Instability likely related to new pmap on Cubieboard A10

[Dmitry Marakasov]

* Ian Lepore (ian at freebsd.org) wrote:

> > I've just tried latest HEAD on cubieboard A10, and discovered that
> > it's completely unstable. Kernel boots without problems, however
> > right after init is started many processes crash with sigsegv and
> > other errors, and it ends with either hang or a panic. Examples
> > below. With kernel built with `nooptions ARM_NEW_PMAP' these
> > problems go away. Feel free to ask for any additional info.
> > 
> > ---
> > [...]
> > ---
> 
> That is, frankly, just hard to accept as real.  ARM_NEW_PMAP has been
> the default for like 8 months.  If it was that broken, the first report
> of it would have come in much earlier than this.  That seems more like
> some sort of broken or corrupted binary, and perhaps the problem went
> away not because the option changed, but because changing the option
> rebuilt whatever was corrupted.
> 
> Actually, given that everything was fine until userland started, and
> "bad syscall" shows up a lot in the errors, a kernel/userland mismatch
> also seems like a good candidate.

I've just rebuilt it again just to be sure, same thing - kernel with
ARM_NEW_PMAP misbehaves. I'd add:
- Both kernels were built with the same commands, from the same revision
  and with the clean /usr/obj; only difference is presence of `nooptions
  ARM_NEW_PMAP`
- Both kernels use the same world, built from the same revision. It was
  not touched when changing kernels

-- 
Dmitry Marakasov   .   55B5 0596 FF1E 8D84 5F56  9510 D35A 80DD F9D2 F77D
amdmi3 at amdmi3.ru  ..:  jabber: amdmi3 at jabber.ru      http://amdmi3.ru