On Tuesday, May 06, 2014 10:59 PM, Arnd Bergmann wrote:

On Tuesday 06 May 2014 19:03:52 Kishon Vijay Abraham I wrote:

quoted

In DRA7, the cpu sees 32bit address, but the pcie controller can see only 28bit
address. So whenever the cpu issues a read/write request, the 4 most
significant bits are used by L3 to determine the target controller.
For example, the cpu reserves 0x2000_0000 - 0x2FFF_FFFF for PCIe controller but
the PCIe controller will see only (0x000_0000 - 0xFFF_FFF). So for programming
the outbound translation window the *base* should be programmed as 0x000_0000.
Whenever we try to write to say 0x2000_0000, it will be translated to whatever
we have programmed in the translation window with base as 0x000_0000.

Cc: Bjorn Helgaas <bhelgaas@google.com>
Cc: Marek Vasut <marex@denx.de>
Signed-off-by: Kishon Vijay Abraham I <redacted>
Acked-by: Jingoo Han <redacted>
Acked-by: Mohit Kumar <redacted>

Sorry, but NAK.

We have a standard 'dma-ranges' property to handle this, so use it.

See the x-gene PCIe driver patches for an example. Please also talk
to Santosh about it, as he is implementing generic support for
parsing dma-ranges in platform devices at the moment.

Hi Arnd,

Do you mean the following patch?
http://www.spinics.net/lists/kernel/msg1737725.html

Thank you.

Best regards,
Jingoo Han

I also suspect you will have to implement swiotlb support to make
generic PCI devices work behind this bridge. Otherwise you end up
with random physical addresses passed into DMA registers.