[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <790ee7c8-a011-431f-e878-636c1e8832fb@intel.com>
Date: Mon, 31 Jul 2017 09:35:53 -0700
From: Dave Jiang <dave.jiang@...el.com>
To: Vinod Koul <vinod.koul@...el.com>,
Abhishek Sahu <absahu@...eaurora.org>
Cc: "andy.gross@...aro.org" <andy.gross@...aro.org>,
"david.brown@...aro.org" <david.brown@...aro.org>,
"Williams, Dan J" <dan.j.williams@...el.com>,
"linux-arm-msm@...r.kernel.org" <linux-arm-msm@...r.kernel.org>,
"linux-soc@...r.kernel.org" <linux-soc@...r.kernel.org>,
"dmaengine@...r.kernel.org" <dmaengine@...r.kernel.org>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH v2 1/3] dmaengine: add DMA_PREP_CMD for non-Data
descriptors.
On 07/31/2017 05:34 AM, Vinod Koul wrote:
> On Fri, Jul 28, 2017 at 09:38:56PM +0530, Abhishek Sahu wrote:
>> On 2017-07-19 17:48, Abhishek Sahu wrote:
>>> On 2017-07-19 15:37, Vinod Koul wrote:
>>>> On Mon, Jun 26, 2017 at 06:19:27PM +0530, Abhishek Sahu wrote:
>>>>> Some of the DMA controllers are capable of issuing the commands
>>>>> to peripheral by the DMA. These commands can be list of register
>>>>> reads/writes and its different from normal data reads/writes.
>>>>> This patch adds new flag DMA_PREP_CMD in DMA API which tells
>>>>> the driver that the data passed to DMA API is in command format
>>>>> and DMA driver will form descriptor in the required format.
>>>>>
>>>>> This flag can be used by any DMA controller driver which requires
>>>>> special handling for non-Data descriptors.
>>>>
>>>> Please add Documentation for this new flag in
>>>> Documentation/dmaengine/provider.txt
>>>>
>>>
>>> Sure. I will add update the documentation in v3.
>>>
>>>>>
>>>>> Signed-off-by: Abhishek Sahu <absahu@...eaurora.org>
>>>>> ---
>>>>> include/linux/dmaengine.h | 3 +++
>>>>> 1 file changed, 3 insertions(+)
>>>>>
>>>>> diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h
>>>>> index 5336808..bbc297e 100644
>>>>> --- a/include/linux/dmaengine.h
>>>>> +++ b/include/linux/dmaengine.h
>>>>> @@ -186,6 +186,8 @@ struct dma_interleaved_template {
>>>>> * on the result of this operation
>>>>> * @DMA_CTRL_REUSE: client can reuse the descriptor and submit again
>>>>> till
>>>>> * cleared or freed
>>>>> + * @DMA_PREP_CMD: tell the driver that the data passed to DMA API is
>>>>> in command
>>>>> + * format and it will be used for configuring the peripheral
>>>>> registers.
>>>>
>>>> Can you explain what is command format..?
>>>>
>>>
>>> The command format is not generic and its format will be dependent
>>> upon DMA engine. The client drivers will give data in its own
>>> command formats and this flag will be passed to DMA API’s to do
>>> the parsing according to its own command format.
>>>
>>> Currently this flag description and name is inclined towards
>>> Qualcomm BAM DMA command flag. We want to make this flag as
>>> generic one so require your suggestion regarding this.
>>> Will renaming this flag as DMA_PREP_NON_DATA or
>>> DMA_PREP_CUSTOM make it more generic?
>>>
>>
>> can we use same flag name DMA_PREP_CMD or should we
>> go for some other name?
>
> Are you asking for using DMA_PREP_CMD, for that I think should be ok
>
> If you asking about adding a new flag with DMA_PREP_CMD, then it would no
Vinod, I wonder if we should introduce a DMA_PREP_CUSTOM flag and then
reserve X number of upper flag bits to vendor specified that only they
care about. That way they can define whatever they want in the upper
bits if they need it.
>
>>
>>>>> */
>>>>> enum dma_ctrl_flags {
>>>>> DMA_PREP_INTERRUPT = (1 << 0),
>>>>> @@ -195,6 +197,7 @@ enum dma_ctrl_flags {
>>>>> DMA_PREP_CONTINUE = (1 << 4),
>>>>> DMA_PREP_FENCE = (1 << 5),
>>>>> DMA_CTRL_REUSE = (1 << 6),
>>>>> + DMA_PREP_CMD = (1 << 7),
>>>>> };
>>>>>
>>
>>
>> --
>> Abhishek Sahu
>
Powered by blists - more mailing lists