Re: ksyms_init considered too heavyweight for early bootstrap

[Andrew Doran <ad%netbsd.org@localhost>]

On Fri, Nov 21, 2008 at 01:22:05PM +0100, Martin Husemann wrote:

> Ok, after more off-list feedback, here is a most intrusive variant that:
> 
>  - makes sure we only have a single function called "init"
>  - initializes the mutex in main w/o exposing it and independend
>    of adding a symbol table
>  - is not in danger of having people add more functions but forgetting
>    the RUN_ONCE() dance
>  - adapts all archs to the renamed functions, removing the now unecessary
>    NULL calls to what used to be ksyms_init()
> 
> Haven't even compile tested most of this; we need to at least test one arch
> that is affected by the NULL-init removals before this could be commited
> [I'll try to check on my dreamcast this weekend].
> 
> Comments?

I don't see a changed call to ksyms_init_explicit() in the diff (x86
multiboot code calls this).

Thanks,
Andrew