Obsolete messages ...

Obsolete messages ...

[Davide Libenzi]

Any CONFIG_DROP_FREAKIN_OBSOLETE_MSGS (SO_BSDCOMPAT,bdflush,...) anywhere
soon in 2.5.67 ? :)



- Davide

Re: Obsolete messages ...

[David S. Miller]

On Wed, 2003-03-26 at 18:57, Davide Libenzi wrote:
> Any CONFIG_DROP_FREAKIN_OBSOLETE_MSGS (SO_BSDCOMPAT,bdflush,...) anywhere
> soon in 2.5.67 ? :)

If you fix the apps, the messages go away.  In fact, you want to know
that you have unfixed apps on your box when you run these kernels so
I'd say the messages should stay even well into early 2.6.x

-- 
David S. Miller <davem@redhat.com>

Re: Obsolete messages ...

[Alan Cox]

On Thu, 2003-03-27 at 14:21, David S. Miller wrote:
> If you fix the apps, the messages go away.  In fact, you want to know
> that you have unfixed apps on your box when you run these kernels so
> I'd say the messages should stay even well into early 2.6.x

In which case they would benefit from net/shut_up sysctl. In lots of
environments they will just be a pain

Re: Obsolete messages ...

[David S. Miller]

   From: Alan Cox <alan@lxorguk.ukuu.org.uk>
   Date: 27 Mar 2003 14:46:11 +0000

   In which case they would benefit from net/shut_up sysctl. In lots of
   environments they will just be a pain
   
Keep in mind we have these kinds of messages in 2.4.x right this
very moment, and nobody complains about them nor asks for sysctls
to shut them off.

See net/socket.c:sock_create() for example.

Now, I'll all for netratelimit()'ing the networking ones.

Re: Obsolete messages ...

[James Morris]

On Thu, 27 Mar 2003, David S. Miller wrote:

> Now, I'll all for netratelimit()'ing the networking ones.

What about warning just once like the SOCK_PACKET one?


- James
-- 
James Morris
<jmorris@intercode.com.au>

--- linux-2.5.66.orig/net/core/sock.c	Tue Mar 25 23:09:41 2003
+++ linux-2.5.66.w2/net/core/sock.c	Fri Mar 28 02:47:31 2003
@@ -155,8 +155,13 @@
 
 static void sock_warn_obsolete_bsdism(const char *name)
 {
-	printk(KERN_WARNING "process `%s' is using obsolete "
-	       "%s SO_BSDCOMPAT\n", current->comm, name);
+	static int warned;
+	
+	if (!warned) {
+		warned = 1;
+		printk(KERN_WARNING "process `%s' is using obsolete "
+		       "%s SO_BSDCOMPAT\n", current->comm, name);
+	}
 }
 
 /*

Re: Obsolete messages ...

[David S. Miller]

   From: James Morris <jmorris@intercode.com.au>
   Date: Fri, 28 Mar 2003 02:51:18 +1100 (EST)

   On Thu, 27 Mar 2003, David S. Miller wrote:
   
   > Now, I'll all for netratelimit()'ing the networking ones.
   
   What about warning just once like the SOCK_PACKET one?
   
Ok, applied.

Re: Obsolete messages ...

[Davide Libenzi]

On Thu, 27 Mar 2003, David S. Miller wrote:

> On Wed, 2003-03-26 at 18:57, Davide Libenzi wrote:
> > Any CONFIG_DROP_FREAKIN_OBSOLETE_MSGS (SO_BSDCOMPAT,bdflush,...) anywhere
> > soon in 2.5.67 ? :)
>
> If you fix the apps, the messages go away.  In fact, you want to know
> that you have unfixed apps on your box when you run these kernels so
> I'd say the messages should stay even well into early 2.6.x

I know David, I already did ( named and /etc/inittab in my case ). My idea
was to have something like warn_obsolete(char *) to be used in all places
where necessary, and have a config option ( on by default ) that chop
messages in off case.



- Davide

Re: Obsolete messages ...

[Dave Jones]

On Thu, Mar 27, 2003 at 06:21:14AM -0800, David S. Miller wrote:

 > > Any CONFIG_DROP_FREAKIN_OBSOLETE_MSGS (SO_BSDCOMPAT,bdflush,...) anywhere
 > > soon in 2.5.67 ? :)
 > 
 > If you fix the apps, the messages go away.  In fact, you want to know
 > that you have unfixed apps on your box when you run these kernels so
 > I'd say the messages should stay even well into early 2.6.x

If folks want to mail me reports of any app (and version, even distro
info) that reports these sorts of messages, I'll add them to the doc at 
http://www.codemonkey.org.uk/post-halloween-2.5.txt

		Dave

Re: Obsolete messages ...

[Davide Libenzi]

On Thu, 27 Mar 2003, Dave Jones wrote:

> On Thu, Mar 27, 2003 at 06:21:14AM -0800, David S. Miller wrote:
>
>  > > Any CONFIG_DROP_FREAKIN_OBSOLETE_MSGS (SO_BSDCOMPAT,bdflush,...) anywhere
>  > > soon in 2.5.67 ? :)
>  >
>  > If you fix the apps, the messages go away.  In fact, you want to know
>  > that you have unfixed apps on your box when you run these kernels so
>  > I'd say the messages should stay even well into early 2.6.x
>
> If folks want to mail me reports of any app (and version, even distro
> info) that reports these sorts of messages, I'll add them to the doc at
> http://www.codemonkey.org.uk/post-halloween-2.5.txt

Well, usually /etc/inittab calls /sbin/update ( bdflush ). About
SO_BSDCOMPAT I can report Bind 9.2.2 but I think their code is right. They
do check for "#ifdef SO_BSDCOMPAT", that is still defined in asm/socket.h.
By removing SO_BSDCOMPAT from asm/socket.h and rebuilding, it should be
fine.



- Davide

Re: Obsolete messages ...

[jjs]

Davide Libenzi wrote:

>Well, usually /etc/inittab calls /sbin/update ( bdflush ). About
>SO_BSDCOMPAT I can report Bind 9.2.2 but I think their code is right. They
>do check for "#ifdef SO_BSDCOMPAT", that is still defined in asm/socket.h.
>By removing SO_BSDCOMPAT from asm/socket.h and rebuilding, it should be
>fine.
>
ACK! - commenting out the SO_BSDCOMPAT
line in asm/socket.h and rebuilding the bind rpms
cured the plague of syslog msgs here (RH 8.0)


Best Regards,

Joe