size = 0;
+	return tok;
+}
+
+/*
+ * Fills token type and boundaries.
+ */
+static void jsmn_fill_token(jsmntok_t *token, jsmntype_t type,
+			    int start, int end)
+{
+	token->type = type;
+	token->start = start;
+	token->end = end;
+	token->size = 0;
+}
+
+/*
+ * Fills next available token with JSON primitive.
+ */
+static jsmnerr_t jsmn_parse_primitive(jsmn_parser *parser, const char *js,
+				      size_t len,
+				      jsmntok_t *tokens, size_t num_tokens)
+{
+	jsmntok_t *token;
+	int start;
+
+	start = parser->pos;
+
+	for (; parser->pos < len; parser->pos++) {
+		switch (js[parser->pos]) {
+#ifndef JSMN_STRICT
+		/*
+		 * In strict mode primitive must be followed by ","
+		 * or "}" or "]"
+		 */
+		case ':':
+#endif
+		case '\t':
+		case '\r':
+		case '\n':
+		case ' ':
+		case ',':
+		case ']':
+		case '}':
+			goto found;
+		default:
+			break;
+		}
+		if (js[parser->pos] < 32 || js[parser->pos] >= 127) {
+			parser->pos = start;
+			return JSMN_ERROR_INVAL;
+		}
+	}
+#ifdef JSMN_STRICT
+	/*
+	 * In strict mode primitive must be followed by a
+	 * comma/object/array.
+	 */
+	parser->pos = start;
+	return JSMN_ERROR_PART;
+#endif
+
+found:
+	token = jsmn_alloc_token(parser, tokens, num_tokens);
+	if (token == NULL) {
+		parser->pos = start;
+		return JSMN_ERROR_NOMEM;
+	}
+	jsmn_fill_token(token, JSMN_PRIMITIVE, start, parser->pos);
+	parser->pos--; /* parent sees closing brackets */
+	return JSMN_SUCCESS;
+}
+
+/*
+ * Fills next token with JSON string.
+ */
+static jsmnerr_t jsmn_parse_string(jsmn_parser *parser, const char *js,
+				   size_t len,
+				   jsmntok_t *tokens, size_t num_tokens)
+{
+	jsmntok_t *token;
+	int start = parser->pos;
+
+	/* Skip starting quote */
+	parser->pos++;
+
+	for (; parser->pos < len; parser->pos++) {
+		char c = js[parser->pos];
+
+		/* Quote: end of string */
+		if (c == '\"') {
+			token = jsmn_alloc_token(parser, tokens, num_tokens);
+			if (token == NULL) {
+				parser->pos = start;
+				return JSMN_ERROR_NOMEM;
+			}
+			jsmn_fill_token(token, JSMN_STRING, start+1,
+					parser->pos);
+			return JSMN_SUCCESS;
+		}
+
+		/* Backslash: Quoted symbol expected */
+		if (c == '\\') {
+			parser->pos++;
+			switch (js[parser->pos]) {
+				/* Allowed escaped symbols */
+			case '\"':
+			case '/':
+			case '\\':
+			case 'b':
+			case 'f':
+			case 'r':
+			case 'n':
+			case 't':
+				break;
+				/* Allows escaped symbol \uXXXX */
+			case 'u':
+				/* TODO */
+				break;
+				/* Unexpected symbol */
+			default:
+				parser->pos = start;
+				return JSMN_ERROR_INVAL;
+			}
+		}
+	}
+	parser->pos = start;
+	return JSMN_ERROR_PART;
+}
+
+/*
+ * Parse JSON string and fill tokens.
+ */
+jsmnerr_t jsmn_parse(jsmn_parser *parser, const char *js, size_t len,
+		     jsmntok_t *tokens, unsigned int num_tokens)
+{
+	jsmnerr_t r;
+	int i;
+	jsmntok_t *token;
+
+	for (; parser->pos < len; parser->pos++) {
+		char c;
+		jsmntype_t type;
+
+		c = js[parser->pos];
+		switch (c) {
+		case '{':
+		case '[':
+			token = jsmn_alloc_token(parser, tokens, num_tokens);
+			if (token == NULL)
+				return JSMN_ERROR_NOMEM;
+			if (parser->toksuper != -1)
+				tokens[parser->toksuper].size++;
+			token->type = (c == '{' ? JSMN_OBJECT : JSMN_ARRAY);
+			token->start = parser->pos;
+			parser->toksuper = parser->toknext - 1;
+			break;
+		case '}':
+		case ']':
+			type = (c == '}' ? JSMN_OBJECT : JSMN_ARRAY);
+			for (i = parser->toknext - 1; i >= 0; i--) {
+				token = &tokens[i];
+				if (token->start != -1 && token->end == -1) {
+					if (token->type != type)
+						return JSMN_ERROR_INVAL;
+					parser->toksuper = -1;
+					token->end = parser->pos + 1;
+					break;
+				}
+			}
+			/* Error if unmatched closing bracket */
+			if (i == -1)
+				return JSMN_ERROR_INVAL;
+			for (; i >= 0; i--) {
+				token = &tokens[i];
+				if (token->start != -1 && token->end == -1) {
+					parser->toksuper = i;
+					break;
+				}
+			}
+			break;
+		case '\"':
+			r = jsmn_parse_string(parser, js, len, tokens,
+					      num_tokens);
+			if (r < 0)
+				return r;
+			if (parser->toksuper != -1)
+				tokens[parser->toksuper].size++;
+			break;
+		case '\t':
+		case '\r':
+		case '\n':
+		case ':':
+		case ',':
+		case ' ':
+			break;
+#ifdef JSMN_STRICT
+			/*
+			 * In strict mode primitives are:
+			 * numbers and booleans.
+			 */
+		case '-':
+		case '0':
+		case '1':
+		case '2':
+		case '3':
+		case '4':
+		case '5':
+		case '6':
+		case '7':
+		case '8':
+		case '9':
+		case 't':
+		case 'f':
+		case 'n':
+#else
+			/*
+			 * In non-strict mode every unquoted value
+			 * is a primitive.
+			 */
+			/*FALL THROUGH */
+		default:
+#endif
+			r = jsmn_parse_primitive(parser, js, len, tokens,
+						 num_tokens);
+			if (r < 0)
+				return r;
+			if (parser->toksuper != -1)
+				tokens[parser->toksuper].size++;
+			break;
+
+#ifdef JSMN_STRICT
+			/* Unexpected char in strict mode */
+		default:
+			return JSMN_ERROR_INVAL;
+#endif
+		}
+	}
+
+	for (i = parser->toknext - 1; i >= 0; i--) {
+		/* Unmatched opened object or array */
+		if (tokens[i].start != -1 && tokens[i].end == -1)
+			return JSMN_ERROR_PART;
+	}
+
+	return JSMN_SUCCESS;
+}
+
+/*
+ * Creates a new parser based over a given  buffer with an array of tokens
+ * available.
+ */
+void jsmn_init(jsmn_parser *parser)
+{
+	parser->pos = 0;
+	parser->toknext = 0;
+	parser->toksuper = -1;
+}
+
+const char *jsmn_strerror(jsmnerr_t err)
+{
+	switch (err) {
+	case JSMN_ERROR_NOMEM:
+		return "No enough tokens";
+	case JSMN_ERROR_INVAL:
+		return "Invalid character inside JSON string";
+	case JSMN_ERROR_PART:
+		return "The string is not a full JSON packet, more bytes expected";
+	case JSMN_SUCCESS:
+		return "Success";
+	default:
+		return "Unknown json error";
+	}
+}
tools/perf/pmu-events/jsmn.h new file +67 -0

67 lines not on this page · raw message

tools/perf/pmu-events/json.c new file +162 -0

162 lines not on this page · raw message

tools/perf/pmu-events/json.h new file +36 -0
new file mode 100644
index 0000000..6b8337e
@@ -0,0 +1,36 @@
+#ifndef JSON_H
+#define JSON_H 1
+
+#include "jsmn.h"
+
+jsmntok_t *parse_json(const char *fn, char **map, size_t *size, int *len);
+void free_json(char *map, size_t size, jsmntok_t *tokens);
+int json_line(char *map, jsmntok_t *t);
+const char *json_name(jsmntok_t *t);
+int json_streq(char *map, jsmntok_t *t, const char *s);
+int json_len(jsmntok_t *t);
+
+extern int verbose;
+
+typedef unsigned int bool;
+
+#ifndef true
+#define	true 1
+#endif
+
+extern int eprintf(int level, int var, const char *fmt, ...);
+#define pr_fmt(fmt)	fmt
+
+#define pr_err(fmt, ...) \
+	eprintf(0, verbose, pr_fmt(fmt), ##__VA_ARGS__)
+
+#ifndef roundup
+#define roundup(x, y) (                                \
+{                                                      \
+        const typeof(y) __y = y;                       \
+        (((x) + (__y - 1)) / __y) * __y;               \
+}                                                      \
+)
+#endif
+
+#endif
Signature · 1 line
1.8.3.1
#2 lore raw

[tip:perf/urgent] perf tools: Add jsmn `jasmine' JSON parser

perf tools: Add jsmn `jasmine' JSON parser
Commit-ID:  867a979a83d38fb82c4e7534a0281a02ef1700a3
Gitweb:     http://git.kernel.org/tip/867a979a83d38fb82c4e7534a0281a02ef1700a3
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:38 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 11:40:36 -0300

perf tools: Add jsmn `jasmine' JSON parser

I need a JSON parser. This adds the simplest JSON parser I could find -- Serge Zaitsev's jsmn `jasmine' -- to the perf library. I merely converted it to (mostly) Linux style and added support for non 0 terminated input.

The parser is quite straight forward and does not copy any data, just returns tokens with offsets into the input buffer. So it's relatively efficient and simple to use.

The code is not fully checkpatch clean, but I didn't want to completely fork the upstream code.

Original source: http://zserge.bitbucket.org/jsmn.html

In addition I added a simple wrapper that mmaps a json file and provides some straight forward access functions.

Used in follow-on patches to parse event files.

[ Use fcntl.h instead of sys/fcntl.h to fix the build on Alpine Linux 3.4/musl libc,
use stdbool.h to avoid clashing with 'bool' typedef there ]

---
 tools/perf/pmu-events/jsmn.c | 313 +++++++++++++++++++++++++++++++++++++++++++
 tools/perf/pmu-events/jsmn.h |  67 +++++++++
 tools/perf/pmu-events/json.c | 162 ++++++++++++++++++++++
 tools/perf/pmu-events/json.h |  32 +++++
 4 files changed, 574 insertions(+)
tools/perf/pmu-events/jsmn.c new file +313 -0
new file mode 100644
index 0000000..11d1fa1
@@ -0,0 +1,313 @@
+/*
+ * Copyright (c) 2010 Serge A. Zaitsev
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a copy
+ * of this software and associated documentation files (the "Software"), to deal
+ * in the Software without restriction, including without limitation the rights
+ * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+ * copies of the Software, and to permit persons to whom the Software is
+ * furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+ * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+ * THE SOFTWARE.
+ *
+ * Slightly modified by AK to not assume 0 terminated input.
+ */
+
+#include <stdlib.h>
+#include "jsmn.h"
+
+/*
+ * Allocates a fresh unused token from the token pool.
+ */
+static jsmntok_t *jsmn_alloc_token(jsmn_parser *parser,
+				   jsmntok_t *tokens, size_t num_tokens)
+{
+	jsmntok_t *tok;
+
+	if ((unsigned)parser->toknext >= num_tokens)
+		return NULL;
+	tok = &tokens[parser->toknext++];
+	tok->start = tok->end = -1;
+	tok->size = 0;
+	return tok;
+}
+
+/*
+ * Fills token type and boundaries.
+ */
+static void jsmn_fill_token(jsmntok_t *token, jsmntype_t type,
+			    int start, int end)
+{
+	token->type = type;
+	token->start = start;
+	token->end = end;
+	token->size = 0;
+}
+
+/*
+ * Fills next available token with JSON primitive.
+ */
+static jsmnerr_t jsmn_parse_primitive(jsmn_parser *parser, const char *js,
+				      size_t len,
+				      jsmntok_t *tokens, size_t num_tokens)
+{
+	jsmntok_t *token;
+	int start;
+
+	start = parser->pos;
+
+	for (; parser->pos < len; parser->pos++) {
+		switch (js[parser->pos]) {
+#ifndef JSMN_STRICT
+		/*
+		 * In strict mode primitive must be followed by ","
+		 * or "}" or "]"
+		 */
+		case ':':
+#endif
+		case '\t':
+		case '\r':
+		case '\n':
+		case ' ':
+		case ',':
+		case ']':
+		case '}':
+			goto found;
+		default:
+			break;
+		}
+		if (js[parser->pos] < 32 || js[parser->pos] >= 127) {
+			parser->pos = start;
+			return JSMN_ERROR_INVAL;
+		}
+	}
+#ifdef JSMN_STRICT
+	/*
+	 * In strict mode primitive must be followed by a
+	 * comma/object/array.
+	 */
+	parser->pos = start;
+	return JSMN_ERROR_PART;
+#endif
+
+found:
+	token = jsmn_alloc_token(parser, tokens, num_tokens);
+	if (token == NULL) {
+		parser->pos = start;
+		return JSMN_ERROR_NOMEM;
+	}
+	jsmn_fill_token(token, JSMN_PRIMITIVE, start, parser->pos);
+	parser->pos--; /* parent sees closing brackets */
+	return JSMN_SUCCESS;
+}
+
+/*
+ * Fills next token with JSON string.
+ */
+static jsmnerr_t jsmn_parse_string(jsmn_parser *parser, const char *js,
+				   size_t len,
+				   jsmntok_t *tokens, size_t num_tokens)
+{
+	jsmntok_t *token;
+	int start = parser->pos;
+
+	/* Skip starting quote */
+	parser->pos++;
+
+	for (; parser->pos < len; parser->pos++) {
+		char c = js[parser->pos];
+
+		/* Quote: end of string */
+		if (c == '\"') {
+			token = jsmn_alloc_token(parser, tokens, num_tokens);
+			if (token == NULL) {
+				parser->pos = start;
+				return JSMN_ERROR_NOMEM;
+			}
+			jsmn_fill_token(token, JSMN_STRING, start+1,
+					parser->pos);
+			return JSMN_SUCCESS;
+		}
+
+		/* Backslash: Quoted symbol expected */
+		if (c == '\\') {
+			parser->pos++;
+			switch (js[parser->pos]) {
+				/* Allowed escaped symbols */
+			case '\"':
+			case '/':
+			case '\\':
+			case 'b':
+			case 'f':
+			case 'r':
+			case 'n':
+			case 't':
+				break;
+				/* Allows escaped symbol \uXXXX */
+			case 'u':
+				/* TODO */
+				break;
+				/* Unexpected symbol */
+			default:
+				parser->pos = start;
+				return JSMN_ERROR_INVAL;
+			}
+		}
+	}
+	parser->pos = start;
+	return JSMN_ERROR_PART;
+}
+
+/*
+ * Parse JSON string and fill tokens.
+ */
+jsmnerr_t jsmn_parse(jsmn_parser *parser, const char *js, size_t len,
+		     jsmntok_t *tokens, unsigned int num_tokens)
+{
+	jsmnerr_t r;
+	int i;
+	jsmntok_t *token;
+
+	for (; parser->pos < len; parser->pos++) {
+		char c;
+		jsmntype_t type;
+
+		c = js[parser->pos];
+		switch (c) {
+		case '{':
+		case '[':
+			token = jsmn_alloc_token(parser, tokens, num_tokens);
+			if (token == NULL)
+				return JSMN_ERROR_NOMEM;
+			if (parser->toksuper != -1)
+				tokens[parser->toksuper].size++;
+			token->type = (c == '{' ? JSMN_OBJECT : JSMN_ARRAY);
+			token->start = parser->pos;
+			parser->toksuper = parser->toknext - 1;
+			break;
+		case '}':
+		case ']':
+			type = (c == '}' ? JSMN_OBJECT : JSMN_ARRAY);
+			for (i = parser->toknext - 1; i >= 0; i--) {
+				token = &tokens[i];
+				if (token->start != -1 && token->end == -1) {
+					if (token->type != type)
+						return JSMN_ERROR_INVAL;
+					parser->toksuper = -1;
+					token->end = parser->pos + 1;
+					break;
+				}
+			}
+			/* Error if unmatched closing bracket */
+			if (i == -1)
+				return JSMN_ERROR_INVAL;
+			for (; i >= 0; i--) {
+				token = &tokens[i];
+				if (token->start != -1 && token->end == -1) {
+					parser->toksuper = i;
+					break;
+				}
+			}
+			break;
+		case '\"':
+			r = jsmn_parse_string(parser, js, len, tokens,
+					      num_tokens);
+			if (r < 0)
+				return r;
+			if (parser->toksuper != -1)
+				tokens[parser->toksuper].size++;
+			break;
+		case '\t':
+		case '\r':
+		case '\n':
+		case ':':
+		case ',':
+		case ' ':
+			break;
+#ifdef JSMN_STRICT
+			/*
+			 * In strict mode primitives are:
+			 * numbers and booleans.
+			 */
+		case '-':
+		case '0':
+		case '1':
+		case '2':
+		case '3':
+		case '4':
+		case '5':
+		case '6':
+		case '7':
+		case '8':
+		case '9':
+		case 't':
+		case 'f':
+		case 'n':
+#else
+			/*
+			 * In non-strict mode every unquoted value
+			 * is a primitive.
+			 */
+			/*FALL THROUGH */
+		default:
+#endif
+			r = jsmn_parse_primitive(parser, js, len, tokens,
+						 num_tokens);
+			if (r < 0)
+				return r;
+			if (parser->toksuper != -1)
+				tokens[parser->toksuper].size++;
+			break;
+
+#ifdef JSMN_STRICT
+			/* Unexpected char in strict mode */
+		default:
+			return JSMN_ERROR_INVAL;
+#endif
+		}
+	}
+
+	for (i = parser->toknext - 1; i >= 0; i--) {
+		/* Unmatched opened object or array */
+		if (tokens[i].start != -1 && tokens[i].end == -1)
+			return JSMN_ERROR_PART;
+	}
+
+	return JSMN_SUCCESS;
+}
+
+/*
+ * Creates a new parser based over a given  buffer with an array of tokens
+ * available.
+ */
+void jsmn_init(jsmn_parser *parser)
+{
+	parser->pos = 0;
+	parser->toknext = 0;
+	parser->toksuper = -1;
+}
+
+const char *jsmn_strerror(jsmnerr_t err)
+{
+	switch (err) {
+	case JSMN_ERROR_NOMEM:
+		return "No enough tokens";
+	case JSMN_ERROR_INVAL:
+		return "Invalid character inside JSON string";
+	case JSMN_ERROR_PART:
+		return "The string is not a full JSON packet, more bytes expected";
+	case JSMN_SUCCESS:
+		return "Success";
+	default:
+		return "Unknown json error";
+	}
+}
tools/perf/pmu-events/jsmn.h new file +67 -0

67 lines not on this page · raw message

tools/perf/pmu-events/json.c new file +162 -0

162 lines not on this page · raw message

tools/perf/pmu-events/json.h new file +32 -0
new file mode 100644
index 0000000..f2745c7
@@ -0,0 +1,32 @@
+#ifndef JSON_H
+#define JSON_H 1
+
+#include "jsmn.h"
+
+jsmntok_t *parse_json(const char *fn, char **map, size_t *size, int *len);
+void free_json(char *map, size_t size, jsmntok_t *tokens);
+int json_line(char *map, jsmntok_t *t);
+const char *json_name(jsmntok_t *t);
+int json_streq(char *map, jsmntok_t *t, const char *s);
+int json_len(jsmntok_t *t);
+
+extern int verbose;
+
+#include <stdbool.h>
+
+extern int eprintf(int level, int var, const char *fmt, ...);
+#define pr_fmt(fmt)	fmt
+
+#define pr_err(fmt, ...) \
+	eprintf(0, verbose, pr_fmt(fmt), ##__VA_ARGS__)
+
+#ifndef roundup
+#define roundup(x, y) (                                \
+{                                                      \
+        const typeof(y) __y = y;                       \
+        (((x) + (__y - 1)) / __y) * __y;               \
+}                                                      \
+)
+#endif
+
+#endif
#3 lore raw

PATCH v21 2/19 perf, tools, jevents: Program to convert JSON file to C style file

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

This is a modified version of an earlier patch by Andi Kleen.

We expect architectures to create JSON files describing the performance monitoring (PMU) events that each CPU model/family of the architecture supports.

Following is an example of the JSON file entry for an x86 event:

    	[
    	...
    	{
    	"EventCode": "0x00",
    	"UMask": "0x01",
    	"EventName": "INST_RETIRED.ANY",
    	"BriefDescription": "Instructions retired from execution.",
    	"PublicDescription": "Instructions retired from execution.",
    	"Counter": "Fixed counter 1",
    	"CounterHTOff": "Fixed counter 1",
    	"SampleAfterValue": "2000003",
    	"SampleAfterValue": "2000003",
    	"MSRIndex": "0",
    	"MSRValue": "0",
    	"TakenAlone": "0",
    	"CounterMask": "0",
    	"Invert": "0",
    	"AnyThread": "0",
    	"EdgeDetect": "0",
    	"PEBS": "0",
    	"PRECISE_STORE": "0",
    	"Errata": "null",
    	"Offcore": "0"
    	},
    	...

    	]

All the PMU events supported by a CPU model/family must be grouped into "topics" such as "Piplelining", "Floating-point", "Virtual-memory" etc.

All events belonging to a topic must be placed in a separate JSON file (eg: "Pipeling.json") and all the topic JSON files for a CPU model must be in a separate directory.

Eg: for the CPU model "Silvermont_core":

$ ls tools/perf/pmu-events/arch/x86/Silvermont_core Floating-point.json
Memory.json
Other.json
Pipelining.json
Virtualmemory.json

Finally, to allow multiple CPU models to share a single set of JSON files, architectures must provide a mapping between a model and its set of events:

    	$ grep Silvermont tools/perf/pmu-events/arch/x86/mapfile.csv
    	GenuineIntel-6-4D,V13,Silvermont_core,core
    	GenuineIntel-6-4C,V13,Silvermont_core,core

which maps each CPU, identified by [vendor, family, model, version, type] to a directory of JSON files. Thus two (or more) CPU models support the set of PMU events listed in the directory.

tools/perf/pmu-events/arch/x86/Silvermont_core/

Given this organization of files, the program, jevents:

- locates all JSON files for each CPU-model of the architecture,

- parses all JSON files for the CPU-model and generates a C-style "PMU-events table" (pmu-events.c) for the model

- locates a mapfile for the architecture

- builds a global table, mapping each model of CPU to the corresponding PMU-events table.

The 'pmu-events.c' is generated when building perf and added to libperf.a. The global table pmu_events_map[] table in this pmu-events.c will be used in perf in a follow-on patch.

If the architecture does not have any JSON files or there is an error in processing them, an empty mapping file is created. This would allow the build of perf to proceed even if we are not able to provide aliases for events.

The parser for JSON files allows parsing Intel style JSON event files. This allows to use an Intel event list directly with perf. The Intel event lists can be quite large and are too big to store in unswappable kernel memory.

The conversion from JSON to C-style is straight forward. The parser knows (very little) Intel specific information, and can be easily extended to handle fields for other CPUs.

The parser code is partially shared with an independent parsing library, which is 2-clause BSD licenced. To avoid any conflicts I marked those files as BSD licenced too. As part of perf they become GPLv2.

---
v2: Address review feedback. Rename option to --event-files
v3: Add JSON example
v4: Update manpages.
v5: Don't remove dot in fixname. Fix compile error. Add include
protection. Comment realloc.
v6: Include debug/util.h
v7: (Sukadev Bhattiprolu)
Rebase to 4.0 and fix some conflicts.
v8: (Sukadev Bhattiprolu)
Move jevents.[hc] to tools/perf/pmu-events/
Rewrite to locate and process arch specific JSON and "map" files; and generate a C file.
(Removed acked-by Namhyung Kim due to modest changes to patch) Compile the generated pmu-events.c and add the pmu-events.o to libperf.a
v9: [Sukadev Bhattiprolu/Andi Kleen] Rename ->vfm to ->cpuid and use
that field to encode the PVR in Power.
Allow blank lines in mapfile.
[Jiri Olsa] Pass ARCH as a parameter to jevents so we don't have to detect it.
[Jiri Olsa] Use the infrastrastructure to build pmu-events/perf (Makefile changes from Jiri included in this patch). [Jiri Olsa, Andi Kleen] Detect changes to JSON files and rebuild pmu-events.o only if necessary.

v11: - [Andi Kleen] Add mapfile, jevents dependency on pmu-events.c
- [Jiri Olsa] Be silient if arch doesn't have JSON files
- Also silence 'jevents' when parsing JSON files unless V=1 is specified during build. Cleanup error messages.

v14:- - [Jiri Olsa] Fix compile error with DEBUG=1; drop unlink() and

	  use "w" mode with fopen(); simplify file_name_to_table_name()

v15: - Fix minor conflict in tools/perf/Makefile.perf when rebasing
to recent perf/core.

v16: - Rebase to upstream; fix conflicts in tools/perf/Makefile.perf

v18: - Rebase to upstream; fix conflicts in tools/perf/Makefile.perf

v20: - Rebase to upstream; rename a local variable to 'ldirname' to avoid
collision with the dirname().

v21: - Breakup the large JSON files into separate topics like
Pipelining.json, Cache.json etc (by Jiri Olsa).
- Ensure BriefDescription field is non-null before adding extra description for PEBS events.

---
 tools/perf/Makefile.perf           |  28 +-
 tools/perf/pmu-events/Build        |  11 +
 tools/perf/pmu-events/jevents.c    | 766 +++++++++++++++++++++++++++++++++++++
 tools/perf/pmu-events/jevents.h    |  17 +
 tools/perf/pmu-events/json.h       |   6 +
 tools/perf/pmu-events/pmu-events.h |  36 ++
 6 files changed, 860 insertions(+), 4 deletions(-)
 create mode 100644 tools/perf/pmu-events/Build
 create mode 100644 tools/perf/pmu-events/jevents.c
 create mode 100644 tools/perf/pmu-events/jevents.h
 create mode 100644 tools/perf/pmu-events/pmu-events.h
tools/perf/Makefile.perf +24 -4
index 828cfd7..0abebcb 100644
@@ -347,6 +347,14 @@ PERF_IN := $(OUTPUT)perf-in.o
 export srctree OUTPUT RM CC LD AR CFLAGS V BISON FLEX AWK
 include $(srctree)/tools/build/Makefile.include
 
+JEVENTS       := $(OUTPUT)pmu-events/jevents
+JEVENTS_IN    := $(OUTPUT)pmu-events/jevents-in.o
+PMU_EVENTS_IN := $(OUTPUT)pmu-events/pmu-events-in.o
+
+export JEVENTS
+
+build := -f $(srctree)/tools/build/Makefile.build dir=. obj
+
 $(PERF_IN): prepare FORCE
 	@(test -f ../../include/uapi/linux/perf_event.h && ( \
         (diff -B ../include/uapi/linux/perf_event.h ../../include/uapi/linux/perf_event.h >/dev/null) \
@@ -434,9 +442,18 @@ $(PERF_IN): prepare FORCE
 	|| echo "Warning: tools/include/linux/coresight-pmu.h differs from kernel" >&2 )) || true
 	$(Q)$(MAKE) $(build)=perf
 
-$(OUTPUT)perf: $(PERFLIBS) $(PERF_IN) $(LIBTRACEEVENT_DYNAMIC_LIST)
+$(JEVENTS_IN): FORCE
+	$(Q)$(MAKE) -f $(srctree)/tools/build/Makefile.build dir=pmu-events obj=jevents
+
+$(JEVENTS): $(JEVENTS_IN)
+	$(QUIET_LINK)$(CC) $(JEVENTS_IN) -o $@
+
+$(PMU_EVENTS_IN): $(JEVENTS) FORCE
+	$(Q)$(MAKE) -f $(srctree)/tools/build/Makefile.build dir=pmu-events obj=pmu-events
+
+$(OUTPUT)perf: $(PERFLIBS) $(PERF_IN) $(PMU_EVENTS_IN) $(LIBTRACEEVENT_DYNAMIC_LIST)
 	$(QUIET_LINK)$(CC) $(CFLAGS) $(LDFLAGS) $(LIBTRACEEVENT_DYNAMIC_LIST_LDFLAGS) \
-		$(PERF_IN) $(LIBS) -o $@
+		$(PERF_IN) $(PMU_EVENTS_IN) $(LIBS) -o $@
 
 $(GTK_IN): fixdep FORCE
 	$(Q)$(MAKE) $(build)=gtk
@@ -465,6 +482,8 @@ perf.spec $(SCRIPTS) \
 ifneq ($(OUTPUT),)
 %.o: $(OUTPUT)%.o
 	@echo "    # Redirected target $@ => $(OUTPUT)$@"
+pmu-events/%.o: $(OUTPUT)pmu-events/%.o
+	@echo "    # Redirected target $@ => $(OUTPUT)$@"
 util/%.o: $(OUTPUT)util/%.o
 	@echo "    # Redirected target $@ => $(OUTPUT)$@"
 bench/%.o: $(OUTPUT)bench/%.o
@@ -720,10 +739,11 @@ clean:: $(LIBTRACEEVENT)-clean $(LIBAPI)-clean $(LIBBPF)-clean $(LIBSUBCMD)-clea
 	$(call QUIET_CLEAN, core-objs)  $(RM) $(LIB_FILE) $(OUTPUT)perf-archive $(OUTPUT)perf-with-kcore $(LANG_BINDINGS)
 	$(Q)find $(if $(OUTPUT),$(OUTPUT),.) -name '*.o' -delete -o -name '\.*.cmd' -delete -o -name '\.*.d' -delete
 	$(Q)$(RM) $(OUTPUT).config-detected
-	$(call QUIET_CLEAN, core-progs) $(RM) $(ALL_PROGRAMS) perf perf-read-vdso32 perf-read-vdsox32
+	$(call QUIET_CLEAN, core-progs) $(RM) $(ALL_PROGRAMS) perf perf-read-vdso32 perf-read-vdsox32 $(OUTPUT)pmu-events/jevents $(srctree)/tools/perf/pmu-events/pmu-events.c
 	$(call QUIET_CLEAN, core-gen)   $(RM)  *.spec *.pyc *.pyo */*.pyc */*.pyo $(OUTPUT)common-cmds.h TAGS tags cscope* $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)FEATURE-DUMP $(OUTPUT)util/*-bison* $(OUTPUT)util/*-flex* \
 		$(OUTPUT)util/intel-pt-decoder/inat-tables.c $(OUTPUT)fixdep \
-		$(OUTPUT)tests/llvm-src-{base,kbuild,prologue,relocation}.c
+		$(OUTPUT)tests/llvm-src-{base,kbuild,prologue,relocation}.c \
+		$(OUTPUT)pmu-events/pmu-events.c
 	$(QUIET_SUBDIR0)Documentation $(QUIET_SUBDIR1) clean
 	$(python-clean)
 
tools/perf/pmu-events/Build new file +11 -0
new file mode 100644
index 0000000..d2f3430
@@ -0,0 +1,11 @@
+jevents-y	+= json.o jsmn.o jevents.o
+pmu-events-y	+= pmu-events.o
+JDIR		=  pmu-events/arch/$(ARCH)
+JSON		=  $(shell [ -d $(JDIR) ] &&				\
+			find $(JDIR) -name '*.json' -o -name 'mapfile.csv')
+#
+# Locate/process JSON files in pmu-events/arch/
+# directory and create tables in pmu-events.c.
+#
+$(OUTPUT)pmu-events/pmu-events.c: $(JSON) $(JEVENTS)
+	$(Q)$(call echo-cmd,gen)$(JEVENTS) $(ARCH) pmu-events/arch $(OUTPUT)pmu-events/pmu-events.c $(V)
tools/perf/pmu-events/jevents.c new file +766 -0

766 lines not on this page · raw message

tools/perf/pmu-events/jevents.h new file +17 -0
new file mode 100644
index 0000000..996601f
@@ -0,0 +1,17 @@
+#ifndef JEVENTS_H
+#define JEVENTS_H 1
+
+int json_events(const char *fn,
+		int (*func)(void *data, char *name, char *event, char *desc),
+		void *data);
+char *get_cpu_str(void);
+
+#ifndef min
+#define min(x, y) ({                            \
+	typeof(x) _min1 = (x);                  \
+	typeof(y) _min2 = (y);                  \
+	(void) (&_min1 == &_min2);              \
+	_min1 < _min2 ? _min1 : _min2; })
+#endif
+
+#endif
tools/perf/pmu-events/json.h +6 -0
index 6b8337e..5339704 100644
@@ -24,6 +24,12 @@ extern int eprintf(int level, int var, const char *fmt, ...);
 #define pr_err(fmt, ...) \
 	eprintf(0, verbose, pr_fmt(fmt), ##__VA_ARGS__)
 
+#define pr_info(fmt, ...) \
+	eprintf(1, verbose, pr_fmt(fmt), ##__VA_ARGS__)
+
+#define pr_debug(fmt, ...) \
+	eprintf(2, verbose, pr_fmt(fmt), ##__VA_ARGS__)
+
 #ifndef roundup
 #define roundup(x, y) (                                \
 {                                                      \
tools/perf/pmu-events/pmu-events.h new file +36 -0
new file mode 100644
index 0000000..70d5479
@@ -0,0 +1,36 @@
+#ifndef PMU_EVENTS_H
+#define PMU_EVENTS_H
+
+/*
+ * Describe each PMU event. Each CPU has a table of PMU events.
+ */
+struct pmu_event {
+	const char *name;
+	const char *event;
+	const char *desc;
+	const char *topic;
+};
+
+/*
+ *
+ * Map a CPU to its table of PMU events. The CPU is identified by the
+ * cpuid field, which is an arch-specific identifier for the CPU.
+ * The identifier specified in tools/perf/pmu-events/arch/xxx/mapfile
+ * must match the get_cpustr() in tools/perf/arch/xxx/util/header.c)
+ *
+ * The  cpuid can contain any character other than the comma.
+ */
+struct pmu_events_map {
+	const char *cpuid;
+	const char *version;
+	const char *type;		/* core, uncore etc */
+	struct pmu_event *table;
+};
+
+/*
+ * Global table mapping each known CPU for the architecture to its
+ * table of PMU events.
+ */
+extern struct pmu_events_map pmu_events_map[];
+
+#endif
Signature · 1 line
1.8.3.1
#4 lore raw

PATCH v21 3/19 perf, tools: Use pmu_events table to create aliases

In reply to Sukadev Bhattiprolu

At run time (when 'perf' is starting up), locate the specific table of PMU events that corresponds to the current CPU. Using that table, create aliases for the each of the PMU events in the CPU. The use these aliases to parse the user specified perf event.

In short this would allow the user to specify events using their aliases rather than raw event codes.

Based on input and some earlier patches from Andi Kleen, Jiri Olsa.

---

Changelog[v4]
- Split off unrelated code into separate patches.
Changelog[v3]
- [Jiri Olsa] Fix memory leak in cpuid
Changelog[v2]
- [Andi Kleen] Replace pmu_events_map->vfm with a generic "cpuid".

---
 tools/perf/util/header.h |  1 +
 tools/perf/util/pmu.c    | 61 ++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 62 insertions(+)
tools/perf/util/header.h +1 -0
index d306ca1..d30109b 100644
@@ -151,4 +151,5 @@ int write_padded(int fd, const void *bf, size_t count, size_t count_aligned);
  */
 int get_cpuid(char *buffer, size_t sz);
 
+char *get_cpuid_str(void);
 #endif /* __PERF_HEADER_H */
tools/perf/util/pmu.c +61 -0

79 lines not on this page · raw message

Signature · 1 line
1.8.3.1
#5 lore raw

Re: [PATCH v21 03/19] perf, tools: Use pmu_events table to create aliases

Em Thu, Sep 15, 2016 at 03:24:40PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 9 lines

At run time (when 'perf' is starting up), locate the specific table of PMU events that corresponds to the current CPU. Using that table, create aliases for the each of the PMU events in the CPU. The use these aliases to parse the user specified perf event.

In short this would allow the user to specify events using their aliases rather than raw event codes.

Based on input and some earlier patches from Andi Kleen, Jiri Olsa.

this took a long time so I'll try to go chainsawing some stuff along the way, like why introduce a function that returns an error just to ignore it, even more using a ugly (void) cast?

- Arnaldo

Quoted 114 lines

---

Changelog[v4]
- Split off unrelated code into separate patches.
Changelog[v3]
- [Jiri Olsa] Fix memory leak in cpuid
Changelog[v2]
- [Andi Kleen] Replace pmu_events_map->vfm with a generic "cpuid".

---
 tools/perf/util/header.h |  1 +
 tools/perf/util/pmu.c    | 61 ++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 62 insertions(+)
tools/perf/util/header.h +1 -0
index d306ca1..d30109b 100644
@@ -151,4 +151,5 @@ int write_padded(int fd, const void *bf, size_t count, size_t count_aligned);
  */
 int get_cpuid(char *buffer, size_t sz);
 
+char *get_cpuid_str(void);
 #endif /* __PERF_HEADER_H */
tools/perf/util/pmu.c +61 -0
index 2babcdf..c842886 100644
@@ -12,6 +12,8 @@
 #include "pmu.h"
 #include "parse-events.h"
 #include "cpumap.h"
+#include "header.h"
+#include "pmu-events/pmu-events.h"
 
 struct perf_pmu_format {
 	char *name;
@@ -473,6 +475,62 @@ static struct cpu_map *pmu_cpumask(const char *name)
 	return cpus;
 }
 
+/*
+ * Return the CPU id as a raw string.
+ *
+ * Each architecture should provide a more precise id string that
+ * can be use to match the architecture's "mapfile".
+ */
+char * __weak get_cpuid_str(void)
+{
+	return NULL;
+}
+
+/*
+ * From the pmu_events_map, find the table of PMU events that corresponds
+ * to the current running CPU. Then, add all PMU events from that table
+ * as aliases.
+ */
+static int pmu_add_cpu_aliases(struct list_head *head)
+{
+	int i;
+	struct pmu_events_map *map;
+	struct pmu_event *pe;
+	char *cpuid;
+
+	cpuid = get_cpuid_str();
+	if (!cpuid)
+		return 0;
+
+	i = 0;
+	while (1) {
+		map = &pmu_events_map[i++];
+		if (!map->table)
+			goto out;
+
+		if (!strcmp(map->cpuid, cpuid))
+			break;
+	}
+
+	/*
+	 * Found a matching PMU events table. Create aliases
+	 */
+	i = 0;
+	while (1) {
+		pe = &map->table[i++];
+		if (!pe->name)
+			break;
+
+		/* need type casts to override 'const' */
+		__perf_pmu__new_alias(head, NULL, (char *)pe->name,
+				(char *)pe->desc, (char *)pe->event);
+	}
+
+out:
+	free(cpuid);
+	return 0;
+}
+
 struct perf_event_attr * __weak
 perf_pmu__get_default_config(struct perf_pmu *pmu __maybe_unused)
 {
@@ -497,6 +555,9 @@ static struct perf_pmu *pmu_lookup(const char *name)
 	if (pmu_aliases(name, &aliases))
 		return NULL;
 
+	if (!strcmp(name, "cpu"))
+		(void)pmu_add_cpu_aliases(&aliases);
+
 	if (pmu_type(name, &type))
 		return NULL;
 
Signature · 1 line
1.8.3.1
#6 lore raw

[tip:perf/urgent] perf pmu: Use pmu_events table to create aliases

In reply to Sukadev Bhattiprolu
perf pmu: Use pmu_events table to create aliases
Commit-ID:  933f82ff72d7d1641663462f61f3056ee1fe3f8b
Gitweb:     http://git.kernel.org/tip/933f82ff72d7d1641663462f61f3056ee1fe3f8b
Author:     Sukadev Bhattiprolu <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:40 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 19:58:00 -0300

perf pmu: Use pmu_events table to create aliases

At run time (when 'perf' is starting up), locate the specific table of PMU events that corresponds to the current CPU. Using that table, create aliases for the each of the PMU events in the CPU. The use these aliases to parse the user specified perf event.

In short this would allow the user to specify events using their aliases rather than raw event codes.

Based on input and some earlier patches from Andi Kleen, Jiri Olsa.

[ Make pmu_add_cpu_aliases() return void, since it was returning just '0' and
furthermore, even that was being discarded via an explicit (void) cast ]

---
 tools/perf/util/header.h |  1 +
 tools/perf/util/pmu.c    | 60 ++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 61 insertions(+)
tools/perf/util/header.h +1 -0
index d306ca1..d30109b 100644
@@ -151,4 +151,5 @@ int write_padded(int fd, const void *bf, size_t count, size_t count_aligned);
  */
 int get_cpuid(char *buffer, size_t sz);
 
+char *get_cpuid_str(void);
 #endif /* __PERF_HEADER_H */
tools/perf/util/pmu.c +60 -0

78 lines not on this page · raw message

#7 lore raw

PATCH v21 4/19 perf, tools: Support CPU ID matching for Powerpc

In reply to Sukadev Bhattiprolu

Implement code that returns the generic CPU ID string for Powerpc. This will be used to identify the specific table of PMU events to parse/compare user specified events against.

---

Changelog[v14]
- [Jiri Olsa] Move this independent code off into a separate patch.

---
 tools/perf/arch/powerpc/util/header.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
tools/perf/arch/powerpc/util/header.c +11 -0
index f8ccee1..9aaa6f5 100644
@@ -32,3 +32,14 @@ get_cpuid(char *buffer, size_t sz)
 	}
 	return -1;
 }
+
+char *
+get_cpuid_str(void)
+{
+	char *bufp;
+
+	if (asprintf(&bufp, "%.8lx", mfspr(SPRN_PVR)) < 0)
+		bufp = NULL;
+
+	return bufp;
+}
Signature · 1 line
1.8.3.1
#8 lore raw

[tip:perf/urgent] perf powerpc: Support CPU ID matching for Powerpc

perf powerpc: Support CPU ID matching for Powerpc
Commit-ID:  ce88f27ccc724fa8e6766fdd4140768931f42f88
Gitweb:     http://git.kernel.org/tip/ce88f27ccc724fa8e6766fdd4140768931f42f88
Author:     Sukadev Bhattiprolu <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:41 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 19:58:01 -0300

perf powerpc: Support CPU ID matching for Powerpc

Implement code that returns the generic CPU ID string for Powerpc. This will be used to identify the specific table of PMU events to parse/compare user specified events against.

---
 tools/perf/arch/powerpc/util/header.c | 11 +++++++++++
 1 file changed, 11 insertions(+)
tools/perf/arch/powerpc/util/header.c +11 -0
index f8ccee1..9aaa6f5 100644
@@ -32,3 +32,14 @@ get_cpuid(char *buffer, size_t sz)
 	}
 	return -1;
 }
+
+char *
+get_cpuid_str(void)
+{
+	char *bufp;
+
+	if (asprintf(&bufp, "%.8lx", mfspr(SPRN_PVR)) < 0)
+		bufp = NULL;
+
+	return bufp;
+}
#9 lore raw

PATCH v21 5/19 perf, tools: Support CPU id matching for x86 v2

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

Implement the code to match CPU types to mapfile types for x86 based on CPUID. This extends an existing similar function, but changes it to use the x86 mapfile cpu description. This allows to resolve event lists generated by jevents.

---

v2: Update to new get_cpuid_str() interface

---
 tools/perf/arch/x86/util/header.c | 24 +++++++++++++++++++++---
 1 file changed, 21 insertions(+), 3 deletions(-)
tools/perf/arch/x86/util/header.c +21 -3
index 146d12a..a74a48d 100644
@@ -19,8 +19,8 @@ cpuid(unsigned int op, unsigned int *a, unsigned int *b, unsigned int *c,
 			: "a" (op));
 }
 
-int
-get_cpuid(char *buffer, size_t sz)
+static int
+__get_cpuid(char *buffer, size_t sz, const char *fmt)
 {
 	unsigned int a, b, c, d, lvl;
 	int family = -1, model = -1, step = -1;
@@ -48,7 +48,7 @@ get_cpuid(char *buffer, size_t sz)
 		if (family >= 0x6)
 			model += ((a >> 16) & 0xf) << 4;
 	}
-	nb = scnprintf(buffer, sz, "%s,%u,%u,%u$", vendor, family, model, step);
+	nb = scnprintf(buffer, sz, fmt, vendor, family, model, step);
 
 	/* look for end marker to ensure the entire data fit */
 	if (strchr(buffer, '$')) {
@@ -57,3 +57,21 @@ get_cpuid(char *buffer, size_t sz)
 	}
 	return -1;
 }
+
+int
+get_cpuid(char *buffer, size_t sz)
+{
+	return __get_cpuid(buffer, sz, "%s,%u,%u,%u$");
+}
+
+char *
+get_cpuid_str(void)
+{
+	char *buf = malloc(128);
+
+	if (__get_cpuid(buf, 128, "%s-%u-%X$") < 0) {
+		free(buf);
+		return NULL;
+	}
+	return buf;
+}
Signature · 1 line
1.8.3.1
#10 lore raw

[tip:perf/urgent] perf tools: Support CPU id matching for x86 v2

perf tools: Support CPU id matching for x86 v2
Commit-ID:  f33d1227197e0fe9d3682c5a766fdc74559da78e
Gitweb:     http://git.kernel.org/tip/f33d1227197e0fe9d3682c5a766fdc74559da78e
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:42 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 19:58:01 -0300

perf tools: Support CPU id matching for x86 v2

Implement the code to match CPU types to mapfile types for x86 based on CPUID. This extends an existing similar function, but changes it to use the x86 mapfile cpu description. This allows to resolve event lists generated by jevents.

---
 tools/perf/arch/x86/util/header.c | 24 +++++++++++++++++++++---
 1 file changed, 21 insertions(+), 3 deletions(-)
tools/perf/arch/x86/util/header.c +21 -3
index 146d12a..a74a48d 100644
@@ -19,8 +19,8 @@ cpuid(unsigned int op, unsigned int *a, unsigned int *b, unsigned int *c,
 			: "a" (op));
 }
 
-int
-get_cpuid(char *buffer, size_t sz)
+static int
+__get_cpuid(char *buffer, size_t sz, const char *fmt)
 {
 	unsigned int a, b, c, d, lvl;
 	int family = -1, model = -1, step = -1;
@@ -48,7 +48,7 @@ get_cpuid(char *buffer, size_t sz)
 		if (family >= 0x6)
 			model += ((a >> 16) & 0xf) << 4;
 	}
-	nb = scnprintf(buffer, sz, "%s,%u,%u,%u$", vendor, family, model, step);
+	nb = scnprintf(buffer, sz, fmt, vendor, family, model, step);
 
 	/* look for end marker to ensure the entire data fit */
 	if (strchr(buffer, '$')) {
@@ -57,3 +57,21 @@ get_cpuid(char *buffer, size_t sz)
 	}
 	return -1;
 }
+
+int
+get_cpuid(char *buffer, size_t sz)
+{
+	return __get_cpuid(buffer, sz, "%s,%u,%u,%u$");
+}
+
+char *
+get_cpuid_str(void)
+{
+	char *buf = malloc(128);
+
+	if (__get_cpuid(buf, 128, "%s-%u-%X$") < 0) {
+		free(buf);
+		return NULL;
+	}
+	return buf;
+}
#11 lore raw

PATCH v21 6/19 perf, tools: Support alias descriptions

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

Add support to print alias descriptions in perf list, which are taken from the generated event files.

The sorting code is changed to put the events with descriptions at the end. The descriptions are printed as possibly multiple word wrapped lines.

Example output:

% perf list
...
arith.fpu_div
[Divide operations executed]
arith.fpu_div_active
[Cycles when divider is busy executing divide operations]

---

Changelog
- Delete a redundant free()

Changelog[v14]
- [Jiri Olsa] Fail, rather than continue if strdup() returns NULL; remove unnecessary __maybe_unused.

---
 tools/perf/util/pmu.c | 83 +++++++++++++++++++++++++++++++++++++++++----------
 tools/perf/util/pmu.h |  1 +
 2 files changed, 68 insertions(+), 16 deletions(-)
tools/perf/util/pmu.c +67 -16
index c842886..af1a612 100644
@@ -222,7 +222,7 @@ static int perf_pmu__parse_snapshot(struct perf_pmu_alias *alias,
 }
 
 static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
-				 char *desc __maybe_unused, char *val)
+				 char *desc, char *val)
 {
 	struct perf_pmu_alias *alias;
 	int ret;
@@ -255,6 +255,8 @@ static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
 		perf_pmu__parse_snapshot(alias, dir, name);
 	}
 
+	alias->desc = desc ? strdup(desc) : NULL;
+
 	list_add_tail(&alias->list, list);
 
 	return 0;
@@ -1044,11 +1046,42 @@ static char *format_alias_or(char *buf, int len, struct perf_pmu *pmu,
 	return buf;
 }
 
-static int cmp_string(const void *a, const void *b)
+struct pair {
+	char *name;
+	char *desc;
+};
+
+static int cmp_pair(const void *a, const void *b)
+{
+	const struct pair *as = a;
+	const struct pair *bs = b;
+
+	/* Put extra events last */
+	if (!!as->desc != !!bs->desc)
+		return !!as->desc - !!bs->desc;
+	return strcmp(as->name, bs->name);
+}
+
+static void wordwrap(char *s, int start, int max, int corr)
 {
-	const char * const *as = a;
-	const char * const *bs = b;
-	return strcmp(*as, *bs);
+	int column = start;
+	int n;
+
+	while (*s) {
+		int wlen = strcspn(s, " \t");
+
+		if (column + wlen >= max && column > start) {
+			printf("\n%*s", start, "");
+			column = start + corr;
+		}
+		n = printf("%s%.*s", column > start ? " " : "", wlen, s);
+		if (n <= 0)
+			break;
+		s += wlen;
+		column += n;
+		while (isspace(*s))
+			s++;
+	}
 }
 
 void print_pmu_events(const char *event_glob, bool name_only)
@@ -1058,7 +1091,9 @@ void print_pmu_events(const char *event_glob, bool name_only)
 	char buf[1024];
 	int printed = 0;
 	int len, j;
-	char **aliases;
+	struct pair *aliases;
+	int numdesc = 0;
+	int columns = 78;
 
 	pmu = NULL;
 	len = 0;
@@ -1068,14 +1103,15 @@ void print_pmu_events(const char *event_glob, bool name_only)
 		if (pmu->selectable)
 			len++;
 	}
-	aliases = zalloc(sizeof(char *) * len);
+	aliases = zalloc(sizeof(struct pair) * len);
 	if (!aliases)
 		goto out_enomem;
 	pmu = NULL;
 	j = 0;
 	while ((pmu = perf_pmu__scan(pmu)) != NULL) {
 		list_for_each_entry(alias, &pmu->aliases, list) {
-			char *name = format_alias(buf, sizeof(buf), pmu, alias);
+			char *name = alias->desc ? alias->name :
+				format_alias(buf, sizeof(buf), pmu, alias);
 			bool is_cpu = !strcmp(pmu->name, "cpu");
 
 			if (event_glob != NULL &&
@@ -1084,12 +1120,19 @@ void print_pmu_events(const char *event_glob, bool name_only)
 						       event_glob))))
 				continue;
 
-			if (is_cpu && !name_only)
+			if (is_cpu && !name_only && !alias->desc)
 				name = format_alias_or(buf, sizeof(buf), pmu, alias);
 
-			aliases[j] = strdup(name);
-			if (aliases[j] == NULL)
+			aliases[j].name = name;
+			if (is_cpu && !name_only && !alias->desc)
+				aliases[j].name = format_alias_or(buf,
+								  sizeof(buf),
+								  pmu, alias);
+			aliases[j].name = strdup(aliases[j].name);
+			if (!aliases[j].name)
 				goto out_enomem;
+
+			aliases[j].desc = alias->desc;
 			j++;
 		}
 		if (pmu->selectable &&
@@ -1097,25 +1140,33 @@ void print_pmu_events(const char *event_glob, bool name_only)
 			char *s;
 			if (asprintf(&s, "%s//", pmu->name) < 0)
 				goto out_enomem;
-			aliases[j] = s;
+			aliases[j].name = s;
 			j++;
 		}
 	}
 	len = j;
-	qsort(aliases, len, sizeof(char *), cmp_string);
+	qsort(aliases, len, sizeof(struct pair), cmp_pair);
 	for (j = 0; j < len; j++) {
 		if (name_only) {
-			printf("%s ", aliases[j]);
+			printf("%s ", aliases[j].name);
 			continue;
 		}
-		printf("  %-50s [Kernel PMU event]\n", aliases[j]);
+		if (aliases[j].desc) {
+			if (numdesc++ == 0)
+				printf("\n");
+			printf("  %-50s\n", aliases[j].name);
+			printf("%*s", 8, "[");
+			wordwrap(aliases[j].desc, 8, columns, 0);
+			printf("]\n");
+		} else
+			printf("  %-50s [Kernel PMU event]\n", aliases[j].name);
 		printed++;
 	}
 	if (printed && pager_in_use())
 		printf("\n");
 out_free:
 	for (j = 0; j < len; j++)
-		zfree(&aliases[j]);
+		zfree(&aliases[j].name);
 	zfree(&aliases);
 	return;
 
tools/perf/util/pmu.h +1 -0
index 5d7e844..11bda57 100644
@@ -38,6 +38,7 @@ struct perf_pmu_info {
 
 struct perf_pmu_alias {
 	char *name;
+	char *desc;
 	struct list_head terms; /* HEAD struct parse_events_term -> list */
 	struct list_head list;  /* ELEM */
 	char unit[UNIT_MAX_LEN+1];
Signature · 1 line
1.8.3.1
#12 lore raw

Re: [PATCH v21 06/19] perf, tools: Support alias descriptions

Em Thu, Sep 15, 2016 at 03:24:43PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 8 lines

From: Andi Kleen <[email protected]>

Add support to print alias descriptions in perf list, which are taken from the generated event files.

The sorting code is changed to put the events with descriptions at the end. The descriptions are printed as possibly multiple word wrapped lines.

So, now I'm trying to reproduce the results below, but I couldn't find a tarball with those .json files for me to use, can you provide me with one?

I've put what I have at:

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tmp.perf/json

Now testing Jiri's cross compiling patch...

Thanks,

- Arnaldo

Quoted 200 lines

Example output:

% perf list
...
arith.fpu_div
[Divide operations executed]
arith.fpu_div_active
[Cycles when divider is busy executing divide operations]

---

Changelog
- Delete a redundant free()

Changelog[v14]
- [Jiri Olsa] Fail, rather than continue if strdup() returns NULL; remove unnecessary __maybe_unused.

---
 tools/perf/util/pmu.c | 83 +++++++++++++++++++++++++++++++++++++++++----------
 tools/perf/util/pmu.h |  1 +
 2 files changed, 68 insertions(+), 16 deletions(-)
tools/perf/util/pmu.c +67 -16
index c842886..af1a612 100644
@@ -222,7 +222,7 @@ static int perf_pmu__parse_snapshot(struct perf_pmu_alias *alias,
 }
 
 static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
-				 char *desc __maybe_unused, char *val)
+				 char *desc, char *val)
 {
 	struct perf_pmu_alias *alias;
 	int ret;
@@ -255,6 +255,8 @@ static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
 		perf_pmu__parse_snapshot(alias, dir, name);
 	}
 
+	alias->desc = desc ? strdup(desc) : NULL;
+
 	list_add_tail(&alias->list, list);
 
 	return 0;
@@ -1044,11 +1046,42 @@ static char *format_alias_or(char *buf, int len, struct perf_pmu *pmu,
 	return buf;
 }
 
-static int cmp_string(const void *a, const void *b)
+struct pair {
+	char *name;
+	char *desc;
+};
+
+static int cmp_pair(const void *a, const void *b)
+{
+	const struct pair *as = a;
+	const struct pair *bs = b;
+
+	/* Put extra events last */
+	if (!!as->desc != !!bs->desc)
+		return !!as->desc - !!bs->desc;
+	return strcmp(as->name, bs->name);
+}
+
+static void wordwrap(char *s, int start, int max, int corr)
 {
-	const char * const *as = a;
-	const char * const *bs = b;
-	return strcmp(*as, *bs);
+	int column = start;
+	int n;
+
+	while (*s) {
+		int wlen = strcspn(s, " \t");
+
+		if (column + wlen >= max && column > start) {
+			printf("\n%*s", start, "");
+			column = start + corr;
+		}
+		n = printf("%s%.*s", column > start ? " " : "", wlen, s);
+		if (n <= 0)
+			break;
+		s += wlen;
+		column += n;
+		while (isspace(*s))
+			s++;
+	}
 }
 
 void print_pmu_events(const char *event_glob, bool name_only)
@@ -1058,7 +1091,9 @@ void print_pmu_events(const char *event_glob, bool name_only)
 	char buf[1024];
 	int printed = 0;
 	int len, j;
-	char **aliases;
+	struct pair *aliases;
+	int numdesc = 0;
+	int columns = 78;
 
 	pmu = NULL;
 	len = 0;
@@ -1068,14 +1103,15 @@ void print_pmu_events(const char *event_glob, bool name_only)
 		if (pmu->selectable)
 			len++;
 	}
-	aliases = zalloc(sizeof(char *) * len);
+	aliases = zalloc(sizeof(struct pair) * len);
 	if (!aliases)
 		goto out_enomem;
 	pmu = NULL;
 	j = 0;
 	while ((pmu = perf_pmu__scan(pmu)) != NULL) {
 		list_for_each_entry(alias, &pmu->aliases, list) {
-			char *name = format_alias(buf, sizeof(buf), pmu, alias);
+			char *name = alias->desc ? alias->name :
+				format_alias(buf, sizeof(buf), pmu, alias);
 			bool is_cpu = !strcmp(pmu->name, "cpu");
 
 			if (event_glob != NULL &&
@@ -1084,12 +1120,19 @@ void print_pmu_events(const char *event_glob, bool name_only)
 						       event_glob))))
 				continue;
 
-			if (is_cpu && !name_only)
+			if (is_cpu && !name_only && !alias->desc)
 				name = format_alias_or(buf, sizeof(buf), pmu, alias);
 
-			aliases[j] = strdup(name);
-			if (aliases[j] == NULL)
+			aliases[j].name = name;
+			if (is_cpu && !name_only && !alias->desc)
+				aliases[j].name = format_alias_or(buf,
+								  sizeof(buf),
+								  pmu, alias);
+			aliases[j].name = strdup(aliases[j].name);
+			if (!aliases[j].name)
 				goto out_enomem;
+
+			aliases[j].desc = alias->desc;
 			j++;
 		}
 		if (pmu->selectable &&
@@ -1097,25 +1140,33 @@ void print_pmu_events(const char *event_glob, bool name_only)
 			char *s;
 			if (asprintf(&s, "%s//", pmu->name) < 0)
 				goto out_enomem;
-			aliases[j] = s;
+			aliases[j].name = s;
 			j++;
 		}
 	}
 	len = j;
-	qsort(aliases, len, sizeof(char *), cmp_string);
+	qsort(aliases, len, sizeof(struct pair), cmp_pair);
 	for (j = 0; j < len; j++) {
 		if (name_only) {
-			printf("%s ", aliases[j]);
+			printf("%s ", aliases[j].name);
 			continue;
 		}
-		printf("  %-50s [Kernel PMU event]\n", aliases[j]);
+		if (aliases[j].desc) {
+			if (numdesc++ == 0)
+				printf("\n");
+			printf("  %-50s\n", aliases[j].name);
+			printf("%*s", 8, "[");
+			wordwrap(aliases[j].desc, 8, columns, 0);
+			printf("]\n");
+		} else
+			printf("  %-50s [Kernel PMU event]\n", aliases[j].name);
 		printed++;
 	}
 	if (printed && pager_in_use())
 		printf("\n");
 out_free:
 	for (j = 0; j < len; j++)
-		zfree(&aliases[j]);
+		zfree(&aliases[j].name);
 	zfree(&aliases);
 	return;
 
tools/perf/util/pmu.h +1 -0
index 5d7e844..11bda57 100644
@@ -38,6 +38,7 @@ struct perf_pmu_info {
 
 struct perf_pmu_alias {
 	char *name;
+	char *desc;
 	struct list_head terms; /* HEAD struct parse_events_term -> list */
 	struct list_head list;  /* ELEM */
 	char unit[UNIT_MAX_LEN+1];
Signature · 1 line
1.8.3.1
#13 lore raw

Re: [PATCH v21 06/19] perf, tools: Support alias descriptions

Quoted 6 lines · Arnaldo Carvalho de Melo [[email protected]] wrote:

Em Thu, Sep 15, 2016 at 03:24:43PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 5 lines

From: Andi Kleen <[email protected]>

Add support to print alias descriptions in perf list, which are taken from the generated event files.

Quoted 7 lines
Quoted 3 lines

The sorting code is changed to put the events with descriptions at the end. The descriptions are printed as possibly multiple word wrapped lines.

So, now I'm trying to reproduce the results below, but I couldn't find a tarball with those .json files for me to use, can you provide me with one?

The data files are in my github, in the json-code+data-v21 branch starting with 23bb101. They are individual commits rather than a tarball though.

	>       https://github.com/sukadev/linux.git
	>
	>       Branch                  Description
	>       ------------------------------------------------------
	>       json-code-v21           Source Code only
	>       json-code+data-v21      Both code and data(for build/test/pull)
	> 

Sukadev

#14 lore raw

Re: [PATCH v21 06/19] perf, tools: Support alias descriptions

Em Tue, Sep 27, 2016 at 11:11:16AM -0700, Sukadev Bhattiprolu escreveu:

Quoted 10 lines
Quoted 9 lines · Arnaldo Carvalho de Melo [[email protected]] wrote:

Em Thu, Sep 15, 2016 at 03:24:43PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 7 lines

From: Andi Kleen <[email protected]>
Add support to print alias descriptions in perf list, which are taken from the generated event files.

The sorting code is changed to put the events with descriptions at the end. The descriptions are printed as possibly multiple word wrapped lines.

Quoted 7 lines
Quoted 3 lines

So, now I'm trying to reproduce the results below, but I couldn't find a tarball with those .json files for me to use, can you provide me with one?

The data files are in my github, in the json-code+data-v21 branch starting with 23bb101. They are individual commits rather than a tarball though.

Ok, I'll pick one for powerpc and another for x86_64 so that I can test it and Jiri's x-compile support.

Refresh my mind, what is the plan on these files? Are we just going to provide pointers to where to get them from vendors, ship it in the kernel, auto-download them as part of the build process?

At least examples that allows to build and have a new 'perf test' entry to check them automatically seems to be in order, no?

- Arnaldo

Quoted 9 lines
	>       https://github.com/sukadev/linux.git
	>
	>       Branch                  Description
	>       ------------------------------------------------------
	>       json-code-v21           Source Code only
	>       json-code+data-v21      Both code and data(for build/test/pull)
	> 

Sukadev

#15 lore raw

Re: [PATCH v21 06/19] perf, tools: Support alias descriptions

Quoted 15 lines · Arnaldo Carvalho de Melo [[email protected]] wrote:

Em Tue, Sep 27, 2016 at 11:11:16AM -0700, Sukadev Bhattiprolu escreveu:

Quoted 14 lines
Quoted 12 lines · Arnaldo Carvalho de Melo [[email protected]] wrote:

Em Thu, Sep 15, 2016 at 03:24:43PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 7 lines

From: Andi Kleen <[email protected]>
Add support to print alias descriptions in perf list, which are taken from the generated event files.

The sorting code is changed to put the events with descriptions at the end. The descriptions are printed as possibly multiple word wrapped lines.

So, now I'm trying to reproduce the results below, but I couldn't find a tarball with those .json files for me to use, can you provide me with one?

Quoted 6 lines
Quoted 3 lines

The data files are in my github, in the json-code+data-v21 branch starting with 23bb101. They are individual commits rather than a tarball though.

Ok, I'll pick one for powerpc and another for x86_64 so that I can test it and Jiri's x-compile support.

Please pull all files if possible, specially on x86, _before_ building the perf binary. If you are going to pull only one, you need to make sure that the file you pull matches the CPU model on the system you are testing.(see below). For Power, you need to test on Power8.

Quoted 4 lines

Refresh my mind, what is the plan on these files? Are we just going to provide pointers to where to get them from vendors, ship it in the kernel, auto-download them as part of the build process?

They are supposed to be committed into the linux kernel tree as shown in the json-code+data-v21 tree and they will be picked up _during build_ of the perf binary. (We are just not mailing those data files as patches since they are large and there is very little value in reviewing them).

When building perf on on say x86, event tables for all the different x86 CPU models will be included in the perf binary. When perf is then executed on an x86 box, it will detect the CPU model of that box and use the set of events corresponding to that model.

If the CPU model does not match the models "known" to the perf binary, then the symbolic names will not work on that system, but there should be no other change in behavior.

Patch 15/19 tries to explain the process.

Quoted 3 lines

At least examples that allows to build and have a new 'perf test' entry to check them automatically seems to be in order, no?

Well, the hope was that build/usage will be transparent! but we did not test in the cross-compile environment. Will think about a test case. Please let me know if we can update the README in Patch 15/19 in any way.

Quoted 12 lines

- Arnaldo

Quoted 9 lines
	>       https://github.com/sukadev/linux.git
	>
	>       Branch                  Description
	>       ------------------------------------------------------
	>       json-code-v21           Source Code only
	>       json-code+data-v21      Both code and data(for build/test/pull)
	> 

Sukadev

#16 lore raw

Re: [PATCH v21 06/19] perf, tools: Support alias descriptions

Em Wed, Sep 28, 2016 at 11:29:16AM -0700, Sukadev Bhattiprolu escreveu:

Quoted 43 lines
Quoted 15 lines · Arnaldo Carvalho de Melo [[email protected]] wrote:

Em Tue, Sep 27, 2016 at 11:11:16AM -0700, Sukadev Bhattiprolu escreveu:

Quoted 14 lines
Quoted 12 lines · Arnaldo Carvalho de Melo [[email protected]] wrote:

Em Thu, Sep 15, 2016 at 03:24:43PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 7 lines

From: Andi Kleen <[email protected]>
Add support to print alias descriptions in perf list, which are taken from the generated event files.

The sorting code is changed to put the events with descriptions at the end. The descriptions are printed as possibly multiple word wrapped lines.

So, now I'm trying to reproduce the results below, but I couldn't find a tarball with those .json files for me to use, can you provide me with one?

Quoted 6 lines
Quoted 3 lines

The data files are in my github, in the json-code+data-v21 branch starting with 23bb101. They are individual commits rather than a tarball though.

Ok, I'll pick one for powerpc and another for x86_64 so that I can test it and Jiri's x-compile support.

Please pull all files if possible, specially on x86, _before_ building the perf binary. If you are going to pull only one, you need to make sure that the file you pull matches the CPU model on the system you are testing.(see below). For Power, you need to test on Power8.

Quoted 4 lines

Refresh my mind, what is the plan on these files? Are we just going to provide pointers to where to get them from vendors, ship it in the kernel, auto-download them as part of the build process?

They are supposed to be committed into the linux kernel tree as shown in the json-code+data-v21 tree and they will be picked up _during build_ of the perf binary. (We are just not mailing those data files as patches since they are large and there is very little value in reviewing them).

When building perf on on say x86, event tables for all the different x86 CPU models will be included in the perf binary. When perf is then executed on an x86 box, it will detect the CPU model of that box and use the set of events corresponding to that model.

Quoted 5 lines

If the CPU model does not match the models "known" to the perf binary, then the symbolic names will not work on that system, but there should be no other change in behavior.

Patch 15/19 tries to explain the process.

I didn't get to that one yet.

Quoted 4 lines
Quoted 2 lines

At least examples that allows to build and have a new 'perf test' entry to check them automatically seems to be in order, no?

Well, the hope was that build/usage will be transparent! but we did not

the 'perf test' entry is to test the whole process of going from a json file to a the perf binary and then asking for one such event and checking if the resulting perf_event_attr is what we expect it to be, see, for instance:

[root@jouet c]# perf test roundtrip
11: roundtrip evsel->name check : Ok

tools/perf/tests/evsel-roundtrip-name.c

Quoted 3 lines

test in the cross-compile environment. Will think about a test case. Please let me know if we can update the README in Patch 15/19 in any way.

I'll let you know when I get to that patch, till then I'll follow the instructions you gave me here.

I.e. I go on testing patch by patch, trying to use the documentation that is available up to that point, trying to reproduce the results described in the patch, to fully validate it.

- Arnaldo

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
	>       https://github.com/sukadev/linux.git
	>
	>       Branch                  Description
	>       ------------------------------------------------------
	>       json-code-v21           Source Code only
	>       json-code+data-v21      Both code and data(for build/test/pull)
	> 

Sukadev

#17 lore raw

Re: [PATCH v21 06/19] perf, tools: Support alias descriptions

In reply to Sukadev Bhattiprolu

Em Wed, Sep 28, 2016 at 11:29:16AM -0700, Sukadev Bhattiprolu escreveu:

Quoted 16 lines
Quoted 15 lines · Arnaldo Carvalho de Melo [[email protected]] wrote:

Em Tue, Sep 27, 2016 at 11:11:16AM -0700, Sukadev Bhattiprolu escreveu:

Quoted 14 lines
Quoted 12 lines · Arnaldo Carvalho de Melo [[email protected]] wrote:

Em Thu, Sep 15, 2016 at 03:24:43PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 7 lines

From: Andi Kleen <[email protected]>
Add support to print alias descriptions in perf list, which are taken from the generated event files.

The sorting code is changed to put the events with descriptions at the end. The descriptions are printed as possibly multiple word wrapped lines.

So, now I'm trying to reproduce the results below, but I couldn't find a tarball with those .json files for me to use, can you provide me with one?

Quoted 8 lines
Quoted 6 lines
Quoted 3 lines

The data files are in my github, in the json-code+data-v21 branch starting with 23bb101. They are individual commits rather than a tarball though.

Ok, I'll pick one for powerpc and another for x86_64 so that I can test it and Jiri's x-compile support.

Please pull all files if possible, specially on x86, _before_ building the

Nah, will leave this for a second step, i.e. you guys can send a pull req after I finish the initial testing.

Quoted 2 lines

perf binary. If you are going to pull only one, you need to make sure that the file you pull matches the CPU model on the system you are testing.(see

Right, right, that was the plan, and I actually did it now, got the mapfile.csv, left only the entry for my test machine:

processor : 3
vendor_id : GenuineIntel
cpu family : 6
model : 61
model name : Intel(R) Core(TM) i7-5600U CPU @ 2.60GHz

Which is 3D, i.e.:

[acme@jouet linux]$ cat tools/perf/pmu-events/arch/x86/mapfile.csv Family-model,Version,Filename,EventType GenuineIntel-6-3D,V16,Broadwell,core
[acme@jouet linux]$

And copied the files in the directory this maps to:

[acme@jouet linux]$ ls -la tools/perf/pmu-events/arch/x86/Broadwell/ total 352

drwxrwxr-x. 2 acme acme   4096 Oct  3 17:23 .
drwxrwxr-x. 3 acme acme   4096 Oct  3 17:23 ..
-rw-r--r--. 1 acme acme 127294 Oct  3 17:20 Cache.json
-rw-r--r--. 1 acme acme  10100 Oct  3 17:21 Floating-point.json
-rw-r--r--. 1 acme acme  16319 Oct  3 17:21 Frontend.json
-rw-r--r--. 1 acme acme 106375 Oct  3 17:21 Memory.json
-rw-r--r--. 1 acme acme   2080 Oct  3 17:22 Other.json
-rw-r--r--. 1 acme acme  64972 Oct  3 17:22 Pipeline.json
-rw-r--r--. 1 acme acme  15356 Oct  3 17:23 Virtual-Memory.json

[acme@jouet linux]$

Then try to build it as usual, i.e. using O=:

make O=/tmp/build/perf -C tools/perf install-bin

And it fails, not building the $(OUTPUT)pmu-events/pmu-events.c somehow...

Ok, I decide to fall back to building it in the same directory as the sources:

[acme@jouet linux]$ cd tools/perf
[acme@jouet perf]$ make

It gets a bit better, but fails as well:

[acme@jouet perf]$ make
BUILD: Doing 'make -j4' parallel build
CC pmu-events/pmu-events.o
pmu-events/pmu-events.c:4430:11: error: ‘pme_Filename’ undeclared here (not in a function)
.table = pme_Filename

           ^~~~~~~~~~~~

mv: cannot stat 'pmu-events/.pmu-events.o.tmp': No such file or directory /home/acme/git/linux/tools/build/Makefile.build:88: recipe for target 'pmu-events/pmu-events.o' failed
make[2]: *** [pmu-events/pmu-events.o] Error 1 Makefile.perf:467: recipe for target 'pmu-events/pmu-events-in.o' failed make[1]: *** [pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
Makefile:68: recipe for target 'all' failed
make: *** [all] Error 2
[acme@jouet perf]$

It finally builds if I ditch the header in

https://raw.githubusercontent.com/sukadev/linux/json-code%2Bdata-v21/tools/perf/pmu-events/arch/x86/mapfile.csv

[acme@jouet perf]$ cat pmu-events/arch/x86/mapfile.csv Family-model,Version,Filename,EventType GenuineIntel-6-3D,V16,Broadwell,core
[acme@jouet perf]$ vim pmu-events/arch/x86/mapfile.csv [acme@jouet perf]$ cat pmu-events/arch/x86/mapfile.csv GenuineIntel-6-3D,V16,Broadwell,core
[acme@jouet perf]$ make
BUILD: Doing 'make -j4' parallel build
GEN pmu-events/pmu-events.c
Warning: tools/include/uapi/linux/bpf.h differs from kernel

  CC       pmu-events/pmu-events.o
  LD       pmu-events/pmu-events-in.o
  AR       libperf.a
  LINK     libperf-gtk.so
  LINK     perf

[acme@jouet perf]$

And then it seems to work, at least for 'perf list', still need to check if reading events with the generic name so far used + the new one looks sane, will be done after this:

[acme@jouet perf]$ perf list | grep arith
arith.fpu_div_active
fp_arith_inst_retired.128b_packed_double
fp_arith_inst_retired.128b_packed_single
fp_arith_inst_retired.256b_packed_double
fp_arith_inst_retired.256b_packed_single
fp_arith_inst_retired.double
fp_arith_inst_retired.packed
fp_arith_inst_retired.scalar
fp_arith_inst_retired.scalar_double
fp_arith_inst_retired.scalar_single
fp_arith_inst_retired.single
[acme@jouet perf]$

Now trying to fix the O= part to continue processing...

- Arnaldo

Quoted 43 lines

below). For Power, you need to test on Power8.

Quoted 4 lines

Refresh my mind, what is the plan on these files? Are we just going to provide pointers to where to get them from vendors, ship it in the kernel, auto-download them as part of the build process?

They are supposed to be committed into the linux kernel tree as shown in the json-code+data-v21 tree and they will be picked up _during build_ of the perf binary. (We are just not mailing those data files as patches since they are large and there is very little value in reviewing them).

When building perf on on say x86, event tables for all the different x86 CPU models will be included in the perf binary. When perf is then executed on an x86 box, it will detect the CPU model of that box and use the set of events corresponding to that model.

If the CPU model does not match the models "known" to the perf binary, then the symbolic names will not work on that system, but there should be no other change in behavior.

Patch 15/19 tries to explain the process.

Quoted 3 lines

At least examples that allows to build and have a new 'perf test' entry to check them automatically seems to be in order, no?

Well, the hope was that build/usage will be transparent! but we did not test in the cross-compile environment. Will think about a test case. Please let me know if we can update the README in Patch 15/19 in any way.

Quoted 12 lines

- Arnaldo

Quoted 9 lines
	>       https://github.com/sukadev/linux.git
	>
	>       Branch                  Description
	>       ------------------------------------------------------
	>       json-code-v21           Source Code only
	>       json-code+data-v21      Both code and data(for build/test/pull)
	> 

Sukadev

#18 lore raw

[tip:perf/urgent] perf pmu: Support alias descriptions

In reply to Sukadev Bhattiprolu
perf pmu: Support alias descriptions
Commit-ID:  08e60ed15d0483be38a87d17538ccf02acff5b1f
Gitweb:     http://git.kernel.org/tip/08e60ed15d0483be38a87d17538ccf02acff5b1f
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:43 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:34:54 -0300

perf pmu: Support alias descriptions

Add support to print alias descriptions in perf list, which are taken from the generated event files.

The sorting code is changed to put the events with descriptions at the end. The descriptions are printed as possibly multiple word wrapped lines.

Example output:

% perf list
...
arith.fpu_div
[Divide operations executed]
arith.fpu_div_active
[Cycles when divider is busy executing divide operations]

Committer notes:

Further testing on a Broadwell machine (ThinkPad t450s), using these files:

  $ find tools/perf/pmu-events/arch/x86/
  tools/perf/pmu-events/arch/x86/
  tools/perf/pmu-events/arch/x86/Broadwell
  tools/perf/pmu-events/arch/x86/Broadwell/Cache.json
  tools/perf/pmu-events/arch/x86/Broadwell/Other.json
  tools/perf/pmu-events/arch/x86/Broadwell/Frontend.json
  tools/perf/pmu-events/arch/x86/Broadwell/Virtual-Memory.json
  tools/perf/pmu-events/arch/x86/Broadwell/Pipeline.json
  tools/perf/pmu-events/arch/x86/Broadwell/Floating-point.json
  tools/perf/pmu-events/arch/x86/Broadwell/Memory.json
  tools/perf/pmu-events/arch/x86/mapfile.csv
  $

Taken from:

https://github.com/sukadev/linux/tree/json-code+data-v21/tools/perf/pmu-events/arch/x86/

to get this machinery to actually parse JSON files, generate $(OUTPUT)pmu-events/pmu-events.c, compile it and link it with perf, that will then use the table it contains, these files will be submitted right after this patchkit.

[acme@jouet linux]$ perf list page_walker

List of pre-defined events (to be used in -e):

page_walker_loads.dtlb_l1
[Number of DTLB page walker hits in the L1+FB]
page_walker_loads.dtlb_l2
[Number of DTLB page walker hits in the L2]
page_walker_loads.dtlb_l3
[Number of DTLB page walker hits in the L3 + XSNP]
page_walker_loads.dtlb_memory
[Number of DTLB page walker hits in Memory]
page_walker_loads.itlb_l1
[Number of ITLB page walker hits in the L1+FB]
page_walker_loads.itlb_l2
[Number of ITLB page walker hits in the L2]
page_walker_loads.itlb_l3
[Number of ITLB page walker hits in the L3 + XSNP]

[acme@jouet linux]$

---
 tools/perf/util/pmu.c | 83 +++++++++++++++++++++++++++++++++++++++++----------
 tools/perf/util/pmu.h |  1 +
 2 files changed, 68 insertions(+), 16 deletions(-)
tools/perf/util/pmu.c +67 -16
index 10668b7..9857fb1 100644
@@ -222,7 +222,7 @@ static int perf_pmu__parse_snapshot(struct perf_pmu_alias *alias,
 }
 
 static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
-				 char *desc __maybe_unused, char *val)
+				 char *desc, char *val)
 {
 	struct perf_pmu_alias *alias;
 	int ret;
@@ -255,6 +255,8 @@ static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
 		perf_pmu__parse_snapshot(alias, dir, name);
 	}
 
+	alias->desc = desc ? strdup(desc) : NULL;
+
 	list_add_tail(&alias->list, list);
 
 	return 0;
@@ -1043,11 +1045,42 @@ static char *format_alias_or(char *buf, int len, struct perf_pmu *pmu,
 	return buf;
 }
 
-static int cmp_string(const void *a, const void *b)
+struct pair {
+	char *name;
+	char *desc;
+};
+
+static int cmp_pair(const void *a, const void *b)
+{
+	const struct pair *as = a;
+	const struct pair *bs = b;
+
+	/* Put extra events last */
+	if (!!as->desc != !!bs->desc)
+		return !!as->desc - !!bs->desc;
+	return strcmp(as->name, bs->name);
+}
+
+static void wordwrap(char *s, int start, int max, int corr)
 {
-	const char * const *as = a;
-	const char * const *bs = b;
-	return strcmp(*as, *bs);
+	int column = start;
+	int n;
+
+	while (*s) {
+		int wlen = strcspn(s, " \t");
+
+		if (column + wlen >= max && column > start) {
+			printf("\n%*s", start, "");
+			column = start + corr;
+		}
+		n = printf("%s%.*s", column > start ? " " : "", wlen, s);
+		if (n <= 0)
+			break;
+		s += wlen;
+		column += n;
+		while (isspace(*s))
+			s++;
+	}
 }
 
 void print_pmu_events(const char *event_glob, bool name_only)
@@ -1057,7 +1090,9 @@ void print_pmu_events(const char *event_glob, bool name_only)
 	char buf[1024];
 	int printed = 0;
 	int len, j;
-	char **aliases;
+	struct pair *aliases;
+	int numdesc = 0;
+	int columns = 78;
 
 	pmu = NULL;
 	len = 0;
@@ -1067,14 +1102,15 @@ void print_pmu_events(const char *event_glob, bool name_only)
 		if (pmu->selectable)
 			len++;
 	}
-	aliases = zalloc(sizeof(char *) * len);
+	aliases = zalloc(sizeof(struct pair) * len);
 	if (!aliases)
 		goto out_enomem;
 	pmu = NULL;
 	j = 0;
 	while ((pmu = perf_pmu__scan(pmu)) != NULL) {
 		list_for_each_entry(alias, &pmu->aliases, list) {
-			char *name = format_alias(buf, sizeof(buf), pmu, alias);
+			char *name = alias->desc ? alias->name :
+				format_alias(buf, sizeof(buf), pmu, alias);
 			bool is_cpu = !strcmp(pmu->name, "cpu");
 
 			if (event_glob != NULL &&
@@ -1083,12 +1119,19 @@ void print_pmu_events(const char *event_glob, bool name_only)
 						       event_glob))))
 				continue;
 
-			if (is_cpu && !name_only)
+			if (is_cpu && !name_only && !alias->desc)
 				name = format_alias_or(buf, sizeof(buf), pmu, alias);
 
-			aliases[j] = strdup(name);
-			if (aliases[j] == NULL)
+			aliases[j].name = name;
+			if (is_cpu && !name_only && !alias->desc)
+				aliases[j].name = format_alias_or(buf,
+								  sizeof(buf),
+								  pmu, alias);
+			aliases[j].name = strdup(aliases[j].name);
+			if (!aliases[j].name)
 				goto out_enomem;
+
+			aliases[j].desc = alias->desc;
 			j++;
 		}
 		if (pmu->selectable &&
@@ -1096,25 +1139,33 @@ void print_pmu_events(const char *event_glob, bool name_only)
 			char *s;
 			if (asprintf(&s, "%s//", pmu->name) < 0)
 				goto out_enomem;
-			aliases[j] = s;
+			aliases[j].name = s;
 			j++;
 		}
 	}
 	len = j;
-	qsort(aliases, len, sizeof(char *), cmp_string);
+	qsort(aliases, len, sizeof(struct pair), cmp_pair);
 	for (j = 0; j < len; j++) {
 		if (name_only) {
-			printf("%s ", aliases[j]);
+			printf("%s ", aliases[j].name);
 			continue;
 		}
-		printf("  %-50s [Kernel PMU event]\n", aliases[j]);
+		if (aliases[j].desc) {
+			if (numdesc++ == 0)
+				printf("\n");
+			printf("  %-50s\n", aliases[j].name);
+			printf("%*s", 8, "[");
+			wordwrap(aliases[j].desc, 8, columns, 0);
+			printf("]\n");
+		} else
+			printf("  %-50s [Kernel PMU event]\n", aliases[j].name);
 		printed++;
 	}
 	if (printed && pager_in_use())
 		printf("\n");
 out_free:
 	for (j = 0; j < len; j++)
-		zfree(&aliases[j]);
+		zfree(&aliases[j].name);
 	zfree(&aliases);
 	return;
 
tools/perf/util/pmu.h +1 -0
index 743422a..51d8d0d 100644
@@ -40,6 +40,7 @@ struct perf_pmu_info {
 
 struct perf_pmu_alias {
 	char *name;
+	char *desc;
 	struct list_head terms; /* HEAD struct parse_events_term -> list */
 	struct list_head list;  /* ELEM */
 	char unit[UNIT_MAX_LEN+1];
#19 lore raw

PATCH v21 7/19 perf, tools: Query terminal width and use in perf list

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

Automatically adapt the now wider and word wrapped perf list output to wider terminals. This requires querying the terminal before the auto pager takes over, and exporting this information from the pager subsystem.

---

Changelog[v20]
- Minor reorg since helpers like setup_pager() are now in tools/lib/subcmd/pager.c

---
 tools/lib/subcmd/pager.c | 16 ++++++++++++++++
 tools/lib/subcmd/pager.h |  1 +
 tools/perf/util/pmu.c    |  3 ++-
 3 files changed, 19 insertions(+), 1 deletion(-)
tools/lib/subcmd/pager.c +16 -0
index d50f3b58..6518bea 100644
@@ -3,6 +3,7 @@
 #include <stdio.h>
 #include <string.h>
 #include <signal.h>
+#include <sys/ioctl.h>
 #include "pager.h"
 #include "run-command.h"
 #include "sigchain.h"
@@ -14,6 +15,7 @@
  */
 
 static int spawned_pager;
+static int pager_columns;
 
 void pager_init(const char *pager_env)
 {
@@ -58,9 +60,12 @@ static void wait_for_pager_signal(int signo)
 void setup_pager(void)
 {
 	const char *pager = getenv(subcmd_config.pager_env);
+	struct winsize sz;
 
 	if (!isatty(1))
 		return;
+	if (ioctl(1, TIOCGWINSZ, &sz) == 0)
+		pager_columns = sz.ws_col;
 	if (!pager)
 		pager = getenv("PAGER");
 	if (!(pager || access("/usr/bin/pager", X_OK)))
@@ -98,3 +103,14 @@ int pager_in_use(void)
 {
 	return spawned_pager;
 }
+
+int pager_get_columns(void)
+{
+	char *s;
+
+	s = getenv("COLUMNS");
+	if (s)
+		return atoi(s);
+
+	return (pager_columns ? pager_columns : 80) - 2;
+}
tools/lib/subcmd/pager.h +1 -0
index 8b83714..623f554 100644
@@ -5,5 +5,6 @@ extern void pager_init(const char *pager_env);
 
 extern void setup_pager(void);
 extern int pager_in_use(void);
+extern int pager_get_columns(void);
 
 #endif /* __SUBCMD_PAGER_H */
tools/perf/util/pmu.c +2 -1
index af1a612..ef9de3e 100644
@@ -14,6 +14,7 @@
 #include "cpumap.h"
 #include "header.h"
 #include "pmu-events/pmu-events.h"
+#include "cache.h"
 
 struct perf_pmu_format {
 	char *name;
@@ -1093,7 +1094,7 @@ void print_pmu_events(const char *event_glob, bool name_only)
 	int len, j;
 	struct pair *aliases;
 	int numdesc = 0;
-	int columns = 78;
+	int columns = pager_get_columns();
 
 	pmu = NULL;
 	len = 0;
Signature · 1 line
1.8.3.1
#20 lore raw

[tip:perf/urgent] perf tools: Query terminal width and use in perf list

perf tools: Query terminal width and use in perf list
Commit-ID:  61eb2eb434b3430c6ef70536eb3d16b616b5ee52
Gitweb:     http://git.kernel.org/tip/61eb2eb434b3430c6ef70536eb3d16b616b5ee52
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:44 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:35:45 -0300

perf tools: Query terminal width and use in perf list

Automatically adapt the now wider and word wrapped perf list output to wider terminals. This requires querying the terminal before the auto pager takes over, and exporting this information from the pager subsystem.

---
 tools/lib/subcmd/pager.c | 16 ++++++++++++++++
 tools/lib/subcmd/pager.h |  1 +
 tools/perf/util/pmu.c    |  3 ++-
 3 files changed, 19 insertions(+), 1 deletion(-)
tools/lib/subcmd/pager.c +16 -0
index d50f3b58..6518bea 100644
@@ -3,6 +3,7 @@
 #include <stdio.h>
 #include <string.h>
 #include <signal.h>
+#include <sys/ioctl.h>
 #include "pager.h"
 #include "run-command.h"
 #include "sigchain.h"
@@ -14,6 +15,7 @@
  */
 
 static int spawned_pager;
+static int pager_columns;
 
 void pager_init(const char *pager_env)
 {
@@ -58,9 +60,12 @@ static void wait_for_pager_signal(int signo)
 void setup_pager(void)
 {
 	const char *pager = getenv(subcmd_config.pager_env);
+	struct winsize sz;
 
 	if (!isatty(1))
 		return;
+	if (ioctl(1, TIOCGWINSZ, &sz) == 0)
+		pager_columns = sz.ws_col;
 	if (!pager)
 		pager = getenv("PAGER");
 	if (!(pager || access("/usr/bin/pager", X_OK)))
@@ -98,3 +103,14 @@ int pager_in_use(void)
 {
 	return spawned_pager;
 }
+
+int pager_get_columns(void)
+{
+	char *s;
+
+	s = getenv("COLUMNS");
+	if (s)
+		return atoi(s);
+
+	return (pager_columns ? pager_columns : 80) - 2;
+}
tools/lib/subcmd/pager.h +1 -0
index 8b83714..623f554 100644
@@ -5,5 +5,6 @@ extern void pager_init(const char *pager_env);
 
 extern void setup_pager(void);
 extern int pager_in_use(void);
+extern int pager_get_columns(void);
 
 #endif /* __SUBCMD_PAGER_H */
tools/perf/util/pmu.c +2 -1
index 9857fb1..7b46e77 100644
@@ -14,6 +14,7 @@
 #include "cpumap.h"
 #include "header.h"
 #include "pmu-events/pmu-events.h"
+#include "cache.h"
 
 struct perf_pmu_format {
 	char *name;
@@ -1092,7 +1093,7 @@ void print_pmu_events(const char *event_glob, bool name_only)
 	int len, j;
 	struct pair *aliases;
 	int numdesc = 0;
-	int columns = 78;
+	int columns = pager_get_columns();
 
 	pmu = NULL;
 	len = 0;
#21 lore raw

PATCH v21 8/19 perf, tools: Add a --no-desc flag to perf list

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

Add a --no-desc flag to perf list to not print the event descriptions that were earlier added for JSON events. This may be useful to get a less crowded listing.

It's still default to print descriptions as that is the more useful default for most users.

---

v2: Rename --quiet to --no-desc. Add option to man page.

v18: Fix minor conflict in tools/perf/builtin-list.c; Add !desc_flag
to the newly introduced print_pmu_events() call site.

v21: Fix minor conflicts in tools/perf/builtin-list.c

---
 tools/perf/Documentation/perf-list.txt |  8 +++++++-
 tools/perf/builtin-list.c              | 14 +++++++++-----
 tools/perf/util/parse-events.c         |  4 ++--
 tools/perf/util/parse-events.h         |  2 +-
 tools/perf/util/pmu.c                  |  4 ++--
 tools/perf/util/pmu.h                  |  2 +-
 6 files changed, 22 insertions(+), 12 deletions(-)
tools/perf/Documentation/perf-list.txt +7 -1
index a126e97..72209bc 100644
@@ -8,13 +8,19 @@ perf-list - List all symbolic event types
 SYNOPSIS
 --------
 [verse]
-'perf list' [hw|sw|cache|tracepoint|pmu|event_glob]
+'perf list' [--no-desc] [hw|sw|cache|tracepoint|pmu|event_glob]
 
 DESCRIPTION
 -----------
 This command displays the symbolic event types which can be selected in the
 various perf commands with the -e option.
 
+OPTIONS
+-------
+--no-desc::
+Don't print descriptions.
+
+
 [[EVENT_MODIFIERS]]
 EVENT MODIFIERS
 ---------------
tools/perf/builtin-list.c +9 -5
index 88ee419..b14cb16 100644
@@ -16,16 +16,20 @@
 #include "util/pmu.h"
 #include <subcmd/parse-options.h>
 
+static bool desc_flag = true;
+
 int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	int i;
 	bool raw_dump = false;
 	struct option list_options[] = {
 		OPT_BOOLEAN(0, "raw-dump", &raw_dump, "Dump raw events"),
+		OPT_BOOLEAN('d', "desc", &desc_flag,
+			    "Print extra event descriptions. --no-desc to not print."),
 		OPT_END()
 	};
 	const char * const list_usage[] = {
-		"perf list [hw|sw|cache|tracepoint|pmu|sdt|event_glob]",
+		"perf list [--no-desc] [hw|sw|cache|tracepoint|pmu|sdt|event_glob]",
 		NULL
 	};
 
@@ -40,7 +44,7 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 		printf("\nList of pre-defined events (to be used in -e):\n\n");
 
 	if (argc == 0) {
-		print_events(NULL, raw_dump);
+		print_events(NULL, raw_dump, !desc_flag);
 		return 0;
 	}
 
@@ -61,14 +65,14 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			 strcmp(argv[i], "hwcache") == 0)
 			print_hwcache_events(NULL, raw_dump);
 		else if (strcmp(argv[i], "pmu") == 0)
-			print_pmu_events(NULL, raw_dump);
+			print_pmu_events(NULL, raw_dump, !desc_flag);
 		else if (strcmp(argv[i], "sdt") == 0)
 			print_sdt_events(NULL, NULL, raw_dump);
 		else if ((sep = strchr(argv[i], ':')) != NULL) {
 			int sep_idx;
 
 			if (sep == NULL) {
-				print_events(argv[i], raw_dump);
+				print_events(argv[i], raw_dump, !desc_flag);
 				continue;
 			}
 			sep_idx = sep - argv[i];
@@ -90,7 +94,7 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			print_symbol_events(s, PERF_TYPE_SOFTWARE,
 					    event_symbols_sw, PERF_COUNT_SW_MAX, raw_dump);
 			print_hwcache_events(s, raw_dump);
-			print_pmu_events(s, raw_dump);
+			print_pmu_events(s, raw_dump, !desc_flag);
 			print_tracepoint_events(NULL, s, raw_dump);
 			print_sdt_events(NULL, s, raw_dump);
 			free(s);
tools/perf/util/parse-events.c +2 -2
index 6c913c3..a3c7739 100644
@@ -2229,7 +2229,7 @@ out_enomem:
 /*
  * Print the help text for the event symbols:
  */
-void print_events(const char *event_glob, bool name_only)
+void print_events(const char *event_glob, bool name_only, bool quiet_flag)
 {
 	print_symbol_events(event_glob, PERF_TYPE_HARDWARE,
 			    event_symbols_hw, PERF_COUNT_HW_MAX, name_only);
@@ -2239,7 +2239,7 @@ void print_events(const char *event_glob, bool name_only)
 
 	print_hwcache_events(event_glob, name_only);
 
-	print_pmu_events(event_glob, name_only);
+	print_pmu_events(event_glob, name_only, quiet_flag);
 
 	if (event_glob != NULL)
 		return;
tools/perf/util/parse-events.h +1 -1
index d1edbf8..795f2579 100644
@@ -171,7 +171,7 @@ void parse_events_update_lists(struct list_head *list_event,
 void parse_events_evlist_error(struct parse_events_evlist *data,
 			       int idx, const char *str);
 
-void print_events(const char *event_glob, bool name_only);
+void print_events(const char *event_glob, bool name_only, bool quiet);
 
 struct event_symbol {
 	const char	*symbol;
tools/perf/util/pmu.c +2 -2
index ef9de3e..cb4c215 100644
@@ -1085,7 +1085,7 @@ static void wordwrap(char *s, int start, int max, int corr)
 	}
 }
 
-void print_pmu_events(const char *event_glob, bool name_only)
+void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
 {
 	struct perf_pmu *pmu;
 	struct perf_pmu_alias *alias;
@@ -1152,7 +1152,7 @@ void print_pmu_events(const char *event_glob, bool name_only)
 			printf("%s ", aliases[j].name);
 			continue;
 		}
-		if (aliases[j].desc) {
+		if (aliases[j].desc && !quiet_flag) {
 			if (numdesc++ == 0)
 				printf("\n");
 			printf("  %-50s\n", aliases[j].name);
tools/perf/util/pmu.h +1 -1
index 11bda57..42999c7 100644
@@ -70,7 +70,7 @@ int perf_pmu__format_parse(char *dir, struct list_head *head);
 
 struct perf_pmu *perf_pmu__scan(struct perf_pmu *pmu);
 
-void print_pmu_events(const char *event_glob, bool name_only);
+void print_pmu_events(const char *event_glob, bool name_only, bool quiet);
 bool pmu_have_event(const char *pname, const char *name);
 
 int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt,
Signature · 1 line
1.8.3.1
#22 lore raw

[tip:perf/urgent] perf list: Add a --no-desc flag

perf list: Add a --no-desc flag
Commit-ID:  1c5f01fe8660fc48625a94a5ed10e4dbaea95d5f
Gitweb:     http://git.kernel.org/tip/1c5f01fe8660fc48625a94a5ed10e4dbaea95d5f
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:45 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:35:45 -0300

perf list: Add a --no-desc flag

Add a --no-desc flag to 'perf list' to not print the event descriptions that were earlier added for JSON events. This may be useful to get a less crowded listing.

It's still default to print descriptions as that is the more useful default for most users.

---
 tools/perf/Documentation/perf-list.txt |  8 +++++++-
 tools/perf/builtin-list.c              | 14 +++++++++-----
 tools/perf/util/parse-events.c         |  4 ++--
 tools/perf/util/parse-events.h         |  2 +-
 tools/perf/util/pmu.c                  |  4 ++--
 tools/perf/util/pmu.h                  |  2 +-
 6 files changed, 22 insertions(+), 12 deletions(-)
tools/perf/Documentation/perf-list.txt +7 -1
index a126e97..72209bc 100644
@@ -8,13 +8,19 @@ perf-list - List all symbolic event types
 SYNOPSIS
 --------
 [verse]
-'perf list' [hw|sw|cache|tracepoint|pmu|event_glob]
+'perf list' [--no-desc] [hw|sw|cache|tracepoint|pmu|event_glob]
 
 DESCRIPTION
 -----------
 This command displays the symbolic event types which can be selected in the
 various perf commands with the -e option.
 
