On 13/11/2019 4:13 pm, Nicolas Saenz Julienne wrote:

Using a mask to represent bus DMA constraints has a set of limitations.
The biggest one being it can only hold a power of two (minus one). The
DMA mapping code is already aware of this and treats dev->bus_dma_mask
as a limit. This quirk is already used by some architectures although
still rare.

With the introduction of the Raspberry Pi 4 we've found a new contender
for the use of bus DMA limits, as its PCIe bus can only address the
lower 3GB of memory (of a total of 4GB). This is impossible to represent
with a mask. To make things worse the device-tree code rounds non power
of two bus DMA limits to the next power of two, which is unacceptable in
this case.

In the light of this, rename dev->bus_dma_mask to dev->bus_dma_limit all
over the tree and treat it as such. Note that dev->bus_dma_limit is
meant to contain the higher accesible DMA address.

Neat, you win a "why didn't I do it that way in the first place?" :)

Looking at it without all the history of previous attempts, this looks 
entirely reasonable, and definitely a step in the right direction.

[...]

Although this preserves the existing behaviour, as in of_dma_configure() 
we can do better here since we have the original address range to hand. 
I think it's worth keeping the ACPI and OF paths in sync for minor 
tweaks like this, rather than letting them diverge unnecessarily.

Otherwise, the rest looks OK to me - in principle we could store it as 
an exclusive limit such that we could then streamline the min_not_zero() 
tests to just min(mask, limit - 1), but that's probably too clever for 
its own good.

Robin.

  		dev->coherent_dma_mask = mask;
  		*dev->dma_mask = mask;
  	}