2009/304 IP PROMISC Flag

Sebastien Roy Sebastien.Roy at Sun.COM
Fri May 15 08:08:56 PDT 2009


FWIW, I'm fine with this flag for familiarity with other OSs, but
strictly speaking, it has very little to do with IP interfaces, and
therefore shouldn't show up in ifconfig output.  The more correct place
to show whether or not a given datalink has had DL_PROMISC_PHYS enabled
would be in dladm output, perhaps through a read-only link property.
That, however, doesn't seem to be this case, but I thought I would note
that this (to me) will be an architectural wart on Solaris that seems to
only be introduced for familiarity.

Given my grumbling above, I'm almost giving my "plus one", but would
like some additional clarifications below.

On Fri, 2009-05-15 at 10:02 -0400, James Carlson wrote:
> I'm sponsoring this fast-track request for Girish Moodalbail.  The
> timer is set to 05/22/2009.
> 
> 
> The release binding is "patch/micro" and consolidation is "ON".  The
> stability level of the ifconfig output with new interface flag is 
> "Committed".
> 
> Problem Definition and Proposal:
> ---------------------------------
> 
> Today the ifconfig(1m) output does not display the promiscuous state
> of an underlying physical interface.  The Solaris kernel has all the
> provisions to record the promiscuous state of the underlying physical
> interface in 'phyint_flags' member variable of struct phyint, but it
> does not expose that state using the existing (unused) IFF_PROMISC
> flag.

Administrators don't really know or care about the "phyint_flags" member
of struct phyint in the kernel.  What they do care about is the meaning
of PROMISC flag when they see it in the ifconfig output.  You'll likely
need to specify something more administratively meaningful in the
ifconfig man page anyway, could that verbiage be specified in this case
please?

Does the flag have any semantics on IPMP interfaces?

What about lo0; should it ever appear there?

(the last two questions relate to my grumblings about this having little
to do with IP interfaces)

-Seb





More information about the opensolaris-arc mailing list