+OPTIONS
+-------
+--no-desc::
+Don't print descriptions.
+
+
 [[EVENT_MODIFIERS]]
 EVENT MODIFIERS
 ---------------
tools/perf/builtin-list.c +9 -5
index 88ee419..b14cb16 100644
@@ -16,16 +16,20 @@
 #include "util/pmu.h"
 #include <subcmd/parse-options.h>
 
+static bool desc_flag = true;
+
 int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	int i;
 	bool raw_dump = false;
 	struct option list_options[] = {
 		OPT_BOOLEAN(0, "raw-dump", &raw_dump, "Dump raw events"),
+		OPT_BOOLEAN('d', "desc", &desc_flag,
+			    "Print extra event descriptions. --no-desc to not print."),
 		OPT_END()
 	};
 	const char * const list_usage[] = {
-		"perf list [hw|sw|cache|tracepoint|pmu|sdt|event_glob]",
+		"perf list [--no-desc] [hw|sw|cache|tracepoint|pmu|sdt|event_glob]",
 		NULL
 	};
 
@@ -40,7 +44,7 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 		printf("\nList of pre-defined events (to be used in -e):\n\n");
 
 	if (argc == 0) {
-		print_events(NULL, raw_dump);
+		print_events(NULL, raw_dump, !desc_flag);
 		return 0;
 	}
 
@@ -61,14 +65,14 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			 strcmp(argv[i], "hwcache") == 0)
 			print_hwcache_events(NULL, raw_dump);
 		else if (strcmp(argv[i], "pmu") == 0)
-			print_pmu_events(NULL, raw_dump);
+			print_pmu_events(NULL, raw_dump, !desc_flag);
 		else if (strcmp(argv[i], "sdt") == 0)
 			print_sdt_events(NULL, NULL, raw_dump);
 		else if ((sep = strchr(argv[i], ':')) != NULL) {
 			int sep_idx;
 
 			if (sep == NULL) {
-				print_events(argv[i], raw_dump);
+				print_events(argv[i], raw_dump, !desc_flag);
 				continue;
 			}
 			sep_idx = sep - argv[i];
@@ -90,7 +94,7 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			print_symbol_events(s, PERF_TYPE_SOFTWARE,
 					    event_symbols_sw, PERF_COUNT_SW_MAX, raw_dump);
 			print_hwcache_events(s, raw_dump);
-			print_pmu_events(s, raw_dump);
+			print_pmu_events(s, raw_dump, !desc_flag);
 			print_tracepoint_events(NULL, s, raw_dump);
 			print_sdt_events(NULL, s, raw_dump);
 			free(s);
tools/perf/util/parse-events.c +2 -2
index 33546c3..3966ad7 100644
@@ -2263,7 +2263,7 @@ out_enomem:
 /*
  * Print the help text for the event symbols:
  */
-void print_events(const char *event_glob, bool name_only)
+void print_events(const char *event_glob, bool name_only, bool quiet_flag)
 {
 	print_symbol_events(event_glob, PERF_TYPE_HARDWARE,
 			    event_symbols_hw, PERF_COUNT_HW_MAX, name_only);
@@ -2273,7 +2273,7 @@ void print_events(const char *event_glob, bool name_only)
 
 	print_hwcache_events(event_glob, name_only);
 
-	print_pmu_events(event_glob, name_only);
+	print_pmu_events(event_glob, name_only, quiet_flag);
 
 	if (event_glob != NULL)
 		return;
tools/perf/util/parse-events.h +1 -1
index 8d09a97..3bf376b 100644
@@ -172,7 +172,7 @@ void parse_events_update_lists(struct list_head *list_event,
 void parse_events_evlist_error(struct parse_events_evlist *data,
 			       int idx, const char *str);
 
-void print_events(const char *event_glob, bool name_only);
+void print_events(const char *event_glob, bool name_only, bool quiet);
 
 struct event_symbol {
 	const char	*symbol;
tools/perf/util/pmu.c +2 -2
index 7b46e77..9dc3506 100644
@@ -1084,7 +1084,7 @@ static void wordwrap(char *s, int start, int max, int corr)
 	}
 }
 
-void print_pmu_events(const char *event_glob, bool name_only)
+void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
 {
 	struct perf_pmu *pmu;
 	struct perf_pmu_alias *alias;
@@ -1151,7 +1151,7 @@ void print_pmu_events(const char *event_glob, bool name_only)
 			printf("%s ", aliases[j].name);
 			continue;
 		}
-		if (aliases[j].desc) {
+		if (aliases[j].desc && !quiet_flag) {
 			if (numdesc++ == 0)
 				printf("\n");
 			printf("  %-50s\n", aliases[j].name);
tools/perf/util/pmu.h +1 -1
index 51d8d0d..7b47192 100644
@@ -72,7 +72,7 @@ int perf_pmu__format_parse(char *dir, struct list_head *head);
 
 struct perf_pmu *perf_pmu__scan(struct perf_pmu *pmu);
 
-void print_pmu_events(const char *event_glob, bool name_only);
+void print_pmu_events(const char *event_glob, bool name_only, bool quiet);
 bool pmu_have_event(const char *pname, const char *name);
 
 int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt,
#23 lore raw

PATCH v21 9/19 perf, tools: Add override support for event list CPUID

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

Add a PERF_CPUID variable to override the CPUID of the current CPU (within the current architecture). This is useful for testing, so that all event lists can be tested on a single system.

---

v2: Fix double free in earlier version.
Print actual CPUID being used with verbose option.

---
 tools/perf/util/pmu.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
tools/perf/util/pmu.c +7 -1
index cb4c215..2291d2a 100644
@@ -501,10 +501,16 @@ static int pmu_add_cpu_aliases(struct list_head *head)
 	struct pmu_event *pe;
 	char *cpuid;
 
-	cpuid = get_cpuid_str();
+	cpuid = getenv("PERF_CPUID");
+	if (cpuid)
+		cpuid = strdup(cpuid);
+	if (!cpuid)
+		cpuid = get_cpuid_str();
 	if (!cpuid)
 		return 0;
 
+	pr_debug("Using CPUID %s\n", cpuid);
+
 	i = 0;
 	while (1) {
 		map = &pmu_events_map[i++];
Signature · 1 line
1.8.3.1
#24 lore raw

[tip:perf/urgent] perf pmu: Add override support for event list CPUID

perf pmu: Add override support for event list CPUID
Commit-ID:  fc06e2a5aad9fcf7efaabd7550ac31f648d2f2bc
Gitweb:     http://git.kernel.org/tip/fc06e2a5aad9fcf7efaabd7550ac31f648d2f2bc
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:46 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:35:46 -0300

perf pmu: Add override support for event list CPUID

Add a PERF_CPUID variable to override the CPUID of the current CPU (within the current architecture). This is useful for testing, so that all event lists can be tested on a single system.

---
 tools/perf/util/pmu.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
tools/perf/util/pmu.c +7 -1
index 9dc3506..79242cf 100644
@@ -501,10 +501,16 @@ static void pmu_add_cpu_aliases(struct list_head *head)
 	struct pmu_event *pe;
 	char *cpuid;
 
-	cpuid = get_cpuid_str();
+	cpuid = getenv("PERF_CPUID");
+	if (cpuid)
+		cpuid = strdup(cpuid);
+	if (!cpuid)
+		cpuid = get_cpuid_str();
 	if (!cpuid)
 		return;
 
+	pr_debug("Using CPUID %s\n", cpuid);
+
 	i = 0;
 	while (1) {
 		map = &pmu_events_map[i++];
#25 lore raw

PATCH v21 10/19 perf, tools, jevents: Add support for long descriptions

In reply to Sukadev Bhattiprolu

Implement support in jevents to parse long descriptions for events that may have them in the JSON files. A follow on patch will make this long description available to user through the 'perf list' command.

---

Changelog[v14]
- [Jiri Olsa] Break up independent parts of the patch into separate patches.

Changelog[v21]
- Fix minor conflicts in tools/perf/pmu-events/jevents.c and tools/perf/pmu-events/pmu-events.h

---
 tools/perf/pmu-events/jevents.c    | 32 ++++++++++++++++++++++++--------
 tools/perf/pmu-events/jevents.h    |  3 ++-
 tools/perf/pmu-events/pmu-events.h |  1 +
 3 files changed, 27 insertions(+), 9 deletions(-)
tools/perf/pmu-events/jevents.c +24 -8
index a9ca86d..f550cad 100644
@@ -268,7 +268,7 @@ static void print_events_table_prefix(FILE *fp, const char *tblname)
 }
 
 static int print_events_table_entry(void *data, char *name, char *event,
-				    char *desc)
+				    char *desc, char *long_desc)
 {
 	struct perf_entry_data *pd = data;
 	FILE *outfp = pd->outfp;
@@ -284,6 +284,8 @@ static int print_events_table_entry(void *data, char *name, char *event,
 	fprintf(outfp, "\t.event = \"%s\",\n", event);
 	fprintf(outfp, "\t.desc = \"%s\",\n", desc);
 	fprintf(outfp, "\t.topic = \"%s\",\n", topic);
+	if (long_desc && long_desc[0])
+		fprintf(outfp, "\t.long_desc = \"%s\",\n", long_desc);
 
 	fprintf(outfp, "},\n");
 
@@ -305,7 +307,8 @@ static void print_events_table_suffix(FILE *outfp)
 
 /* Call func with each event in the json file */
 int json_events(const char *fn,
-	  int (*func)(void *data, char *name, char *event, char *desc),
+	  int (*func)(void *data, char *name, char *event, char *desc,
+		      char *long_desc),
 	  void *data)
 {
 	int err = -EIO;
@@ -324,6 +327,8 @@ int json_events(const char *fn,
 	tok = tokens + 1;
 	for (i = 0; i < tokens->size; i++) {
 		char *event = NULL, *desc = NULL, *name = NULL;
+		char *long_desc = NULL;
+		char *extra_desc = NULL;
 		struct msrmap *msr = NULL;
 		jsmntok_t *msrval = NULL;
 		jsmntok_t *precise = NULL;
@@ -349,6 +354,10 @@ int json_events(const char *fn,
 			} else if (json_streq(map, field, "BriefDescription")) {
 				addfield(map, &desc, "", "", val);
 				fixdesc(desc);
+			} else if (json_streq(map, field,
+					     "PublicDescription")) {
+				addfield(map, &long_desc, "", "", val);
+				fixdesc(long_desc);
 			} else if (json_streq(map, field, "PEBS") && nz) {
 				precise = val;
 			} else if (json_streq(map, field, "MSRIndex") && nz) {
@@ -357,10 +366,10 @@ int json_events(const char *fn,
 				msrval = val;
 			} else if (json_streq(map, field, "Errata") &&
 				   !json_streq(map, val, "null")) {
-				addfield(map, &desc, ". ",
+				addfield(map, &extra_desc, ". ",
 					" Spec update: ", val);
 			} else if (json_streq(map, field, "Data_LA") && nz) {
-				addfield(map, &desc, ". ",
+				addfield(map, &extra_desc, ". ",
 					" Supports address when precise",
 					NULL);
 			}
@@ -368,19 +377,26 @@ int json_events(const char *fn,
 		}
 		if (precise && desc && !strstr(desc, "(Precise Event)")) {
 			if (json_streq(map, precise, "2"))
-				addfield(map, &desc, " ", "(Must be precise)",
-						NULL);
+				addfield(map, &extra_desc, " ",
+						"(Must be precise)", NULL);
 			else
-				addfield(map, &desc, " ",
+				addfield(map, &extra_desc, " ",
 						"(Precise event)", NULL);
 		}
+		if (desc && extra_desc)
+			addfield(map, &desc, " ", extra_desc, NULL);
+		if (long_desc && extra_desc)
+			addfield(map, &long_desc, " ", extra_desc, NULL);
 		if (msr != NULL)
 			addfield(map, &event, ",", msr->pname, msrval);
 		fixname(name);
-		err = func(data, name, event, desc);
+
+		err = func(data, name, event, desc, long_desc);
 		free(event);
 		free(desc);
 		free(name);
+		free(long_desc);
+		free(extra_desc);
 		if (err)
 			break;
 		tok += j;
tools/perf/pmu-events/jevents.h +2 -1
index 996601f..b0eb274 100644
@@ -2,7 +2,8 @@
 #define JEVENTS_H 1
 
 int json_events(const char *fn,
-		int (*func)(void *data, char *name, char *event, char *desc),
+		int (*func)(void *data, char *name, char *event, char *desc,
+				char *long_desc),
 		void *data);
 char *get_cpu_str(void);
 
tools/perf/pmu-events/pmu-events.h +1 -0
index 70d5479..2eaef59 100644
@@ -9,6 +9,7 @@ struct pmu_event {
 	const char *event;
 	const char *desc;
 	const char *topic;
+	const char *long_desc;
 };
 
 /*
Signature · 1 line
1.8.3.1
#26 lore raw

[tip:perf/urgent] perf jevents: Add support for long descriptions

perf jevents: Add support for long descriptions
Commit-ID:  794ba54a8393456d503958d6217874e455b6a771
Gitweb:     http://git.kernel.org/tip/794ba54a8393456d503958d6217874e455b6a771
Author:     Sukadev Bhattiprolu <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:47 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:35:46 -0300

perf jevents: Add support for long descriptions

Implement support in jevents to parse long descriptions for events that may have them in the JSON files. A follow on patch will make this long description available to user through the 'perf list' command.

---
 tools/perf/pmu-events/jevents.c    | 32 ++++++++++++++++++++++++--------
 tools/perf/pmu-events/jevents.h    |  3 ++-
 tools/perf/pmu-events/pmu-events.h |  1 +
 3 files changed, 27 insertions(+), 9 deletions(-)
tools/perf/pmu-events/jevents.c +24 -8
index c9bf9a7..13f4284 100644
@@ -269,7 +269,7 @@ static void print_events_table_prefix(FILE *fp, const char *tblname)
 }
 
 static int print_events_table_entry(void *data, char *name, char *event,
-				    char *desc)
+				    char *desc, char *long_desc)
 {
 	struct perf_entry_data *pd = data;
 	FILE *outfp = pd->outfp;
@@ -285,6 +285,8 @@ static int print_events_table_entry(void *data, char *name, char *event,
 	fprintf(outfp, "\t.event = \"%s\",\n", event);
 	fprintf(outfp, "\t.desc = \"%s\",\n", desc);
 	fprintf(outfp, "\t.topic = \"%s\",\n", topic);
+	if (long_desc && long_desc[0])
+		fprintf(outfp, "\t.long_desc = \"%s\",\n", long_desc);
 
 	fprintf(outfp, "},\n");
 
@@ -306,7 +308,8 @@ static void print_events_table_suffix(FILE *outfp)
 
 /* Call func with each event in the json file */
 int json_events(const char *fn,
-	  int (*func)(void *data, char *name, char *event, char *desc),
+	  int (*func)(void *data, char *name, char *event, char *desc,
+		      char *long_desc),
 	  void *data)
 {
 	int err = -EIO;
@@ -325,6 +328,8 @@ int json_events(const char *fn,
 	tok = tokens + 1;
 	for (i = 0; i < tokens->size; i++) {
 		char *event = NULL, *desc = NULL, *name = NULL;
+		char *long_desc = NULL;
+		char *extra_desc = NULL;
 		struct msrmap *msr = NULL;
 		jsmntok_t *msrval = NULL;
 		jsmntok_t *precise = NULL;
@@ -350,6 +355,10 @@ int json_events(const char *fn,
 			} else if (json_streq(map, field, "BriefDescription")) {
 				addfield(map, &desc, "", "", val);
 				fixdesc(desc);
+			} else if (json_streq(map, field,
+					     "PublicDescription")) {
+				addfield(map, &long_desc, "", "", val);
+				fixdesc(long_desc);
 			} else if (json_streq(map, field, "PEBS") && nz) {
 				precise = val;
 			} else if (json_streq(map, field, "MSRIndex") && nz) {
@@ -358,10 +367,10 @@ int json_events(const char *fn,
 				msrval = val;
 			} else if (json_streq(map, field, "Errata") &&
 				   !json_streq(map, val, "null")) {
-				addfield(map, &desc, ". ",
+				addfield(map, &extra_desc, ". ",
 					" Spec update: ", val);
 			} else if (json_streq(map, field, "Data_LA") && nz) {
-				addfield(map, &desc, ". ",
+				addfield(map, &extra_desc, ". ",
 					" Supports address when precise",
 					NULL);
 			}
@@ -369,19 +378,26 @@ int json_events(const char *fn,
 		}
 		if (precise && desc && !strstr(desc, "(Precise Event)")) {
 			if (json_streq(map, precise, "2"))
-				addfield(map, &desc, " ", "(Must be precise)",
-						NULL);
+				addfield(map, &extra_desc, " ",
+						"(Must be precise)", NULL);
 			else
-				addfield(map, &desc, " ",
+				addfield(map, &extra_desc, " ",
 						"(Precise event)", NULL);
 		}
+		if (desc && extra_desc)
+			addfield(map, &desc, " ", extra_desc, NULL);
+		if (long_desc && extra_desc)
+			addfield(map, &long_desc, " ", extra_desc, NULL);
 		if (msr != NULL)
 			addfield(map, &event, ",", msr->pname, msrval);
 		fixname(name);
-		err = func(data, name, event, desc);
+
+		err = func(data, name, event, desc, long_desc);
 		free(event);
 		free(desc);
 		free(name);
+		free(long_desc);
+		free(extra_desc);
 		if (err)
 			break;
 		tok += j;
tools/perf/pmu-events/jevents.h +2 -1
index 996601f..b0eb274 100644
@@ -2,7 +2,8 @@
 #define JEVENTS_H 1
 
 int json_events(const char *fn,
-		int (*func)(void *data, char *name, char *event, char *desc),
+		int (*func)(void *data, char *name, char *event, char *desc,
+				char *long_desc),
 		void *data);
 char *get_cpu_str(void);
 
tools/perf/pmu-events/pmu-events.h +1 -0
index 70d5479..2eaef59 100644
@@ -9,6 +9,7 @@ struct pmu_event {
 	const char *event;
 	const char *desc;
 	const char *topic;
+	const char *long_desc;
 };
 
 /*
#27 lore raw

PATCH v21 11/19 perf, tools: Add alias support for long descriptions

In reply to Sukadev Bhattiprolu

Previously we were dropping the useful longer descriptions that some events have in the event list completely. Now that jevents provides support for longer descriptions (see previous patch), add support for parsing the long descriptions

---

Changelog[v14]
- [Jiri Olsa] Break up independent parts of the patch into separate patches.

---
 tools/perf/util/parse-events.c |  5 +++--
 tools/perf/util/parse-events.h |  3 ++-
 tools/perf/util/pmu.c          | 15 ++++++++++-----
 tools/perf/util/pmu.h          |  4 +++-
 4 files changed, 18 insertions(+), 9 deletions(-)
tools/perf/util/parse-events.c +3 -2
index a3c7739..9abd60d 100644
@@ -2229,7 +2229,8 @@ out_enomem:
 /*
  * Print the help text for the event symbols:
  */
-void print_events(const char *event_glob, bool name_only, bool quiet_flag)
+void print_events(const char *event_glob, bool name_only, bool quiet_flag,
+			bool long_desc)
 {
 	print_symbol_events(event_glob, PERF_TYPE_HARDWARE,
 			    event_symbols_hw, PERF_COUNT_HW_MAX, name_only);
@@ -2239,7 +2240,7 @@ void print_events(const char *event_glob, bool name_only, bool quiet_flag)
 
 	print_hwcache_events(event_glob, name_only);
 
-	print_pmu_events(event_glob, name_only, quiet_flag);
+	print_pmu_events(event_glob, name_only, quiet_flag, long_desc);
 
 	if (event_glob != NULL)
 		return;
tools/perf/util/parse-events.h +2 -1
index 795f2579..7efde4a 100644
@@ -171,7 +171,8 @@ void parse_events_update_lists(struct list_head *list_event,
 void parse_events_evlist_error(struct parse_events_evlist *data,
 			       int idx, const char *str);
 
-void print_events(const char *event_glob, bool name_only, bool quiet);
+void print_events(const char *event_glob, bool name_only, bool quiet,
+		  bool long_desc);
 
 struct event_symbol {
 	const char	*symbol;
tools/perf/util/pmu.c +10 -5
index 2291d2a..43838b3 100644
@@ -223,7 +223,7 @@ static int perf_pmu__parse_snapshot(struct perf_pmu_alias *alias,
 }
 
 static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
-				 char *desc, char *val)
+				 char *desc, char *val, char *long_desc)
 {
 	struct perf_pmu_alias *alias;
 	int ret;
@@ -257,6 +257,8 @@ static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
 	}
 
 	alias->desc = desc ? strdup(desc) : NULL;
+	alias->long_desc = long_desc ? strdup(long_desc) :
+				desc ? strdup(desc) : NULL;
 
 	list_add_tail(&alias->list, list);
 
@@ -274,7 +276,7 @@ static int perf_pmu__new_alias(struct list_head *list, char *dir, char *name, FI
 
 	buf[ret] = 0;
 
-	return __perf_pmu__new_alias(list, dir, name, NULL, buf);
+	return __perf_pmu__new_alias(list, dir, name, NULL, buf, NULL);
 }
 
 static inline bool pmu_alias_info_file(char *name)
@@ -532,7 +534,8 @@ static int pmu_add_cpu_aliases(struct list_head *head)
 
 		/* need type casts to override 'const' */
 		__perf_pmu__new_alias(head, NULL, (char *)pe->name,
-				(char *)pe->desc, (char *)pe->event);
+				(char *)pe->desc, (char *)pe->event,
+				(char *)pe->long_desc);
 	}
 
 out:
@@ -1091,7 +1094,8 @@ static void wordwrap(char *s, int start, int max, int corr)
 	}
 }
 
-void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
+void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
+			bool long_desc)
 {
 	struct perf_pmu *pmu;
 	struct perf_pmu_alias *alias;
@@ -1139,7 +1143,8 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
 			if (!aliases[j].name)
 				goto out_enomem;
 
-			aliases[j].desc = alias->desc;
+			aliases[j].desc = long_desc ? alias->long_desc :
+						alias->desc;
 			j++;
 		}
 		if (pmu->selectable &&
tools/perf/util/pmu.h +3 -1
index 42999c7..1aa614e 100644
@@ -39,6 +39,7 @@ struct perf_pmu_info {
 struct perf_pmu_alias {
 	char *name;
 	char *desc;
+	char *long_desc;
 	struct list_head terms; /* HEAD struct parse_events_term -> list */
 	struct list_head list;  /* ELEM */
 	char unit[UNIT_MAX_LEN+1];
@@ -70,7 +71,8 @@ int perf_pmu__format_parse(char *dir, struct list_head *head);
 
 struct perf_pmu *perf_pmu__scan(struct perf_pmu *pmu);
 
-void print_pmu_events(const char *event_glob, bool name_only, bool quiet);
+void print_pmu_events(const char *event_glob, bool name_only, bool quiet,
+		      bool long_desc);
 bool pmu_have_event(const char *pname, const char *name);
 
 int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt,
Signature · 1 line
1.8.3.1
#28 lore raw

Re: [PATCH v21 11/19] perf, tools: Add alias support for long descriptions

Em Thu, Sep 15, 2016 at 03:24:48PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 10 lines

Previously we were dropping the useful longer descriptions that some events have in the event list completely. Now that jevents provides support for longer descriptions (see previous patch), add support for parsing the long descriptions

---

This patch doesn't build, fixing... probably needs stuff that is in a following patch.

- Arnaldo

Quoted 132 lines

Changelog[v14]
- [Jiri Olsa] Break up independent parts of the patch into separate patches.

---
 tools/perf/util/parse-events.c |  5 +++--
 tools/perf/util/parse-events.h |  3 ++-
 tools/perf/util/pmu.c          | 15 ++++++++++-----
 tools/perf/util/pmu.h          |  4 +++-
 4 files changed, 18 insertions(+), 9 deletions(-)
tools/perf/util/parse-events.c +3 -2
index a3c7739..9abd60d 100644
@@ -2229,7 +2229,8 @@ out_enomem:
 /*
  * Print the help text for the event symbols:
  */
-void print_events(const char *event_glob, bool name_only, bool quiet_flag)
+void print_events(const char *event_glob, bool name_only, bool quiet_flag,
+			bool long_desc)
 {
 	print_symbol_events(event_glob, PERF_TYPE_HARDWARE,
 			    event_symbols_hw, PERF_COUNT_HW_MAX, name_only);
@@ -2239,7 +2240,7 @@ void print_events(const char *event_glob, bool name_only, bool quiet_flag)
 
 	print_hwcache_events(event_glob, name_only);
 
-	print_pmu_events(event_glob, name_only, quiet_flag);
+	print_pmu_events(event_glob, name_only, quiet_flag, long_desc);
 
 	if (event_glob != NULL)
 		return;
tools/perf/util/parse-events.h +2 -1
index 795f2579..7efde4a 100644
@@ -171,7 +171,8 @@ void parse_events_update_lists(struct list_head *list_event,
 void parse_events_evlist_error(struct parse_events_evlist *data,
 			       int idx, const char *str);
 
-void print_events(const char *event_glob, bool name_only, bool quiet);
+void print_events(const char *event_glob, bool name_only, bool quiet,
+		  bool long_desc);
 
 struct event_symbol {
 	const char	*symbol;
tools/perf/util/pmu.c +10 -5
index 2291d2a..43838b3 100644
@@ -223,7 +223,7 @@ static int perf_pmu__parse_snapshot(struct perf_pmu_alias *alias,
 }
 
 static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
-				 char *desc, char *val)
+				 char *desc, char *val, char *long_desc)
 {
 	struct perf_pmu_alias *alias;
 	int ret;
@@ -257,6 +257,8 @@ static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
 	}
 
 	alias->desc = desc ? strdup(desc) : NULL;
+	alias->long_desc = long_desc ? strdup(long_desc) :
+				desc ? strdup(desc) : NULL;
 
 	list_add_tail(&alias->list, list);
 
@@ -274,7 +276,7 @@ static int perf_pmu__new_alias(struct list_head *list, char *dir, char *name, FI
 
 	buf[ret] = 0;
 
-	return __perf_pmu__new_alias(list, dir, name, NULL, buf);
+	return __perf_pmu__new_alias(list, dir, name, NULL, buf, NULL);
 }
 
 static inline bool pmu_alias_info_file(char *name)
@@ -532,7 +534,8 @@ static int pmu_add_cpu_aliases(struct list_head *head)
 
 		/* need type casts to override 'const' */
 		__perf_pmu__new_alias(head, NULL, (char *)pe->name,
-				(char *)pe->desc, (char *)pe->event);
+				(char *)pe->desc, (char *)pe->event,
+				(char *)pe->long_desc);
 	}
 
 out:
@@ -1091,7 +1094,8 @@ static void wordwrap(char *s, int start, int max, int corr)
 	}
 }
 
-void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
+void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
+			bool long_desc)
 {
 	struct perf_pmu *pmu;
 	struct perf_pmu_alias *alias;
@@ -1139,7 +1143,8 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
 			if (!aliases[j].name)
 				goto out_enomem;
 
-			aliases[j].desc = alias->desc;
+			aliases[j].desc = long_desc ? alias->long_desc :
+						alias->desc;
 			j++;
 		}
 		if (pmu->selectable &&
tools/perf/util/pmu.h +3 -1
index 42999c7..1aa614e 100644
@@ -39,6 +39,7 @@ struct perf_pmu_info {
 struct perf_pmu_alias {
 	char *name;
 	char *desc;
+	char *long_desc;
 	struct list_head terms; /* HEAD struct parse_events_term -> list */
 	struct list_head list;  /* ELEM */
 	char unit[UNIT_MAX_LEN+1];
@@ -70,7 +71,8 @@ int perf_pmu__format_parse(char *dir, struct list_head *head);
 
 struct perf_pmu *perf_pmu__scan(struct perf_pmu *pmu);
 
-void print_pmu_events(const char *event_glob, bool name_only, bool quiet);
+void print_pmu_events(const char *event_glob, bool name_only, bool quiet,
+		      bool long_desc);
 bool pmu_have_event(const char *pname, const char *name);
 
 int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt,
Signature · 1 line
1.8.3.1
#29 lore raw

PATCH v21 12/19 perf, tools: Support long descriptions with perf list

In reply to Sukadev Bhattiprolu

Previously we were dropping the useful longer descriptions that some events have in the event list completely. This patch makes them appear with perf list.

Old perf list:

baclears:
baclears.all
[Counts the number of baclears]

vs new:

perf list -v:
...
baclears:
baclears.all
[The BACLEARS event counts the number of times the front end is
resteered, mainly when the Branch Prediction Unit cannot provide a correct prediction and this is corrected by the Branch Address Calculator at the front end. The BACLEARS.ANY event counts the number of baclears for any type of branch]

---

Changelog[v15]
- [Jir Olsa, Andi Kleen] Fix usage strings; update man page.

Changelog[v14]
- [Jiri Olsa] Break up independent parts of the patch into separate patches.

Changelog[v18]:
- Fix minor conflict in tools/perf/builtin-list.c; add long_desc_flag parameter to new print_pmu_events() call site.

Changelog[v21]
- Fix minor conflicts in tools/perf/builtin-list.c

---
 tools/perf/Documentation/perf-list.txt |  6 +++++-
 tools/perf/builtin-list.c              | 16 +++++++++++-----
 2 files changed, 16 insertions(+), 6 deletions(-)
tools/perf/Documentation/perf-list.txt +5 -1
index 72209bc..41857cc 100644
@@ -8,7 +8,7 @@ perf-list - List all symbolic event types
 SYNOPSIS
 --------
 [verse]
-'perf list' [--no-desc] [hw|sw|cache|tracepoint|pmu|event_glob]
+'perf list' [--no-desc] [--long-desc] [hw|sw|cache|tracepoint|pmu|event_glob]
 
 DESCRIPTION
 -----------
@@ -20,6 +20,10 @@ OPTIONS
 --no-desc::
 Don't print descriptions.
 
+-v::
+--long-desc::
+Print longer event descriptions.
+
 
 [[EVENT_MODIFIERS]]
 EVENT MODIFIERS
tools/perf/builtin-list.c +11 -5
index b14cb16..ba9322f 100644
@@ -22,14 +22,17 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	int i;
 	bool raw_dump = false;
+	bool long_desc_flag = false;
 	struct option list_options[] = {
 		OPT_BOOLEAN(0, "raw-dump", &raw_dump, "Dump raw events"),
 		OPT_BOOLEAN('d', "desc", &desc_flag,
 			    "Print extra event descriptions. --no-desc to not print."),
+		OPT_BOOLEAN('v', "long-desc", &long_desc_flag,
+			    "Print longer event descriptions."),
 		OPT_END()
 	};
 	const char * const list_usage[] = {
-		"perf list [--no-desc] [hw|sw|cache|tracepoint|pmu|sdt|event_glob]",
+		"perf list [<options>] [hw|sw|cache|tracepoint|pmu|sdt|event_glob]",
 		NULL
 	};
 
@@ -44,7 +47,7 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 		printf("\nList of pre-defined events (to be used in -e):\n\n");
 
 	if (argc == 0) {
-		print_events(NULL, raw_dump, !desc_flag);
+		print_events(NULL, raw_dump, !desc_flag, long_desc_flag);
 		return 0;
 	}
 
@@ -65,14 +68,16 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			 strcmp(argv[i], "hwcache") == 0)
 			print_hwcache_events(NULL, raw_dump);
 		else if (strcmp(argv[i], "pmu") == 0)
-			print_pmu_events(NULL, raw_dump, !desc_flag);
+			print_pmu_events(NULL, raw_dump, !desc_flag,
+						long_desc_flag);
 		else if (strcmp(argv[i], "sdt") == 0)
 			print_sdt_events(NULL, NULL, raw_dump);
 		else if ((sep = strchr(argv[i], ':')) != NULL) {
 			int sep_idx;
 
 			if (sep == NULL) {
-				print_events(argv[i], raw_dump, !desc_flag);
+				print_events(argv[i], raw_dump, !desc_flag,
+							long_desc_flag);
 				continue;
 			}
 			sep_idx = sep - argv[i];
@@ -94,7 +99,8 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			print_symbol_events(s, PERF_TYPE_SOFTWARE,
 					    event_symbols_sw, PERF_COUNT_SW_MAX, raw_dump);
 			print_hwcache_events(s, raw_dump);
-			print_pmu_events(s, raw_dump, !desc_flag);
+			print_pmu_events(s, raw_dump, !desc_flag,
+						long_desc_flag);
 			print_tracepoint_events(NULL, s, raw_dump);
 			print_sdt_events(NULL, s, raw_dump);
 			free(s);
Signature · 1 line
1.8.3.1
#30 lore raw

Re: [PATCH v21 12/19] perf, tools: Support long descriptions with perf list

Em Thu, Sep 15, 2016 at 03:24:49PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 3 lines

Previously we were dropping the useful longer descriptions that some events have in the event list completely. This patch makes them appear with perf list.

Squashed this with 11/19 so that we don't break bisect.

Quoted 129 lines

Old perf list:

baclears:
baclears.all
[Counts the number of baclears]

vs new:

perf list -v:
...
baclears:
baclears.all
[The BACLEARS event counts the number of times the front end is
resteered, mainly when the Branch Prediction Unit cannot provide a correct prediction and this is corrected by the Branch Address Calculator at the front end. The BACLEARS.ANY event counts the number of baclears for any type of branch]

---

Changelog[v15]
- [Jir Olsa, Andi Kleen] Fix usage strings; update man page.

Changelog[v14]
- [Jiri Olsa] Break up independent parts of the patch into separate patches.

Changelog[v18]:
- Fix minor conflict in tools/perf/builtin-list.c; add long_desc_flag parameter to new print_pmu_events() call site.

