[U-Boot] [PATCH 4/6] dma: ti: add driver to K3 UDMA

[Vignesh R]

On 23/01/19 7:26 PM, Tom Rini wrote:
> On Wed, Jan 23, 2019 at 07:13:11PM +0530, Vignesh R wrote:
>> Tom,
>>
>> On 23/01/19 4:05 PM, Peter Ujfalusi wrote:
>>> Tom,
>>>
>>> On 22/01/2019 20.56, Tom Rini wrote:
>>>> On Tue, Jan 22, 2019 at 08:33:57PM +0530, Vignesh R wrote:
>>>>
>>>>> The UDMA-P is intended to perform similar (but significantly upgraded) functions
>>>>> as the packet-oriented DMA used on previous SoC devices. The UDMA-P module
>>>>> supports the transmission and reception of various packet types.
>>>>> The UDMA-P also supports acting as both a UTC and UDMA-C for its internal
>>>>> channels. Channels in the UDMA-P can be configured to be either Packet-Based or
>>>>> Third-Party channels on a channel by channel basis.
>>>>>
>>>>> The initial driver supports:
>>>>> - MEM_TO_MEM (TR mode)
>>>>> - DEV_TO_MEM (Packet mode)
>>>>> - MEM_TO_DEV (Packet mode)
>>>>>
>>>>> Signed-off-by: Peter Ujfalusi <peter.ujfalusi at ti.com>
>>>>> Signed-off-by: Grygorii Strashko <grygorii.strashko at ti.com>
>>>>> Signed-off-by: Vignesh R <vigneshr at ti.com>
>>>>
>>>> Reviewed-by: Tom Rini <trini at konsulko.com>
>>>>
>>>> And the DT binding is common to Linux, and been reviewed there?  Or?
>>
>> As Peter pointed out DT bindings are not yet frozen and may change.
>> Hence, I have added the nodes in -u-boot.dtsi. As and when DT bindings
>> are accepted, these would be moved to base dtsi.
>>
>> UDMA support in kernel is currently blocked due to on going discussions
>> on how UDMA/ring interrupts needs to be modeled and supported (not a so
>> important for U-Boot UDMA support). But, I submitted patches for U-Boot
>> as UDMA is support is required to support networking on AM654 which is
>> essential for ease of booting the platform with U-Boot.
>>
>>>
>>> The binding is the same for Linux but unfortunately it has not went
>>> through a proper review yet due to the fact that I need to wait for the
>>> interrupt support to arrive to mainline.
>>>
>>> However I have sent an earlier version as RFC:
>>> https://www.spinics.net/lists/dmaengine/msg16661.html
>>>
>>> As for the bindings (and code):
>>> The linux bindings are different:
>>>
>>> - there is no ti,psi-proxy anymore.
>>
>> Will drop this as U-Boot driver does not use them either.
>>
>>> - ringacc uses tisci to get GP ring range and we need
>>>   ti,sci-rm-range-gp-rings property in DT for it
>>> - UDMA also uses tisci to get resource ranges and needs:
>>>   ti,sci-rm-range-tchan, ti,sci-rm-range-rchan, ti,sci-rm-range-rflow in DT
>>> - UDMA does not have/need dma-channels property
>>
>> I plan to align above bindings with Linux as when bindings are accepted
>> into kernel and then port it to U-Boot along with relevant TI-SCI driver
>> code.
>>
>> Meanwhile would it be acceptable to add these nodes in -u-boot.dtsi to
>> get closer to enable tftp?
> 
> ... I knew I should have read the whole thread first.  So, how likely do
> we think the kernel bindings are to see big changes?  I wonder if we
> should re-align things now, or not.
> 

I have posted v2 with bindings that are in full alignment with proposed
Linux bindings that Peter pointed out above. IMO, I dont expect any big
change in these bindings in upstream.

-- 
Regards
Vignesh