[PATCH] b44: allow ethtool get_settings when down

[PATCH] b44: allow ethtool get_settings when down

[Stephen Hemminger]

The FC and Suse startup scripts use ethtool to check for link present. This has 
problems on my laptop with Broadcom because it quieries settings before
bringing link up. The problem is driver returns EAGAIN when queried for
settings but not up. Just go ahead and return values anyway, the supported and link
state values will be correct, speed will end up being 10BaseT/Half which is a
reasonable default.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>

diff -Nru a/drivers/net/b44.c b/drivers/net/b44.c
--- a/drivers/net/b44.c	2004-11-29 09:41:27 -08:00
+++ b/drivers/net/b44.c	2004-11-29 09:41:27 -08:00
@@ -1487,8 +1487,6 @@
 {
 	struct b44 *bp = netdev_priv(dev);
 
-	if (!(bp->flags & B44_FLAG_INIT_COMPLETE))
-		return -EAGAIN;
 	cmd->supported = (SUPPORTED_Autoneg);
 	cmd->supported |= (SUPPORTED_100baseT_Half |
 			  SUPPORTED_100baseT_Full |

Re: [PATCH] b44: allow ethtool get_settings when down

[Jeff Garzik]

Stephen Hemminger wrote:
> The FC and Suse startup scripts use ethtool to check for link present. This has 
> problems on my laptop with Broadcom because it quieries settings before
> bringing link up. The problem is driver returns EAGAIN when queried for
> settings but not up. Just go ahead and return values anyway, the supported and link
> state values will be correct, speed will end up being 10BaseT/Half which is a
> reasonable default.
> 
> Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
> 
> diff -Nru a/drivers/net/b44.c b/drivers/net/b44.c
> --- a/drivers/net/b44.c	2004-11-29 09:41:27 -08:00
> +++ b/drivers/net/b44.c	2004-11-29 09:41:27 -08:00
> @@ -1487,8 +1487,6 @@
>  {
>  	struct b44 *bp = netdev_priv(dev);
>  
> -	if (!(bp->flags & B44_FLAG_INIT_COMPLETE))
> -		return -EAGAIN;
>  	cmd->supported = (SUPPORTED_Autoneg);
>  	cmd->supported |= (SUPPORTED_100baseT_Half |
>  			  SUPPORTED_100baseT_Full |

I'm not so sure about this one...

This sounds like working around stupid userland in the kernel?

	Jeff

Re: [PATCH] b44: allow ethtool get_settings when down

[Stephen Hemminger]

On Thu, 02 Dec 2004 06:25:09 -0500
Jeff Garzik <jgarzik@pobox.com> wrote:

> Stephen Hemminger wrote:
> > The FC and Suse startup scripts use ethtool to check for link present. This has 
> > problems on my laptop with Broadcom because it quieries settings before
> > bringing link up. The problem is driver returns EAGAIN when queried for
> > settings but not up. Just go ahead and return values anyway, the supported and link
> > state values will be correct, speed will end up being 10BaseT/Half which is a
> > reasonable default.
> > 
> > Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
> > 
> > diff -Nru a/drivers/net/b44.c b/drivers/net/b44.c
> > --- a/drivers/net/b44.c	2004-11-29 09:41:27 -08:00
> > +++ b/drivers/net/b44.c	2004-11-29 09:41:27 -08:00
> > @@ -1487,8 +1487,6 @@
> >  {
> >  	struct b44 *bp = netdev_priv(dev);
> >  
> > -	if (!(bp->flags & B44_FLAG_INIT_COMPLETE))
> > -		return -EAGAIN;
> >  	cmd->supported = (SUPPORTED_Autoneg);
> >  	cmd->supported |= (SUPPORTED_100baseT_Half |
> >  			  SUPPORTED_100baseT_Full |
> 
> I'm not so sure about this one...
> 
> This sounds like working around stupid userland in the kernel?
> 
> 	Jeff

Don't bother with the patch, if I use smart user land code like NetworkManager
then there is no problem.  Although EAGAIN seems like a poor choice for errno
how about ENETDOWN or ENONET