Changelog[v21]
- Fix minor conflicts in tools/perf/builtin-list.c

---
 tools/perf/Documentation/perf-list.txt |  6 +++++-
 tools/perf/builtin-list.c              | 16 +++++++++++-----
 2 files changed, 16 insertions(+), 6 deletions(-)
tools/perf/Documentation/perf-list.txt +5 -1
index 72209bc..41857cc 100644
@@ -8,7 +8,7 @@ perf-list - List all symbolic event types
 SYNOPSIS
 --------
 [verse]
-'perf list' [--no-desc] [hw|sw|cache|tracepoint|pmu|event_glob]
+'perf list' [--no-desc] [--long-desc] [hw|sw|cache|tracepoint|pmu|event_glob]
 
 DESCRIPTION
 -----------
@@ -20,6 +20,10 @@ OPTIONS
 --no-desc::
 Don't print descriptions.
 
+-v::
+--long-desc::
+Print longer event descriptions.
+
 
 [[EVENT_MODIFIERS]]
 EVENT MODIFIERS
tools/perf/builtin-list.c +11 -5
index b14cb16..ba9322f 100644
@@ -22,14 +22,17 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	int i;
 	bool raw_dump = false;
+	bool long_desc_flag = false;
 	struct option list_options[] = {
 		OPT_BOOLEAN(0, "raw-dump", &raw_dump, "Dump raw events"),
 		OPT_BOOLEAN('d', "desc", &desc_flag,
 			    "Print extra event descriptions. --no-desc to not print."),
+		OPT_BOOLEAN('v', "long-desc", &long_desc_flag,
+			    "Print longer event descriptions."),
 		OPT_END()
 	};
 	const char * const list_usage[] = {
-		"perf list [--no-desc] [hw|sw|cache|tracepoint|pmu|sdt|event_glob]",
+		"perf list [<options>] [hw|sw|cache|tracepoint|pmu|sdt|event_glob]",
 		NULL
 	};
 
@@ -44,7 +47,7 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 		printf("\nList of pre-defined events (to be used in -e):\n\n");
 
 	if (argc == 0) {
-		print_events(NULL, raw_dump, !desc_flag);
+		print_events(NULL, raw_dump, !desc_flag, long_desc_flag);
 		return 0;
 	}
 
@@ -65,14 +68,16 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			 strcmp(argv[i], "hwcache") == 0)
 			print_hwcache_events(NULL, raw_dump);
 		else if (strcmp(argv[i], "pmu") == 0)
-			print_pmu_events(NULL, raw_dump, !desc_flag);
+			print_pmu_events(NULL, raw_dump, !desc_flag,
+						long_desc_flag);
 		else if (strcmp(argv[i], "sdt") == 0)
 			print_sdt_events(NULL, NULL, raw_dump);
 		else if ((sep = strchr(argv[i], ':')) != NULL) {
 			int sep_idx;
 
 			if (sep == NULL) {
-				print_events(argv[i], raw_dump, !desc_flag);
+				print_events(argv[i], raw_dump, !desc_flag,
+							long_desc_flag);
 				continue;
 			}
 			sep_idx = sep - argv[i];
@@ -94,7 +99,8 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			print_symbol_events(s, PERF_TYPE_SOFTWARE,
 					    event_symbols_sw, PERF_COUNT_SW_MAX, raw_dump);
 			print_hwcache_events(s, raw_dump);
-			print_pmu_events(s, raw_dump, !desc_flag);
+			print_pmu_events(s, raw_dump, !desc_flag,
+						long_desc_flag);
 			print_tracepoint_events(NULL, s, raw_dump);
 			print_sdt_events(NULL, s, raw_dump);
 			free(s);
Signature · 1 line
1.8.3.1
#31 lore raw

[tip:perf/urgent] perf list: Support long jevents descriptions

In reply to Sukadev Bhattiprolu
perf list: Support long jevents descriptions
Commit-ID:  c8d6828a6510c9363180ebf662b51d032e0eb85a
Gitweb:     http://git.kernel.org/tip/c8d6828a6510c9363180ebf662b51d032e0eb85a
Author:     Sukadev Bhattiprolu <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:48 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:35:47 -0300

perf list: Support long jevents descriptions

Previously we were dropping the useful longer descriptions that some events have in the event list completely. This patch makes them appear with perf list.

Old perf list:

baclears:
baclears.all
[Counts the number of baclears]

vs new:

perf list -v:
...
baclears:
baclears.all
[The BACLEARS event counts the number of times the front end is
resteered, mainly when the Branch Prediction Unit cannot provide a correct prediction and this is corrected by the Branch Address Calculator at the front end. The BACLEARS.ANY event counts the number of baclears for any type of branch]

---
 tools/perf/Documentation/perf-list.txt |  6 +++++-
 tools/perf/builtin-list.c              | 16 +++++++++++-----
 tools/perf/util/parse-events.c         |  5 +++--
 tools/perf/util/parse-events.h         |  3 ++-
 tools/perf/util/pmu.c                  | 15 ++++++++++-----
 tools/perf/util/pmu.h                  |  4 +++-
 6 files changed, 34 insertions(+), 15 deletions(-)
tools/perf/Documentation/perf-list.txt +5 -1
index 72209bc..41857cc 100644
@@ -8,7 +8,7 @@ perf-list - List all symbolic event types
 SYNOPSIS
 --------
 [verse]
-'perf list' [--no-desc] [hw|sw|cache|tracepoint|pmu|event_glob]
+'perf list' [--no-desc] [--long-desc] [hw|sw|cache|tracepoint|pmu|event_glob]
 
 DESCRIPTION
 -----------
@@ -20,6 +20,10 @@ OPTIONS
 --no-desc::
 Don't print descriptions.
 
+-v::
+--long-desc::
+Print longer event descriptions.
+
 
 [[EVENT_MODIFIERS]]
 EVENT MODIFIERS
tools/perf/builtin-list.c +11 -5
index b14cb16..ba9322f 100644
@@ -22,14 +22,17 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	int i;
 	bool raw_dump = false;
+	bool long_desc_flag = false;
 	struct option list_options[] = {
 		OPT_BOOLEAN(0, "raw-dump", &raw_dump, "Dump raw events"),
 		OPT_BOOLEAN('d', "desc", &desc_flag,
 			    "Print extra event descriptions. --no-desc to not print."),
+		OPT_BOOLEAN('v', "long-desc", &long_desc_flag,
+			    "Print longer event descriptions."),
 		OPT_END()
 	};
 	const char * const list_usage[] = {
-		"perf list [--no-desc] [hw|sw|cache|tracepoint|pmu|sdt|event_glob]",
+		"perf list [<options>] [hw|sw|cache|tracepoint|pmu|sdt|event_glob]",
 		NULL
 	};
 
@@ -44,7 +47,7 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 		printf("\nList of pre-defined events (to be used in -e):\n\n");
 
 	if (argc == 0) {
-		print_events(NULL, raw_dump, !desc_flag);
+		print_events(NULL, raw_dump, !desc_flag, long_desc_flag);
 		return 0;
 	}
 
@@ -65,14 +68,16 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			 strcmp(argv[i], "hwcache") == 0)
 			print_hwcache_events(NULL, raw_dump);
 		else if (strcmp(argv[i], "pmu") == 0)
-			print_pmu_events(NULL, raw_dump, !desc_flag);
+			print_pmu_events(NULL, raw_dump, !desc_flag,
+						long_desc_flag);
 		else if (strcmp(argv[i], "sdt") == 0)
 			print_sdt_events(NULL, NULL, raw_dump);
 		else if ((sep = strchr(argv[i], ':')) != NULL) {
 			int sep_idx;
 
 			if (sep == NULL) {
-				print_events(argv[i], raw_dump, !desc_flag);
+				print_events(argv[i], raw_dump, !desc_flag,
+							long_desc_flag);
 				continue;
 			}
 			sep_idx = sep - argv[i];
@@ -94,7 +99,8 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			print_symbol_events(s, PERF_TYPE_SOFTWARE,
 					    event_symbols_sw, PERF_COUNT_SW_MAX, raw_dump);
 			print_hwcache_events(s, raw_dump);
-			print_pmu_events(s, raw_dump, !desc_flag);
+			print_pmu_events(s, raw_dump, !desc_flag,
+						long_desc_flag);
 			print_tracepoint_events(NULL, s, raw_dump);
 			print_sdt_events(NULL, s, raw_dump);
 			free(s);
tools/perf/util/parse-events.c +3 -2
index 3966ad7..8f88f63 100644
@@ -2263,7 +2263,8 @@ out_enomem:
 /*
  * Print the help text for the event symbols:
  */
-void print_events(const char *event_glob, bool name_only, bool quiet_flag)
+void print_events(const char *event_glob, bool name_only, bool quiet_flag,
+			bool long_desc)
 {
 	print_symbol_events(event_glob, PERF_TYPE_HARDWARE,
 			    event_symbols_hw, PERF_COUNT_HW_MAX, name_only);
@@ -2273,7 +2274,7 @@ void print_events(const char *event_glob, bool name_only, bool quiet_flag)
 
 	print_hwcache_events(event_glob, name_only);
 
-	print_pmu_events(event_glob, name_only, quiet_flag);
+	print_pmu_events(event_glob, name_only, quiet_flag, long_desc);
 
 	if (event_glob != NULL)
 		return;
tools/perf/util/parse-events.h +2 -1
index 3bf376b..da246a3 100644
@@ -172,7 +172,8 @@ void parse_events_update_lists(struct list_head *list_event,
 void parse_events_evlist_error(struct parse_events_evlist *data,
 			       int idx, const char *str);
 
-void print_events(const char *event_glob, bool name_only, bool quiet);
+void print_events(const char *event_glob, bool name_only, bool quiet,
+		  bool long_desc);
 
 struct event_symbol {
 	const char	*symbol;
tools/perf/util/pmu.c +10 -5
index 79242cf..8ff382c 100644
@@ -223,7 +223,7 @@ static int perf_pmu__parse_snapshot(struct perf_pmu_alias *alias,
 }
 
 static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
-				 char *desc, char *val)
+				 char *desc, char *val, char *long_desc)
 {
 	struct perf_pmu_alias *alias;
 	int ret;
@@ -257,6 +257,8 @@ static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
 	}
 
 	alias->desc = desc ? strdup(desc) : NULL;
+	alias->long_desc = long_desc ? strdup(long_desc) :
+				desc ? strdup(desc) : NULL;
 
 	list_add_tail(&alias->list, list);
 
@@ -274,7 +276,7 @@ static int perf_pmu__new_alias(struct list_head *list, char *dir, char *name, FI
 
 	buf[ret] = 0;
 
-	return __perf_pmu__new_alias(list, dir, name, NULL, buf);
+	return __perf_pmu__new_alias(list, dir, name, NULL, buf, NULL);
 }
 
 static inline bool pmu_alias_info_file(char *name)
@@ -532,7 +534,8 @@ static void pmu_add_cpu_aliases(struct list_head *head)
 
 		/* need type casts to override 'const' */
 		__perf_pmu__new_alias(head, NULL, (char *)pe->name,
-				(char *)pe->desc, (char *)pe->event);
+				(char *)pe->desc, (char *)pe->event,
+				(char *)pe->long_desc);
 	}
 
 out:
@@ -1090,7 +1093,8 @@ static void wordwrap(char *s, int start, int max, int corr)
 	}
 }
 
-void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
+void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
+			bool long_desc)
 {
 	struct perf_pmu *pmu;
 	struct perf_pmu_alias *alias;
@@ -1138,7 +1142,8 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag)
 			if (!aliases[j].name)
 				goto out_enomem;
 
-			aliases[j].desc = alias->desc;
+			aliases[j].desc = long_desc ? alias->long_desc :
+						alias->desc;
 			j++;
 		}
 		if (pmu->selectable &&
tools/perf/util/pmu.h +3 -1
index 7b47192..2fb8aa0 100644
@@ -41,6 +41,7 @@ struct perf_pmu_info {
 struct perf_pmu_alias {
 	char *name;
 	char *desc;
+	char *long_desc;
 	struct list_head terms; /* HEAD struct parse_events_term -> list */
 	struct list_head list;  /* ELEM */
 	char unit[UNIT_MAX_LEN+1];
@@ -72,7 +73,8 @@ int perf_pmu__format_parse(char *dir, struct list_head *head);
 
 struct perf_pmu *perf_pmu__scan(struct perf_pmu *pmu);
 
-void print_pmu_events(const char *event_glob, bool name_only, bool quiet);
+void print_pmu_events(const char *event_glob, bool name_only, bool quiet,
+		      bool long_desc);
 bool pmu_have_event(const char *pname, const char *name);
 
 int perf_pmu__scan_file(struct perf_pmu *pmu, const char *name, const char *fmt,
#32 lore raw

PATCH v21 13/19 perf, tools: Add support for event list topics

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

Add support to group the output of perf list by the Topic field in the JSON file.

Example output:

% perf list
...
Cache:
l1d.replacement
[L1D data line replacements]
l1d_pend_miss.pending
[L1D miss oustandings duration in cycles]
l1d_pend_miss.pending_cycles
[Cycles with L1D load Misses outstanding]
l2_l1d_wb_rqsts.all
[Not rejected writebacks from L1D to L2 cache lines in any state]
l2_l1d_wb_rqsts.hit_e
[Not rejected writebacks from L1D to L2 cache lines in E state]
l2_l1d_wb_rqsts.hit_m
[Not rejected writebacks from L1D to L2 cache lines in M state]

...
Pipeline:
arith.fpu_div
[Divide operations executed]
arith.fpu_div_active
[Cycles when divider is busy executing divide operations]
baclears.any
[Counts the total number when the front end is resteered, mainly when the BPU cannot provide a correct prediction and this is corrected by other branch handling mechanisms at the front end]
br_inst_exec.all_branches
[Speculative and retired branches]
br_inst_exec.all_conditional
[Speculative and retired macro-conditional branches]
br_inst_exec.all_direct_jmp
[Speculative and retired macro-unconditional branches excluding calls and indirects]
br_inst_exec.all_direct_near_call
[Speculative and retired direct near calls]
br_inst_exec.all_indirect_jump_non_call_ret

---

Changelog[v14]
- [Jiri Olsa] Move jevents support for Topic to a separate patch.

---
 tools/perf/util/pmu.c | 37 +++++++++++++++++++++++++++----------
 tools/perf/util/pmu.h |  1 +
 2 files changed, 28 insertions(+), 10 deletions(-)
tools/perf/util/pmu.c +27 -10
index 43838b3..ac097fc 100644
@@ -223,7 +223,8 @@ static int perf_pmu__parse_snapshot(struct perf_pmu_alias *alias,
 }
 
 static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
-				 char *desc, char *val, char *long_desc)
+				 char *desc, char *val, char *long_desc,
+				 char *topic)
 {
 	struct perf_pmu_alias *alias;
 	int ret;
@@ -259,6 +260,7 @@ static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
 	alias->desc = desc ? strdup(desc) : NULL;
 	alias->long_desc = long_desc ? strdup(long_desc) :
 				desc ? strdup(desc) : NULL;
+	alias->topic = topic ? strdup(topic) : NULL;
 
 	list_add_tail(&alias->list, list);
 
@@ -276,7 +278,7 @@ static int perf_pmu__new_alias(struct list_head *list, char *dir, char *name, FI
 
 	buf[ret] = 0;
 
-	return __perf_pmu__new_alias(list, dir, name, NULL, buf, NULL);
+	return __perf_pmu__new_alias(list, dir, name, NULL, buf, NULL, NULL);
 }
 
 static inline bool pmu_alias_info_file(char *name)
@@ -535,7 +537,7 @@ static int pmu_add_cpu_aliases(struct list_head *head)
 		/* need type casts to override 'const' */
 		__perf_pmu__new_alias(head, NULL, (char *)pe->name,
 				(char *)pe->desc, (char *)pe->event,
-				(char *)pe->long_desc);
+				(char *)pe->long_desc, (char *)pe->topic);
 	}
 
 out:
@@ -1056,19 +1058,26 @@ static char *format_alias_or(char *buf, int len, struct perf_pmu *pmu,
 	return buf;
 }
 
-struct pair {
+struct sevent {
 	char *name;
 	char *desc;
+	char *topic;
 };
 
-static int cmp_pair(const void *a, const void *b)
+static int cmp_sevent(const void *a, const void *b)
 {
-	const struct pair *as = a;
-	const struct pair *bs = b;
+	const struct sevent *as = a;
+	const struct sevent *bs = b;
 
 	/* Put extra events last */
 	if (!!as->desc != !!bs->desc)
 		return !!as->desc - !!bs->desc;
+	if (as->topic && bs->topic) {
+		int n = strcmp(as->topic, bs->topic);
+
+		if (n)
+			return n;
+	}
 	return strcmp(as->name, bs->name);
 }
 
@@ -1102,9 +1111,10 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
 	char buf[1024];
 	int printed = 0;
 	int len, j;
-	struct pair *aliases;
+	struct sevent *aliases;
 	int numdesc = 0;
 	int columns = pager_get_columns();
+	char *topic = NULL;
 
 	pmu = NULL;
 	len = 0;
@@ -1114,7 +1124,7 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
 		if (pmu->selectable)
 			len++;
 	}
-	aliases = zalloc(sizeof(struct pair) * len);
+	aliases = zalloc(sizeof(struct sevent) * len);
 	if (!aliases)
 		goto out_enomem;
 	pmu = NULL;
@@ -1145,6 +1155,7 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
 
 			aliases[j].desc = long_desc ? alias->long_desc :
 						alias->desc;
+			aliases[j].topic = alias->topic;
 			j++;
 		}
 		if (pmu->selectable &&
@@ -1157,7 +1168,7 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
 		}
 	}
 	len = j;
-	qsort(aliases, len, sizeof(struct pair), cmp_pair);
+	qsort(aliases, len, sizeof(struct sevent), cmp_sevent);
 	for (j = 0; j < len; j++) {
 		if (name_only) {
 			printf("%s ", aliases[j].name);
@@ -1166,6 +1177,12 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
 		if (aliases[j].desc && !quiet_flag) {
 			if (numdesc++ == 0)
 				printf("\n");
+			if (aliases[j].topic && (!topic ||
+					strcmp(topic, aliases[j].topic))) {
+				printf("%s%s:\n", topic ? "\n" : "",
+						aliases[j].topic);
+				topic = aliases[j].topic;
+			}
 			printf("  %-50s\n", aliases[j].name);
 			printf("%*s", 8, "[");
 			wordwrap(aliases[j].desc, 8, columns, 0);
tools/perf/util/pmu.h +1 -0
index 1aa614e..de26919 100644
@@ -40,6 +40,7 @@ struct perf_pmu_alias {
 	char *name;
 	char *desc;
 	char *long_desc;
+	char *topic;
 	struct list_head terms; /* HEAD struct parse_events_term -> list */
 	struct list_head list;  /* ELEM */
 	char unit[UNIT_MAX_LEN+1];
Signature · 1 line
1.8.3.1
#33 lore raw

[tip:perf/urgent] perf list jevents: Add support for event list topics

perf list jevents: Add support for event list topics
Commit-ID:  dd5f10368a5f27ec926395056d959513ca4578cc
Gitweb:     http://git.kernel.org/tip/dd5f10368a5f27ec926395056d959513ca4578cc
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:50 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:35:47 -0300

perf list jevents: Add support for event list topics

Add support to group the output of perf list by the Topic field in the JSON file.

Example output:

% perf list
...
Cache:
l1d.replacement
[L1D data line replacements]
l1d_pend_miss.pending
[L1D miss oustandings duration in cycles]
l1d_pend_miss.pending_cycles
[Cycles with L1D load Misses outstanding]
l2_l1d_wb_rqsts.all
[Not rejected writebacks from L1D to L2 cache lines in any state]
l2_l1d_wb_rqsts.hit_e
[Not rejected writebacks from L1D to L2 cache lines in E state]
l2_l1d_wb_rqsts.hit_m
[Not rejected writebacks from L1D to L2 cache lines in M state]

...
Pipeline:
arith.fpu_div
[Divide operations executed]
arith.fpu_div_active
[Cycles when divider is busy executing divide operations]
baclears.any
[Counts the total number when the front end is resteered, mainly when the BPU cannot provide a correct prediction and this is corrected by other branch handling mechanisms at the front end]
br_inst_exec.all_branches
[Speculative and retired branches]
br_inst_exec.all_conditional
[Speculative and retired macro-conditional branches]
br_inst_exec.all_direct_jmp
[Speculative and retired macro-unconditional branches excluding calls and indirects]
br_inst_exec.all_direct_near_call
[Speculative and retired direct near calls]
br_inst_exec.all_indirect_jump_non_call_ret

---
 tools/perf/util/pmu.c | 37 +++++++++++++++++++++++++++----------
 tools/perf/util/pmu.h |  1 +
 2 files changed, 28 insertions(+), 10 deletions(-)
tools/perf/util/pmu.c +27 -10
index 8ff382c..b1474dc 100644
@@ -223,7 +223,8 @@ static int perf_pmu__parse_snapshot(struct perf_pmu_alias *alias,
 }
 
 static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
-				 char *desc, char *val, char *long_desc)
+				 char *desc, char *val, char *long_desc,
+				 char *topic)
 {
 	struct perf_pmu_alias *alias;
 	int ret;
@@ -259,6 +260,7 @@ static int __perf_pmu__new_alias(struct list_head *list, char *dir, char *name,
 	alias->desc = desc ? strdup(desc) : NULL;
 	alias->long_desc = long_desc ? strdup(long_desc) :
 				desc ? strdup(desc) : NULL;
+	alias->topic = topic ? strdup(topic) : NULL;
 
 	list_add_tail(&alias->list, list);
 
@@ -276,7 +278,7 @@ static int perf_pmu__new_alias(struct list_head *list, char *dir, char *name, FI
 
 	buf[ret] = 0;
 
-	return __perf_pmu__new_alias(list, dir, name, NULL, buf, NULL);
+	return __perf_pmu__new_alias(list, dir, name, NULL, buf, NULL, NULL);
 }
 
 static inline bool pmu_alias_info_file(char *name)
@@ -535,7 +537,7 @@ static void pmu_add_cpu_aliases(struct list_head *head)
 		/* need type casts to override 'const' */
 		__perf_pmu__new_alias(head, NULL, (char *)pe->name,
 				(char *)pe->desc, (char *)pe->event,
-				(char *)pe->long_desc);
+				(char *)pe->long_desc, (char *)pe->topic);
 	}
 
 out:
@@ -1055,19 +1057,26 @@ static char *format_alias_or(char *buf, int len, struct perf_pmu *pmu,
 	return buf;
 }
 
-struct pair {
+struct sevent {
 	char *name;
 	char *desc;
+	char *topic;
 };
 
-static int cmp_pair(const void *a, const void *b)
+static int cmp_sevent(const void *a, const void *b)
 {
-	const struct pair *as = a;
-	const struct pair *bs = b;
+	const struct sevent *as = a;
+	const struct sevent *bs = b;
 
 	/* Put extra events last */
 	if (!!as->desc != !!bs->desc)
 		return !!as->desc - !!bs->desc;
+	if (as->topic && bs->topic) {
+		int n = strcmp(as->topic, bs->topic);
+
+		if (n)
+			return n;
+	}
 	return strcmp(as->name, bs->name);
 }
 
@@ -1101,9 +1110,10 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
 	char buf[1024];
 	int printed = 0;
 	int len, j;
-	struct pair *aliases;
+	struct sevent *aliases;
 	int numdesc = 0;
 	int columns = pager_get_columns();
+	char *topic = NULL;
 
 	pmu = NULL;
 	len = 0;
@@ -1113,7 +1123,7 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
 		if (pmu->selectable)
 			len++;
 	}
-	aliases = zalloc(sizeof(struct pair) * len);
+	aliases = zalloc(sizeof(struct sevent) * len);
 	if (!aliases)
 		goto out_enomem;
 	pmu = NULL;
@@ -1144,6 +1154,7 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
 
 			aliases[j].desc = long_desc ? alias->long_desc :
 						alias->desc;
+			aliases[j].topic = alias->topic;
 			j++;
 		}
 		if (pmu->selectable &&
@@ -1156,7 +1167,7 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
 		}
 	}
 	len = j;
-	qsort(aliases, len, sizeof(struct pair), cmp_pair);
+	qsort(aliases, len, sizeof(struct sevent), cmp_sevent);
 	for (j = 0; j < len; j++) {
 		if (name_only) {
 			printf("%s ", aliases[j].name);
@@ -1165,6 +1176,12 @@ void print_pmu_events(const char *event_glob, bool name_only, bool quiet_flag,
 		if (aliases[j].desc && !quiet_flag) {
 			if (numdesc++ == 0)
 				printf("\n");
+			if (aliases[j].topic && (!topic ||
+					strcmp(topic, aliases[j].topic))) {
+				printf("%s%s:\n", topic ? "\n" : "",
+						aliases[j].topic);
+				topic = aliases[j].topic;
+			}
 			printf("  %-50s\n", aliases[j].name);
 			printf("%*s", 8, "[");
 			wordwrap(aliases[j].desc, 8, columns, 0);
tools/perf/util/pmu.h +1 -0
index 2fb8aa0..2571203 100644
@@ -42,6 +42,7 @@ struct perf_pmu_alias {
 	char *name;
 	char *desc;
 	char *long_desc;
+	char *topic;
 	struct list_head terms; /* HEAD struct parse_events_term -> list */
 	struct list_head list;  /* ELEM */
 	char unit[UNIT_MAX_LEN+1];
#34 lore raw

PATCH v21 14/19 perf, tools, jevents: Handle header line in mapfile

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

To work with existing mapfiles, assume that the first line in 'mapfile.csv' is a header line and skip over it.

---

Changelog[v2]
All architectures may not use the "Family" to identify. So, assume first line is header.

---
 tools/perf/pmu-events/jevents.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
tools/perf/pmu-events/jevents.c +7 -2
index f550cad..9cdfbaa 100644
@@ -492,7 +492,12 @@ static int process_mapfile(FILE *outfp, char *fpath)
 
 	print_mapping_table_prefix(outfp);
 
-	line_num = 0;
+	/* Skip first line (header) */
+	p = fgets(line, n, mapfp);
+	if (!p)
+		goto out;
+
+	line_num = 1;
 	while (1) {
 		char *cpuid, *version, *type, *fname;
 
@@ -536,8 +541,8 @@ static int process_mapfile(FILE *outfp, char *fpath)
 		fprintf(outfp, "},\n");
 	}
 
+out:
 	print_mapping_table_suffix(outfp);
-
 	return 0;
 }
 
Signature · 1 line
1.8.3.1
#35 lore raw

Re: [PATCH v21 14/19] perf, tools, jevents: Handle header line in mapfile

Em Thu, Sep 15, 2016 at 03:24:51PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 4 lines

From: Andi Kleen <[email protected]>

To work with existing mapfiles, assume that the first line in 'mapfile.csv' is a header line and skip over it.

Moved this more to the front of this patchkit so that when testing it, reproducing the tests performed earlier, one don't have to remove the header.

- Arnaldo

Quoted 43 lines

---

Changelog[v2]
All architectures may not use the "Family" to identify. So, assume first line is header.

---
 tools/perf/pmu-events/jevents.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
tools/perf/pmu-events/jevents.c +7 -2
index f550cad..9cdfbaa 100644
@@ -492,7 +492,12 @@ static int process_mapfile(FILE *outfp, char *fpath)
 
 	print_mapping_table_prefix(outfp);
 
-	line_num = 0;
+	/* Skip first line (header) */
+	p = fgets(line, n, mapfp);
+	if (!p)
+		goto out;
+
+	line_num = 1;
 	while (1) {
 		char *cpuid, *version, *type, *fname;
 
@@ -536,8 +541,8 @@ static int process_mapfile(FILE *outfp, char *fpath)
 		fprintf(outfp, "},\n");
 	}
 
+out:
 	print_mapping_table_suffix(outfp);
-
 	return 0;
 }
 
Signature · 1 line
1.8.3.1
#36 lore raw

[tip:perf/urgent] perf jevents: Handle header line in mapfile

In reply to Sukadev Bhattiprolu
perf jevents: Handle header line in mapfile
Commit-ID:  dc720ffc9863b618a192f5949234c54aad00ed24
Gitweb:     http://git.kernel.org/tip/dc720ffc9863b618a192f5949234c54aad00ed24
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:51 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:33:54 -0300

perf jevents: Handle header line in mapfile

To work with existing mapfiles, assume that the first line in 'mapfile.csv' is a header line and skip over it.

---
 tools/perf/pmu-events/jevents.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)
tools/perf/pmu-events/jevents.c +7 -2
index c4c074a..c9bf9a7 100644
@@ -477,7 +477,12 @@ static int process_mapfile(FILE *outfp, char *fpath)
 
 	print_mapping_table_prefix(outfp);
 
-	line_num = 0;
+	/* Skip first line (header) */
+	p = fgets(line, n, mapfp);
+	if (!p)
+		goto out;
+
+	line_num = 1;
 	while (1) {
 		char *cpuid, *version, *type, *fname;
 
@@ -521,8 +526,8 @@ static int process_mapfile(FILE *outfp, char *fpath)
 		fprintf(outfp, "},\n");
 	}
 
+out:
 	print_mapping_table_suffix(outfp);
-
 	return 0;
 }
 
#37 lore raw

PATCH v21 15/19 perf, tools: Add README for info on parsing JSON/map files

In reply to Sukadev Bhattiprolu

---
Changelog[v21]
- Update README to reflect the Topics.json directory tree layout.

---
 tools/perf/pmu-events/README | 148 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 148 insertions(+)
 create mode 100644 tools/perf/pmu-events/README
tools/perf/pmu-events/README new file +148 -0
new file mode 100644
index 0000000..c5ee208e
@@ -0,0 +1,148 @@
+
+The contents of this directory allow users to specify PMU events in their
+CPUs by their symbolic names rather than raw event codes (see example below).
+
+The main program in this directory, is the 'jevents', which is built and
+executed _BEFORE_ the perf binary itself is built.
+
+The 'jevents' program tries to locate and process JSON files in the directory
+tree tools/perf/pmu-events/arch/foo.
+
+	- Regular files with '.json' extension in the name are assumed to be
+	  JSON files, each of which describes a set of PMU events.
+
+	- Regular files with basename starting with 'mapfile.csv' are assumed
+	  to be a CSV file that maps a specific CPU to its set of PMU events.
+	  (see below for mapfile format)
+
+	- Directories are traversed, but all other files are ignored.
+
+The PMU events supported by a CPU model are expected to grouped into topics
+such as Pipelining, Cache, Memory, Floating-point etc. All events for a topic
+should be placed in a separate JSON file - where the file name identifies
+the topic. Eg: "Floating-point.json".
+
+All the topic JSON files for a CPU model/family should be in a separate
+sub directory. Thus for the Silvermont X86 CPU:
+
+	$ ls tools/perf/pmu-events/arch/x86/Silvermont_core
+	Cache.json 	Memory.json 	Virtual-Memory.json
+	Frontend.json 	Pipeline.json
+
+Using the JSON files and the mapfile, 'jevents' generates the C source file,
+'pmu-events.c', which encodes the two sets of tables:
+
+	- Set of 'PMU events tables' for all known CPUs in the architecture,
+	  (one table like the following, per JSON file; table name 'pme_power8'
+	  is derived from JSON file name, 'power8.json').
+
+		struct pmu_event pme_power8[] = {
+
+			...
+
+			{
+				.name = "pm_1plus_ppc_cmpl",
+				.event = "event=0x100f2",
+				.desc = "1 or more ppc insts finished,",
+			},
+
+			...
+		}
+
+	- A 'mapping table' that maps each CPU of the architecture, to its
+	  'PMU events table'
+
+		struct pmu_events_map pmu_events_map[] = {
+		{
+			.cpuid = "004b0000",
+			.version = "1",
+			.type = "core",
+			.table = pme_power8
+		},
+			...
+
+		};
+
+After the 'pmu-events.c' is generated, it is compiled and the resulting
+'pmu-events.o' is added to 'libperf.a' which is then used to build perf.
+
+NOTES:
+	1. Several CPUs can support same set of events and hence use a common
+	   JSON file. Hence several entries in the pmu_events_map[] could map
+	   to a single 'PMU events table'.
+
+	2. The 'pmu-events.h' has an extern declaration for the mapping table
+	   and the generated 'pmu-events.c' defines this table.
+
+	3. _All_ known CPU tables for architecture are included in the perf
+	   binary.
+
+At run time, perf determines the actual CPU it is running on, finds the
+matching events table and builds aliases for those events. This allows
+users to specify events by their name:
+
+	$ perf stat -e pm_1plus_ppc_cmpl sleep 1
+
+where 'pm_1plus_ppc_cmpl' is a Power8 PMU event.
+
+In case of errors when processing files in the tools/perf/pmu-events/arch
+directory, 'jevents' tries to create an empty mapping file to allow the perf
+build to succeed even if the PMU event aliases cannot be used.
+
+However some errors in processing may cause the perf build to fail.
+
+Mapfile format
+===============
+
+The mapfile enables multiple CPU models to share a single set of PMU events.
+It is required even if such mapping is 1:1.
+
+The mapfile.csv format is expected to be:
+
+	Header line
+	CPUID,Version,Dir/path/name,Type
+
+where:
+
+	Comma:
+		is the required field delimiter (i.e other fields cannot
+		have commas within them).
+
+	Comments:
+		Lines in which the first character is either '\n' or '#'
+		are ignored.
+
+	Header line
+		The header line is the first line in the file, which is
+		always _IGNORED_. It can empty.
+
+	CPUID:
+		CPUID is an arch-specific char string, that can be used
+		to identify CPU (and associate it with a set of PMU events
+		it supports). Multiple CPUIDS can point to the same
+		File/path/name.json.
+
+		Example:
+			CPUID == 'GenuineIntel-6-2E' (on x86).
+			CPUID == '004b0100' (PVR value in Powerpc)
+	Version:
+		is the Version of the mapfile.
+
+	Dir/path/name:
+		is the pathname to the directory containing the CPU's JSON
+		files, relative to the directory containing the mapfile.csv
+
+	Type:
+		indicates whether the events or "core" or "uncore" events.
+
+
+	Eg:
+
+	$ grep Silvermont tools/perf/pmu-events/arch/x86/mapfile.csv
+	GenuineIntel-6-37,V13,Silvermont_core,core
+	GenuineIntel-6-4D,V13,Silvermont_core,core
+	GenuineIntel-6-4C,V13,Silvermont_core,core
+
+	i.e the three CPU models use the JSON files (i.e PMU events) listed
+	in the directory 'tools/perf/pmu-events/arch/x86/Silvermont_core'.
+
Signature · 1 line
1.8.3.1
#38 lore raw

Re: [PATCH v21 15/19] perf, tools: Add README for info on parsing JSON/map files

Em Thu, Sep 15, 2016 at 03:24:52PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 6 lines

---
Changelog[v21]
- Update README to reflect the Topics.json directory tree layout.

