Date: Tue, 13 Jan 2009 15:21:43 +0100 From: Luigi Rizzo <rizzo@iet.unipi.it> To: geom@freebsd.org, Luigi Rizzo <rizzo@iet.unipi.it> Subject: Re: geom 'taste' vs. manual creation ? Message-ID: <20090113142143.GA92704@onelab2.iet.unipi.it> In-Reply-To: <20090113122111.GA89189@onelab2.iet.unipi.it> References: <20090113122111.GA89189@onelab2.iet.unipi.it>
next in thread | previous in thread | raw e-mail | index | archive | help
Ivan Voras said: > On Tue, Jan 13, 2009 at 01:21:11PM +0100, Luigi Rizzo wrote: > > geom(4) says: > > > > A geom which came into being as a result of a normal taste operation > > should self-destruct... > > > > Now I wonder: > > does the GEOM infrastructure record whether a geom has been created > > by a 'taste' call, or manually through a 'geom xxx create ..', or > > this info should be managed directly by the individual implementation ? > > I'm not sure if this is what you're asking, but when a new geom is > created (for example by a complex / transformation GEOM class...), it is > automatically tested again, so why would it be different now? i am looking at the module unloading (i.e.. destroy) case. I have a geom_sched module that uses the 'taste' method to automatically create instances to selected devices (listed in a kenv variable for what matters). The user can still manually issue "geom sched create ..." on other providers. When I unload the module, right now i call destroy() on all instances (including those created manually), but i wonder if the autodestruction should be limited to only the auto-created entities, as the manpage seems to suggest, and in this case whether there is any flag that records manual vs. automatic creation. cheers luigi
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20090113142143.GA92704>