Skip to contentSkip to conversation

perf list: fix --raw-dump

PATCH In review Ramkumar Ramachandra 4 participants 8 messages Dec 11, 2013 – Dec 30, 2013 perf events

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 perf list: fix --raw-dump

While adding usage information, 44d742e (perf list: Add usage, 2013-10-30) broke

  $ perf list --raw-dump

by asking parse_options() to stop at
non-option (PARSE_OPT_STOP_AT_NON_OPTION), hence parsing the option --raw-dump and barfing. Ask it to keep unknown options unparsed (PARSE_OPT_KEEP_UNKNOWN) so that we have a chance to handle it later. The trade-off that this patch makes is that

  $ perf list --foo

will not error out like it used to.

Fixes 44d742e01e6d (perf list: Add usage)
Cc David Ahern <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>

---
I spotted this while trying to complete

   $ perf record -e <TAB>
 tools/perf/builtin-list.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/builtin-list.c +1 -1
index 011195e..8b076f3 100644
@@ -28,7 +28,7 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 	};
 
 	argc = parse_options(argc, argv, list_options, list_usage,
-			     PARSE_OPT_STOP_AT_NON_OPTION);
+			PARSE_OPT_KEEP_UNKNOWN);
 
 	setup_pager();
 
Signature · 1 line
1.8.5.rc0.5.g70ebc73.dirty
#1 lore raw
Quoted 11 lines · Ramkumar Ramachandra

While adding usage information, 44d742e (perf list: Add usage, 2013-10-30) broke

   $ perf list --raw-dump

by asking parse_options() to stop at
non-option (PARSE_OPT_STOP_AT_NON_OPTION), hence parsing the option --raw-dump and barfing. Ask it to keep unknown options unparsed (PARSE_OPT_KEEP_UNKNOWN) so that we have a chance to handle it later. The trade-off that this patch makes is that

Quoted 3 lines
   $ perf list --foo

will not error out like it used to.

Why not make raw_dump a proper argument?

Quoted 1 line

From b4444edaeffba7ebc459ee1b2cbe4f9cbd057476 Mon Sep 17 00:00:00 2001

From: David Ahern <[email protected]>
Date: Wed, 11 Dec 2013 14:00:20 -0700
Subject: [PATCH] perf list: Fix raw-dump arg

Ramkumar reported that perf list --raw-dump was broken by 44d742e. Fix by making raw-dump a proper argument.

Cc Ingo Molnar <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Ramkumar Ramachandra <[email protected]>
---
 tools/perf/builtin-list.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
tools/perf/builtin-list.c +6 -2
index 011195e38f21..b553d0c4ca82 100644
@@ -19,7 +19,9 @@
 int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	int i;
+	bool raw_dump = false;
 	const struct option list_options[] = {
+		OPT_BOOLEAN(0, "raw-dump", &raw_dump, "raw dump for completion"),
 		OPT_END()
 	};
 	const char * const list_usage[] = {
@@ -36,6 +38,10 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 		print_events(NULL, false);
 		return 0;
 	}
+	if (raw_dump) {
+		print_events(NULL, true);
+		return 0;
+	}
 
 	for (i = 0; i < argc; ++i) {
 		if (i)
@@ -53,8 +59,6 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			print_hwcache_events(NULL, false);
 		else if (strcmp(argv[i], "pmu") == 0)
 			print_pmu_events(NULL, false);
-		else if (strcmp(argv[i], "--raw-dump") == 0)
-			print_events(NULL, true);
 		else {
 			char *sep = strchr(argv[i], ':'), *s;
 			int sep_idx;
Signature · 1 line
1.8.3.4 (Apple Git-47)
#2 lore raw
Quoted 1 line · David Ahern

Why not make raw_dump a proper argument?

