Re: [PATCH 3/3] EDAC: mv64x60: replace in_le32/out_le32 with ioread32/iowrite32

[Michael Ellerman]

Arnd Bergmann <arnd@xxxxxxxx> writes:

> On Thu, May 18, 2017 at 7:36 AM, Michael Ellerman <mpe@xxxxxxxxxxxxxx> wrote:
>> Borislav Petkov <bp@xxxxxxxxx> writes:
>>
>>> Top-posting so that the PPC list can see the whole patch below.
>>>
>>> Since I don't know PPC, let me add PPC ML to CC for a confirmation this
>>> change is correct.
>>>
>>> Which brings me to the tangential: this driver is from 2006-ish and
>>> is for some "Marvell MV64x60 Memory Controller kernel module for PPC
>>> platforms". If you're going to touch it, then you should test on the PPC
>>> hardware too, so that you don't break the driver there.
>>>
>>> Unless that hardware is obsolete now and we don't care and, and ..?
>>>
>>> Maybe someone has some insights...
>>
>> Not really sorry.
>>
>> I don't have one of those boards, so I can't test. Maybe someone else on
>> the list does?
>>
>> I'd err on the side of the PPC hardware being obsolete and no one really
>> caring. If the driver is helpful on ARM then we may as well use it
>> there, if we can also avoid breaking it on PPC then great.
>
> I never had one myself, but tried to figure out what is still there to be
> supported. In 2014, we removed one platform (PrPMC2800) that was
> obsolete. There were eight boards that didn't make the cut from
> arch/ppc32 to arch/powerpc. The C2K is the last one and it was
> added in 2008 with this comment:
>
>     Support for the C2K cPCI Single Board Computer from GEFanuc
>     (PowerPC MPC7448 with a Marvell MV64460 chipset).
>     All features of the board are not supported yet, but the board
>     boots, flash works, all Ethernet ports are working and PCI
>     devices are all found (USB and SATA on PCI1 do not work yet).
>
>     Part 3 of 5: driver for the board.  At this time it is very generic
>     and similar to its original, the driver for the prpmc2800.
>
> The original submitter never followed up on it and neither the
> board code not the DTS was ever updated to include additional
> features, so I assume it only got worse from there.
>
> According to https://www.abaco.com/products/c2ka-compactpci-sbc
> the end-of-life date for the product was in 2015, presumably 10 years
> after it got introduced.
>
> This is definitely obsolete by now, and given the missing features
> I would assume that nobody is running mainline kernels on it and
> it can be removed.

Great, thanks for doing all that digging.

I'll cook up a patch to remove c2k next week.

Then we can drop CONFIG_MV64X60 from PPC and see where that leaves
things, depending on what's useful on ARM.

cheers