Fixed this:

[acme@jouet linux]$ am /wb/1.patch
Applying: perf, tools: Add README for info on parsing JSON/map files .git/rebase-apply/patch:162: new blank line at EOF.
+
warning: 1 line adds whitespace errors. tools/perf/pmu-events/README:148: new blank line at EOF. [acme@jouet linux]$

Quoted 161 lines
---
 tools/perf/pmu-events/README | 148 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 148 insertions(+)
 create mode 100644 tools/perf/pmu-events/README
tools/perf/pmu-events/README new file +148 -0
new file mode 100644
index 0000000..c5ee208e
@@ -0,0 +1,148 @@
+
+The contents of this directory allow users to specify PMU events in their
+CPUs by their symbolic names rather than raw event codes (see example below).
+
+The main program in this directory, is the 'jevents', which is built and
+executed _BEFORE_ the perf binary itself is built.
+
+The 'jevents' program tries to locate and process JSON files in the directory
+tree tools/perf/pmu-events/arch/foo.
+
+	- Regular files with '.json' extension in the name are assumed to be
+	  JSON files, each of which describes a set of PMU events.
+
+	- Regular files with basename starting with 'mapfile.csv' are assumed
+	  to be a CSV file that maps a specific CPU to its set of PMU events.
+	  (see below for mapfile format)
+
+	- Directories are traversed, but all other files are ignored.
+
+The PMU events supported by a CPU model are expected to grouped into topics
+such as Pipelining, Cache, Memory, Floating-point etc. All events for a topic
+should be placed in a separate JSON file - where the file name identifies
+the topic. Eg: "Floating-point.json".
+
+All the topic JSON files for a CPU model/family should be in a separate
+sub directory. Thus for the Silvermont X86 CPU:
+
+	$ ls tools/perf/pmu-events/arch/x86/Silvermont_core
+	Cache.json 	Memory.json 	Virtual-Memory.json
+	Frontend.json 	Pipeline.json
+
+Using the JSON files and the mapfile, 'jevents' generates the C source file,
+'pmu-events.c', which encodes the two sets of tables:
+
+	- Set of 'PMU events tables' for all known CPUs in the architecture,
+	  (one table like the following, per JSON file; table name 'pme_power8'
+	  is derived from JSON file name, 'power8.json').
+
+		struct pmu_event pme_power8[] = {
+
+			...
+
+			{
+				.name = "pm_1plus_ppc_cmpl",
+				.event = "event=0x100f2",
+				.desc = "1 or more ppc insts finished,",
+			},
+
+			...
+		}
+
+	- A 'mapping table' that maps each CPU of the architecture, to its
+	  'PMU events table'
+
+		struct pmu_events_map pmu_events_map[] = {
+		{
+			.cpuid = "004b0000",
+			.version = "1",
+			.type = "core",
+			.table = pme_power8
+		},
+			...
+
+		};
+
+After the 'pmu-events.c' is generated, it is compiled and the resulting
+'pmu-events.o' is added to 'libperf.a' which is then used to build perf.
+
+NOTES:
+	1. Several CPUs can support same set of events and hence use a common
+	   JSON file. Hence several entries in the pmu_events_map[] could map
+	   to a single 'PMU events table'.
+
+	2. The 'pmu-events.h' has an extern declaration for the mapping table
+	   and the generated 'pmu-events.c' defines this table.
+
+	3. _All_ known CPU tables for architecture are included in the perf
+	   binary.
+
+At run time, perf determines the actual CPU it is running on, finds the
+matching events table and builds aliases for those events. This allows
+users to specify events by their name:
+
+	$ perf stat -e pm_1plus_ppc_cmpl sleep 1
+
+where 'pm_1plus_ppc_cmpl' is a Power8 PMU event.
+
+In case of errors when processing files in the tools/perf/pmu-events/arch
+directory, 'jevents' tries to create an empty mapping file to allow the perf
+build to succeed even if the PMU event aliases cannot be used.
+
+However some errors in processing may cause the perf build to fail.
+
+Mapfile format
+===============
+
+The mapfile enables multiple CPU models to share a single set of PMU events.
+It is required even if such mapping is 1:1.
+
+The mapfile.csv format is expected to be:
+
+	Header line
+	CPUID,Version,Dir/path/name,Type
+
+where:
+
+	Comma:
+		is the required field delimiter (i.e other fields cannot
+		have commas within them).
+
+	Comments:
+		Lines in which the first character is either '\n' or '#'
+		are ignored.
+
+	Header line
+		The header line is the first line in the file, which is
+		always _IGNORED_. It can empty.
+
+	CPUID:
+		CPUID is an arch-specific char string, that can be used
+		to identify CPU (and associate it with a set of PMU events
+		it supports). Multiple CPUIDS can point to the same
+		File/path/name.json.
+
+		Example:
+			CPUID == 'GenuineIntel-6-2E' (on x86).
+			CPUID == '004b0100' (PVR value in Powerpc)
+	Version:
+		is the Version of the mapfile.
+
+	Dir/path/name:
+		is the pathname to the directory containing the CPU's JSON
+		files, relative to the directory containing the mapfile.csv
+
+	Type:
+		indicates whether the events or "core" or "uncore" events.
+
+
+	Eg:
+
+	$ grep Silvermont tools/perf/pmu-events/arch/x86/mapfile.csv
+	GenuineIntel-6-37,V13,Silvermont_core,core
+	GenuineIntel-6-4D,V13,Silvermont_core,core
+	GenuineIntel-6-4C,V13,Silvermont_core,core
+
+	i.e the three CPU models use the JSON files (i.e PMU events) listed
+	in the directory 'tools/perf/pmu-events/arch/x86/Silvermont_core'.
+
Signature · 1 line
1.8.3.1
#39 lore raw

[tip:perf/urgent] perf tools: Add README for info on parsing JSON/map files

In reply to Sukadev Bhattiprolu
perf tools: Add README for info on parsing JSON/map files
Commit-ID:  06d839b4f712065d76cba92e7e10e9d771771c5a
Gitweb:     http://git.kernel.org/tip/06d839b4f712065d76cba92e7e10e9d771771c5a
Author:     Sukadev Bhattiprolu <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:52 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:37:48 -0300

perf tools: Add README for info on parsing JSON/map files

---
 tools/perf/pmu-events/README | 147 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 147 insertions(+)
tools/perf/pmu-events/README new file +147 -0
new file mode 100644
index 0000000..1408ade
@@ -0,0 +1,147 @@
+
+The contents of this directory allow users to specify PMU events in their
+CPUs by their symbolic names rather than raw event codes (see example below).
+
+The main program in this directory, is the 'jevents', which is built and
+executed _BEFORE_ the perf binary itself is built.
+
+The 'jevents' program tries to locate and process JSON files in the directory
+tree tools/perf/pmu-events/arch/foo.
+
+	- Regular files with '.json' extension in the name are assumed to be
+	  JSON files, each of which describes a set of PMU events.
+
+	- Regular files with basename starting with 'mapfile.csv' are assumed
+	  to be a CSV file that maps a specific CPU to its set of PMU events.
+	  (see below for mapfile format)
+
+	- Directories are traversed, but all other files are ignored.
+
+The PMU events supported by a CPU model are expected to grouped into topics
+such as Pipelining, Cache, Memory, Floating-point etc. All events for a topic
+should be placed in a separate JSON file - where the file name identifies
+the topic. Eg: "Floating-point.json".
+
+All the topic JSON files for a CPU model/family should be in a separate
+sub directory. Thus for the Silvermont X86 CPU:
+
+	$ ls tools/perf/pmu-events/arch/x86/Silvermont_core
+	Cache.json 	Memory.json 	Virtual-Memory.json
+	Frontend.json 	Pipeline.json
+
+Using the JSON files and the mapfile, 'jevents' generates the C source file,
+'pmu-events.c', which encodes the two sets of tables:
+
+	- Set of 'PMU events tables' for all known CPUs in the architecture,
+	  (one table like the following, per JSON file; table name 'pme_power8'
+	  is derived from JSON file name, 'power8.json').
+
+		struct pmu_event pme_power8[] = {
+
+			...
+
+			{
+				.name = "pm_1plus_ppc_cmpl",
+				.event = "event=0x100f2",
+				.desc = "1 or more ppc insts finished,",
+			},
+
+			...
+		}
+
+	- A 'mapping table' that maps each CPU of the architecture, to its
+	  'PMU events table'
+
+		struct pmu_events_map pmu_events_map[] = {
+		{
+			.cpuid = "004b0000",
+			.version = "1",
+			.type = "core",
+			.table = pme_power8
+		},
+			...
+
+		};
+
+After the 'pmu-events.c' is generated, it is compiled and the resulting
+'pmu-events.o' is added to 'libperf.a' which is then used to build perf.
+
+NOTES:
+	1. Several CPUs can support same set of events and hence use a common
+	   JSON file. Hence several entries in the pmu_events_map[] could map
+	   to a single 'PMU events table'.
+
+	2. The 'pmu-events.h' has an extern declaration for the mapping table
+	   and the generated 'pmu-events.c' defines this table.
+
+	3. _All_ known CPU tables for architecture are included in the perf
+	   binary.
+
+At run time, perf determines the actual CPU it is running on, finds the
+matching events table and builds aliases for those events. This allows
+users to specify events by their name:
+
+	$ perf stat -e pm_1plus_ppc_cmpl sleep 1
+
+where 'pm_1plus_ppc_cmpl' is a Power8 PMU event.
+
+In case of errors when processing files in the tools/perf/pmu-events/arch
+directory, 'jevents' tries to create an empty mapping file to allow the perf
+build to succeed even if the PMU event aliases cannot be used.
+
+However some errors in processing may cause the perf build to fail.
+
+Mapfile format
+===============
+
+The mapfile enables multiple CPU models to share a single set of PMU events.
+It is required even if such mapping is 1:1.
+
+The mapfile.csv format is expected to be:
+
+	Header line
+	CPUID,Version,Dir/path/name,Type
+
+where:
+
+	Comma:
+		is the required field delimiter (i.e other fields cannot
+		have commas within them).
+
+	Comments:
+		Lines in which the first character is either '\n' or '#'
+		are ignored.
+
+	Header line
+		The header line is the first line in the file, which is
+		always _IGNORED_. It can empty.
+
+	CPUID:
+		CPUID is an arch-specific char string, that can be used
+		to identify CPU (and associate it with a set of PMU events
+		it supports). Multiple CPUIDS can point to the same
+		File/path/name.json.
+
+		Example:
+			CPUID == 'GenuineIntel-6-2E' (on x86).
+			CPUID == '004b0100' (PVR value in Powerpc)
+	Version:
+		is the Version of the mapfile.
+
+	Dir/path/name:
+		is the pathname to the directory containing the CPU's JSON
+		files, relative to the directory containing the mapfile.csv
+
+	Type:
+		indicates whether the events or "core" or "uncore" events.
+
+
+	Eg:
+
+	$ grep Silvermont tools/perf/pmu-events/arch/x86/mapfile.csv
+	GenuineIntel-6-37,V13,Silvermont_core,core
+	GenuineIntel-6-4D,V13,Silvermont_core,core
+	GenuineIntel-6-4C,V13,Silvermont_core,core
+
+	i.e the three CPU models use the JSON files (i.e PMU events) listed
+	in the directory 'tools/perf/pmu-events/arch/x86/Silvermont_core'.
#40 lore raw

PATCH v21 16/19 perf, tools: Make alias matching case-insensitive

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

Make alias matching the events parser case-insensitive. This is useful with the JSON events. perf uses lower case events, but the CPU manuals generally use upper case event names. The JSON files use lower case by default too. But if we search case insensitively then users can cut-n-paste the upper case event names.

So the following works:

% perf stat -e BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL true

Performance counter stats for 'true':

               305      BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL

       0.000492799 seconds time elapsed
---
 tools/perf/util/parse-events.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/parse-events.c +1 -1
index 9abd60d..1abda10 100644
@@ -1453,7 +1453,7 @@ comp_pmu(const void *p1, const void *p2)
 	struct perf_pmu_event_symbol *pmu1 = (struct perf_pmu_event_symbol *) p1;
 	struct perf_pmu_event_symbol *pmu2 = (struct perf_pmu_event_symbol *) p2;
 
-	return strcmp(pmu1->symbol, pmu2->symbol);
+	return strcasecmp(pmu1->symbol, pmu2->symbol);
 }
 
 static void perf_pmu__parse_cleanup(void)
Signature · 1 line
1.8.3.1
#41 lore raw

Re: [PATCH v21 16/19] perf, tools: Make alias matching case-insensitive

Em Thu, Sep 15, 2016 at 03:24:53PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 10 lines

From: Andi Kleen <[email protected]>

Make alias matching the events parser case-insensitive. This is useful with the JSON events. perf uses lower case events, but the CPU manuals generally use upper case event names. The JSON files use lower case by default too. But if we search case insensitively then users can cut-n-paste the upper case event names.

So the following works:

Quoted 7 lines

% perf stat -e BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL true

Performance counter stats for 'true':

               305      BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL

       0.000492799 seconds time elapsed

So now trying to figure this out:

[acme@jouet linux]$ perf stat -e br_inst_exec.all_direct_near_call true event syntax error: 'br_inst_exec.all_direct_near_call'
\___ 'period' is not usable in 'perf stat'
Run 'perf list' for a list of valid events

Usage: perf stat [<options>] [<command>]

-e, --event <event> event selector. use 'perf list' to list available events
[acme@jouet linux]$

[acme@jouet linux]$ perf stat -e cycles true

Performance counter stats for 'true':

           228,093      cycles:u                                                    

       0.002678362 seconds time elapsed

[acme@jouet linux]$

Who is setting period in that case?

For cycles we get in verbose mode:

[acme@jouet linux]$ perf stat -vv -e cycles true
Using CPUID GenuineIntel-6-3D
intel_pt default config: tsc
------------------------------------------------------------
perf_event_attr:

  size                             112
  sample_type                      IDENTIFIER
  read_format                      TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING
  disabled                         1
  inherit                          1
  enable_on_exec                   1
  exclude_guest                    1

------------------------------------------------------------
sys_perf_event_open: pid 13346 cpu -1 group_fd -1 flags 0x8 sys_perf_event_open failed, error -13
Warning:
kernel.perf_event_paranoid=2, trying to fall back to excluding kernel samples
------------------------------------------------------------
perf_event_attr:

  size                             112
  sample_type                      IDENTIFIER
  read_format                      TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING
  disabled                         1
  inherit                          1
  exclude_kernel                   1
  enable_on_exec                   1
  exclude_guest                    1

------------------------------------------------------------
sys_perf_event_open: pid 13346 cpu -1 group_fd -1 flags 0x8 cycles:u: 0: 322539 536424 536424
cycles:u: 322539 536424 536424

Performance counter stats for 'true':

           322,539      cycles:u                                                    

       0.001061349 seconds time elapsed

And for br_inst_exec.all_direct_near_call:

Strange:

[acme@jouet linux]$ perf stat -vvv -e 'br_inst_exec.all_direct_near_call' true Using CPUID GenuineIntel-6-3D
event syntax error: 'br_inst_exec.all_direct_near_call'
\___ 'period' is not usable in 'perf stat'
Run 'perf list' for a list of valid events

Usage: perf stat [<options>] [<command>]

-e, --event <event> event selector. use 'perf list' to list available events
[acme@jouet linux]$
[acme@jouet linux]$ perf record -e 'br_inst_exec.all_direct_near_call' true
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.020 MB perf.data ]
[acme@jouet linux]$ perf evlist -v
br_inst_exec.all_direct_near_call:u: type: 4, size: 112, config: 0xd088, { sample_period, sample_freq }: 200003, sample_type: IP|TID|TIME|PERIOD, disabled: 1, inherit: 1, exclude_kernel: 1, mmap: 1, comm: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1 [acme@jouet linux]$

I've put what I have on a tmp.perf/json branch, will run it thru my tests and continue tomorrow, when I'll try to investigate this problem with 'perf stat' and these events, unless someone does it while I'm asleep. :-)

- Arnaldo

Quoted 21 lines
---
 tools/perf/util/parse-events.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/parse-events.c +1 -1
index 9abd60d..1abda10 100644
@@ -1453,7 +1453,7 @@ comp_pmu(const void *p1, const void *p2)
 	struct perf_pmu_event_symbol *pmu1 = (struct perf_pmu_event_symbol *) p1;
 	struct perf_pmu_event_symbol *pmu2 = (struct perf_pmu_event_symbol *) p2;
 
-	return strcmp(pmu1->symbol, pmu2->symbol);
+	return strcasecmp(pmu1->symbol, pmu2->symbol);
 }
 
 static void perf_pmu__parse_cleanup(void)
Signature · 1 line
1.8.3.1
#42 lore raw

Re: [PATCH v21 16/19] perf, tools: Make alias matching case-insensitive

Em Mon, Oct 03, 2016 at 09:47:06PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 13 lines

Em Thu, Sep 15, 2016 at 03:24:53PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 12 lines

From: Andi Kleen <[email protected]>

Make alias matching the events parser case-insensitive. This is useful with the JSON events. perf uses lower case events, but the CPU manuals generally use upper case event names. The JSON files use lower case by default too. But if we search case insensitively then users can cut-n-paste the upper case event names.

So the following works:

% perf stat -e BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL true

Quoted 7 lines
Quoted 5 lines

Performance counter stats for 'true':

               305      BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL

       0.000492799 seconds time elapsed

So now trying to figure this out:

Ok, so this is just another case of bad patch ordering: this uses 'perf stat' as the example, but it doesn't work at this point, one needs to go to the last patch, apply it and then test it again, when it will work. So just moved the last patch to before this one, sigh.

- Arnaldo

Quoted 111 lines

[acme@jouet linux]$ perf stat -e br_inst_exec.all_direct_near_call true event syntax error: 'br_inst_exec.all_direct_near_call'
\___ 'period' is not usable in 'perf stat'
Run 'perf list' for a list of valid events

Usage: perf stat [<options>] [<command>]

-e, --event <event> event selector. use 'perf list' to list available events
[acme@jouet linux]$

[acme@jouet linux]$ perf stat -e cycles true

Performance counter stats for 'true':

           228,093      cycles:u                                                    

       0.002678362 seconds time elapsed

[acme@jouet linux]$

Who is setting period in that case?

For cycles we get in verbose mode:

[acme@jouet linux]$ perf stat -vv -e cycles true
Using CPUID GenuineIntel-6-3D
intel_pt default config: tsc
------------------------------------------------------------
perf_event_attr:

  size                             112
  sample_type                      IDENTIFIER
  read_format                      TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING
  disabled                         1
  inherit                          1
  enable_on_exec                   1
  exclude_guest                    1

------------------------------------------------------------
sys_perf_event_open: pid 13346 cpu -1 group_fd -1 flags 0x8 sys_perf_event_open failed, error -13
Warning:
kernel.perf_event_paranoid=2, trying to fall back to excluding kernel samples
------------------------------------------------------------
perf_event_attr:

  size                             112
  sample_type                      IDENTIFIER
  read_format                      TOTAL_TIME_ENABLED|TOTAL_TIME_RUNNING
  disabled                         1
  inherit                          1
  exclude_kernel                   1
  enable_on_exec                   1
  exclude_guest                    1

------------------------------------------------------------
sys_perf_event_open: pid 13346 cpu -1 group_fd -1 flags 0x8 cycles:u: 0: 322539 536424 536424
cycles:u: 322539 536424 536424

Performance counter stats for 'true':

           322,539      cycles:u                                                    

       0.001061349 seconds time elapsed

And for br_inst_exec.all_direct_near_call:

Strange:

[acme@jouet linux]$ perf stat -vvv -e 'br_inst_exec.all_direct_near_call' true Using CPUID GenuineIntel-6-3D
event syntax error: 'br_inst_exec.all_direct_near_call'
\___ 'period' is not usable in 'perf stat'
Run 'perf list' for a list of valid events

Usage: perf stat [<options>] [<command>]

-e, --event <event> event selector. use 'perf list' to list available events
[acme@jouet linux]$
[acme@jouet linux]$ perf record -e 'br_inst_exec.all_direct_near_call' true
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.020 MB perf.data ]
[acme@jouet linux]$ perf evlist -v
br_inst_exec.all_direct_near_call:u: type: 4, size: 112, config: 0xd088, { sample_period, sample_freq }: 200003, sample_type: IP|TID|TIME|PERIOD, disabled: 1, inherit: 1, exclude_kernel: 1, mmap: 1, comm: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1 [acme@jouet linux]$

I've put what I have on a tmp.perf/json branch, will run it thru my tests and continue tomorrow, when I'll try to investigate this problem with 'perf stat' and these events, unless someone does it while I'm asleep. :-)

- Arnaldo

Quoted 21 lines
---
 tools/perf/util/parse-events.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/parse-events.c +1 -1
index 9abd60d..1abda10 100644
@@ -1453,7 +1453,7 @@ comp_pmu(const void *p1, const void *p2)
 	struct perf_pmu_event_symbol *pmu1 = (struct perf_pmu_event_symbol *) p1;
 	struct perf_pmu_event_symbol *pmu2 = (struct perf_pmu_event_symbol *) p2;
 
-	return strcmp(pmu1->symbol, pmu2->symbol);
+	return strcasecmp(pmu1->symbol, pmu2->symbol);
 }
 
 static void perf_pmu__parse_cleanup(void)
Signature · 1 line
1.8.3.1
#43 lore raw

Re: [PATCH v21 16/19] perf, tools: Make alias matching case-insensitive

Quoted 18 lines · Arnaldo Carvalho de Melo

Em Mon, Oct 03, 2016 at 09:47:06PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 17 lines

Em Thu, Sep 15, 2016 at 03:24:53PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 16 lines

From: Andi Kleen <[email protected]>

Make alias matching the events parser case-insensitive. This is useful with the JSON events. perf uses lower case events, but the CPU manuals generally use upper case event names. The JSON files use lower case by default too. But if we search case insensitively then users can cut-n-paste the upper case event names.

So the following works:

% perf stat -e BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL true

Performance counter stats for 'true':

305 BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL

Quoted 8 lines
Quoted 3 lines
Quoted 1 line
       0.000492799 seconds time elapsed

So now trying to figure this out:

Ok, so this is just another case of bad patch ordering: this uses 'perf stat' as the example, but it doesn't work at this point, one needs to go to the last patch, apply it and then test it again, when it will work. So just moved the last patch to before this one, sigh.

ok, you found it.. sry ;-)

jirka

#44 lore raw

Re: [PATCH v21 16/19] perf, tools: Make alias matching case-insensitive

In reply to Arnaldo Carvalho de Melo
Quoted 17 lines · Arnaldo Carvalho de Melo

Em Thu, Sep 15, 2016 at 03:24:53PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 16 lines

From: Andi Kleen <[email protected]>

Make alias matching the events parser case-insensitive. This is useful with the JSON events. perf uses lower case events, but the CPU manuals generally use upper case event names. The JSON files use lower case by default too. But if we search case insensitively then users can cut-n-paste the upper case event names.

So the following works:

% perf stat -e BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL true

Performance counter stats for 'true':

305 BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL

Quoted 8 lines
Quoted 1 line
       0.000492799 seconds time elapsed

So now trying to figure this out:

[acme@jouet linux]$ perf stat -e br_inst_exec.all_direct_near_call true event syntax error: 'br_inst_exec.all_direct_near_call'
\___ 'period' is not usable in 'perf stat'
Run 'perf list' for a list of valid events

looks like you need also this one:
155128eaa62f Allow period= in perf stat CPU event descriptions.

jirka

#45 lore raw

[tip:perf/urgent] perf tools: Make alias matching case-insensitive

In reply to Sukadev Bhattiprolu
perf tools: Make alias matching case-insensitive
Commit-ID:  e312bcf13053970c0f07ec2c02d7d9be1a036ce0
Gitweb:     http://git.kernel.org/tip/e312bcf13053970c0f07ec2c02d7d9be1a036ce0
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:53 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:51:48 -0300

perf tools: Make alias matching case-insensitive

Make alias matching the events parser case-insensitive. This is useful with the JSON events. perf uses lower case events, but the CPU manuals generally use upper case event names. The JSON files use lower case by default too. But if we search case insensitively then users can cut-n-paste the upper case event names.

So the following works:

% perf stat -e BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL true

Performance counter stats for 'true':

               305      BR_INST_EXEC.TAKEN_INDIRECT_NEAR_CALL

       0.000492799 seconds time elapsed
---
 tools/perf/util/parse-events.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/util/parse-events.c +1 -1
index d445b11..4e778ea 100644
@@ -1459,7 +1459,7 @@ comp_pmu(const void *p1, const void *p2)
 	struct perf_pmu_event_symbol *pmu1 = (struct perf_pmu_event_symbol *) p1;
 	struct perf_pmu_event_symbol *pmu2 = (struct perf_pmu_event_symbol *) p2;
 
-	return strcmp(pmu1->symbol, pmu2->symbol);
+	return strcasecmp(pmu1->symbol, pmu2->symbol);
 }
 
 static void perf_pmu__parse_cleanup(void)
#46 lore raw

PATCH v21 17/19 perf, tools, pmu-events: Fix fixed counters on Intel

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

The JSON event lists use a different encoding for fixed counters than perf for instructions and cycles (ref-cycles is ok)

This lead to some common events like inst_retired.any or cpu_clk_unhalted.thread not counting, when specified with their JSON name.

Special case these events in the jevents conversion process. I prefer to not touch the JSON files for this, as it's intended that standard JSON files can be just dropped into the perf build without changes.

---
Changelog[v21]:
Fix minor conflict in tools/perf/pmu-events/jevents.c

---
 tools/perf/pmu-events/jevents.c | 25 ++++++++++++++++++++++++-
 1 file changed, 24 insertions(+), 1 deletion(-)
tools/perf/pmu-events/jevents.c +24 -1
index 9cdfbaa..e8e2a87 100644
@@ -305,6 +305,29 @@ static void print_events_table_suffix(FILE *outfp)
 	close_table = 0;
 }
 
+static struct fixed {
+	const char *name;
+	const char *event;
+} fixed[] = {
+	{ "inst_retired.any", "event=0xc0" },
+	{ "cpu_clk_unhalted.thread", "event=0x3c" },
+	{ "cpu_clk_unhalted.thread_any", "event=0x3c,any=1" },
+	{ NULL, NULL},
+};
+
+/*
+ * Handle different fixed counter encodings between JSON and perf.
+ */
+static char *real_event(const char *name, char *event)
+{
+	int i;
+
+	for (i = 0; fixed[i].name; i++)
+		if (!strcasecmp(name, fixed[i].name))
+			return (char *)fixed[i].event;
+	return event;
+}
+
 /* Call func with each event in the json file */
 int json_events(const char *fn,
 	  int (*func)(void *data, char *name, char *event, char *desc,
@@ -391,7 +414,7 @@ int json_events(const char *fn,
 			addfield(map, &event, ",", msr->pname, msrval);
 		fixname(name);
 
-		err = func(data, name, event, desc, long_desc);
+		err = func(data, name, real_event(name, event), desc, long_desc);
 		free(event);
 		free(desc);
 		free(name);
Signature · 1 line
1.8.3.1
#47 lore raw

[tip:perf/urgent] perf pmu-events: Fix fixed counters on Intel

perf pmu-events: Fix fixed counters on Intel
Commit-ID:  0b1db474314c883d0bfc6720278667a6155e995a
Gitweb:     http://git.kernel.org/tip/0b1db474314c883d0bfc6720278667a6155e995a
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:54 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:52:00 -0300

perf pmu-events: Fix fixed counters on Intel

The JSON event lists use a different encoding for fixed counters than perf for instructions and cycles (ref-cycles is ok)

This lead to some common events like inst_retired.any or cpu_clk_unhalted.thread not counting, when specified with their JSON name.

Special case these events in the jevents conversion process. I prefer to not touch the JSON files for this, as it's intended that standard JSON files can be just dropped into the perf build without changes.

---
 tools/perf/pmu-events/jevents.c | 25 ++++++++++++++++++++++++-
 1 file changed, 24 insertions(+), 1 deletion(-)
tools/perf/pmu-events/jevents.c +24 -1
index 13f4284..04e106e 100644
@@ -306,6 +306,29 @@ static void print_events_table_suffix(FILE *outfp)
 	close_table = 0;
 }
 
+static struct fixed {
+	const char *name;
+	const char *event;
+} fixed[] = {
+	{ "inst_retired.any", "event=0xc0" },
+	{ "cpu_clk_unhalted.thread", "event=0x3c" },
+	{ "cpu_clk_unhalted.thread_any", "event=0x3c,any=1" },
+	{ NULL, NULL},
+};
+
+/*
+ * Handle different fixed counter encodings between JSON and perf.
+ */
+static char *real_event(const char *name, char *event)
+{
+	int i;
+
+	for (i = 0; fixed[i].name; i++)
+		if (!strcasecmp(name, fixed[i].name))
+			return (char *)fixed[i].event;
+	return event;
+}
+
 /* Call func with each event in the json file */
 int json_events(const char *fn,
 	  int (*func)(void *data, char *name, char *event, char *desc,
@@ -392,7 +415,7 @@ int json_events(const char *fn,
 			addfield(map, &event, ",", msr->pname, msrval);
 		fixname(name);
 
-		err = func(data, name, event, desc, long_desc);
+		err = func(data, name, real_event(name, event), desc, long_desc);
 		free(event);
 		free(desc);
 		free(name);
#48 lore raw

PATCH v21 18/19 perf, tools, pmu-events: Add Skylake frontend MSR support

In reply to Sukadev Bhattiprolu

From: Andi Kleen <[email protected]>

Add support for the "frontend" extra MSR on Skylake in the JSON conversion.

---
 tools/perf/pmu-events/jevents.c | 1 +
 1 file changed, 1 insertion(+)
tools/perf/pmu-events/jevents.c +1 -0
index e8e2a87..5846057 100644
@@ -126,6 +126,7 @@ static struct msrmap {
 	{ "0x3F6", "ldlat=" },
 	{ "0x1A6", "offcore_rsp=" },
 	{ "0x1A7", "offcore_rsp=" },
+	{ "0x3F7", "frontend=" },
 	{ NULL, NULL }
 };
 
Signature · 1 line
1.8.3.1
#49 lore raw

[tip:perf/urgent] perf pmu-events: Add Skylake frontend MSR support

perf pmu-events: Add Skylake frontend MSR support
Commit-ID:  b42c7369e3f451e22c2b0be5d193955498d37546
Gitweb:     http://git.kernel.org/tip/b42c7369e3f451e22c2b0be5d193955498d37546
Author:     Andi Kleen <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:55 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:52:01 -0300

perf pmu-events: Add Skylake frontend MSR support

Add support for the "frontend" extra MSR on Skylake in the JSON conversion.

---
 tools/perf/pmu-events/jevents.c | 1 +
 1 file changed, 1 insertion(+)
tools/perf/pmu-events/jevents.c +1 -0
index 04e106e..79c2133 100644
@@ -127,6 +127,7 @@ static struct msrmap {
 	{ "0x3F6", "ldlat=" },
 	{ "0x1A6", "offcore_rsp=" },
 	{ "0x1A7", "offcore_rsp=" },
+	{ "0x3F7", "frontend=" },
 	{ NULL, NULL }
 };
 
#50 lore raw

PATCH v21 19/19 perf, tools: Allow period= in perf stat CPU event descriptions.

In reply to Sukadev Bhattiprolu

This avoids the JSON PMU events parser having to know whether its aliases are for perf stat or perf record.

