On Sat, 24 Sep 2016, Manuel Bouyer wrote:
On Sat, Sep 24, 2016 at 01:29:02PM +0800, Paul Goyette wrote:On Sat, 24 Sep 2016, Masao Uebayashi wrote:At a glance, getdevi() can't return pseudo devices. You have to refactor `delpseudo()' to have `getpseudo()', then try it from `addpseudoroot()'. I'm not sure what will happen after that. (These two different meanings of "pseudo" are confusing.)It turns out that I simply have to use pseudo-device ataraid ld* at ataraid? No need to use pseudo-root at all!Doens't this means that, without pseudo-root, ataraid can't be a boot device anymore ?
Totally unrelated.This ioconf stuff applies only to loadable (ie, non-built-in) modules. If you continue to use a built-in copy of ataraid it will continue to work the same way it does now. (In other words, the boot capability of the ld device attached to ataraid will not be changed.)
Currently, built-in is the only option for ataraid. When my current changes are done, it will be possible for ataraid to be loadable separately from the kernel. If you want to boot from such a ataraid, you'll need to make sure the module gets "pushed" to the boot loader, either using "load" commands at the boot prompt, or placing appropriate entries in your /boot.conf file.
But I repeat: None of these changes should have any impact to ataraid when it is built-in. If any surprises occur, they will either be fixed quickly or reverted.
+------------------+--------------------------+------------------------+ | Paul Goyette | PGP Key fingerprint: | E-mail addresses: | | (Retired) | FA29 0E3B 35AF E8AE 6651 | paul at whooppee.com | | Kernel Developer | 0786 F758 55DE 53BA 7731 | pgoyette at netbsd.org | +------------------+--------------------------+------------------------+