Date: Thu, 08 May 2003 11:56:35 -0400 From: Jeff Garzik <jgarzik@pobox.com>
> That is not necessarily the case. What is the case is that > you absolutely cannot assume that a non-UP interface will give > a valid link status.
Correct. And this will become -more- true as time goes on. Ideally we will be powering-down the NICs if the interface has been down more than a few seconds. It's just luck and (in some cases) ancient drivers, that getting link-status when interface is down works at all.
OK. So why are you testing for "running" rather than "up"? Are these two synonymous in the kernel?
I don't test either of these in my drivers.
More generally, I need some help understanding how to configure a
"down" interface to detect link beat.
Like I just said, this is mostly impossible.
If the interface is down, consider the NIC powered down. There is obviously no link beat.
My program polls the known interfaces to see if there are any changes to link beat. Right now I'm just ORing in IFF_UP to the existing set of flags but I worry that by doing so I'm doing too much. Is there a minimum set of flags that will give me access to link beat without making the interface fully active?
See what I said, above :)
Jeff
- : send the line "unsubscribe linux-net" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
