In message: <200605032151.k43Lpjg2009960_at_ns.init-main.com> takawata_at_jp.freebsd.org writes: : In message <20060501.222743.32503989.imp_at_bsdimp.com>, "M. Warner Losh" さんいわ : く: : >In message: <200605020815.31012.doconnor_at_gsoft.com.au> : > "Daniel O'Connor" <doconnor_at_gsoft.com.au> writes: : >: On Tuesday 02 May 2006 01:23, M. Warner Losh wrote: : >: > : However you wouldn't expect that using it as a module would result in : >: > : reduced functionality. (The same as how you wouldn't expect compiling : >: > : something into your kernel would result in reduced functionality) : >: > : >: > At the same time, you'd expect it to behave like every other 'bus' in : >: > the tree. We omit the attachments for drivers to that bus when the : >: > kernel is built w/o that bus. This is why if you have, say, ep in the : >: > kernel, but pccard loaded as a module, the 3c589 you just inserted : >: > into the pccard slot won't work. : >: > : >: > It is a minor imperfection in the config system that no one has taken : >: > on as a Problem To Solve. Solving it turns out to be somewhat tricky. : >: : >: Hmm, but I load acpi as a module and get acpi attachments.. (for sio and ppc : >) : >: : >: <shrugs> it's black magic to me anyway :) : >: : >: I didn't mean to belittle Marcel's work, I was just suprised that it would : >: cause a functionality loss like that. : > : >I think it just points out a weakness in the underlying system... : : Is it time to hide isa_dma interface under isa_if ? : All acpi capable archictectures links isa_if.c, I think. Yes. I think it would be worthwhile to have all the ISA symbols hidden, like we do for PC Card. This would be a step in that direction... Maybe the last step? WarnerReceived on Wed May 03 2006 - 20:02:32 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:38:55 UTC