Re: [PATCH v3] dt-bindings: timer: Add SiFive CLINT2

[Charles Perry <charles.perry@microchip.com>]

On Thu, Apr 09, 2026 at 01:37:46PM +0800, Nick Hu wrote:
> On Wed, Apr 8, 2026 at 10:41 PM Charles Perry
> <charles.perry@microchip.com> wrote:
> >
> > On Wed, Apr 08, 2026 at 02:15:19PM +0800, Nick Hu wrote:
> > > On Fri, Mar 27, 2026 at 4:56 AM Charles Perry
> > > <charles.perry@microchip.com> wrote:
> > > >
> > > > On Fri, Mar 21, 2025 at 04:35:06PM +0800, Nick Hu wrote:
> > > > > Add compatible string and property for the SiFive CLINT v2. The SiFive
> > > > > CLINT v2 is incompatible with the SiFive CLINT v0 due to differences
> > > > > in their control methods.
> > > >
> > > > Hello Nick,
> > > >
> > > > Can you help me understand what is this different control method? I've
> > > > found that both OpenSBI [1] and U-Boot [2] use the same match data in their
> > > > clint driver which would indicate that they are compatible.
> > > >
> > > Sorry for the late reply.
> >
> > Hello Nick, no worries.
> >
> > > Unlike v0, v2 requires a write to the mtime register to kick the timer.
> >
> > A kick once at the beginning or every time a machine timer interrupt fires?
> >
> A kick once at the beginning

That would explain what I observed in U-Boot and OpenSBI. The kick might be
in some arch specific code or a prior boot stage.

> 
> > >
> > > > Also, do you know if there's an easy way to tell if a sifive clint is a v0
> > > > or v2?
> > > >
> > > sifive,clint2 introduces additional MMIO registers
> >
> > Is that the high resolution timers (HRT) mentionned in the description?
> >
> > And last question, would you happen to know if sifive made a clint v1 that
> > does NOT require the mtime kick but does have the HRT?
> >
> Yes, that's correct.

Ok.

Thanks,
Charles

> 
> > I'm trying to figure out what should be the compatible for PIC64-HPSC.
> > Some code from a repository called "FSFM" used "sifive,clint1" for the
> > compatible. I'm wondering if I should add the clint v1 to this file.
> >
> > Thanks,
> > Charles
> >