On Tue, May 16, 2006, Andrew de Quincey wrote: > In the meantime, I've done a partial version with js's suggestions - it is at > > http://linuxtv.org/hg/~quincy/v4l-dvb-attach > > This removes chaining, and adds support for the SEC devices into the > frontend_ops. I have not implemented any frontend changes yet. Those two changes: http://linuxtv.org/hg/~quincy/v4l-dvb-attach?cmd=changeset;node=66280c078abf;style=gitweb http://linuxtv.org/hg/~quincy/v4l-dvb-attach?cmd=changeset;node=600622e5a61e;style=gitweb Look good to me, but should be folded into one before asking Mauro to pull (they are not independent). Maybe Oliver can ACK. I think they completely remove the chaining problem. > Also: there are no symbol_put_addrs() done yet. I've not yet had time to think > on how do to do that - any suggestions? We could use a dvb_detach as js > suggests, passing it the release() pointer to detach and then call > symbol_put_adddr. ... > Here is a first stab at dvb_detach (sorry if you already posted one johannes; > I have zapped the message :( > > #define dvb_detach(FUNCTION, FE) \ > do { \ > if (FUNCTION) { \ > FUNCTION(FE); \ > symbol_put_addr(FUNCTION); \ > } \ > } while(0); > > We could either make the card drivers call this directly, or (for convenience) > call it from dvb_unregister_frontend() instead of directly calling the > release() methods like we do just now It cannot be called from dvb_unregister_frontend(), as the whole point of this exercise is that dvb_attach/detach is optional. So, it must be called by the card drivers, right after the dvb_unregister_frontend() call, and it should simply do: #define dvb_detach(function) symbol_put_addr(function) (As a minor complication, fe->ops.release needs to be saved in a temporary between dvb_unregister_frontend() and dvb_detach().) We should also add a BUG_ON(!fe->ops.release) to dvb_register_frontend(), and remove the bogus warning printk from dvb_unregister_frontend(). Or, probably more correct, change dvb_attach() to return the foo_attach function pointer, and save that in a new field in the car driver struct, and pass that to dvb_detach(). Or, maybe we should handle this via Kconfig? So that you can add the dvb_detach() to dvb_unregister_frontend(), but one can disable dvb_detach/dvb_detach via CONFIG_DVB_FRONTEND_STATIC_DEPS? Vote for your favourite solution... Johannes _______________________________________________ linux-dvb@xxxxxxxxxxx http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb
