[Xenomai-help] Re-2: rt_printf with daemonized task

[Xenomai-help] Re-2: rt_printf with daemonized task

[oliver.schlenker]

After tracking down the issue a little bit further, it seems not to be a problem of daemonizing a server task,
but to change a line of code in rtprint.c

What I would like to achieve :
- My server process is setting up two Xenomai threads
- Both Xenomai should report some messages with rt_printf()
- rt_printf-function should write not to stdout or stderr but to the syslog

Therefor I changed the code for output the message in rt_print.c from
   fprintf(head->dest, "%s", head->text); 
to
   syslog( LOG_ERR | LOG_LOCAL4, head->text ); 

Problem :
The syslog() call is not doing any output the syslog file, it doesn't matter
if the process is daemonized or not.




Oliver

To: gilles.chanteperdrix@xenomai.org
Cc: xenomai@xenomai.org

Re: [Xenomai-help] Re-2: rt_printf with daemonized task

[Gilles Chanteperdrix]

oliver.schlenker@domain.hid wrote:
> The syslog() call is not doing any output the syslog file, it doesn't matter
> if the process is daemonized or not.

The syslog() syscall is handled by the syslogd daemon. So, you should
check if the syslogd daemon is running and if it is correctly configured.

-- 
                                          Gilles

Re: [Xenomai-help] Re-2: rt_printf with daemonized task

[Andreas Glatz]

Hi,

On Wed, 2009-10-07 at 13:13 +0000, oliver.schlenker@domain.hid
wrote:
> 
> 
> 
> After tracking down the issue a little bit further, it seems not to be a problem of daemonizing a server task,
> but to change a line of code in rtprint.c
> 
> What I would like to achieve :
> - My server process is setting up two Xenomai threads
> - Both Xenomai should report some messages with rt_printf()
> - rt_printf-function should write not to stdout or stderr but to the syslog
> 

FYI: That's how we tackled this problem:

We created a RT_PIPE object. Both threads are writing their messages to
the pipe. Together with the Xenomai application we have 'logger' 
running: 'logger -f /dev/rtpXX -t <your process name> &' which takes 
everything which gets written to RT_PIPE and sends it to the syslog
daemon. 

Andreas

[Xenomai-help] Re-2: :: rt_printf with daemonized task

[oliver.schlenker]

-------- Original Message --------
Subject: Re: :: rt_printf with daemonized task (19-Okt-2009 16:00)
From:    Jan Kiszka <jan.kiszka@siemens.com>
To:      oliver.schlenker@smmotioncontrol.de

> If you move the definition of print_to_buffer right below this prototype
> block, you would avoid having to add this prototype and you would have
> likely generated a more reviewable patch. The large code movement below
> is very hard to check regarding correctness as I do not see the actually
> modified lines.

Ok, moved back definition of print_to_buffer to avoid large code movement

> 
> >  static void cleanup_buffer(struct print_buffer *buffer);
> >  static void print_buffers(void);
> > +static void forked_child_init(void);
> 
> Stray prototype.

deleted

> >  
> > +void rt_syslog(int priority, char *format, ...)
> > +{
> > +	va_list args;
> > +
> > +	va_start(args, format);
> > +	print_to_buffer(RT_PRINT_SYSLOG_STREAM, priority, format, args);
> > +	va_end(args);
> > +
> > +	return;
> 
> Unneeded return.

Deleted

> 
> > +}
> > +
> > +void rt_vsyslog(int priority, char *format, va_list args )
> > +{
> > +	print_to_buffer(RT_PRINT_SYSLOG_STREAM, priority, format, args);
> > +
> > +	return;
> 
> Here too.

Also deleted

> 
> > +}
> > +
> >  static void set_buffer_name(struct print_buffer *buffer, const char *name)
> >  {
> >  	int n;
> >                      
> >  
> >  /* *** Deferred Output Management *** */
> >  
> > +static int print_to_buffer(FILE *stream, int priority,
> > +							const char *format,va_list args)
> 
> When moving this function, please also indent this properly and take
> care for a space after the ','.

should be correct now.

> 
> > 
> 
> Jan
> 
> -- 
> Siemens AG, Corporate Technology, CT SE 2
> Corporate Competence Center Embedded Linux


when patches are acked by you, should I do something with the patches
like sending to a special mailing list or is it ok to have them posted in
this list ?


Oliver



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

Subject : Syslog extension of rt_print

Extension of rt_print library with rt_syslog() and rt_vsyslog() to print messages
rt-safe directly to syslog

Signed-off-by: Oliver Schlenker <oliver.schlenker <at> sm motion control.de>
---


--- rtdk.h.original	2007-12-09 11:46:36.000000000 +0100
+++ rtdk.h	2009-10-09 16:30:57.759836300 +0200
                 
 int rt_vprintf(const char *format, va_list args);
 int rt_fprintf(FILE *stream, const char *format, ...);
 int rt_printf(const char *format, ...);
+void rt_syslog(int priority, char *format, ...);
+void rt_vsyslog(int priority, char *format, va_list args);
+
 
 int rt_print_init(size_t buffer_size, const char *name);
 void rt_print_cleanup(void); 


--- rt_print.c.forkonly	2009-10-16 16:19:25.949289800 +0200
+++ rt_print.c.wsyslog	2009-10-21 13:46:35.580985700 +0200
                 
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
+#include <syslog.h>
 
 #include <rtdk.h>
 #include <asm/xenomai/system.h>
                  
 
 #define RT_PRINT_LINE_BREAK		256
 
+#define RT_PRINT_SYSLOG_STREAM 	NULL
+
 struct entry_head {
 	FILE *dest;
 	uint32_t seq_no;
+	int priority;
 	char text[1];
 } __attribute__((packed));
 
                 
 
 /* *** rt_print API *** */
 
-int rt_vfprintf(FILE *stream, const char *format, va_list args)
+static int print_to_buffer(FILE *stream, int priority,
+				const char *format, va_list args)
 {
 	struct print_buffer *buffer = pthread_getspecific(__buffer_key);
 	off_t write_pos, read_pos;
                   
 			/* Write out empty entry */
 			head = buffer->ring + write_pos;
 			head->seq_no = __seq_no;
+			head->priority = 0;
 			head->text[0] = 0;
 
 			/* Forward to the ring buffer start */
                   
 	/* If we were able to write some text, finalise the entry */
 	if (len > 0) {
 		head->seq_no = ++__seq_no;
+		head->priority = priority;
 		head->dest = stream;
 
 		/* Move forward by text and head length */
                   
 		/* An empty entry marks the wrap-around */
 		head = buffer->ring + write_pos;
 		head->seq_no = __seq_no;
+		head->priority = priority;
 		head->text[0] = 0;
 
 		write_pos = 0;
                    
 	return res;
 }
 
+int rt_vfprintf(FILE *stream, const char *format, va_list args)
+{
+	return print_to_buffer(stream, 0, format, args);
+}
+
 int rt_vprintf(const char *format, va_list args)
 {
 	return rt_vfprintf(stdout, format, args);
                    
 	return n;
 }
 
+void rt_syslog(int priority, char *format, ...)
+{
+	va_list args;
+
+	va_start(args, format);
+	print_to_buffer(RT_PRINT_SYSLOG_STREAM, priority, format, args);
+	va_end(args);
+}
+
+void rt_vsyslog(int priority, char *format, va_list args)
+{
+	print_to_buffer(RT_PRINT_SYSLOG_STREAM, priority, format, args);
+}
+
 static void set_buffer_name(struct print_buffer *buffer, const char *name)
 {
 	int n;
                   
 
 /* *** Deferred Output Management *** */
 
+
 static void cleanup_buffer(struct print_buffer *buffer)
 {
 	struct print_buffer *prev, *next;
                    
 
 		if (len) {
 			/* Print out non-empty entry and proceed */
-			fprintf(head->dest, "%s", head->text);
+			/* Check if output goes to syslog */
+			if (head->dest == RT_PRINT_SYSLOG_STREAM) {
+			   	syslog( head->priority, "%s", head->text );
+			} else {
+				/* Output goes to specified stream */
+				fprintf(head->dest, "%s", head->text);
+			}
+
 			read_pos += sizeof(*head) + len;
 		} else {
 			/* Emptry entries mark the wrap-around */ 



To: jan.kiszka@siemens.com
Cc: gilles.chanteperdrix@xenomai.org
    xenomai-help@gna.org

Re: [Xenomai-help] Re-2: :: rt_printf with daemonized task

[Jan Kiszka]

oliver.schlenker@domain.hid wrote:
> -------- Original Message --------
> Subject: Re: :: rt_printf with daemonized task (19-Okt-2009 16:00)
> From:    Jan Kiszka <jan.kiszka@domain.hid>
> To:      oliver.schlenker@domain.hid
> 
>> If you move the definition of print_to_buffer right below this prototype
>> block, you would avoid having to add this prototype and you would have
>> likely generated a more reviewable patch. The large code movement below
>> is very hard to check regarding correctness as I do not see the actually
>> modified lines.
> 
> Ok, moved back definition of print_to_buffer to avoid large code movement
> 
>>>  static void cleanup_buffer(struct print_buffer *buffer);
>>>  static void print_buffers(void);
>>> +static void forked_child_init(void);
>> Stray prototype.
> 
> deleted
> 
>>>  
>>> +void rt_syslog(int priority, char *format, ...)
>>> +{
>>> +	va_list args;
>>> +
>>> +	va_start(args, format);
>>> +	print_to_buffer(RT_PRINT_SYSLOG_STREAM, priority, format, args);
>>> +	va_end(args);
>>> +
>>> +	return;
>> Unneeded return.
> 
> Deleted
> 
>>> +}
>>> +
>>> +void rt_vsyslog(int priority, char *format, va_list args )
>>> +{
>>> +	print_to_buffer(RT_PRINT_SYSLOG_STREAM, priority, format, args);
>>> +
>>> +	return;
>> Here too.
> 
> Also deleted
> 
>>> +}
>>> +
>>>  static void set_buffer_name(struct print_buffer *buffer, const char *name)
>>>  {
>>>  	int n;
>>>                      
>>>  
>>>  /* *** Deferred Output Management *** */
>>>  
>>> +static int print_to_buffer(FILE *stream, int priority,
>>> +							const char *format,va_list args)
>> When moving this function, please also indent this properly and take
>> care for a space after the ','.
> 
> should be correct now.
> 
>> Jan
>>
>> -- 
>> Siemens AG, Corporate Technology, CT SE 2
>> Corporate Competence Center Embedded Linux
> 
> 
> when patches are acked by you, should I do something with the patches
> like sending to a special mailing list or is it ok to have them posted in
> this list ?

I was about to apply them to my tree and send out a git pull request to
Philippe. Unfortunately I missed so far that your mail client or gateway
damaged the patch format seriously (the '@@' markers are all missing).
So if you have a clean "diff -up" of your changes, please sent them in a
tarball to me or via a proper mail client / account that keeps emails
intact. Otherwise I would have to reconstruct all changes by hand...

Jan

-- 
Siemens AG, Corporate Technology, CT SE 2
Corporate Competence Center Embedded Linux

Re: [Xenomai-help] Re-2: :: rt_printf with daemonized task

[oliver.schlenker]

[-- Attachment #1: Type: text/plain, Size: 1289 bytes --]






-------- Original Message --------
Subject: Re: Re-2: :: rt_printf with daemonized task (22-Okt-2009 18:34)
From:    Jan Kiszka <jan.kiszka@siemens.com>
To:      oliver.schlenker@smmotioncontrol.de

> 
> I was about to apply them to my tree and send out a git pull request to
> Philippe. Unfortunately I missed so far that your mail client or gateway
> damaged the patch format seriously (the '> So if you have a clean "diff -up" of your changes, please sent them in a
> tarball to me or via a proper mail client / account that keeps emails
> intact. Otherwise I would have to reconstruct all changes by hand...
> 
> Jan
> 
> -- 
> Siemens AG, Corporate Technology, CT SE 2
> Corporate Competence Center Embedded Linux

I wasn't aware of such a behaviour of my mail client, but it's indeed
like that. All 'commands and filtered out of the mail. I could'nt find a possibility to switch off
this features ...
Therefor I did attach to this mail a tarball with all changes. These patches
have been created with "diff -up". Former ones have been created with just
"diff -u".

Hopefully it's possible now to integrate it without any further problems.


Oliver

To: jan.kiszka@siemens.com
Cc: gilles.chanteperdrix@xenomai.org
    xenomai-help@gna.org


[-- Attachment #2: rt_print.patch.tgz --]
[-- Type: application/octet-stream, Size: 1852 bytes --]