Skip to contentSkip to conversation

perf-record: auto detect when stdout is a pipe

PATCH Applied Franck Bui-Huu 3 participants 4 messages Jan 16, 2011 – Jan 26, 2011 perf events

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 perf-record: auto detect when stdout is a pipe

From: Franck Bui-Huu <[email protected]>

This patch gives the ability to perf-record to detect when its stdout has been redirected to a pipe. There's now no more need to add '-o -' switch in this case.

However '-o <path>' option has always precedence, that is if specified and stdout has been connected via a pipe then the output will go into the specified output.

---
 tools/perf/builtin-record.c |   32 ++++++++++++++++++++------------
 1 files changed, 20 insertions(+), 12 deletions(-)
tools/perf/builtin-record.c +20 -12
index 7069bd3..4f86354 100644
@@ -46,7 +46,7 @@ static unsigned int		user_freq 			= UINT_MAX;
 static int			freq				=   1000;
 static int			output;
 static int			pipe_output			=      0;
-static const char		*output_name			= "perf.data";
+static const char		*output_name			= NULL;
 static int			group				=      0;
 static int			realtime_prio			=      0;
 static bool			raw_samples			=  false;
@@ -562,18 +562,26 @@ static int __cmd_record(int argc, const char **argv)
 		exit(-1);
 	}
 
-	if (!strcmp(output_name, "-"))
-		pipe_output = 1;
-	else if (!stat(output_name, &st) && st.st_size) {
-		if (write_mode == WRITE_FORCE) {
-			char oldname[PATH_MAX];
-			snprintf(oldname, sizeof(oldname), "%s.old",
-				 output_name);
-			unlink(oldname);
-			rename(output_name, oldname);
+	if (!output_name) {
+		if (!fstat(STDOUT_FILENO, &st) && S_ISFIFO(st.st_mode))
+			pipe_output = 1;
+		else
+			output_name = "perf.data";
+	}
+	if (output_name) {
+		if (!strcmp(output_name, "-"))
+			pipe_output = 1;
+		else if (!stat(output_name, &st) && st.st_size) {
+			if (write_mode == WRITE_FORCE) {
+				char oldname[PATH_MAX];
+				snprintf(oldname, sizeof(oldname), "%s.old",
+					 output_name);
+				unlink(oldname);
+				rename(output_name, oldname);
+			}
+		} else if (write_mode == WRITE_APPEND) {
+			write_mode = WRITE_FORCE;
 		}
-	} else if (write_mode == WRITE_APPEND) {
-		write_mode = WRITE_FORCE;
 	}
 
 	flags = O_CREAT|O_RDWR;
Signature · 1 line
1.7.3.2
#1 lore raw

Em Sun, Jan 16, 2011 at 05:14:45PM +0100, Franck Bui-Huu escreveu:

Quoted 10 lines

From: Franck Bui-Huu <[email protected]>

This patch gives the ability to perf-record to detect when its stdout has been redirected to a pipe. There's now no more need to add '-o -' switch in this case.

However '-o <path>' option has always precedence, that is if specified and stdout has been connected via a pipe then the output will go into the specified output.

I guess we could have that in perf report as well? I.e. if stdin is a FIFO...

Thanks, applied,

- Arnaldo

#2 lore raw
Quoted 7 lines · Arnaldo Carvalho de Melo

Em Sun, Jan 16, 2011 at 05:14:45PM +0100, Franck Bui-Huu escreveu:

Quoted 6 lines

From: Franck Bui-Huu <[email protected]>

This patch gives the ability to perf-record to detect when its stdout has been redirected to a pipe. There's now no more need to add '-o -' switch in this case.

Quoted 8 lines
Quoted 4 lines

However '-o <path>' option has always precedence, that is if specified and stdout has been connected via a pipe then the output will go into the specified output.

I guess we could have that in perf report as well? I.e. if stdin is a FIFO...

Yes, I noticed that too, perf-script is also a candidate.

I'll do.

Franck

#3 lore raw

[tip:perf/core] perf record: auto detect when stdout is a pipe

In reply to Franck Bui-Huu
Commit-ID:  d7065adb9b4f3384c2615f0a3dbdb6c3aae1eb18
Gitweb:     http://git.kernel.org/tip/d7065adb9b4f3384c2615f0a3dbdb6c3aae1eb18
Author:     Franck Bui-Huu <[email protected]>
AuthorDate: Sun, 16 Jan 2011 17:14:45 +0100
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Mon, 24 Jan 2011 10:58:27 -0200

perf record: auto detect when stdout is a pipe

This patch gives the ability to 'perf record' to detect when its stdout has been redirected to a pipe. There's now no more need to add '-o -' switch in this case.

However '-o <path>' option has always precedence, that is if specified and stdout has been connected via a pipe then the output will go into the specified output.

LKML-Reference: <[email protected]>

---
 tools/perf/builtin-record.c |   32 ++++++++++++++++++++------------
 1 files changed, 20 insertions(+), 12 deletions(-)
tools/perf/builtin-record.c +20 -12
index 45a3689..1346d42 100644
@@ -48,7 +48,7 @@ static unsigned int		user_freq 			= UINT_MAX;
 static int			freq				=   1000;
 static int			output;
 static int			pipe_output			=      0;
-static const char		*output_name			= "perf.data";
+static const char		*output_name			= NULL;
 static int			group				=      0;
 static int			realtime_prio			=      0;
 static bool			nodelay				=  false;
@@ -497,18 +497,26 @@ static int __cmd_record(int argc, const char **argv)
 		exit(-1);
 	}
 
-	if (!strcmp(output_name, "-"))
-		pipe_output = 1;
-	else if (!stat(output_name, &st) && st.st_size) {
-		if (write_mode == WRITE_FORCE) {
-			char oldname[PATH_MAX];
-			snprintf(oldname, sizeof(oldname), "%s.old",
-				 output_name);
-			unlink(oldname);
-			rename(output_name, oldname);
+	if (!output_name) {
+		if (!fstat(STDOUT_FILENO, &st) && S_ISFIFO(st.st_mode))
+			pipe_output = 1;
+		else
+			output_name = "perf.data";
+	}
+	if (output_name) {
+		if (!strcmp(output_name, "-"))
+			pipe_output = 1;
+		else if (!stat(output_name, &st) && st.st_size) {
+			if (write_mode == WRITE_FORCE) {
+				char oldname[PATH_MAX];
+				snprintf(oldname, sizeof(oldname), "%s.old",
+					 output_name);
+				unlink(oldname);
+				rename(output_name, oldname);
+			}
+		} else if (write_mode == WRITE_APPEND) {
+			write_mode = WRITE_FORCE;
 		}
-	} else if (write_mode == WRITE_APPEND) {
-		write_mode = WRITE_FORCE;
 	}
 
 	flags = O_CREAT|O_RDWR;