netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* Re: [PATCH] gc_min_interval in milleseconds via /proc (fwd)
@ 2005-01-27 12:13 Robert Olsson
  2005-01-31  6:56 ` David S. Miller
  0 siblings, 1 reply; 4+ messages in thread
From: Robert Olsson @ 2005-01-27 12:13 UTC (permalink / raw)
  To: David S.Miller; +Cc: Robert.Olsson, netdev


 > David S. Miller wrote:

 > It would, but since this is a user visible API we really can't
 > change it.
 
  Ok! But I haven't heard of any successful use of this paticaul API. :-)

 > We could instead make another sysctl (perhaps name it something
 > like "gc_min_interval_ms") to do what you propose, and then
 > we'd keep the existing one around for compatibility.

 All-right a new patch below. 
					--ro
  

--- net/ipv4/route.c.orig	2005-01-27 11:19:37.551668112 +0100
+++ net/ipv4/route.c	2005-01-27 12:34:14.089130976 +0100
@@ -2529,6 +2529,8 @@
 		.proc_handler	= &proc_dointvec,
 	},
 	{
+		/*  Deprecated. Use gc_min_interval_ms */
+ 
 		.ctl_name	= NET_IPV4_ROUTE_GC_MIN_INTERVAL,
 		.procname	= "gc_min_interval",
 		.data		= &ip_rt_gc_min_interval,
@@ -2538,6 +2540,15 @@
 		.strategy	= &sysctl_jiffies,
 	},
 	{
+		.ctl_name	= NET_IPV4_ROUTE_GC_MIN_INTERVAL,
+		.procname	= "gc_min_interval_ms",
+		.data		= &ip_rt_gc_min_interval,
+		.maxlen		= sizeof(unsigned long),
+		.mode		= 0644,
+		.proc_handler	= &proc_doulongvec_ms_jiffies_minmax,
+		.strategy	= &sysctl_jiffies,
+	},
+	{
 		.ctl_name	= NET_IPV4_ROUTE_GC_TIMEOUT,
 		.procname	= "gc_timeout",
 		.data		= &ip_rt_gc_timeout,
--- Documentation/filesystems/proc.txt.orig	2005-01-27 12:35:02.460777368 +0100
+++ Documentation/filesystems/proc.txt	2005-01-27 12:46:30.981106368 +0100
@@ -1709,12 +1709,13 @@
 
 Writing to this file results in a flush of the routing cache.
 
-gc_elasticity, gc_interval, gc_min_interval, gc_tresh, gc_timeout,
-gc_thresh, gc_thresh1, gc_thresh2, gc_thresh3
---------------------------------------------------------------
+gc_elasticity, gc_interval, gc_min_interval_ms, gc_timeout, gc_thresh
+---------------------------------------------------------------------
 
 Values to  control  the  frequency  and  behavior  of  the  garbage collection
-algorithm for the routing cache.
+algorithm for the routing cache. gc_min_interval is deprecated and replaced
+by gc_min_interval_ms.
+
 
 max_size
 --------

 

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] gc_min_interval in milleseconds via /proc (fwd)
  2005-01-27 12:13 [PATCH] gc_min_interval in milleseconds via /proc (fwd) Robert Olsson
@ 2005-01-31  6:56 ` David S. Miller
  2005-01-31 18:01   ` Robert Olsson
  0 siblings, 1 reply; 4+ messages in thread
From: David S. Miller @ 2005-01-31  6:56 UTC (permalink / raw)
  To: Robert Olsson; +Cc: Robert.Olsson, netdev

On Thu, 27 Jan 2005 13:13:04 +0100
Robert Olsson <Robert.Olsson@data.slu.se> wrote:

>   Ok! But I haven't heard of any successful use of this paticaul API. :-)

Are you really so sure?

>  > We could instead make another sysctl (perhaps name it something
>  > like "gc_min_interval_ms") to do what you propose, and then
>  > we'd keep the existing one around for compatibility.
> 
>  All-right a new patch below. 

We must pick a new sysctl number too, for sysctl() system call.
If the only access possible were through /proc then yes your
change would be valid, but due to sysctl() system call the
.ctl_name defines that dimension of the sysctl namespace.

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] gc_min_interval in milleseconds via /proc (fwd)
  2005-01-31  6:56 ` David S. Miller
@ 2005-01-31 18:01   ` Robert Olsson
  2005-02-02 21:09     ` David S. Miller
  0 siblings, 1 reply; 4+ messages in thread
From: Robert Olsson @ 2005-01-31 18:01 UTC (permalink / raw)
  To: David S. Miller; +Cc: Robert Olsson, netdev


David S. Miller writes:

 > We must pick a new sysctl number too, for sysctl() system call.
 > If the only access possible were through /proc then yes your
 > change would be valid, but due to sysctl() system call the
 > .ctl_name defines that dimension of the sysctl namespace.