Sure, that'd work too. I was thinking of a minimal way to fix the problem myself.

tools/perf/builtin-list.c int cmd_list(int argc, const char · line 44

Quoted 12 lines
tools/perf/builtin-list.c +4 -0
index 011195e38f21..b553d0c4ca82 100644
@@ -36,6 +38,10 @@ int cmd_list(int argc, const char
             print_events(NULL, false);
             return 0;
       }
+      if (raw_dump) {
+            print_events(NULL, true);
+            return 0;
+      }

This won't work because you've put it right below the if (argc == 0), which executes print_events(). You could move it up and get it to work.

From 7198a494cfef43395e8683ac3a0576277b8d1d80 Mon Sep 17 00:00:00 2001
From: David Ahern <[email protected]>
Date: Wed, 11 Dec 2013 14:00:20 -0700
Subject: [PATCH] perf list: Fix raw-dump arg

Ramkumar reported that perf list --raw-dump was broken by 44d742e. Fix by making raw-dump a proper argument.

Cc Ingo Molnar <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Ramkumar Ramachandra <[email protected]>
---
 tools/perf/builtin-list.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
tools/perf/builtin-list.c +6 -2
index 011195e..2629c24 100644
@@ -19,7 +19,9 @@
 int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	int i;
+	bool raw_dump = false;
 	const struct option list_options[] = {
+		OPT_BOOLEAN(0, "raw-dump", &raw_dump, "raw dump for completion"),
 		OPT_END()
 	};
 	const char * const list_usage[] = {
@@ -32,6 +34,10 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 
 	setup_pager();
 
+	if (raw_dump) {
+		print_events(NULL, true);
+		return 0;
+	}
 	if (argc == 0) {
 		print_events(NULL, false);
 		return 0;
@@ -53,8 +59,6 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			print_hwcache_events(NULL, false);
 		else if (strcmp(argv[i], "pmu") == 0)
 			print_pmu_events(NULL, false);
-		else if (strcmp(argv[i], "--raw-dump") == 0)
-			print_events(NULL, true);
 		else {
 			char *sep = strchr(argv[i], ':'), *s;
 			int sep_idx;
Signature · 1 line
1.8.5.1.113.g8cb5bef.dirty
#3 lore raw
Quoted 6 lines · Ramkumar Ramachandra
Quoted 1 line · David Ahern

Why not make raw_dump a proper argument?

Sure, that'd work too. I was thinking of a minimal way to fix the problem myself.

Quoted 17 lines

tools/perf/builtin-list.c int cmd_list(int argc, const char · line 44

Quoted 12 lines
tools/perf/builtin-list.c +4 -0
index 011195e38f21..b553d0c4ca82 100644
@@ -36,6 +38,10 @@ int cmd_list(int argc, const char
              print_events(NULL, false);
              return 0;
        }
+      if (raw_dump) {
+            print_events(NULL, true);
+            return 0;
+      }

This won't work because you've put it right below the if (argc == 0), which executes print_events(). You could move it up and get it to work.

your updated patch looks good to me.

David

#4 lore raw
In reply to Ramkumar Ramachandra
Quoted 4 lines · Ramkumar Ramachandra
Quoted 1 line · David Ahern

Why not make raw_dump a proper argument?

Sure, that'd work too. I was thinking of a minimal way to fix the problem myself.

Hi Ramkumar and David,
If this argument is only used for perf complement, how about make
it hidden to user? Please refer https://lkml.org/lkml/2013/12/12/52 and https://lkml.org/lkml/2013/12/12/53 The two patches make --raw-dump used internally but hidden in output of '-h'.

If you think we should expose it to user, I think we should make it
work as an option rather than an argument, and make the output of it more readable, how about the patch attached in this mail?

