[PATCH v5 3/6] riscv: errata: Add Andes alternative ports

Lad, Prabhakar prabhakar.csengg at gmail.com
Tue Dec 20 16:31:16 PST 2022


Hi Conor,

On Mon, Dec 19, 2022 at 4:20 PM Conor Dooley <conor at kernel.org> wrote:
>
> On Mon, Dec 19, 2022 at 11:19:13AM +0000, Lad, Prabhakar wrote:
> > Hi Conor,
> >
> > Thank you for the review.
> >
> > On Sat, Dec 17, 2022 at 9:19 PM Conor Dooley <conor at kernel.org> wrote:
> > >
> > > On Mon, Dec 12, 2022 at 11:55:02AM +0000, Prabhakar wrote:
> > > > From: Lad Prabhakar <prabhakar.mahadev-lad.rj at bp.renesas.com>
> > > >
> > > > Add required ports of the Alternative scheme for Andes CPU cores.
> > > >
> > > > I/O Coherence Port (IOCP) provides an AXI interface for connecting external
> > > > non-caching masters, such as DMA controllers. IOCP is a specification
> > > > option and is disabled on the Renesas RZ/Five SoC due to this reason cache
> > > > management needs a software workaround.
> > > >
> > > > Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj at bp.renesas.com>
> > > > ---
> > > > v4 -> v5
> > > > * Sorted the Kconfig/Makefile/Switch based on Core name
> > > > * Added a comments
> > > > * Introduced RZFIVE_SBI_EXT_IOCP_SW_WORKAROUND SBI EXT ID to check if
> > > >   CMO needs to be applied. Is there a way we can access the DTB while patching
> > > >   as we can drop this SBI EXT ID and add a DT property instead for cmo?
> > > >
<snip>
> > > Seeing as you need a new version for some of the other bits, I think it
> > > would be good to add a minor comment here somewhere (be it here or the
> > > commit message) that links to the SBI specs for this.
> > > I think this looks pretty good though.
> > Sure I'll add a comment here.
> >
> > I was wondering if we can get rid of this vendor specific extension
> > here if we get access to the DT here (for example having a DT property
> > which would indicate if IOCP CMO should be applied or not). Do you
> > think that would be good approach?  ATM we dont have a pointer here
> > for FDT whie early patching.
>
> I dunno. I think it is fine to use the ECALL to be honest - I'd rather
> that than a property that someone may omit.
>
Ok, I was so I will stick with the current implementation.

> That said, for the cache management stuff we are gonna need for
> PolarFire SoC, we will need to have info from the DT AFAICT - marchid
> etc are all set to zero on our platform so cannot be used.
>
Aha so while patching you will need a pointer to FDT node.

> I was thinking about using the compatible instead, but...
> we've not tried to "forward"-port our stuff from 5.15 yet as we have
> not yet completed testing testing on our vendor tree (and need some PCI
> changes accepted upstream first anyway), as a result I have not looked
> into what's needed there for use with alternatives. We've been using a
> pre-alternatives version of that patchset from around the 5.15
> development point in time instead.
>
Good to know. Let me know if you plan to implement the patching
mechanism based on FDT soon. I can give it a test.

Cheers,
Prabhakar



More information about the linux-riscv mailing list