REGRESSION in 6.0-rc7 caused by patch "ARM/dma-mapping: use dma-direct unconditionally"

[Pali Rohár]

On Monday 03 October 2022 21:04:46 Marek Behún wrote:
> On Mon, 3 Oct 2022 18:09:50 +0200
> Pali Rohár <pali at kernel.org> wrote:
> 
> > On Monday 03 October 2022 17:25:33 Marek Behún wrote:
> > > I seem to remember Pali talking about how the ranges defined in some
> > > upstream mvebu-tree, using MBUS_ID() macros, are incorrect.
> > > Pali, what do you remember about this?  
> > 
> > This is "incorrect" in respect to the standard DT translation of
> > "ranges" property. And applies for PCIe. Anyway, it works correctly just
> > because mvebu PCIe drivers (both u-boot and linux) have workarounds for
> > those issues, and expects that DT are "incorrect". So lets say mvebu
> > PCIe defined _new DT standard_ for PCIe incompatible with DT standard.
> 
> So this doesn't concert non-PCIe peripherals?

I have not spotted such giant hack for other peripherals. But I have not
tested it deeply. So it is possible that there are issues also with
non-PCIe peripherals. I found that PCIe issue only during review of
Orion support when half of ranges started to be translated to nonsense
values if we tried to use kernel of_* functions instead of home-made dts
address parsers.