Example:

    # ./perf list cache --raw-dump
    L1-dcache-loads
    L1-dcache-load-misses
    L1-dcache-stores
    L1-dcache-store-misses
    L1-dcache-prefetch-misses
    L1-icache-load-misses
    LLC-loads
    LLC-stores
    LLC-prefetches
    dTLB-loads
    dTLB-load-misses
    dTLB-stores
    dTLB-store-misses
    iTLB-loads
    iTLB-load-misses
    branch-loads
    branch-load-misses
Quoted 1 line

From 7bd20eb147783e0c1e659b3bceeebb75aad8436c Mon Sep 17 00:00:00 2001

From: David Ahern <[email protected]>
Date: Wed, 11 Dec 2013 14:00:20 -0700
Subject: [PATCH] perf list: Fix raw-dump arg

Ramkumar reported that perf list --raw-dump was broken by 44d742e. Fix by making raw-dump a proper option.

Cc Ingo Molnar <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Ramkumar Ramachandra <[email protected]>
---
 tools/perf/builtin-list.c      | 23 +++++++++++------------
 tools/perf/util/parse-events.c | 19 +++++++++++--------
 tools/perf/util/parse-events.h |  2 +-
 tools/perf/util/pmu.c          |  2 +-
 4 files changed, 24 insertions(+), 22 deletions(-)
tools/perf/builtin-list.c +11 -12
index 011195e..9cf12f2 100644
@@ -19,7 +19,9 @@
 int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	int i;
+	bool raw_dump = false;
 	const struct option list_options[] = {
+		OPT_BOOLEAN(0, "raw-dump", &raw_dump, "raw dump for completion"),
 		OPT_END()
 	};
 	const char * const list_usage[] = {
@@ -27,13 +29,12 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 		NULL
 	};
 
-	argc = parse_options(argc, argv, list_options, list_usage,
-			     PARSE_OPT_STOP_AT_NON_OPTION);
+	argc = parse_options(argc, argv, list_options, list_usage, 0);
 
 	setup_pager();
 
 	if (argc == 0) {
-		print_events(NULL, false);
+		print_events(NULL, raw_dump);
 		return 0;
 	}
 
@@ -41,26 +42,24 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 		if (i)
 			putchar('\n');
 		if (strncmp(argv[i], "tracepoint", 10) == 0)
-			print_tracepoint_events(NULL, NULL, false);
+			print_tracepoint_events(NULL, NULL, raw_dump);
 		else if (strcmp(argv[i], "hw") == 0 ||
 			 strcmp(argv[i], "hardware") == 0)
-			print_events_type(PERF_TYPE_HARDWARE);
+			print_events_type(PERF_TYPE_HARDWARE, raw_dump);
 		else if (strcmp(argv[i], "sw") == 0 ||
 			 strcmp(argv[i], "software") == 0)
-			print_events_type(PERF_TYPE_SOFTWARE);
+			print_events_type(PERF_TYPE_SOFTWARE, raw_dump);
 		else if (strcmp(argv[i], "cache") == 0 ||
 			 strcmp(argv[i], "hwcache") == 0)
-			print_hwcache_events(NULL, false);
+			print_hwcache_events(NULL, raw_dump);
 		else if (strcmp(argv[i], "pmu") == 0)
-			print_pmu_events(NULL, false);
-		else if (strcmp(argv[i], "--raw-dump") == 0)
-			print_events(NULL, true);
+			print_pmu_events(NULL, raw_dump);
 		else {
 			char *sep = strchr(argv[i], ':'), *s;
 			int sep_idx;
 
 			if (sep == NULL) {
-				print_events(argv[i], false);
+				print_events(argv[i], raw_dump);
 				continue;
 			}
 			sep_idx = sep - argv[i];
@@ -69,7 +68,7 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 				return -1;
 
 			s[sep_idx] = '\0';
-			print_tracepoint_events(s, s + sep_idx + 1, false);
+			print_tracepoint_events(s, s + sep_idx + 1, raw_dump);
 			free(s);
 		}
 	}
tools/perf/util/parse-events.c +11 -8
index 969cb8f..ded4a4e 100644
@@ -1034,7 +1034,7 @@ void print_tracepoint_events(const char *subsys_glob, const char *event_glob,
 				continue;
 
 			if (name_only) {
-				printf("%s:%s ", sys_dirent.d_name, evt_dirent.d_name);
+				printf("%s:%s\n", sys_dirent.d_name, evt_dirent.d_name);
 				continue;
 			}
 
@@ -1117,7 +1117,7 @@ static bool is_event_supported(u8 type, unsigned config)
 }
 
 static void __print_events_type(u8 type, struct event_symbol *syms,
-				unsigned max)
+				unsigned max, bool name_only)
 {
 	char name[64];
 	unsigned i;
@@ -1132,16 +1132,19 @@ static void __print_events_type(u8 type, struct event_symbol *syms,
 		else
 			snprintf(name, sizeof(name), "%s", syms->symbol);
 
-		printf("  %-50s [%s]\n", name, event_type_descriptors[type]);
+		if (name_only)
+            printf("  %-50s\n", name);
+        else
+            printf("  %-50s [%s]\n", name, event_type_descriptors[type]);
 	}
 }
 
-void print_events_type(u8 type)
+void print_events_type(u8 type, bool name_only)
 {
 	if (type == PERF_TYPE_SOFTWARE)
-		__print_events_type(type, event_symbols_sw, PERF_COUNT_SW_MAX);
+		__print_events_type(type, event_symbols_sw, PERF_COUNT_SW_MAX, name_only);
 	else
-		__print_events_type(type, event_symbols_hw, PERF_COUNT_HW_MAX);
+		__print_events_type(type, event_symbols_hw, PERF_COUNT_HW_MAX, name_only);
 }
 
 int print_hwcache_events(const char *event_glob, bool name_only)
@@ -1166,7 +1169,7 @@ int print_hwcache_events(const char *event_glob, bool name_only)
 					continue;
 
 				if (name_only)
-					printf("%s ", name);
+					printf("%s\n", name);
 				else
 					printf("  %-50s [%s]\n", name,
 					       event_type_descriptors[PERF_TYPE_HW_CACHE]);
@@ -1198,7 +1201,7 @@ static void print_symbol_events(const char *event_glob, unsigned type,
 			continue;
 
 		if (name_only) {
-			printf("%s ", syms->symbol);
+			printf("%s\n", syms->symbol);
 			continue;
 		}
 
tools/perf/util/parse-events.h +1 -1
index f1cb4c4..8f991e2 100644
@@ -102,7 +102,7 @@ void parse_events_update_lists(struct list_head *list_event,
 void parse_events_error(void *data, void *scanner, char const *msg);
 
 void print_events(const char *event_glob, bool name_only);
-void print_events_type(u8 type);
+void print_events_type(u8 type, bool name_only);
 void print_tracepoint_events(const char *subsys_glob, const char *event_glob,
 			     bool name_only);
 int print_hwcache_events(const char *event_glob, bool name_only);
tools/perf/util/pmu.c +1 -1
index 56fc10a..1c57356 100644
@@ -751,7 +751,7 @@ void print_pmu_events(const char *event_glob, bool name_only)
 	qsort(aliases, len, sizeof(char *), cmp_string);
 	for (j = 0; j < len; j++) {
 		if (name_only) {
-			printf("%s ", aliases[j]);
+			printf("%s\n", aliases[j]);
 			continue;
 		}
 		printf("  %-50s [Kernel PMU event]\n", aliases[j]);
Signature · 1 line
1.8.2.1
#5 lore raw
In reply to Ramkumar Ramachandra

Em Thu, Dec 12, 2013 at 01:04:04PM +0530, Ramkumar Ramachandra escreveu:

Quoted 5 lines
Quoted 1 line · David Ahern

Why not make raw_dump a proper argument?

Sure, that'd work too. I was thinking of a minimal way to fix the problem myself.

