mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Guoniu Zhou <guoniu.zhou@oss.nxp.com>
To: Laurent Pinchart <laurent.pinchart@ideasonboard.com>,
	 Frank Li <Frank.Li@nxp.com>,
	Mauro Carvalho Chehab <mchehab@kernel.org>,
	 Sascha Hauer <s.hauer@pengutronix.de>,
	 Pengutronix Kernel Team <kernel@pengutronix.de>,
	 Fabio Estevam <festevam@gmail.com>,
	Rob Herring <robh@kernel.org>,
	 Krzysztof Kozlowski <krzk+dt@kernel.org>,
	 Conor Dooley <conor+dt@kernel.org>,
	 Loic Poulain <loic.poulain@oss.qualcomm.com>
Cc: linux-media@vger.kernel.org, imx@lists.linux.dev,
	 linux-arm-kernel@lists.infradead.org,
	linux-kernel@vger.kernel.org,  devicetree@vger.kernel.org,
	Guoniu Zhou <guoniu.zhou@oss.nxp.com>
Subject: [PATCH v4 11/11] media: imx8-isi: Enable RAW10/12/14 output LSB alignment on i.MX952
Date: Tue, 29 Sep 2026 16:52:32 +0800	[thread overview]
Message-ID: <20260929-isi_imx952-v4-11-e69f89c7c651@oss.nxp.com> (raw)
In-Reply-To: <20260929-isi_imx952-v4-0-e69f89c7c651@oss.nxp.com>

Before i.MX952, the ISI RAW10/12/14 output formats are always
MSB-aligned, which is not compatible with the V4L2 RAW format
definition that requires LSB alignment.

From i.MX952 onwards, the ISI introduces CHNL_CTRL_RAW_OUT_LSB_ALIGN
to select whether RAW10/12/14 outputs are MSB or LSB aligned. Enable
it so the driver produces correct V4L2-compatible LSB-aligned output.

Reviewed-by: Frank Li <Frank.Li@nxp.com>
Signed-off-by: Guoniu Zhou <guoniu.zhou@oss.nxp.com>
---
Changes in v4:
- Split RAW_IN_MSB_ALIGN into a separate preceding patch
- Split RAW format table and RAW14 changes into separate preceding patches
- Move isi_out_raw_shifted() to regs.h in the preceding workaround patch
- Rewrite commit message

Changes in v3:
- Use CHNL_IMG_CTRL_FORMAT_MASK instead of the removed per-platform format_mask (Laurent)
- Derive raw_out_lsb from the model field instead of dedicated platform data (Laurent)
- Clarify the CHNL_IMG_CTRL_FORMAT_RAW14 comment: the format is
  documented on i.MX{8ULP,91,93,95,952} but only used on i.MX952
- Update the mxc_isi_channel_set_output_format() comment wording
- Rewrite commit message
- Add Reviewed-by tag from Frank
---
 drivers/media/platform/nxp/imx8-isi/imx8-isi-hw.c  | 37 +++++++++++++++++++---
 .../media/platform/nxp/imx8-isi/imx8-isi-regs.h    |  5 +--
 2 files changed, 36 insertions(+), 6 deletions(-)

diff --git a/drivers/media/platform/nxp/imx8-isi/imx8-isi-hw.c b/drivers/media/platform/nxp/imx8-isi/imx8-isi-hw.c
index b3f32b9796d5..6733776e72aa 100644
--- a/drivers/media/platform/nxp/imx8-isi/imx8-isi-hw.c
+++ b/drivers/media/platform/nxp/imx8-isi/imx8-isi-hw.c
@@ -311,12 +311,24 @@ static void mxc_isi_channel_set_panic_threshold(struct mxc_isi_pipe *pipe)
  * RAW samples from either the high or low bits, as both LSB and MSB
  * alignment can be selected in the gasket.
  *
- * On i.MX95, the CSI formatter provides MSB-aligned RAW data, so the ISI
- * must select RAW samples from the high bits of the Pixel Link bus.
+ * Starting with i.MX95, the CSI formatter provides MSB-aligned RAW data,
+ * so the ISI must select RAW samples from the high bits of the Pixel Link
+ * bus.
  */
 static bool isi_raw_in_msb(const struct mxc_isi_plat_data *pdata)
 {
-	return pdata->model == MXC_ISI_IMX95;
+	return pdata->model == MXC_ISI_IMX95 ||
+	       pdata->model == MXC_ISI_IMX952;
+}
+
+/*
+ * Starting with i.MX952, the ISI can output RAW10/12/14 formats with LSB
+ * alignment, matching the V4L2 requirement. Older platforms lack this
+ * capability and need the RAW16 workaround.
+ */
+static bool isi_raw_out_lsb(const struct mxc_isi_plat_data *pdata)
+{
+	return pdata->model == MXC_ISI_IMX952;
 }
 
