'default' vs. 'all'

'default' vs. 'all'

[David Miller]

Herbert, Yoshifuji and I were just discussing the
sysfs device attribute issue.

It's seems sane to me that if we had some kind of
'dirty' bit per attribute we could propagate default
settings everywhere except where the dirty bit had
been set.

The question is how to implement this nicely.

What do you think?

Re: 'default' vs. 'all'

[YOSHIFUJI Hideaki / 吉藤英明]

In article <20071208.000429.66028544.davem@davemloft.net> (at Sat, 08 Dec 2007 00:04:29 -0800 (PST)), David Miller <davem@davemloft.net> says:

> Herbert, Yoshifuji and I were just discussing the
> sysfs device attribute issue.
> 
> It's seems sane to me that if we had some kind of
> 'dirty' bit per attribute we could propagate default
> settings everywhere except where the dirty bit had
> been set.

One good event to propagate is the NETDEV_UP; if the dirty bit is
not set when the device is brought up, copy the default values to
the device.

Regards,

--yoshfuji @ Seattle

Re: 'default' vs. 'all'

[Herbert Xu]

On Sat, Dec 08, 2007 at 12:14:11AM -0800, YOSHIFUJI Hideaki / 吉藤英明 wrote:
>
> One good event to propagate is the NETDEV_UP; if the dirty bit is
> not set when the device is brought up, copy the default values to
> the device.

Yeah I already did that for IPv4 except that for backwards
compatibility it's set on address addition not device up.

I was in the process of doing it for IPv6 too but got distracted
by other things.  I might still have that tree on my hard drive
if anyone is interested.

Cheers,
-- 
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

Re: 'default' vs. 'all'

[Herbert Xu]

On Sat, Dec 08, 2007 at 12:04:29AM -0800, David Miller wrote:
> 
> Herbert, Yoshifuji and I were just discussing the
> sysfs device attribute issue.
> 
> It's seems sane to me that if we had some kind of
> 'dirty' bit per attribute we could propagate default
> settings everywhere except where the dirty bit had
> been set.

Yep.

> The question is how to implement this nicely.
> 
> What do you think?

Well this is how it works on IPv4 already :)

The only thing is that we forcibly set the dirty bit on everything
when an address is added for backwards compatibility.

If you delete that call then it's as you described.

Cheers,
-- 
Visit Openswan at http://www.openswan.org/
Email: Herbert Xu ~{PmV>HI~} <herbert@gondor.apana.org.au>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt