http://defect.opensolaris.org/bz/show_bug.cgi?id=10653
--- Comment #5 from amaguire <alan.maguire at sun.com> 2009-08-14 12:20:13 UTC
---
(In reply to comment #4)
> (In reply to comment #3)
> > (In reply to comment #2)
> > > The DHCP flag in the ifconfig output does not necessarily mean that
> > > the address was acquired from DHCP; the address is not the only
> > > configuration data that can be obtained from the DHCP server. Given
> > > that there is no dns-nameservice-configsrc value specified for this
> > > location, the default value would be used, and the default is to get
> > > this information from the DHCP server. Thus nwamd would have initiated
> > > a DHCP inform transaction to get that info, which would cause the DHCP
> > > flag to be turned on.
> > >
> > > Bottom line is that the presence of the DHCP flag does not necessarily
> > > mean that the address itself was DHCP-assigned; it simply means that a
> > > DHCP transaction of some sort has occurred.
> >
> > Ah, didn't realise that!
> >
> > So I'm assuming there is no way to validate where the address really came
> > from,
> > other than to remember what the saved configuration is vs the in-memory
> > configuration (which could differ).
>
> I _think_ the right answer here is to do a DHCP_STATUS request via
> libdhcpagent. The problem is that libdhcpagent is project-private to ON. In an
> ideal world, there would be a libnwam function to request detailed interface
> status (addresses configured and sources). Don't know how feasible that is
> given the time pressures.
>
> An ugly hack however could possibly make use of the fact that at the moment -
> while we only support 1 DHCP interface per-link - DHCP is always configured on
> the physical interface (i.e. ath0 rather than the logical ath0:1 etc). Any
> static addresses are added as logical interfaces. So (again this is very
> hacky)
> you could conclude that any address that appears on ath0 is from DHCP
> (provided
> that DHCP is an address source for the IP NCU of course).
>
> This will break if a user starts using ifconfig, but as an ugly band-aid it
> might just about do it.
>
> As a middleground we could cook up a limited interface status request that
> stated whether static/dhcp addresses or both were configured on the link.
>
> nwam_error_t nwam_interface_get_status(const char *linkname, uint64_t
> *statusp);
>
> The status flag could be a combination of:
>
> typedef enum {
> NWAM_IF_STATUS_DHCP_FAILED = 0x1,
> NWAM_IF_STATUS_DHCP_STARTED = 0x2,
> NWAM_IF_STATUS_DHCP_ACQUIRED = 0x4,
> NWAM_IF_STATUS_STATIC_SET = 0x8
> } nwam_if_status_t;
>
> Comments anyone?
following up, maybe a better model would be to add a flags field to
nwam_*_get_state, i.e.
nwam_error_t nwam_*_get_state(handle, statep, aux_statep, uint64_t *flagsp);
It'd be unused for other objects, and interface NCUs could fill it in...
--
Configure bugmail: http://defect.opensolaris.org/bz/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