 static void mxc_isi_channel_set_control(struct mxc_isi_pipe *pipe,
@@ -377,6 +389,13 @@ static void mxc_isi_channel_set_control(struct mxc_isi_pipe *pipe,
 	if (isi_raw_in_msb(pipe->isi->pdata))
 		val |= CHNL_CTRL_RAW_IN_MSB_ALIGN;
 
+	/*
+	 * Enable RAW10/12/14 output LSB alignment to match the
+	 * V4L2 requirement that RAW formats are LSB-aligned.
+	 */
+	if (isi_raw_out_lsb(pipe->isi->pdata))
+		val |= CHNL_CTRL_RAW_OUT_LSB_ALIGN;
+
 	mxc_isi_write(pipe, CHNL_CTRL, val);
 
 	mutex_unlock(&pipe->lock);
@@ -430,13 +449,23 @@ void mxc_isi_channel_set_output_format(struct mxc_isi_pipe *pipe,
 				       const struct mxc_isi_format_info *info,
 				       struct v4l2_pix_format_mplane *format)
 {
+	const struct mxc_isi_plat_data *pdata = pipe->isi->pdata;
 	u32 val;
 	u32 fmt;
 
 	/* set outbuf format */
 	dev_dbg(pipe->isi->dev, "output format %p4cc", &format->pixelformat);
 
-	fmt = isi_out_raw_shifted(info->isi_out_format)
+	/*
+	 * Before i.MX952, the ISI shifts the 10/12/14-bit formats left
+	 * by 6, 4 and 2 bits when using CHNL_IMG_CTRL_FORMAT_RAW10/12/14
+	 * respectively, to align the bits to the left and pad with zeros in
+	 * the LSBs. The corresponding V4L2 formats are however right-aligned,
+	 * we have to use CHNL_IMG_CTRL_FORMAT_RAW16 to avoid the left shift.
+	 * Starting with i.MX952, the ISI supports RAW10/12/14 LSB output
+	 * alignment, so skip the above workaround.
+	 */
+	fmt = !isi_raw_out_lsb(pdata) && isi_out_raw_shifted(info->isi_out_format)
 	    ? CHNL_IMG_CTRL_FORMAT_RAW16
 	    : info->isi_out_format;
 
diff --git a/drivers/media/platform/nxp/imx8-isi/imx8-isi-regs.h b/drivers/media/platform/nxp/imx8-isi/imx8-isi-regs.h
index d2b8a26b7c21..075fc6848213 100644
--- a/drivers/media/platform/nxp/imx8-isi/imx8-isi-regs.h
+++ b/drivers/media/platform/nxp/imx8-isi/imx8-isi-regs.h
@@ -20,7 +20,8 @@
 #define CHNL_CTRL_CHAIN_BUF_NO_CHAIN				0
 #define CHNL_CTRL_CHAIN_BUF_2_CHAIN				1
 #define CHNL_CTRL_SW_RST					BIT(24)
-#define CHNL_CTRL_RAW_IN_MSB_ALIGN				BIT(22)	/* i.MX{91,93,95} */
+#define CHNL_CTRL_RAW_OUT_LSB_ALIGN				BIT(23)	/* i.MX952 only */
+#define CHNL_CTRL_RAW_IN_MSB_ALIGN				BIT(22)	/* i.MX{91,93,95,952} */
 /*
  * CHNL_CTRL_BLANK_PXL: i.MX8{QM,QXP} only
  * CHNL_CTRL_VC_ID_1, CHNL_CTRL_VC_ID_1_MASK: i.MX{95,952} only
@@ -90,7 +91,7 @@
 #define CHNL_IMG_CTRL_FORMAT_YUV420_3P10P			0x3a
 #define CHNL_IMG_CTRL_FORMAT_YUV420_2P12			0x3d
 #define CHNL_IMG_CTRL_FORMAT_YUV420_3P12			0x3e
-#define CHNL_IMG_CTRL_FORMAT_RAW14				0x41	/* i.MX{8ULP,91,93,95} */
+#define CHNL_IMG_CTRL_FORMAT_RAW14				0x41	/* i.MX{8ULP,91,93,95,952}, used on i.MX952 only */
 
