Re: [PATCH] hw/ppc: fix decrementer with BookE timers

["Clément Chigot" <chigot@adacore.com>]

On Thu, Aug 29, 2024 at 2:33 PM Cédric Le Goater <clg@kaod.org> wrote:
>
> On 8/28/24 09:21, Clément Chigot wrote:
> > On Tue, Aug 27, 2024 at 7:40 PM Cédric Le Goater <clg@kaod.org> wrote:
> >>
> >> Hello Clément,
> >>
> >> On 7/15/24 10:46, Clément Chigot wrote:
> >>> The BookE decrementer stops at 0, meaning that it won't decremented
> >>> towards "negative" values.
> >>> However, the current logic is inverted: decr is updated solely when
> >>> the resulting value would be negative.
> >>
> >> How did you hit the issue ? which machine ? I didn't see any error
> >> when booting Linux 6.6.3 on mpc8544ds, e500mc, e5500 and e6500.
> >
> > I hit this issue while running some version of VxWorks on a custom
> > machine: p3041ds (description [1] and our local implementation [2]).
> > So, I'm not that surprised you were not able to reproduce.
> >
> >>> Signed-off-by: Clément Chigot <chigot@adacore.com>
> >>> Fixed: 8e0a5ac87800 ("hw/ppc: Avoid decrementer rounding errors")
> >>
> >> LGTM,
> >>
> >> Reviewed-by: Cédric Le Goater <clg@redhat.com>
> >>
> >> We have some automated tests with the ppce500 machine which it would be
> >> interesting  to extend to have a better coverage of booke.
> >
> > Thanks for the pointer, I'll see if I can extend them.
> >
> >> Thanks,
> >>
> >> C.
> >>
> >
> > [1] https://www.nxp.com/design/design-center/software/qoriq-developer-resources/p3041-qoriq-development-system:P3041DS
> > [2] https://github.com/AdaCore/qemu/blob/qemu-stable-9.0.0/hw/ppc/p3041ds.c
>
> Nice. Do you have any plans to upstream the machine ?

That was the plan. However, we are using that emulation in order to
test VxWorks on PPC.  Recent versions have now officially provided a
BSP for QEmu using ppce500 board. We are planning to transition to it,
mainly to avoid having to maintain that P3041DS home-made
implementation. Thus, I'm not sure we'll spend time on upstreaming it
:(
That being said, if you heard people being interested in that board,
please do warn me.

Thanks,
Clément

> Thanks,
>
> C.
>