 Ok!
 NET_IPV4_ROUTE_GC_MIN_INTERVAL_MS was added last in list to avoid
 renumbering. Feel free to change.

					--ro

--- include/linux/sysctl.h.orig	2005-01-31 18:36:11.524393024 +0100
+++ include/linux/sysctl.h	2005-01-31 18:37:41.843662408 +0100
@@ -365,6 +365,7 @@
 	NET_IPV4_ROUTE_MIN_PMTU=16,
 	NET_IPV4_ROUTE_MIN_ADVMSS=17,
 	NET_IPV4_ROUTE_SECRET_INTERVAL=18,
+	NET_IPV4_ROUTE_GC_MIN_INTERVAL_MS=19
 };
 
 enum
--- net/ipv4/route.c.orig	2005-01-27 11:19:37.000000000 +0100
+++ net/ipv4/route.c	2005-01-31 18:35:20.552141984 +0100
@@ -2529,6 +2529,8 @@
 		.proc_handler	= &proc_dointvec,
 	},
 	{
+		/*  Deprecated. Use gc_min_interval_ms */
+ 
 		.ctl_name	= NET_IPV4_ROUTE_GC_MIN_INTERVAL,
 		.procname	= "gc_min_interval",
 		.data		= &ip_rt_gc_min_interval,
@@ -2538,6 +2540,15 @@
 		.strategy	= &sysctl_jiffies,
 	},
 	{
+		.ctl_name	= NET_IPV4_ROUTE_GC_MIN_INTERVAL_MS,
+		.procname	= "gc_min_interval_ms",
+		.data		= &ip_rt_gc_min_interval,
+		.maxlen		= sizeof(unsigned long),
+		.mode		= 0644,
+		.proc_handler	= &proc_doulongvec_ms_jiffies_minmax,
+		.strategy	= &sysctl_jiffies,
+	},
+	{
 		.ctl_name	= NET_IPV4_ROUTE_GC_TIMEOUT,
 		.procname	= "gc_timeout",
 		.data		= &ip_rt_gc_timeout,
--- Documentation/filesystems/proc.txt.orig	2005-01-27 12:35:02.000000000 +0100
+++ Documentation/filesystems/proc.txt	2005-01-27 12:46:30.000000000 +0100
@@ -1709,12 +1709,13 @@
 
 Writing to this file results in a flush of the routing cache.
 
-gc_elasticity, gc_interval, gc_min_interval, gc_tresh, gc_timeout,
-gc_thresh, gc_thresh1, gc_thresh2, gc_thresh3
---------------------------------------------------------------
+gc_elasticity, gc_interval, gc_min_interval_ms, gc_timeout, gc_thresh
+---------------------------------------------------------------------
 
 Values to  control  the  frequency  and  behavior  of  the  garbage collection
-algorithm for the routing cache.
+algorithm for the routing cache. gc_min_interval is deprecated and replaced
+by gc_min_interval_ms.
+
 
 max_size
 --------

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH] gc_min_interval in milleseconds via /proc (fwd)
  2005-01-31 18:01   ` Robert Olsson
@ 2005-02-02 21:09     ` David S. Miller
  0 siblings, 0 replies; 4+ messages in thread
From: David S. Miller @ 2005-02-02 21:09 UTC (permalink / raw)
  To: Robert Olsson; +Cc: Robert.Olsson, netdev

On Mon, 31 Jan 2005 19:01:05 +0100
Robert Olsson <Robert.Olsson@data.slu.se> wrote:

> David S. Miller writes:
> 
>  > We must pick a new sysctl number too, for sysctl() system call.
>  > If the only access possible were through /proc then yes your
>  > change would be valid, but due to sysctl() system call the
>  > .ctl_name defines that dimension of the sysctl namespace.
> 
>  Ok!
>  NET_IPV4_ROUTE_GC_MIN_INTERVAL_MS was added last in list to avoid
>  renumbering. Feel free to change.

Applied, thanks Robert.

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2005-02-02 21:09 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2005-01-27 12:13 [PATCH] gc_min_interval in milleseconds via /proc (fwd) Robert Olsson
2005-01-31  6:56 ` David S. Miller
2005-01-31 18:01   ` Robert Olsson
2005-02-02 21:09     ` David S. Miller

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).