 #define CHNL_IMG_CTRL_GBL_ALPHA_VAL(n)				((n) << 16)
 #define CHNL_IMG_CTRL_GBL_ALPHA_VAL_MASK			GENMASK(23, 16)

-- 
2.34.1


      parent reply	other threads:[~2026-09-29  8:49 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-29  8:52 [PATCH v4 00/11] media: imx8-isi: Add i.MX952 ISI support Guoniu Zhou
2026-09-29  8:52 ` [PATCH v4 01/11] media: imx8-isi: Reorder mxc_imx95_data to follow mxc_imx93_data Guoniu Zhou
2026-09-29  8:52 ` [PATCH v4 02/11] media: imx8-isi: Extend output format mask to 7 bits Guoniu Zhou
2026-09-29 14:20   ` Frank Li
2026-09-29  8:52 ` [PATCH v4 03/11] media: imx8-isi: Support mbus_code filtering in format enumeration Guoniu Zhou
2026-09-29 14:22   ` Frank Li
2026-09-29  8:52 ` [PATCH v4 04/11] media: imx8-isi: Reuse mxc_isi_format_enum() in capture " Guoniu Zhou
2026-09-29 14:23   ` Frank Li
2026-09-29  8:52 ` [PATCH v4 05/11] media: imx8-isi: Apply RAW output format workaround at runtime Guoniu Zhou
2026-09-29 14:26   ` Frank Li
2026-09-29  8:52 ` [PATCH v4 06/11] media: imx8-isi: Add RAW14 output format support Guoniu Zhou
2026-09-29 14:28   ` Frank Li
2026-09-29  8:52 ` [PATCH v4 07/11] media: imx8-isi: Select MSB-aligned RAW input from pixel link on i.MX95 Guoniu Zhou
2026-09-29 14:31   ` Frank Li
2026-09-30  2:39     ` G.N. Zhou (OSS)
2026-09-29  8:52 ` [PATCH v4 08/11] media: imx8-isi: Exclude RAW10/12/14 output formats " Guoniu Zhou
2026-09-29 14:33   ` Frank Li
2026-09-29  8:52 ` [PATCH v4 09/11] media: dt-bindings: nxp,imx8-isi: Add i.MX952 ISI compatible string Guoniu Zhou
2026-09-29  8:52 ` [PATCH v4 10/11] media: imx8-isi: Add i.MX952 ISI support Guoniu Zhou
2026-09-29  8:52 ` Guoniu Zhou [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260929-isi_imx952-v4-11-e69f89c7c651@oss.nxp.com \
    --to=guoniu.zhou@oss.nxp.com \
    --cc=Frank.Li@nxp.com \
    --cc=conor+dt@kernel.org \
    --cc=devicetree@vger.kernel.org \
    --cc=festevam@gmail.com \
    --cc=imx@lists.linux.dev \
    --cc=kernel@pengutronix.de \
    --cc=krzk+dt@kernel.org \
    --cc=laurent.pinchart@ideasonboard.com \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-media@vger.kernel.org \
    --cc=loic.poulain@oss.qualcomm.com \
    --cc=mchehab@kernel.org \
    --cc=robh@kernel.org \
    --cc=s.hauer@pengutronix.de \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®