[PATCH v2 5/5] lib: utils/reset: Add generic da9063 reset driver

[Nikita Shubin <nikita.shubin@maquefel.me>]

On Wed, 20 Oct 2021 10:33:26 +0200
Alexandre Ghiti <alexandre.ghiti@canonical.com> wrote:

> On Wed, Oct 20, 2021 at 9:00 AM Nikita Shubin
> <nikita.shubin@maquefel.me> wrote:
> >
> > Hello Alexandre!
> >
> > On Tue, 19 Oct 2021 14:39:16 +0200
> > Alexandre Ghiti <alexandre.ghiti@canonical.com> wrote:
> >  
> > > > +
> > > > +static const struct fdt_match da9063_reset_match[] = {
> > > > +       { .compatible = "da9063-reset", .data = (void *)TRUE },
> > > > +       { },
> > > > +};  
> > >
> > > Agree with Jessica here, as per Adam, the reset as we implement it
> > > here should remain a board specific thing, so we should not rely
> > > on this device tree node. I was hoping to make something that
> > > would work in general for da9063 users, but no luck.
> > >
> > > And finally I think we should listen to Adam and go for a full
> > > reset using the RTC, instead of the partial reset we implemented
> > > here. We sacrifice the RTC but anyway, it was not even exposed in
> > > the device tree, we should add a comment there so that people
> > > don't get surprised if they enable it. I will post the right
> > > sequence adapted to your sources soon.  
> >
> > Well the RTC is much more valuable than reset actually and used for
> > many tests.
> >
> > Indeed da9063 reset can be moved totally to "platform" code, but do
> > you think that leaving users with hard-coded reset cutting them off
> > from RTC and no other possibility, other than fixing OpenSBI code
> > in-place is really a good option ?
> >
> > With DTS approuch it can be as simple as removing the da9063-reset
> > from DTS file.
> >  
> 
> Arf, I change my mind every two days regarding this...I'm just trying
> to have a reliable reset, and to quote Adam below, the reset as
> implemented here is only partial:
> 
> "With the sequence you provided this is only a partial reset whereby
> all of the output rails are sequenced down then up again and restored
> to OTP voltages. However the remainder of the chip settings aren't
> reset as this isn't a true reset of the device going through full
> reload from OTP, so for example settings of regulator mode GPIO
> states, or IRQ mask bits would persist on the restart, which could
> have implications on system operation."

Actually i don't think battling over a board which isn't supposed to be
more than a development board and won't be massed used makes any sense.

It's more like QoL for those who are currently porting software or
experimenting with RISC-V.

Before this we used openocd reset, which can be greatly automated with
expect.

> 
> I imagine the partial reset could have side-effects on the RTC
> configuration too. So we might end up with a partial reset and a
> broken RTC...
> 

Well we should really investigate this.

As well as i don't quite understand this "sacrifice RTC" thing.

> and again from Adam:
> 
> "Personally, if it was possible I think the RTC approach would be
> best as it's a full reset and to me is far safer with regards to
> potential side effects"
> 
> 
> > >
> > > Thanks,
> > >
> > > Alex
> > >
> > >  
> > > > +
> > > > +struct fdt_reset fdt_reset_da9063 = {
> > > > +       .match_table = da9063_reset_match,
> > > > +       .init = da9063_reset_init,
> > > > +};
> > > > diff --git a/lib/utils/reset/objects.mk
> > > > b/lib/utils/reset/objects.mk index 6c95db3..cfe4c09 100644
> > > > --- a/lib/utils/reset/objects.mk
> > > > +++ b/lib/utils/reset/objects.mk
> > > > @@ -14,3 +14,4 @@ libsbiutils-objs-y +=
> > > > reset/fdt_reset_sifive_test.o libsbiutils-objs-y +=
> > > > reset/fdt_reset_sunxi_wdt.o libsbiutils-objs-y +=
> > > > reset/fdt_reset_thead.o libsbiutils-objs-y +=
> > > > reset/fdt_reset_thead_asm.o +libsbiutils-objs-y +=
> > > > reset/fdt_reset_da9063.o --
> > > > 2.31.1
> > > >
> > > >
> > > > --
> > > > opensbi mailing list
> > > > opensbi at lists.infradead.org
> > > > http://lists.infradead.org/mailman/listinfo/opensbi  
> >