Re: can't reboot after running a 5.0 kernel

["Valeriy E. Ushakov" <uwe%stderr.spb.ru@localhost>]

On Mon, Jan 31, 2011 at 16:04:59 -0500, Chuck Cranor wrote:

> I also did a survey of some of the ports in the tree, and it looks
> some port's cpu_setfunc() still call lwp_startup() while other ports
> have been modified (like the alpha) to not call it:

Ok, glancing at the code I think that cpu_setfunc must arrange for the
trampoline to call lwp_startup, since, even though we "recycle" an
existing lwp it still acts like a "new" one since it's not resumed in
mi_switch() - so (as you discovered) we need lwp_startup to run the
code that for an "old" lwp we do at kern_synch.c:791

I guess what happened is that

. at the time of yamt-idlelwp SA were already dead

. when I adapted sh3 to yamt-idlelwp I didn't bother to investigate
  and just conservatively left cpu_setfunc as it was (i.e. w/out call
  to lwp_startup)

. that didn't matter since SA were dead

. I got mrg@ confused with my question
  http://mail-index.netbsd.org/source-changes-d/2009/05/27/msg000502.html

. mrg changed sparc and sparc64

. martin propagated that change to other ports

. all that still didn't matter since SA were dead

. SA were resurrected, but, I guess, not tested that much on !x86


Sorry :)

-uwe