Re: [PATCH] KERN_INFO 2.4.19-pre2 IP/TCP hash table size printks

Re: [PATCH] KERN_INFO 2.4.19-pre2 IP/TCP hash table size printks

[David S. Miller]

   From: Denis Vlasenko <vda@port.imtp.ilyichevsk.odessa.ua>
   Date: Mon, 11 Mar 2002 15:58:15 -0200

   Primary purpose of this patch is to make KERN_WARNING and
   KERN_INFO log levels closer to their original meaning.
   Today they are quite far from what was intended.
   Just look what kernel writes at the WARNING level
   each time you boot your box!

Maybe it is even better idea to change default kernel printk
logging level which is used when no KERN_* is specified, eh?

[PATCH] KERN_INFO 2.4.19-pre2 IP/TCP hash table size printks

[Denis Vlasenko]

Primary purpose of this patch is to make KERN_WARNING and
KERN_INFO log levels closer to their original meaning.
Today they are quite far from what was intended.
Just look what kernel writes at the WARNING level
each time you boot your box!

This patch is for IPv4 IP/TCP hash table size printks

diff -u --recursive -x *.orig -x *.rej linux-2.4.19-pre2/net/ipv4/route.c 
linux-new/net/ipv4/route.c
--- linux-2.4.19-pre2/net/ipv4/route.c	Mon Feb 25 17:38:14 2002
+++ linux-new/net/ipv4/route.c	Mon Mar 11 12:26:37 2002
@@ -2487,9 +2487,10 @@
 	if (!rt_hash_table)
 		panic("Failed to allocate IP route cache hash table\n");

-	printk("IP: routing cache hash table of %u buckets, %ldKbytes\n",
-	       rt_hash_mask,
-	       (long) (rt_hash_mask * sizeof(struct rt_hash_bucket)) / 1024);
+	printk(KERN_INFO "IP: routing cache hash table of"
+		" %u buckets, %ldKbytes\n",
+		rt_hash_mask,
+		(long) (rt_hash_mask * sizeof(struct rt_hash_bucket)) / 1024);

 	for (rt_hash_log = 0; (1 << rt_hash_log) != rt_hash_mask; rt_hash_log++)
 		/* NOTHING */;
diff -u --recursive -x *.orig -x *.rej linux-2.4.19-pre2/net/ipv4/tcp.c 
linux-new/net/ipv4/tcp.c
--- linux-2.4.19-pre2/net/ipv4/tcp.c	Fri Dec 21 15:42:05 2001
+++ linux-new/net/ipv4/tcp.c	Mon Mar 11 12:28:10 2002
@@ -2552,8 +2552,9 @@
 		sysctl_tcp_rmem[2] = 2*43689;
 	}

-	printk("TCP: Hash tables configured (established %d bind %d)\n",
-	       tcp_ehash_size<<1, tcp_bhash_size);
+	printk(KERN_INFO "TCP: Hash tables configured"
+		" (established %d bind %d)\n",
+	        tcp_ehash_size<<1, tcp_bhash_size);

 	tcpdiag_init();
 }

Re: [PATCH] KERN_INFO 2.4.19-pre2 IP/TCP hash table size printks

[Denis Vlasenko]

On 11 March 2002 12:03, David S. Miller wrote:
>    From: Denis Vlasenko <vda@port.imtp.ilyichevsk.odessa.ua>
>    Date: Mon, 11 Mar 2002 15:58:15 -0200
>
>    Primary purpose of this patch is to make KERN_WARNING and
>    KERN_INFO log levels closer to their original meaning.
>    Today they are quite far from what was intended.
>    Just look what kernel writes at the WARNING level
>    each time you boot your box!
>
> Maybe it is even better idea to change default kernel printk
> logging level which is used when no KERN_* is specified, eh?

Not "eh", "yeah!" instead :-) I like this idea.
But how to convince everybody else?
--
vda

Re: [PATCH] KERN_INFO 2.4.19-pre2 IP/TCP hash table size printks

[Robert Love]

On Mon, 2002-03-11 at 09:03, David S. Miller wrote:

> Maybe it is even better idea to change default kernel printk
> logging level which is used when no KERN_* is specified, eh?

Only if we really consider the printks in question "whatever the default
is."  The default has changed before and may change again (as you are
suggesting).  I have no problem changing the default - I just think if
Denis suggests a printk should be classified as "informative" and we
agree, let's change it.  Although I don't think it would fly with Linus,
any printk we can solidly classify we should, imo.

This doesn't preclude changing the default, as you suggest, especially
if its clearly too high.  But I don't see why we shouldn't be explicit
where possible.  No?

	Robert Love