So is the patch below the one to apply? When these things happen, it helps if you resubmit with a v2 in the subject line and then an explicit Ack is provided, be it from, in this case, David, or Ramkumar could just add it since David provided it in free form, ok?

- Arnaldo

Quoted 71 lines

tools/perf/builtin-list.c int cmd_list(int argc, const char · line 44

Quoted 12 lines
tools/perf/builtin-list.c +4 -0
index 011195e38f21..b553d0c4ca82 100644
@@ -36,6 +38,10 @@ int cmd_list(int argc, const char
             print_events(NULL, false);
             return 0;
       }
+      if (raw_dump) {
+            print_events(NULL, true);
+            return 0;
+      }

This won't work because you've put it right below the if (argc == 0), which executes print_events(). You could move it up and get it to work.

From 7198a494cfef43395e8683ac3a0576277b8d1d80 Mon Sep 17 00:00:00 2001
From: David Ahern <[email protected]>
Date: Wed, 11 Dec 2013 14:00:20 -0700
Subject: [PATCH] perf list: Fix raw-dump arg

Ramkumar reported that perf list --raw-dump was broken by 44d742e. Fix by making raw-dump a proper argument.

Cc Ingo Molnar <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Ramkumar Ramachandra <[email protected]>
---
 tools/perf/builtin-list.c | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)
tools/perf/builtin-list.c +6 -2
index 011195e..2629c24 100644
@@ -19,7 +19,9 @@
 int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 {
 	int i;
+	bool raw_dump = false;
 	const struct option list_options[] = {
+		OPT_BOOLEAN(0, "raw-dump", &raw_dump, "raw dump for completion"),
 		OPT_END()
 	};
 	const char * const list_usage[] = {
@@ -32,6 +34,10 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 
 	setup_pager();
 
+	if (raw_dump) {
+		print_events(NULL, true);
+		return 0;
+	}
 	if (argc == 0) {
 		print_events(NULL, false);
 		return 0;
@@ -53,8 +59,6 @@ int cmd_list(int argc, const char **argv, const char *prefix __maybe_unused)
 			print_hwcache_events(NULL, false);
 		else if (strcmp(argv[i], "pmu") == 0)
 			print_pmu_events(NULL, false);
-		else if (strcmp(argv[i], "--raw-dump") == 0)
-			print_events(NULL, true);
 		else {
 			char *sep = strchr(argv[i], ':'), *s;
 			int sep_idx;
Signature · 1 line
1.8.5.1.113.g8cb5bef.dirty
#6 lore raw
Quoted 4 lines · Arnaldo Carvalho de Melo

So is the patch below the one to apply? When these things happen, it helps if you resubmit with a v2 in the subject line and then an explicit Ack is provided, be it from, in this case, David, or Ramkumar could just add it since David provided it in free form, ok?

Is any action required on my part: should I resubmit this, or has it already been picked up?

p.s- Sorry about the terribly late response; was on vacation.

#7 lore raw
Quoted 7 lines · Ramkumar Ramachandra
Quoted 4 lines · Arnaldo Carvalho de Melo

So is the patch below the one to apply? When these things happen, it helps if you resubmit with a v2 in the subject line and then an explicit Ack is provided, be it from, in this case, David, or Ramkumar could just add it since David provided it in free form, ok?

Is any action required on my part: should I resubmit this, or has it already been picked up?

Hi Ramkumar,
I sent a patch set for enhancement of perf list recently,
[PATCH 0/8 V2] Enhancement for perf list.
<http://lkml.kernel.org/r/[email protected]>.

It contains a patch [6/8] is very similar with yours here. But I make it
works as a option for other argument, such as hw, too. And I add a signed-off-by Ramkumar and signed-off-by David in that patch.

Do you mind we turn to that thread to fix --raw-dump ??

Could you help to take a look at it??

Yang

Quoted 8 lines

p.s- Sorry about the terribly late response; was on vacation. --
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/