Date:	Wed, 25 Jan 2012 18:41:26 +0530
From:	Ravi Kumar V <kumarrav@...eaurora.org>
To:	Vinod Koul <vinod.koul@...el.com>
CC:	linux-arch@...r.kernel.org, linux@....linux.org.uk, arnd@...db.de,
	linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org,
	bryanh@...eaurora.org, tsoni@...eaurora.org,
	Daniel Walker <dwalker@...o99.com>, dan.j.williams@...el.com,
	davidb@...eaurora.org, linux-arm-kernel@...ts.infradead.org,
	johlstei@...eaurora.org
Subject: Re: [PATCH v2 0/2] Add Qualcomm MSM ADM DMAEngine driver

On 1/23/2012 7:21 PM, Vinod Koul wrote:
> On Mon, 2012-01-23 at 16:41 +0530, Ravi Kumar V wrote:
>>
>> If some changes are made in interleave API then it can support our BOX
>> mode. Here in interleaved template he is assuming destination pattern as
>> can be contiguous or same as source pattern, but in our case destination
>> pattern is different from source pattern.
>> So if a new parameter destination data chunk is added in "struct
>> dma_interleaved_template" structure then it can support different
>> destination pattern.
> do you mean you have cases where you are doing a "memcpy" from one
> interleaved memory to another?
> Can you provide me with a scenario where this maybe helpful?

Presently we are transferring data from interleaved memory tho 
contagious memory and vice-verse.
We can use the interleaved API for present scenario, but it will 
restrict the HW capability of transferring data from one interleaved 
pattern to other interleaved pattern.

>
> The reason why the API was designed like this was to give ability to
> take these kind of interleaved memory and copy them to peripheral
> (constant addr) or memory (typically contagious).
>
> In case it is just a pattern I wonder why it cannot be described in
> standard scatter gather definitions as you can split the block further
> down to copy from one respective block to somewhere else in memory.

We can use scatter gather but it will be extra burden on software to 
create those many SG list unlike in box mode just a single command 
serves the purpose.

>> Also it will good if you can provide another parameter for passing
>> private data to dma driver.
> 1. what does this parameter do?

Private parameter in our case will be command configuration parameter 
where we are passing information to HW like endianness, synchronization 
& acknowledge mechanism between DMA HW and peripherals running with 
different clock than DMA.

> 2. is this parameter static for a channel or it changes per transfer?
>
This parameter changes per each transfer.

We can see these possibilities to implement our DMA driver but still 
have to add new parameter in "scatterlist" & "interleaved template" for 
supporting per transfer private data like our command configuration 
parameter.

1.We have to add new parameter "struct data_chunk" in 
"interleaved_template" for supporting different destination pattern.
It helps a lot for implementing our total HW capability.

2.We have to use present API's for box mode.
--scatter gather API for different interleaved source & destination 
patterns.
--interleaved API for interleaved source pattern to destination 
contiguous pattern/vice-verse.
It causes some extra burden on our software to create sglist.

3.Implement New API for BOX mode.

I feel if we can go with first option then it helps a lot.

Please can you suggest a better way to solve our problem.

Thanks,
Ravi Kumar

-- 
Sent by a consultant of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists