[Patch 1/5] Network Drop Monitor: Add netlink protocol identifer

[Patch 1/5] Network Drop Monitor: Add netlink protocol identifer

[Neil Horman]

Network Drop Monitor: Add Netlink protocol identifier
    
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>


 netlink.h |    1 +
 1 file changed, 1 insertion(+)

diff --git a/include/linux/netlink.h b/include/linux/netlink.h
index 1e6bf99..d4ac636 100644
--- a/include/linux/netlink.h
+++ b/include/linux/netlink.h
@@ -24,6 +24,7 @@
 /* leave room for NETLINK_DM (DM Events) */
 #define NETLINK_SCSITRANSPORT	18	/* SCSI Transports */
 #define NETLINK_ECRYPTFS	19
+#define NETLINK_DRPMON		20	/* Netork packet drop alerts */
 
 #define MAX_LINKS 32		
 

Re: [Patch 1/5] Network Drop Monitor: Add netlink protocol identifer

[Evgeniy Polyakov]

Hi Neil.

On Tue, Mar 03, 2009 at 12:00:14PM -0500, Neil Horman (nhorman@tuxdriver.com) wrote:
> 
> Network Drop Monitor: Add Netlink protocol identifier
>     
> Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
> 
> 
>  netlink.h |    1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/include/linux/netlink.h b/include/linux/netlink.h
> index 1e6bf99..d4ac636 100644
> --- a/include/linux/netlink.h
> +++ b/include/linux/netlink.h
> @@ -24,6 +24,7 @@
>  /* leave room for NETLINK_DM (DM Events) */
>  #define NETLINK_SCSITRANSPORT	18	/* SCSI Transports */
>  #define NETLINK_ECRYPTFS	19
> +#define NETLINK_DRPMON		20	/* Netork packet drop alerts */
>  
>  #define MAX_LINKS 32		
>  

Why do you want to use own family instead of using existing netlink
helpers like connector or genetlink?

-- 
	Evgeniy Polyakov

Re: [Patch 1/5] Network Drop Monitor: Add netlink protocol identifer

[Neil Horman]

On Tue, Mar 03, 2009 at 09:20:06PM +0300, Evgeniy Polyakov wrote:
> Hi Neil.
> 
> On Tue, Mar 03, 2009 at 12:00:14PM -0500, Neil Horman (nhorman@tuxdriver.com) wrote:
> > 
> > Network Drop Monitor: Add Netlink protocol identifier
> >     
> > Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
> > 
> > 
> >  netlink.h |    1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/include/linux/netlink.h b/include/linux/netlink.h
> > index 1e6bf99..d4ac636 100644
> > --- a/include/linux/netlink.h
> > +++ b/include/linux/netlink.h
> > @@ -24,6 +24,7 @@
> >  /* leave room for NETLINK_DM (DM Events) */
> >  #define NETLINK_SCSITRANSPORT	18	/* SCSI Transports */
> >  #define NETLINK_ECRYPTFS	19
> > +#define NETLINK_DRPMON		20	/* Netork packet drop alerts */
> >  
> >  #define MAX_LINKS 32		
> >  
> 
> Why do you want to use own family instead of using existing netlink
> helpers like connector or genetlink?
> 
Because this is its own protocol.  I could have used NETLINK_GENERIC I suppose,
but this is a feature that, if used is going to be commonly built into the
kernel, I see no need to add an additional step of making userspace search for a
protocol registered to NETLINK_GENERIC when we dont need to.  In other words,
why not define my own protocol?
Neil

> -- 
> 	Evgeniy Polyakov
> 

Re: [Patch 1/5] Network Drop Monitor: Add netlink protocol identifer

[Evgeniy Polyakov]

On Tue, Mar 03, 2009 at 02:00:54PM -0500, Neil Horman (nhorman@tuxdriver.com) wrote:
> Because this is its own protocol.  I could have used NETLINK_GENERIC I suppose,
> but this is a feature that, if used is going to be commonly built into the
> kernel, I see no need to add an additional step of making userspace search for a
> protocol registered to NETLINK_GENERIC when we dont need to.  In other words,
> why not define my own protocol?

genetlink is alwats built into the kernel, there are access libraries
helpers and there is really no userspace overhead. It was really
specially designed for this kind of the tasks to eliminate netlink
internal structure access and provide simple high-level interface.

-- 
	Evgeniy Polyakov

Re: [Patch 1/5] Network Drop Monitor: Add netlink protocol identifer

[Neil Horman]

On Tue, Mar 03, 2009 at 10:04:57PM +0300, Evgeniy Polyakov wrote:
> On Tue, Mar 03, 2009 at 02:00:54PM -0500, Neil Horman (nhorman@tuxdriver.com) wrote:
> > Because this is its own protocol.  I could have used NETLINK_GENERIC I suppose,
> > but this is a feature that, if used is going to be commonly built into the
> > kernel, I see no need to add an additional step of making userspace search for a
> > protocol registered to NETLINK_GENERIC when we dont need to.  In other words,
> > why not define my own protocol?
> 
> genetlink is alwats built into the kernel, there are access libraries
> helpers and there is really no userspace overhead. It was really
> specially designed for this kind of the tasks to eliminate netlink
> internal structure access and provide simple high-level interface.
> 
Not really about performance overhead in my view, its about programming
overhead.  Its just more work, and without any really tangible gain IMO.  Could
I use it?  Sure, but I don't really get any particular benefit out of doing so,
and from the kernel standpoint, it makes no difference either, except that its
one more header to parse before I get to my data.  I guess it just boils down to
choice for me
Neil

> -- 
> 	Evgeniy Polyakov
> 

Re: [Patch 1/5] Network Drop Monitor: Add netlink protocol identifer

[David Miller]

From: Evgeniy Polyakov <zbr@ioremap.net>
Date: Tue, 3 Mar 2009 21:20:06 +0300

> Why do you want to use own family instead of using existing netlink
> helpers like connector or genetlink?

Yes, indeed, use genetlink for this.

We don't add new core netlink protocols any more with it's
fixed address space of link types.

Re: [Patch 1/5] Network Drop Monitor: Add netlink protocol identifer

[Neil Horman]

On Tue, Mar 03, 2009 at 02:17:49PM -0800, David Miller wrote:
> From: Evgeniy Polyakov <zbr@ioremap.net>
> Date: Tue, 3 Mar 2009 21:20:06 +0300
> 
> > Why do you want to use own family instead of using existing netlink
> > helpers like connector or genetlink?
> 
> Yes, indeed, use genetlink for this.
> 
> We don't add new core netlink protocols any more with it's
> fixed address space of link types.
> 

very well, I'll make the change, and repost in a few days when I have it all
tested.
Neil

Re: [Patch 1/5] Network Drop Monitor: Add netlink protocol identifer

[Neil Horman]

On Tue, Mar 03, 2009 at 12:00:14PM -0500, Neil Horman wrote:
> 
> Network Drop Monitor: Add Netlink protocol identifier
>     
> Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
> 
> 
>  netlink.h |    1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/include/linux/netlink.h b/include/linux/netlink.h
> index 1e6bf99..d4ac636 100644
> --- a/include/linux/netlink.h
> +++ b/include/linux/netlink.h
> @@ -24,6 +24,7 @@
>  /* leave room for NETLINK_DM (DM Events) */
>  #define NETLINK_SCSITRANSPORT	18	/* SCSI Transports */
>  #define NETLINK_ECRYPTFS	19
> +#define NETLINK_DRPMON		20	/* Netork packet drop alerts */
>  
>  #define MAX_LINKS 32		
>  
> --
> To unsubscribe from this list: send the line "unsubscribe netdev" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 


Ok, as requested by David, respinning entire set against latest net-next-2.6.
tree.  Patch 1/5, in the course of updates in this thread, is rescinded.

Neil