Received-SPF: pass (google.com: domain of linux-kernel+bounces-129895-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
From: Sugar Zhang <sugar.zhang@rock-chips.com>
To: heiko@sntech.de,
	vkoul@kernel.org
Cc: linux-rockchip@lists.infradead.org,
	Sugar Zhang <sugar.zhang@rock-chips.com>,
	dmaengine@vger.kernel.org,
	linux-kernel@vger.kernel.org
Subject: [PATCH v2 1/2] dmaengine: Add support for audio interleaved transfer
Date: Wed,  3 Apr 2024 21:18:22 +0800
Message-Id: <20240403211810.v2.1.I502ea9c86c8403dc5b1f38abf40be8b6ee13c1dc@changeid>
In-Reply-To: <1712150304-60832-1-git-send-email-sugar.zhang@rock-chips.com>
References: <1712150304-60832-1-git-send-email-sugar.zhang@rock-chips.com>
Precedence: bulk

This patch add support for interleaved transfer which used
for interleaved audio or 2d video data transfer.

for audio situation, we add 'nump' for number of period frames.

e.g. combine 2 stream into a union one by 2D dma.

DAI0: 16CH
+-------------------------------------------------------------+
| Frame-1 | Frame-2 | Frame-3 | Frame-4 | ...... Frame-'numf' |
+-------------------------------------------------------------+

DAI1: 16CH
+-------------------------------------------------------------+
| Frame-1 | Frame-2 | Frame-3 | Frame-4 | ...... Frame-'numf' |
+-------------------------------------------------------------+

DAI0 + DAI1: 32CH

+-------------------------------------------------------------+
| DAI0-F1 | DAI1-F1 | DAI0-F2 | DAI1-F2 | ......              |
+-------------------------------------------------------------+
|      Frame-1      |      Frame-2      | ...... Frame-'numf' |

For audio situation, we have buffer_size and period_size,
the 'numf' is the buffer_size. so, we need another one for
period_size, e.g. 'nump'.

| Frame-1 | ~ | Frame-'nump' | ~ | Frame-'nump+1' | ~ |  Frame-'numf' |
|

As the above shown:

each DAI0 transfer 1 Frame, should skip a gap size (DAI1-F1)
each DAI1 transfer 1 Frame, should skip a gap size (DAI0-F1)

So, the interleaved template describe as follows:

DAI0:

struct dma_interleaved_template *xt;

xt->sgl[0].size = DAI0-F1;
xt->sgl[0].icg =  DAI1-F1;
xt->nump = nump; //the period_size in frames
xt->numf = numf; //the buffer_size in frames

DAI1:

struct dma_interleaved_template *xt;

xt->sgl[0].size = DAI1-F1;
xt->sgl[0].icg =  DAI0-F1;
xt->nump = nump; //the period_size in frames
xt->numf = numf; //the buffer_size in frames

Signed-off-by: Sugar Zhang <sugar.zhang@rock-chips.com>
---

Changes in v2:
- Add the pl330 interleaved transfer

 include/linux/dmaengine.h | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/include/linux/dmaengine.h b/include/linux/dmaengine.h
index 752dbde..5263cde 100644
--- a/include/linux/dmaengine.h
+++ b/include/linux/dmaengine.h
@@ -144,6 +144,7 @@ struct data_chunk {
  *		Otherwise, destination is filled contiguously (icg ignored).
  *		Ignored if dst_inc is false.
  * @numf: Number of frames in this template.
+ * @nump: Number of period frames in this template.
  * @frame_size: Number of chunks in a frame i.e, size of sgl[].
  * @sgl: Array of {chunk,icg} pairs that make up a frame.
  */
@@ -156,6 +157,7 @@ struct dma_interleaved_template {
 	bool src_sgl;
 	bool dst_sgl;
 	size_t numf;
+	size_t nump;
 	size_t frame_size;
 	struct data_chunk sgl[];
 };
-- 
2.7.4