Looking for architecture papers

[Raz]

Both systemsim and my powerpc server boots with MSR_HV=1, i.e, hypervisor state.
Is there away to fix that ? writing to the MSR cannot work according
the documentation ( and reality ).



On Sat, Oct 6, 2018 at 3:27 PM Segher Boessenkool
<segher at kernel.crashing.org> wrote:
>
> On Sat, Oct 06, 2018 at 12:19:45PM +0300, Raz wrote:
> > Hey
> > How does HVSC works ?
> > I looked in the code and LoPAR documentation. It looks like there is
> > vector called
> > system_call_pSeries ( at 0xc00 ) that is supposed to be called when we
> > invoke HVSC from kernel
> > mode.
> > Now, I wrote a NULL call HSVC and patched the exceptions-64s.S to
> > return RFID immediately.
> > This does not work.
> > Would you be so kind to explain how HVSC works ?
> > thank you
>
> If your kernel is not running in hypervisor mode, sc 1 does not call the
> kernel (but the hypervisor, instead).  If your kernel _is_ running in
> hypervisor mode, sc 1 does the same as sc 0, a normal system call.
>
> I don't know which it is for you; you didn't say.
>
> I have no idea what "a NULL call HSVC" means.  If you make exception c00
> return immediately (as you suggest) then you have made all system calls
> non-functional, which indeed is unlikely to work as you want.
>
>
> Segher