---
 tools/perf/util/parse-events.c | 1 +
 1 file changed, 1 insertion(+)
tools/perf/util/parse-events.c +1 -0
index 1abda10..b675273 100644
@@ -923,6 +923,7 @@ config_term_avail(int term_type, struct parse_events_error *err)
 	case PARSE_EVENTS__TERM_TYPE_CONFIG1:
 	case PARSE_EVENTS__TERM_TYPE_CONFIG2:
 	case PARSE_EVENTS__TERM_TYPE_NAME:
+	case PARSE_EVENTS__TERM_TYPE_SAMPLE_PERIOD:
 		return true;
 	default:
 		if (!err)
Signature · 1 line
1.8.3.1
#51 lore raw

[tip:perf/urgent] perf tools: Allow period= in perf stat CPU event descriptions.

perf tools: Allow period= in perf stat CPU event descriptions.
Commit-ID:  06835545b144cab6e7748748576145ea15f68dff
Gitweb:     http://git.kernel.org/tip/06835545b144cab6e7748748576145ea15f68dff
Author:     Sukadev Bhattiprolu <[email protected]>
AuthorDate: Thu, 15 Sep 2016 15:24:56 -0700
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 21:50:56 -0300

perf tools: Allow period= in perf stat CPU event descriptions.

This avoids the JSON PMU events parser having to know whether its aliases are for perf stat or perf record.

---
 tools/perf/util/parse-events.c | 1 +
 1 file changed, 1 insertion(+)
tools/perf/util/parse-events.c +1 -0
index 8f88f63..d445b11 100644
@@ -924,6 +924,7 @@ config_term_avail(int term_type, struct parse_events_error *err)
 	case PARSE_EVENTS__TERM_TYPE_CONFIG1:
 	case PARSE_EVENTS__TERM_TYPE_CONFIG2:
 	case PARSE_EVENTS__TERM_TYPE_NAME:
+	case PARSE_EVENTS__TERM_TYPE_SAMPLE_PERIOD:
 		return true;
 	default:
 		if (!err)
#52 lore raw
In reply to Sukadev Bhattiprolu

I messed up the Cc list. Fixing it now.

Quoted 173 lines · Sukadev Bhattiprolu [[email protected]] wrote:

CPUs support a large number of performance monitoring events (PMU events) and often these events are very specific to an architecture/model of the CPU. To use most of these PMU events with perf, we currently have to identify them by their raw codes:

perf stat -e r100f2 sleep 1

This patchset allows architectures to specify these PMU events in JSON files located in 'tools/perf/pmu-events/arch/' of the mainline tree. The events from the JSON files for the architecture are then built into the perf binary.

At run time, perf identifies the specific set of events for the CPU and creates "event aliases". These aliases allow users to specify events by "name" as:

perf stat -e pm_1plus_ppc_cmpl sleep 1

The file, 'tools/perf/pmu-events/README' in [PATCH 15/19] gives more details.

Note:
- All known events tables for the architecture are included in the perf binary.

- For architectures that don't have any JSON files, an empty mapping table is created and they should continue to build.

Thanks to input from Andi Kleen, Jiri Olsa, Namhyung Kim and Ingo Molnar.

These patches are available from:

https://github.com/sukadev/linux.git

Branch Description
------------------------------------------------------
json-code-v21 Source Code only
json-code+data-v21 Both code and data (for build/test/pull)

NOTE: Only "source code" patches (i.e those in json-code-v21) are being
emailed. Please pull the json-code+data-v21 branch for build/test.

Changelog[v21]
- Rebase to recent perf/core
- Group the PMU events supported by a CPU model into topics and create a separate JSON file for each topic for each CPU (code and input from Jiri Olsa).

Changelog[v20]
- Rebase to recent perf/core
- Add Patch 20/20 to allow perf-stat to work with the period= field

Changelog[v19]
Rebase to recent perf/core; fix couple lines >80 chars.

Changelog[v18]
Rebase to recent perf/core; fix minor merge conflicts.

Changelog[v17]
Rebase to recent perf/core; couple of small fixes to processing Intel JSON files; allow case-insensitive PMU event names.

Changelog[v16]
Rebase to recent perf/core; fix minor merge conflicts; drop 3 patches that were merged into perf/core.

Changelog[v15]
Code changes:
- Fix 'perf list' usage string and update man page.
- Remove a redundant __maybe_unused tag.
- Rebase to recent perf/core branch.

Data files updates: json-files-5 branch
- Rebase to perf/intel-json-files-5 from Andi Kleen
- Add patch from Madhavan Srinivasan for couple more Powerpc models

Changelog[v14]
Comments from Jiri Olsa:
- Change parameter name/type for pmu_add_cpu_aliases (from void *data to list_head *head)
- Use asprintf() in file_name_to_tablename() and simplify/reorg code.
- Use __weak definition from <linux/compile.h>
- Use fopen() with mode "w" and eliminate unlink()
- Remove minor TODO.
- Add error check for return value from strdup() in print_pmu_events().
- Move independent changes from patches 3,11,12 .. to separate patches for easier review/backport.
- Clarify mapfile's "header line support" in patch description.
- Fix build failure with DEBUG=1

Comment from Andi Kleen:
- In tools/perf/pmu-events/Build, check for 'mapfile.csv' rather than 'mapfile*'

Misc:
- Minor changes/clarifications to tools/perf/pmu-events/README.

Changelog[v13]
Version: Individual patches have their own history :-) that I am preserving. Patchset version (v13) is for overall patchset and is somewhat arbitrary.

- Added support for "categories" of events to perf
- Add mapfile, jevents build dependency on pmu-events.c
- Silence jevents when parsing JSON files unless V=1 is specified
- Cleanup error messages
- Fix memory leak with ->cpuid
- Rebase to Arnaldo's tree
- Allow overriding CPUID via environment variable
- Support long descriptions for events
- Handle header line in mapfile.csv
- Cleanup JSON files (trim PublicDescription if identical to/prefix of BriefDescription field)

Andi Kleen (12):
perf, tools: Add jsmn `jasmine' JSON parser
perf, tools, jevents: Program to convert JSON file to C style file perf, tools: Support CPU id matching for x86 v2
perf, tools: Support alias descriptions
perf, tools: Query terminal width and use in perf list perf, tools: Add a --no-desc flag to perf list
perf, tools: Add override support for event list CPUID perf, tools: Add support for event list topics
perf, tools, jevents: Handle header line in mapfile perf, tools: Make alias matching case-insensitive
perf, tools, pmu-events: Fix fixed counters on Intel perf, tools, pmu-events: Add Skylake frontend MSR support

Sukadev Bhattiprolu (7):
perf, tools: Use pmu_events table to create aliases perf, tools: Support CPU ID matching for Powerpc perf, tools, jevents: Add support for long descriptions perf, tools: Add alias support for long descriptions perf, tools: Support long descriptions with perf list perf, tools: Add README for info on parsing JSON/map files Allow period= in perf stat CPU event descriptions.

 tools/lib/subcmd/pager.c               |  16 +
 tools/lib/subcmd/pager.h               |   1 +
 tools/perf/Documentation/perf-list.txt |  12 +-
 tools/perf/Makefile.perf               |  28 +-
 tools/perf/arch/powerpc/util/header.c  |  11 +
 tools/perf/arch/x86/util/header.c      |  24 +-
 tools/perf/builtin-list.c              |  20 +-
 tools/perf/pmu-events/Build            |  11 +
 tools/perf/pmu-events/README           | 148 ++++++
 tools/perf/pmu-events/jevents.c        | 811 +++++++++++++++++++++++++++++++++
 tools/perf/pmu-events/jevents.h        |  18 +
 tools/perf/pmu-events/jsmn.c           | 313 +++++++++++++
 tools/perf/pmu-events/jsmn.h           |  67 +++
 tools/perf/pmu-events/json.c           | 162 +++++++
 tools/perf/pmu-events/json.h           |  42 ++
 tools/perf/pmu-events/pmu-events.h     |  37 ++
 tools/perf/util/header.h               |   1 +
 tools/perf/util/parse-events.c         |   8 +-
 tools/perf/util/parse-events.h         |   3 +-
 tools/perf/util/pmu.c                  | 177 ++++++-
 tools/perf/util/pmu.h                  |   6 +-
 21 files changed, 1880 insertions(+), 36 deletions(-)
 create mode 100644 tools/perf/pmu-events/Build
 create mode 100644 tools/perf/pmu-events/README
 create mode 100644 tools/perf/pmu-events/jevents.c
 create mode 100644 tools/perf/pmu-events/jevents.h
 create mode 100644 tools/perf/pmu-events/jsmn.c
 create mode 100644 tools/perf/pmu-events/jsmn.h
 create mode 100644 tools/perf/pmu-events/json.c
 create mode 100644 tools/perf/pmu-events/json.h
 create mode 100644 tools/perf/pmu-events/pmu-events.h
Signature · 1 line
1.8.3.1
#53 lore raw
In reply to Sukadev Bhattiprolu

Em Thu, Sep 15, 2016 at 03:24:37PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 6 lines

CPUs support a large number of performance monitoring events (PMU events) and often these events are very specific to an architecture/model of the CPU. To use most of these PMU events with perf, we currently have to identify them by their raw codes:

perf stat -e r100f2 sleep 1

So, trying to build this with my set of containers I get lots of failures, double checking running on another machine with those containers for the segfault cases, changing sys/fcntl.h to fcntl.h cures the build on alpine:3.4 (musl libc), will try to fix all of them.

The list is incomplete, lots of other systems failed as well.

- Arnaldo

1 alpine:3.4: FAIL
CC /tmp/build/perf/pmu-events/jsmn.o
In file included from pmu-events/json.c:35:0:
/usr/include/sys/fcntl.h:1:2: error: #warning redirecting incorrect #include <sys/fcntl.h> to <fcntl.h> [-Werror=cpp]
#warning redirecting incorrect #include <sys/fcntl.h> to <fcntl.h>
-----------------------------------------------------------------------
android-ndk:r12b-arm: FAIL

CC /tmp/build/perf/event-plugin.o
pmu-events/json.c:35:23: fatal error: sys/fcntl.h: No such file or directory

 #include <sys/fcntl.h>
                       ^

compilation terminated.
-----------------------------------------------------------------------
archlinux:latest: FAIL
/bin/sh: line 1: 1408 Segmentation fault (core dumped) /tmp/build/perf/pmu-events/jevents x86 pmu-events/arch /tmp/build/perf/pmu-events/pmu-events.c
make[2]: *** [pmu-events/Build:11: /tmp/build/perf/pmu-events/pmu-events.c] Error 139 make[1]: *** [Makefile.perf:461: /tmp/build/perf/pmu-events/pmu-events-in.o] Error 2 make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------
centos:5: FAIL
/bin/sh: line 1: 1336 Segmentation fault (core dumped) /tmp/build/perf/pmu-events/jevents x86 pmu-events/arch /tmp/build/perf/pmu-events/pmu-events.c
make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139
make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------
centos:6: FAIL
/bin/sh: line 1: 1633 Segmentation fault (core dumped) /tmp/build/perf/pmu-events/jevents x86 pmu-events/arch /tmp/build/perf/pmu-events/pmu-events.c
make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139
make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------
centos:7: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
/bin/sh: line 1: 1548 Segmentation fault (core dumped) /tmp/build/perf/pmu-events/jevents x86 pmu-events/arch /tmp/build/perf/pmu-events/pmu-events.c
make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139
make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------
debian:7: Ok
-----------------------------------------------------------------------
debian:8: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
Segmentation fault (core dumped)
pmu-events/Build:11: recipe for target '/tmp/build/perf/pmu-events/pmu-events.c' failed make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139 Makefile.perf:461: recipe for target '/tmp/build/perf/pmu-events/pmu-events-in.o' failed make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------
debian:experimental: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
Segmentation fault (core dumped)
pmu-events/Build:11: recipe for target '/tmp/build/perf/pmu-events/pmu-events.c' failed make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139 Makefile.perf:461: recipe for target '/tmp/build/perf/pmu-events/pmu-events-in.o' failed make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
-----------------------------------------------------------------------
fedora:20: FAIL
/bin/sh: line 1: 1460 Segmentation fault (core dumped) /tmp/build/perf/pmu-events/jevents x86 pmu-events/arch /tmp/build/perf/pmu-events/pmu-events.c
make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139
make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------

#54 lore raw

Em Mon, Sep 19, 2016 at 06:20:17PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 8 lines

Em Thu, Sep 15, 2016 at 03:24:37PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 6 lines

CPUs support a large number of performance monitoring events (PMU events) and often these events are very specific to an architecture/model of the CPU. To use most of these PMU events with perf, we currently have to identify them by their raw codes:

perf stat -e r100f2 sleep 1

Quoted 9 lines

So, trying to build this with my set of containers I get lots of failures, double checking running on another machine with those containers for the segfault cases, changing sys/fcntl.h to fcntl.h cures the build on alpine:3.4 (musl libc), will try to fix all of them.

The list is incomplete, lots of other systems failed as well.

- Arnaldo

1 alpine:3.4: Ok

Fixed with fcntl.h + capping the maxfds parameter to nftw to avoid it exploding on alloca() in environments where rlim_max is set to a high value, like in docker.

Now looking at:

CC /tmp/build/perf/fs/tracing_path.o
In file included from /git/linux/tools/include/linux/types.h:4:0,

                 from /opt/android-ndk-r12b/platforms/android-24/arch-arm/usr/include/sys/types.h:35,
                 from /opt/android-ndk-r12b/platforms/android-24/arch-arm/usr/include/strings.h:42,
                 from /opt/android-ndk-r12b/platforms/android-24/arch-arm/usr/include/stdlib.h:36,
                 from pmu-events/json.c:31:

pmu-events/json.h:15:22: error: two or more data types in declaration specifiers
typedef unsigned int bool;

Which looks like clashing with stdbool.h...

- Arnaldo

Quoted 58 lines

android-ndk:r12b-arm: FAIL

CC /tmp/build/perf/event-plugin.o
pmu-events/json.c:35:23: fatal error: sys/fcntl.h: No such file or directory

 #include <sys/fcntl.h>
                       ^

compilation terminated.
-----------------------------------------------------------------------
archlinux:latest: FAIL
/bin/sh: line 1: 1408 Segmentation fault (core dumped) /tmp/build/perf/pmu-events/jevents x86 pmu-events/arch /tmp/build/perf/pmu-events/pmu-events.c
make[2]: *** [pmu-events/Build:11: /tmp/build/perf/pmu-events/pmu-events.c] Error 139 make[1]: *** [Makefile.perf:461: /tmp/build/perf/pmu-events/pmu-events-in.o] Error 2 make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------
centos:5: FAIL
/bin/sh: line 1: 1336 Segmentation fault (core dumped) /tmp/build/perf/pmu-events/jevents x86 pmu-events/arch /tmp/build/perf/pmu-events/pmu-events.c
make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139
make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------
centos:6: FAIL
/bin/sh: line 1: 1633 Segmentation fault (core dumped) /tmp/build/perf/pmu-events/jevents x86 pmu-events/arch /tmp/build/perf/pmu-events/pmu-events.c
make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139
make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------
centos:7: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
/bin/sh: line 1: 1548 Segmentation fault (core dumped) /tmp/build/perf/pmu-events/jevents x86 pmu-events/arch /tmp/build/perf/pmu-events/pmu-events.c
make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139
make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------
debian:7: Ok
-----------------------------------------------------------------------
debian:8: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
Segmentation fault (core dumped)
pmu-events/Build:11: recipe for target '/tmp/build/perf/pmu-events/pmu-events.c' failed make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139 Makefile.perf:461: recipe for target '/tmp/build/perf/pmu-events/pmu-events-in.o' failed make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------
debian:experimental: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
Segmentation fault (core dumped)
pmu-events/Build:11: recipe for target '/tmp/build/perf/pmu-events/pmu-events.c' failed make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139 Makefile.perf:461: recipe for target '/tmp/build/perf/pmu-events/pmu-events-in.o' failed make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
-----------------------------------------------------------------------
fedora:20: FAIL
/bin/sh: line 1: 1460 Segmentation fault (core dumped) /tmp/build/perf/pmu-events/jevents x86 pmu-events/arch /tmp/build/perf/pmu-events/pmu-events.c
make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 139
make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
-----------------------------------------------------------------------

#55 lore raw

Em Mon, Sep 19, 2016 at 08:31:13PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 32 lines

Em Mon, Sep 19, 2016 at 06:20:17PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 8 lines

Em Thu, Sep 15, 2016 at 03:24:37PM -0700, Sukadev Bhattiprolu escreveu:

Quoted 6 lines

CPUs support a large number of performance monitoring events (PMU events) and often these events are very specific to an architecture/model of the CPU. To use most of these PMU events with perf, we currently have to identify them by their raw codes:

perf stat -e r100f2 sleep 1

Quoted 9 lines

So, trying to build this with my set of containers I get lots of failures, double checking running on another machine with those containers for the segfault cases, changing sys/fcntl.h to fcntl.h cures the build on alpine:3.4 (musl libc), will try to fix all of them.

The list is incomplete, lots of other systems failed as well.

- Arnaldo

1 alpine:3.4: Ok

Fixed with fcntl.h + capping the maxfds parameter to nftw to avoid it exploding on alloca() in environments where rlim_max is set to a high value, like in docker.

Now looking at:

CC /tmp/build/perf/fs/tracing_path.o
In file included from /git/linux/tools/include/linux/types.h:4:0,

                 from /opt/android-ndk-r12b/platforms/android-24/arch-arm/usr/include/sys/types.h:35,
                 from /opt/android-ndk-r12b/platforms/android-24/arch-arm/usr/include/strings.h:42,
                 from /opt/android-ndk-r12b/platforms/android-24/arch-arm/usr/include/stdlib.h:36,
                 from pmu-events/json.c:31:
Quoted 5 lines

pmu-events/json.h:15:22: error: two or more data types in declaration specifiers
typedef unsigned int bool;

Which looks like clashing with stdbool.h...

yeah, changing that typedef + true def to plain include <stdbool.h> makes it progress to the next failure, which is in cross compilation environments, such as using fedora 24 + the Android NDK to try to build a ARM android binary.

On the bright side, in addition to alpine:3.4 now these are building ok:

3 archlinux:latest: Ok
4 centos:5: Ok
5 centos:6: Ok
6 centos:7: Ok
7 debian:7: Ok
8 debian:8: Ok
9 fedora:20: Ok

Waiting for some extra cross compilation envs to check that hunch...

- Arnaldo

#56 lore raw

Em Mon, Sep 19, 2016 at 08:37:53PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 15 lines

yeah, changing that typedef + true def to plain include <stdbool.h> makes it progress to the next failure, which is in cross compilation environments, such as using fedora 24 + the Android NDK to try to build a ARM android binary.

On the bright side, in addition to alpine:3.4 now these are building ok:

3 archlinux:latest: Ok
4 centos:5: Ok
5 centos:6: Ok
6 centos:7: Ok
7 debian:7: Ok
8 debian:8: Ok
9 fedora:20: Ok

Quoted 1 line

Waiting for some extra cross compilation envs to check that hunch...

Yeap:

10 fedora:21: Ok
11 fedora:22: Ok
12 fedora:23: Ok
13 fedora:24: Ok
14 fedora:24-x-ARC-uClibc: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
/bin/sh: /tmp/build/perf/pmu-events/jevents: cannot execute binary file: Exec format error pmu-events/Build:11: recipe for target '/tmp/build/perf/pmu-events/pmu-events.c' failed make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 126 Makefile.perf:461: recipe for target '/tmp/build/perf/pmu-events/pmu-events-in.o' failed make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....
15 fedora:rawhide: Ok
16 mageia:5: Ok
17 opensuse:13.2: Ok
18 opensuse:42.1: Ok
19 86.364838418 opensuse:tumbleweed: Ok
20 62.414227297 ubuntu:12.04.5: Ok
21 41.172970676 ubuntu:14.04: Ok
22 74.243049939 ubuntu:14.04.4: Ok
23 75.634449405 ubuntu:15.10: Ok
24 70.487913217 ubuntu:16.04: Ok
25 33.938554395 ubuntu:16.04-x-arm: FAIL
26 31.902507224 ubuntu:16.04-x-arm64: FAIL
27 32.903876355 ubuntu:16.04-x-powerpc64: FAIL
28 32.849412876 ubuntu:16.04-x-powerpc64el: FAIL
29 79.446950856 ubuntu:16.10: Ok
30 33.443948539 ubuntu:16.10-x-arm64: FAIL
31 33.801496151 ubuntu:16.10-x-powerpc: FAIL
32 32.860975730 ubuntu:16.10-x-s390: FAIL

#57 lore raw

Em Mon, Sep 19, 2016 at 09:02:58PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 6 lines

Em Mon, Sep 19, 2016 at 08:37:53PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 4 lines

yeah, changing that typedef + true def to plain include <stdbool.h> makes it progress to the next failure, which is in cross compilation environments, such as using fedora 24 + the Android NDK to try to build a ARM android binary.

Quoted 8 lines

14 fedora:24-x-ARC-uClibc: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
/bin/sh: /tmp/build/perf/pmu-events/jevents: cannot execute binary file: Exec format error pmu-events/Build:11: recipe for target '/tmp/build/perf/pmu-events/pmu-events.c' failed make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 126 Makefile.perf:461: recipe for target '/tmp/build/perf/pmu-events/pmu-events-in.o' failed make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....

Jiri, we need something similar to scripts/Makefile.host :-\

Calling it a day, perhaps, for now, we should just detect that it is a corss compile env (CROSS_COMPILE is set) and exclude all this code from the build, emitting a warning.

I left what I did at the tmp.perf/core branch of my repo at
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git.

- Arnaldo

#58 lore raw

oops, Jiri wasn't CCed, fixing it...

Em Mon, Sep 19, 2016 at 09:28:20PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 26 lines

Em Mon, Sep 19, 2016 at 09:02:58PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 6 lines

Em Mon, Sep 19, 2016 at 08:37:53PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 4 lines

yeah, changing that typedef + true def to plain include <stdbool.h> makes it progress to the next failure, which is in cross compilation environments, such as using fedora 24 + the Android NDK to try to build a ARM android binary.

Quoted 8 lines

14 fedora:24-x-ARC-uClibc: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
/bin/sh: /tmp/build/perf/pmu-events/jevents: cannot execute binary file: Exec format error pmu-events/Build:11: recipe for target '/tmp/build/perf/pmu-events/pmu-events.c' failed make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 126 Makefile.perf:461: recipe for target '/tmp/build/perf/pmu-events/pmu-events-in.o' failed make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....

Jiri, we need something similar to scripts/Makefile.host :-\

Calling it a day, perhaps, for now, we should just detect that it is a corss compile env (CROSS_COMPILE is set) and exclude all this code from the build, emitting a warning.

I left what I did at the tmp.perf/core branch of my repo at
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git.

- Arnaldo

#59 lore raw
In reply to Arnaldo Carvalho de Melo
Quoted 16 lines · Arnaldo Carvalho de Melo

Em Mon, Sep 19, 2016 at 09:02:58PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 6 lines

Em Mon, Sep 19, 2016 at 08:37:53PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 4 lines

yeah, changing that typedef + true def to plain include <stdbool.h> makes it progress to the next failure, which is in cross compilation environments, such as using fedora 24 + the Android NDK to try to build a ARM android binary.

Quoted 8 lines

14 fedora:24-x-ARC-uClibc: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
/bin/sh: /tmp/build/perf/pmu-events/jevents: cannot execute binary file: Exec format error pmu-events/Build:11: recipe for target '/tmp/build/perf/pmu-events/pmu-events.c' failed make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 126 Makefile.perf:461: recipe for target '/tmp/build/perf/pmu-events/pmu-events-in.o' failed make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....

Quoted 8 lines

Jiri, we need something similar to scripts/Makefile.host :-\

Calling it a day, perhaps, for now, we should just detect that it is a corss compile env (CROSS_COMPILE is set) and exclude all this code from the build, emitting a warning.

I left what I did at the tmp.perf/core branch of my repo at
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git.

as discussed on irc we will disable it for cross builds now, because we dont have good solution at the moment.. it's similar case as for fixdep tool:

3a70fcd3a4db tools build: Fix cross compile build
...
We need to add support for host side tools build, meanwhile disabling fixdep usage for cross arch builds.

I'll make a change to disable this for crossbuild and work on common solution later

thanks,
jirka

#60 lore raw
Quoted 30 lines · Jiri Olsa
Quoted 16 lines · Arnaldo Carvalho de Melo

Em Mon, Sep 19, 2016 at 09:02:58PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 6 lines

Em Mon, Sep 19, 2016 at 08:37:53PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 4 lines

yeah, changing that typedef + true def to plain include <stdbool.h> makes it progress to the next failure, which is in cross compilation environments, such as using fedora 24 + the Android NDK to try to build a ARM android binary.

Quoted 8 lines

14 fedora:24-x-ARC-uClibc: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
/bin/sh: /tmp/build/perf/pmu-events/jevents: cannot execute binary file: Exec format error pmu-events/Build:11: recipe for target '/tmp/build/perf/pmu-events/pmu-events.c' failed make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 126 Makefile.perf:461: recipe for target '/tmp/build/perf/pmu-events/pmu-events-in.o' failed make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....

Quoted 8 lines

Jiri, we need something similar to scripts/Makefile.host :-\

Calling it a day, perhaps, for now, we should just detect that it is a corss compile env (CROSS_COMPILE is set) and exclude all this code from the build, emitting a warning.

I left what I did at the tmp.perf/core branch of my repo at
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git.

as discussed on irc we will disable it for cross builds now, because we dont have good solution at the moment.. it's similar case as for fixdep tool:

Quoted 7 lines

3a70fcd3a4db tools build: Fix cross compile build
...
We need to add support for host side tools build, meanwhile disabling fixdep usage for cross arch builds.

I'll make a change to disable this for crossbuild and work on common solution later

could you please give it a try with patch below?
I tested but not with properly cross building...

also, did you want some message during the cross build that pmu-events are not included?

thanks,
jirka

---

tools/perf/Makefile.config +4 -0
index 24803c58049a..ec3a59d9a56b 100644
@@ -755,6 +755,10 @@ ifndef NO_AUXTRACE
   endif
 endif
 
+ifndef CROSS_COMPILE
+  CFLAGS += -DHAVE_PMU_EVENTS_SUPPORT
+endif
+
 # Among the variables below, these:
 #   perfexecdir
 #   template_dir
tools/perf/Makefile.perf +7 -0
index 0abebcba849f..85257c49d5ab 100644
@@ -349,7 +349,14 @@ include $(srctree)/tools/build/Makefile.include
 
 JEVENTS       := $(OUTPUT)pmu-events/jevents
 JEVENTS_IN    := $(OUTPUT)pmu-events/jevents-in.o
+
+#
+# Disabling pmu-events for cross compile, as
+# we dont support host CC tools building yet.
+#
+ifndef CROSS_COMPILE
 PMU_EVENTS_IN := $(OUTPUT)pmu-events/pmu-events-in.o
+endif
 
 export JEVENTS
 
tools/perf/util/pmu.c +9 -0
index ac097fcaba7b..21a3f21a6f6f 100644
@@ -482,6 +482,8 @@ static struct cpu_map *pmu_cpumask(const char *name)
 	return cpus;
 }
 
+#ifdef HAVE_PMU_EVENTS_SUPPORT
+
 /*
  * Return the CPU id as a raw string.
  *
@@ -545,6 +547,13 @@ out:
 	return 0;
 }
 
+#else
+static int pmu_add_cpu_aliases(struct list_head *head __maybe_unused)
+{
+	return 0;
+}
+#endif /* HAVE_PMU_EVENTS_SUPPORT */
+
 struct perf_event_attr * __weak
 perf_pmu__get_default_config(struct perf_pmu *pmu __maybe_unused)
 {
#61 lore raw
Quoted 36 lines · Jiri Olsa
Quoted 35 lines · Jiri Olsa
Quoted 16 lines · Arnaldo Carvalho de Melo

Em Mon, Sep 19, 2016 at 09:02:58PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 6 lines

Em Mon, Sep 19, 2016 at 08:37:53PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 4 lines

yeah, changing that typedef + true def to plain include <stdbool.h> makes it progress to the next failure, which is in cross compilation environments, such as using fedora 24 + the Android NDK to try to build a ARM android binary.

Quoted 8 lines

14 fedora:24-x-ARC-uClibc: FAIL
GEN /tmp/build/perf/pmu-events/pmu-events.c
/bin/sh: /tmp/build/perf/pmu-events/jevents: cannot execute binary file: Exec format error pmu-events/Build:11: recipe for target '/tmp/build/perf/pmu-events/pmu-events.c' failed make[2]: *** [/tmp/build/perf/pmu-events/pmu-events.c] Error 126 Makefile.perf:461: recipe for target '/tmp/build/perf/pmu-events/pmu-events-in.o' failed make[1]: *** [/tmp/build/perf/pmu-events/pmu-events-in.o] Error 2
make[1]: *** Waiting for unfinished jobs....

Quoted 8 lines

Jiri, we need something similar to scripts/Makefile.host :-\

Calling it a day, perhaps, for now, we should just detect that it is a corss compile env (CROSS_COMPILE is set) and exclude all this code from the build, emitting a warning.

I left what I did at the tmp.perf/core branch of my repo at
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git.

as discussed on irc we will disable it for cross builds now, because we dont have good solution at the moment.. it's similar case as for fixdep tool:

3a70fcd3a4db tools build: Fix cross compile build
...
We need to add support for host side tools build, meanwhile disabling fixdep usage for cross arch builds.

Quoted 7 lines
Quoted 2 lines

I'll make a change to disable this for crossbuild and work on common solution later

could you please give it a try with patch below?
I tested but not with properly cross building...

also, did you want some message during the cross build that pmu-events are not included?

ping.. is that working for you? IMO we can include this as additional patch to the set..

thanks,
jirka

#62 lore raw

Em Mon, Sep 26, 2016 at 10:35:33AM +0200, Jiri Olsa escreveu:

Quoted 2 lines

ping.. is that working for you? IMO we can include this as additional patch to the set..

No, it doesn't fails to build on the first cross env I tried, fixing it now, resulting patch:

tools/perf/Makefile.config +4 -0
index 72edf83d76b7..9365c155c6f3 100644
@@ -758,6 +758,10 @@ ifndef NO_AUXTRACE
   endif
 endif
 
+ifndef CROSS_COMPILE
+  CFLAGS += -DHAVE_PMU_EVENTS_SUPPORT
+endif
+
 # Among the variables below, these:
 #   perfexecdir
 #   template_dir
tools/perf/Makefile.perf +7 -0
index 26dbee50b36c..ee86dbf2814e 100644
@@ -349,7 +349,14 @@ include $(srctree)/tools/build/Makefile.include
 
 JEVENTS       := $(OUTPUT)pmu-events/jevents
 JEVENTS_IN    := $(OUTPUT)pmu-events/jevents-in.o
+
+#
+# Disabling pmu-events for cross compile, as
+# we dont support host CC tools building yet.
+#
+ifndef CROSS_COMPILE
 PMU_EVENTS_IN := $(OUTPUT)pmu-events/pmu-events-in.o
+endif
 
 export JEVENTS
 
tools/perf/util/pmu.c +19 -0
index 2babcdf62839..37f74fcc9ca2 100644
@@ -377,6 +377,8 @@ static int pmu_alias_terms(struct perf_pmu_alias *alias,
 	return 0;
 }
 
+#ifdef HAVE_PMU_EVENTS_SUPPORT
+
 /*
  * Reading/parsing the default pmu type value, which should be
  * located at:
@@ -473,6 +475,23 @@ static struct cpu_map *pmu_cpumask(const char *name)
 	return cpus;
 }
 
+#else
+static int pmu_type(const char *name __maybe_unused, __u32 *type)
+{
+	*type = 0;
+	return 0;
+}
+
+static void pmu_read_sysfs(void)
+{
+}
+
+static struct cpu_map *pmu_cpumask(const char *name __maybe_unused)
+{
+	return NULL;
+}
+#endif /* HAVE_PMU_EVENTS_SUPPORT */
+
 struct perf_event_attr * __weak
 perf_pmu__get_default_config(struct perf_pmu *pmu __maybe_unused)
 {
#63 lore raw
Quoted 6 lines · Arnaldo Carvalho de Melo

Em Mon, Sep 26, 2016 at 10:35:33AM +0200, Jiri Olsa escreveu:

Quoted 2 lines

ping.. is that working for you? IMO we can include this as additional patch to the set..

No, it doesn't fails to build on the first cross env I tried, fixing it now, resulting patch:

Yes it shouldn't be difficult to fix cross building. I don't think there are any fundamental problems.

-Andi

#64 lore raw
Quoted 10 lines · Andi Kleen
Quoted 6 lines · Arnaldo Carvalho de Melo

Em Mon, Sep 26, 2016 at 10:35:33AM +0200, Jiri Olsa escreveu:

Quoted 2 lines

ping.. is that working for you? IMO we can include this as additional patch to the set..

No, it doesn't fails to build on the first cross env I tried, fixing it now, resulting patch:

Yes it shouldn't be difficult to fix cross building. I don't think there are any fundamental problems.

right, how about attached patch

Arnaldo,
could you please try it on cross build.. I still dont have setup for that :-\

thanks,
jirka

---

tools/build/Build +2 -0
index 63a6c34c0c88..76d1a4960973 100644
@@ -1 +1,3 @@
+hostprogs := fixdep
+
 fixdep-y := fixdep.o
tools/build/Makefile +6 -0
index 0d5a0e3a8fa9..653faee2a055 100644
@@ -14,6 +14,12 @@ endef
 $(call allow-override,CC,$(CROSS_COMPILE)gcc)
 $(call allow-override,LD,$(CROSS_COMPILE)ld)
 
+HOSTCC ?= gcc
+HOSTLD ?= ld
+HOSTAR ?= ar
+
+export HOSTCC HOSTLD HOSTAR
+
 ifeq ($(V),1)
   Q =
 else
tools/build/Makefile.build +15 -4
index 27f3583193e6..031c5631cc21 100644
@@ -58,6 +58,9 @@ quiet_cmd_mkdir = MKDIR    $(dir $@)
 quiet_cmd_cc_o_c = CC       $@
       cmd_cc_o_c = $(CC) $(c_flags) -c -o $@ $<
 
+quiet_cmd_host_cc_o_c = HOSTCC   $@
+      cmd_host_cc_o_c = $(HOSTCC) $(c_flags) -c -o $@ $<
+
 quiet_cmd_cpp_i_c = CPP      $@
       cmd_cpp_i_c = $(CC) $(c_flags) -E -o $@ $<
 
@@ -70,16 +73,24 @@ quiet_cmd_gen = GEN      $@
 # If there's nothing to link, create empty $@ object.
 quiet_cmd_ld_multi = LD       $@
       cmd_ld_multi = $(if $(strip $(obj-y)),\
-		       $(LD) -r -o $@  $(filter $(obj-y),$^),rm -f $@; $(AR) rcs $@)
+                     $(LD) -r -o $@  $(filter $(obj-y),$^),rm -f $@; $(AR) rcs $@)
+
+quiet_cmd_host_ld_multi = HOSTLD   $@
+      cmd_host_ld_multi = $(if $(strip $(obj-y)),\
+                          $(HOSTLD) -r -o $@  $(filter $(obj-y),$^),rm -f $@; $(HOSTAR) rcs $@)
+
+ifneq ($(filter $(obj),$(hostprogs)),)
+  host = host_
+endif
 
 # Build rules
 $(OUTPUT)%.o: %.c FORCE
 	$(call rule_mkdir)
-	$(call if_changed_dep,cc_o_c)
+	$(call if_changed_dep,$(host)cc_o_c)
 
 $(OUTPUT)%.o: %.S FORCE
 	$(call rule_mkdir)
-	$(call if_changed_dep,cc_o_c)
+	$(call if_changed_dep,$(host)cc_o_c)
 
 $(OUTPUT)%.i: %.c FORCE
 	$(call rule_mkdir)
@@ -119,7 +130,7 @@ $(sort $(subdir-obj-y)): $(subdir-y) ;
 
 $(in-target): $(obj-y) FORCE
 	$(call rule_mkdir)
-	$(call if_changed,ld_multi)
+	$(call if_changed,$(host)ld_multi)
 
 __build: $(in-target)
 	@:
tools/build/Makefile.include +0 -4
index be630bed66d2..ad22e4e7bc59 100644
@@ -1,10 +1,6 @@
 build := -f $(srctree)/tools/build/Makefile.build dir=. obj
 
-ifdef CROSS_COMPILE
-fixdep:
-else
 fixdep:
 	$(Q)$(MAKE) -C $(srctree)/tools/build CFLAGS= LDFLAGS= $(OUTPUT)fixdep
-endif
 
 .PHONY: fixdep
tools/perf/Makefile.perf +5 -0
index 0abebcba849f..1347b5de3669 100644
@@ -144,6 +144,10 @@ $(call allow-override,LD,$(CROSS_COMPILE)ld)
 
 LD += $(EXTRA_LDFLAGS)
 
+HOSTCC  ?= gcc
+HOSTLD  ?= ld
+HOSTAR  ?= ar
+
 PKG_CONFIG = $(CROSS_COMPILE)pkg-config
 
 RM      = rm -f
@@ -345,6 +349,7 @@ strip: $(PROGRAMS) $(OUTPUT)perf
 PERF_IN := $(OUTPUT)perf-in.o
 
 export srctree OUTPUT RM CC LD AR CFLAGS V BISON FLEX AWK
+export HOSTCC HOSTLD HOSTAR
 include $(srctree)/tools/build/Makefile.include
 
 JEVENTS       := $(OUTPUT)pmu-events/jevents
tools/perf/pmu-events/Build +2 -0
index d2f34307ae79..9213a1273697 100644
@@ -1,3 +1,5 @@
+hostprogs := jevents
+
 jevents-y	+= json.o jsmn.o jevents.o
 pmu-events-y	+= pmu-events.o
 JDIR		=  pmu-events/arch/$(ARCH)
#65 lore raw

Em Tue, Sep 27, 2016 at 04:18:46PM +0200, Jiri Olsa escreveu:

Quoted 12 lines · Andi Kleen
Quoted 10 lines · Arnaldo Carvalho de Melo
Quoted 6 lines

Em Mon, Sep 26, 2016 at 10:35:33AM +0200, Jiri Olsa escreveu:

Quoted 2 lines

ping.. is that working for you? IMO we can include this as additional patch to the set..

No, it doesn't fails to build on the first cross env I tried, fixing it now, resulting patch:

Yes it shouldn't be difficult to fix cross building. I don't think there are any fundamental problems.

Quoted 7 lines

right, how about attached patch

Arnaldo,
could you please try it on cross build.. I still dont have setup for that :-\

thanks,
jirka

So, this makes it work for me in one of the cross build envs I have (all in https://hub.docker.com/r/acmel/) if I apply this patch on top:

tools/build/Makefile +1 -1
index 653faee2a055..8332959fbca4 100644
@@ -42,7 +42,7 @@ $(OUTPUT)fixdep-in.o: FORCE
 	$(Q)$(MAKE) $(build)=fixdep
 
 $(OUTPUT)fixdep: $(OUTPUT)fixdep-in.o
-	$(QUIET_LINK)$(CC) $(LDFLAGS) -o $@ $<
+	$(QUIET_LINK)$(HOSTCC) $(LDFLAGS) -o $@ $<
 
 FORCE:
 

-------------------

I've broken up the patch into multiple ones, to get first fixdep working, then to move to jevents, I'm putting this on a tmp.perf/hostprog branch till I've tested it all.

- Arnaldo

Quoted 126 lines

---

tools/build/Build +2 -0
index 63a6c34c0c88..76d1a4960973 100644
@@ -1 +1,3 @@
+hostprogs := fixdep
+
 fixdep-y := fixdep.o
tools/build/Makefile +6 -0
index 0d5a0e3a8fa9..653faee2a055 100644
@@ -14,6 +14,12 @@ endef
 $(call allow-override,CC,$(CROSS_COMPILE)gcc)
 $(call allow-override,LD,$(CROSS_COMPILE)ld)
 
+HOSTCC ?= gcc
+HOSTLD ?= ld
+HOSTAR ?= ar
+
+export HOSTCC HOSTLD HOSTAR
+
 ifeq ($(V),1)
   Q =
 else
tools/build/Makefile.build +15 -4
index 27f3583193e6..031c5631cc21 100644
@@ -58,6 +58,9 @@ quiet_cmd_mkdir = MKDIR    $(dir $@)
 quiet_cmd_cc_o_c = CC       $@
       cmd_cc_o_c = $(CC) $(c_flags) -c -o $@ $<
 
+quiet_cmd_host_cc_o_c = HOSTCC   $@
+      cmd_host_cc_o_c = $(HOSTCC) $(c_flags) -c -o $@ $<
+
 quiet_cmd_cpp_i_c = CPP      $@
       cmd_cpp_i_c = $(CC) $(c_flags) -E -o $@ $<
 
@@ -70,16 +73,24 @@ quiet_cmd_gen = GEN      $@
 # If there's nothing to link, create empty $@ object.
 quiet_cmd_ld_multi = LD       $@
       cmd_ld_multi = $(if $(strip $(obj-y)),\
-		       $(LD) -r -o $@  $(filter $(obj-y),$^),rm -f $@; $(AR) rcs $@)
+                     $(LD) -r -o $@  $(filter $(obj-y),$^),rm -f $@; $(AR) rcs $@)
+
+quiet_cmd_host_ld_multi = HOSTLD   $@
+      cmd_host_ld_multi = $(if $(strip $(obj-y)),\
+                          $(HOSTLD) -r -o $@  $(filter $(obj-y),$^),rm -f $@; $(HOSTAR) rcs $@)
+
+ifneq ($(filter $(obj),$(hostprogs)),)
+  host = host_
+endif
 
 # Build rules
 $(OUTPUT)%.o: %.c FORCE
 	$(call rule_mkdir)
-	$(call if_changed_dep,cc_o_c)
+	$(call if_changed_dep,$(host)cc_o_c)
 
 $(OUTPUT)%.o: %.S FORCE
 	$(call rule_mkdir)
-	$(call if_changed_dep,cc_o_c)
+	$(call if_changed_dep,$(host)cc_o_c)
 
 $(OUTPUT)%.i: %.c FORCE
 	$(call rule_mkdir)
@@ -119,7 +130,7 @@ $(sort $(subdir-obj-y)): $(subdir-y) ;
 
 $(in-target): $(obj-y) FORCE
 	$(call rule_mkdir)
-	$(call if_changed,ld_multi)
+	$(call if_changed,$(host)ld_multi)
 
 __build: $(in-target)
 	@:
tools/build/Makefile.include +0 -4
index be630bed66d2..ad22e4e7bc59 100644
@@ -1,10 +1,6 @@
 build := -f $(srctree)/tools/build/Makefile.build dir=. obj
 
-ifdef CROSS_COMPILE
-fixdep:
-else
 fixdep:
 	$(Q)$(MAKE) -C $(srctree)/tools/build CFLAGS= LDFLAGS= $(OUTPUT)fixdep
-endif
 
 .PHONY: fixdep
tools/perf/Makefile.perf +5 -0
index 0abebcba849f..1347b5de3669 100644
@@ -144,6 +144,10 @@ $(call allow-override,LD,$(CROSS_COMPILE)ld)
 
 LD += $(EXTRA_LDFLAGS)
 
+HOSTCC  ?= gcc
+HOSTLD  ?= ld
+HOSTAR  ?= ar
+
 PKG_CONFIG = $(CROSS_COMPILE)pkg-config
 
 RM      = rm -f
@@ -345,6 +349,7 @@ strip: $(PROGRAMS) $(OUTPUT)perf
 PERF_IN := $(OUTPUT)perf-in.o
 
 export srctree OUTPUT RM CC LD AR CFLAGS V BISON FLEX AWK
+export HOSTCC HOSTLD HOSTAR
 include $(srctree)/tools/build/Makefile.include
 
 JEVENTS       := $(OUTPUT)pmu-events/jevents
tools/perf/pmu-events/Build +2 -0
index d2f34307ae79..9213a1273697 100644
@@ -1,3 +1,5 @@
+hostprogs := jevents
+
 jevents-y	+= json.o jsmn.o jevents.o
 pmu-events-y	+= pmu-events.o
 JDIR		=  pmu-events/arch/$(ARCH)
#66 lore raw
Quoted 39 lines · Arnaldo Carvalho de Melo

Em Tue, Sep 27, 2016 at 04:18:46PM +0200, Jiri Olsa escreveu:

Quoted 12 lines · Andi Kleen
Quoted 10 lines · Arnaldo Carvalho de Melo
Quoted 6 lines

Em Mon, Sep 26, 2016 at 10:35:33AM +0200, Jiri Olsa escreveu:

Quoted 2 lines

ping.. is that working for you? IMO we can include this as additional patch to the set..

No, it doesn't fails to build on the first cross env I tried, fixing it now, resulting patch:

Yes it shouldn't be difficult to fix cross building. I don't think there are any fundamental problems.

Quoted 7 lines

right, how about attached patch

Arnaldo,
could you please try it on cross build.. I still dont have setup for that :-\

thanks,
jirka

So, this makes it work for me in one of the cross build envs I have (all in https://hub.docker.com/r/acmel/) if I apply this patch on top:

tools/build/Makefile +1 -2
index 653faee2a055..8332959fbca4 100644
@@ -42,7 +42,7 @@ $(OUTPUT)fixdep-in.o: FORCE
 	$(Q)$(MAKE) $(build)=fixdep
 
 $(OUTPUT)fixdep: $(OUTPUT)fixdep-in.o
-	$(QUIET_LINK)$(CC) $(LDFLAGS) -o $@ $<
+	$(QUIET_LINK)$(HOSTCC) $(LDFLAGS) -o $@ $<
 
 FORCE:
 
-------------------
Quoted 3 lines

I've broken up the patch into multiple ones, to get first fixdep working, then to move to jevents, I'm putting this on a tmp.perf/hostprog branch till I've tested it all.

looks great, thanks

jirka

#67 lore raw

[tip:perf/urgent] tools build: Add support for host programs format

In reply to Jiri Olsa
tools build: Add support for host programs format
Commit-ID:  0c3b7e42616f1f6084cfeb0d443cbff0b2c424a9
Gitweb:     http://git.kernel.org/tip/0c3b7e42616f1f6084cfeb0d443cbff0b2c424a9
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 27 Sep 2016 16:18:46 +0200
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 11:39:09 -0300

tools build: Add support for host programs format

In some cases, like for fixdep and shortly for jevents, we need to build a tool to run on the host that will be used in building a tool, such as perf, that is being cross compiled, so do like the kernel and provide HOSTCC, HOSTLD and HOSTAR to do that.

---
 tools/build/Build.include  |  5 +++++
 tools/build/Makefile       |  6 ++++++
 tools/build/Makefile.build | 19 +++++++++++++++----
 3 files changed, 26 insertions(+), 4 deletions(-)
tools/build/Build.include +5 -0
index 4d000bc..0248938 100644
@@ -90,3 +90,8 @@ if_changed = $(if $(strip $(any-prereq) $(arg-check)),             \
 # - per object C flags
 # - BUILD_STR macro to allow '-D"$(variable)"' constructs
 c_flags = -Wp,-MD,$(depfile),-MT,$@ $(CFLAGS) -D"BUILD_STR(s)=\#s" $(CFLAGS_$(basetarget).o) $(CFLAGS_$(obj))
+
+###
+## HOSTCC C flags
+
+host_c_flags = -Wp,-MD,$(depfile),-MT,$@ $(CHOSTFLAGS) -D"BUILD_STR(s)=\#s" $(CHOSTFLAGS_$(basetarget).o) $(CHOSTFLAGS_$(obj))
tools/build/Makefile +6 -0
index 0d5a0e3..653faee 100644
@@ -14,6 +14,12 @@ endef
 $(call allow-override,CC,$(CROSS_COMPILE)gcc)
 $(call allow-override,LD,$(CROSS_COMPILE)ld)
 
+HOSTCC ?= gcc
+HOSTLD ?= ld
+HOSTAR ?= ar
+
+export HOSTCC HOSTLD HOSTAR
+
 ifeq ($(V),1)
   Q =
 else
tools/build/Makefile.build +15 -4
index 27f3583..190519a 100644
@@ -58,6 +58,9 @@ quiet_cmd_mkdir = MKDIR    $(dir $@)
 quiet_cmd_cc_o_c = CC       $@
       cmd_cc_o_c = $(CC) $(c_flags) -c -o $@ $<
 
+quiet_cmd_host_cc_o_c = HOSTCC   $@
+      cmd_host_cc_o_c = $(HOSTCC) $(host_c_flags) -c -o $@ $<
+
 quiet_cmd_cpp_i_c = CPP      $@
       cmd_cpp_i_c = $(CC) $(c_flags) -E -o $@ $<
 
@@ -70,16 +73,24 @@ quiet_cmd_gen = GEN      $@
 # If there's nothing to link, create empty $@ object.
 quiet_cmd_ld_multi = LD       $@
       cmd_ld_multi = $(if $(strip $(obj-y)),\
-		       $(LD) -r -o $@  $(filter $(obj-y),$^),rm -f $@; $(AR) rcs $@)
+                     $(LD) -r -o $@  $(filter $(obj-y),$^),rm -f $@; $(AR) rcs $@)
+
+quiet_cmd_host_ld_multi = HOSTLD   $@
+      cmd_host_ld_multi = $(if $(strip $(obj-y)),\
+                          $(HOSTLD) -r -o $@  $(filter $(obj-y),$^),rm -f $@; $(HOSTAR) rcs $@)
+
+ifneq ($(filter $(obj),$(hostprogs)),)
+  host = host_
+endif
 
 # Build rules
 $(OUTPUT)%.o: %.c FORCE
 	$(call rule_mkdir)
-	$(call if_changed_dep,cc_o_c)
+	$(call if_changed_dep,$(host)cc_o_c)
 
 $(OUTPUT)%.o: %.S FORCE
 	$(call rule_mkdir)
-	$(call if_changed_dep,cc_o_c)
+	$(call if_changed_dep,$(host)cc_o_c)
 
 $(OUTPUT)%.i: %.c FORCE
 	$(call rule_mkdir)
@@ -119,7 +130,7 @@ $(sort $(subdir-obj-y)): $(subdir-y) ;
 
 $(in-target): $(obj-y) FORCE
 	$(call rule_mkdir)
-	$(call if_changed,ld_multi)
+	$(call if_changed,$(host)ld_multi)
 
 __build: $(in-target)
 	@:
#68 lore raw

[tip:perf/urgent] tools build: Make fixdep a hostprog

In reply to Jiri Olsa
tools build: Make fixdep a hostprog
Commit-ID:  6b3db6f9b970001b751a6cd001bc2ab581ce0fb3
Gitweb:     http://git.kernel.org/tip/6b3db6f9b970001b751a6cd001bc2ab581ce0fb3
Author:     Jiri Olsa <[email protected]>
AuthorDate: Tue, 27 Sep 2016 16:18:46 +0200
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 11:40:35 -0300

tools build: Make fixdep a hostprog

It is used in the build process, so stop suppressing its build in tools cross builds.

[ Use HOSTCC on the $(OUTPUT)fixdep target, it was using the x-compiler
to link fixdep-in.o, that was correctly built with HOSTCC and thus failing ]

---
 tools/build/Build            | 2 ++
 tools/build/Makefile         | 2 +-
 tools/build/Makefile.include | 4 ----
 3 files changed, 3 insertions(+), 5 deletions(-)
tools/build/Build +2 -0
index 63a6c34..76d1a49 100644
@@ -1 +1,3 @@
+hostprogs := fixdep
+
 fixdep-y := fixdep.o
tools/build/Makefile +1 -1
index 653faee..8332959 100644
@@ -42,7 +42,7 @@ $(OUTPUT)fixdep-in.o: FORCE
 	$(Q)$(MAKE) $(build)=fixdep
 
 $(OUTPUT)fixdep: $(OUTPUT)fixdep-in.o
-	$(QUIET_LINK)$(CC) $(LDFLAGS) -o $@ $<
+	$(QUIET_LINK)$(HOSTCC) $(LDFLAGS) -o $@ $<
 
 FORCE:
 
tools/build/Makefile.include +0 -4
index be630be..ad22e4e 100644
@@ -1,10 +1,6 @@
 build := -f $(srctree)/tools/build/Makefile.build dir=. obj
 
-ifdef CROSS_COMPILE
-fixdep:
-else
 fixdep:
 	$(Q)$(MAKE) -C $(srctree)/tools/build CFLAGS= LDFLAGS= $(OUTPUT)fixdep
-endif
 
 .PHONY: fixdep
#69 lore raw

[tip:perf/urgent] perf jevents: Program to convert JSON file

In reply to Jiri Olsa
perf jevents: Program to convert JSON file
Commit-ID:  80eeb67fe577aa76b2d1bb5b029bca097f0f25bc
Gitweb:     http://git.kernel.org/tip/80eeb67fe577aa76b2d1bb5b029bca097f0f25bc
Author:     Andi Kleen <[email protected]>
AuthorDate: Mon, 19 Sep 2016 17:39:33 -0300
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 3 Oct 2016 19:55:55 -0300

perf jevents: Program to convert JSON file

This is a modified version of an earlier patch by Andi Kleen.

We expect architectures to create JSON files describing the performance monitoring (PMU) events that each CPU model/family of the architecture supports.

Following is an example of the JSON file entry for an x86 event:

    	[
    	...
    	{
    	"EventCode": "0x00",
    	"UMask": "0x01",
    	"EventName": "INST_RETIRED.ANY",
    	"BriefDescription": "Instructions retired from execution.",
    	"PublicDescription": "Instructions retired from execution.",
    	"Counter": "Fixed counter 1",
    	"CounterHTOff": "Fixed counter 1",
    	"SampleAfterValue": "2000003",
    	"SampleAfterValue": "2000003",
    	"MSRIndex": "0",
    	"MSRValue": "0",
    	"TakenAlone": "0",
    	"CounterMask": "0",
    	"Invert": "0",
    	"AnyThread": "0",
    	"EdgeDetect": "0",
    	"PEBS": "0",
    	"PRECISE_STORE": "0",
    	"Errata": "null",
    	"Offcore": "0"
    	},
    	...

    	]

All the PMU events supported by a CPU model/family must be grouped into "topics" such as "Pipelining", "Floating-point", "Virtual-memory" etc.

All events belonging to a topic must be placed in a separate JSON file (eg: "Pipelining.json") and all the topic JSON files for a CPU model must be in a separate directory.

Eg: for the CPU model "Silvermont_core":

$ ls tools/perf/pmu-events/arch/x86/Silvermont_core Floating-point.json
Memory.json
Other.json
Pipelining.json
Virtualmemory.json

Finally, to allow multiple CPU models to share a single set of JSON files, architectures must provide a mapping between a model and its set of events:

    	$ grep Silvermont tools/perf/pmu-events/arch/x86/mapfile.csv
    	GenuineIntel-6-4D,V13,Silvermont_core,core
    	GenuineIntel-6-4C,V13,Silvermont_core,core

which maps each CPU, identified by [vendor, family, model, version, type] to a directory of JSON files. Thus two (or more) CPU models support the set of PMU events listed in the directory.

tools/perf/pmu-events/arch/x86/Silvermont_core/

Given this organization of files, the program, jevents:

- locates all JSON files for each CPU-model of the architecture,

- parses all JSON files for the CPU-model and generates a C-style "PMU-events table" (pmu-events.c) for the model

- locates a mapfile for the architecture

- builds a global table, mapping each model of CPU to the corresponding PMU-events table.

The 'pmu-events.c' is generated when building perf and added to libperf.a. The global table pmu_events_map[] table in this pmu-events.c will be used in perf in a follow-on patch.

If the architecture does not have any JSON files or there is an error in processing them, an empty mapping file is created. This would allow the build of perf to proceed even if we are not able to provide aliases for events.

The parser for JSON files allows parsing Intel style JSON event files. This allows to use an Intel event list directly with perf. The Intel event lists can be quite large and are too big to store in unswappable kernel memory.

The conversion from JSON to C-style is straight forward. The parser knows (very little) Intel specific information, and can be easily extended to handle fields for other CPUs.

The parser code is partially shared with an independent parsing library, which is 2-clause BSD licensed. To avoid any conflicts I marked those files as BSD licensed too. As part of perf they become GPLv2.

Committer notes:

Fixes:

1) Limit maxfds to 512 to avoid nftd() segfaulting on alloca() with a big rlim_max, as in docker containers - acme

2) Make jevents a hostprog, supporting cross compilation - jolsa

3) Use HOSTCC for jevents final step - acme

4) Define _GNU_SOURCE for asprintf, as we can't use CC's EXTRA_CFLAGS, that has to have --sysroot on the Android NDK 24 - acme

5) Removed $(srctree)/tools/perf/pmu-events/pmu-events.c from the 'clean' target, it is generated on $(OUTPUT)pmu-events/pmu-events.c, which is already taken care of in the original patch - acme

---
 tools/perf/Makefile.perf           |  34 +-
 tools/perf/pmu-events/Build        |  13 +
 tools/perf/pmu-events/jevents.c    | 767 +++++++++++++++++++++++++++++++++++++
 tools/perf/pmu-events/jevents.h    |  17 +
 tools/perf/pmu-events/json.h       |   6 +
 tools/perf/pmu-events/pmu-events.h |  36 ++
 6 files changed, 869 insertions(+), 4 deletions(-)
tools/perf/Makefile.perf +30 -4
index d710db1..982d643 100644
@@ -144,6 +144,10 @@ $(call allow-override,LD,$(CROSS_COMPILE)ld)
 
 LD += $(EXTRA_LDFLAGS)
 
+HOSTCC  ?= gcc
+HOSTLD  ?= ld
+HOSTAR  ?= ar
+
 PKG_CONFIG = $(CROSS_COMPILE)pkg-config
 
 RM      = rm -f
@@ -345,8 +349,18 @@ strip: $(PROGRAMS) $(OUTPUT)perf
 PERF_IN := $(OUTPUT)perf-in.o
 
 export srctree OUTPUT RM CC LD AR CFLAGS V BISON FLEX AWK
+export HOSTCC HOSTLD HOSTAR
 include $(srctree)/tools/build/Makefile.include
 
+JEVENTS       := $(OUTPUT)pmu-events/jevents
+JEVENTS_IN    := $(OUTPUT)pmu-events/jevents-in.o
+
+PMU_EVENTS_IN := $(OUTPUT)pmu-events/pmu-events-in.o
+
+export JEVENTS
+
+build := -f $(srctree)/tools/build/Makefile.build dir=. obj
+
 $(PERF_IN): prepare FORCE
 	@(test -f ../../include/uapi/linux/perf_event.h && ( \
         (diff -B ../include/uapi/linux/perf_event.h ../../include/uapi/linux/perf_event.h >/dev/null) \
@@ -443,9 +457,18 @@ $(PERF_IN): prepare FORCE
 	|| echo "Warning: tools/include/uapi/linux/mman.h differs from kernel" >&2 )) || true
 	$(Q)$(MAKE) $(build)=perf
 
-$(OUTPUT)perf: $(PERFLIBS) $(PERF_IN) $(LIBTRACEEVENT_DYNAMIC_LIST)
+$(JEVENTS_IN): FORCE
+	$(Q)$(MAKE) -f $(srctree)/tools/build/Makefile.build dir=pmu-events obj=jevents
+
+$(JEVENTS): $(JEVENTS_IN)
+	$(QUIET_LINK)$(HOSTCC) $(JEVENTS_IN) -o $@
+
+$(PMU_EVENTS_IN): $(JEVENTS) FORCE
+	$(Q)$(MAKE) -f $(srctree)/tools/build/Makefile.build dir=pmu-events obj=pmu-events
+
+$(OUTPUT)perf: $(PERFLIBS) $(PERF_IN) $(PMU_EVENTS_IN) $(LIBTRACEEVENT_DYNAMIC_LIST)
 	$(QUIET_LINK)$(CC) $(CFLAGS) $(LDFLAGS) $(LIBTRACEEVENT_DYNAMIC_LIST_LDFLAGS) \
-		$(PERF_IN) $(LIBS) -o $@
+		$(PERF_IN) $(PMU_EVENTS_IN) $(LIBS) -o $@
 
 $(GTK_IN): fixdep FORCE
 	$(Q)$(MAKE) $(build)=gtk
@@ -474,6 +497,8 @@ perf.spec $(SCRIPTS) \
 ifneq ($(OUTPUT),)
 %.o: $(OUTPUT)%.o
 	@echo "    # Redirected target $@ => $(OUTPUT)$@"
+pmu-events/%.o: $(OUTPUT)pmu-events/%.o
+	@echo "    # Redirected target $@ => $(OUTPUT)$@"
 util/%.o: $(OUTPUT)util/%.o
 	@echo "    # Redirected target $@ => $(OUTPUT)$@"
 bench/%.o: $(OUTPUT)bench/%.o
@@ -729,10 +754,11 @@ clean:: $(LIBTRACEEVENT)-clean $(LIBAPI)-clean $(LIBBPF)-clean $(LIBSUBCMD)-clea
 	$(call QUIET_CLEAN, core-objs)  $(RM) $(LIB_FILE) $(OUTPUT)perf-archive $(OUTPUT)perf-with-kcore $(LANG_BINDINGS)
 	$(Q)find $(if $(OUTPUT),$(OUTPUT),.) -name '*.o' -delete -o -name '\.*.cmd' -delete -o -name '\.*.d' -delete
 	$(Q)$(RM) $(OUTPUT).config-detected
-	$(call QUIET_CLEAN, core-progs) $(RM) $(ALL_PROGRAMS) perf perf-read-vdso32 perf-read-vdsox32
+	$(call QUIET_CLEAN, core-progs) $(RM) $(ALL_PROGRAMS) perf perf-read-vdso32 perf-read-vdsox32 $(OUTPUT)pmu-events/jevents
 	$(call QUIET_CLEAN, core-gen)   $(RM)  *.spec *.pyc *.pyo */*.pyc */*.pyo $(OUTPUT)common-cmds.h TAGS tags cscope* $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)FEATURE-DUMP $(OUTPUT)util/*-bison* $(OUTPUT)util/*-flex* \
 		$(OUTPUT)util/intel-pt-decoder/inat-tables.c $(OUTPUT)fixdep \
-		$(OUTPUT)tests/llvm-src-{base,kbuild,prologue,relocation}.c
+		$(OUTPUT)tests/llvm-src-{base,kbuild,prologue,relocation}.c \
+		$(OUTPUT)pmu-events/pmu-events.c
 	$(QUIET_SUBDIR0)Documentation $(QUIET_SUBDIR1) clean
 	$(python-clean)
 
tools/perf/pmu-events/Build new file +13 -0
new file mode 100644
index 0000000..9213a12
@@ -0,0 +1,13 @@
+hostprogs := jevents
+
+jevents-y	+= json.o jsmn.o jevents.o
+pmu-events-y	+= pmu-events.o
+JDIR		=  pmu-events/arch/$(ARCH)
+JSON		=  $(shell [ -d $(JDIR) ] &&				\
+			find $(JDIR) -name '*.json' -o -name 'mapfile.csv')
+#
+# Locate/process JSON files in pmu-events/arch/
+# directory and create tables in pmu-events.c.
+#
+$(OUTPUT)pmu-events/pmu-events.c: $(JSON) $(JEVENTS)
+	$(Q)$(call echo-cmd,gen)$(JEVENTS) $(ARCH) pmu-events/arch $(OUTPUT)pmu-events/pmu-events.c $(V)
tools/perf/pmu-events/jevents.c new file +767 -0

767 lines not on this page · raw message

tools/perf/pmu-events/jevents.h new file +17 -0
new file mode 100644
index 0000000..996601f
@@ -0,0 +1,17 @@
+#ifndef JEVENTS_H
+#define JEVENTS_H 1
+
+int json_events(const char *fn,
+		int (*func)(void *data, char *name, char *event, char *desc),
+		void *data);
+char *get_cpu_str(void);
+
+#ifndef min
+#define min(x, y) ({                            \
+	typeof(x) _min1 = (x);                  \
+	typeof(y) _min2 = (y);                  \
+	(void) (&_min1 == &_min2);              \
+	_min1 < _min2 ? _min1 : _min2; })
+#endif
+
+#endif
tools/perf/pmu-events/json.h +6 -0
index f2745c7..278ebd3 100644
@@ -20,6 +20,12 @@ extern int eprintf(int level, int var, const char *fmt, ...);
 #define pr_err(fmt, ...) \
 	eprintf(0, verbose, pr_fmt(fmt), ##__VA_ARGS__)
 
+#define pr_info(fmt, ...) \
+	eprintf(1, verbose, pr_fmt(fmt), ##__VA_ARGS__)
+
+#define pr_debug(fmt, ...) \
+	eprintf(2, verbose, pr_fmt(fmt), ##__VA_ARGS__)
+
 #ifndef roundup
 #define roundup(x, y) (                                \
 {                                                      \
tools/perf/pmu-events/pmu-events.h new file +36 -0
new file mode 100644
index 0000000..70d5479
@@ -0,0 +1,36 @@
+#ifndef PMU_EVENTS_H
+#define PMU_EVENTS_H
+
+/*
+ * Describe each PMU event. Each CPU has a table of PMU events.
+ */
+struct pmu_event {
+	const char *name;
+	const char *event;
+	const char *desc;
+	const char *topic;
+};
+
+/*
+ *
+ * Map a CPU to its table of PMU events. The CPU is identified by the
+ * cpuid field, which is an arch-specific identifier for the CPU.
+ * The identifier specified in tools/perf/pmu-events/arch/xxx/mapfile
+ * must match the get_cpustr() in tools/perf/arch/xxx/util/header.c)
+ *
+ * The  cpuid can contain any character other than the comma.
+ */
+struct pmu_events_map {
+	const char *cpuid;
+	const char *version;
+	const char *type;		/* core, uncore etc */
+	struct pmu_event *table;
+};
+
+/*
+ * Global table mapping each known CPU for the architecture to its
+ * table of PMU events.
+ */
+extern struct pmu_events_map pmu_events_map[];
+
+#endif