linux-kernel - [PATCH] media: imx-jpeg: Remove soft reset between frames encoding

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-Id: <20210927185719.3094796-1-mirela.rabulea@nxp.com>
Date:   Mon, 27 Sep 2021 21:57:19 +0300
From:   Mirela Rabulea <mirela.rabulea@....com>
To:     mchehab@...nel.org, hverkuil-cisco@...all.nl,
        robert.chiras@....com, laurentiu.palcu@....com
Cc:     linux-imx@....com, shawnguo@...nel.org, s.hauer@...gutronix.de,
        kernel@...gutronix.de, festevam@...il.com,
        linux-media@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
        linux-kernel@...r.kernel.org
Subject: [PATCH] media: imx-jpeg: Remove soft reset between frames encoding

Remove soft reset between frames, which was merely a workaround,
actually replace it with something less aggressive: clearing of jpeg stream
buffer pointer, via STM_CTRL[BITBUF_PTR_CLR].
According to the reference manual, the BITBUF_PTR_CLR is performed on
context switch, but, per hardware team recommendation, it is also working
without context switch, which is the case for this driver, for now.

Without soft reset and without BITBUF_PTR_CLR, successive encoding jobs
continue from where the previous left off, basically writing past the
v4l2 buffer.

Signed-off-by: Mirela Rabulea <mirela.rabulea@....com>
Reviewed-by: Robert Chiras <robert.chiras@....com>
Reviewed-by: Laurentiu Palcu <laurentiu.palcu@....nxp.com>
---
 drivers/media/platform/imx-jpeg/mxc-jpeg.c | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/drivers/media/platform/imx-jpeg/mxc-jpeg.c b/drivers/media/platform/imx-jpeg/mxc-jpeg.c
index bd19a4b1bf08..f18b67453152 100644
--- a/drivers/media/platform/imx-jpeg/mxc-jpeg.c
+++ b/drivers/media/platform/imx-jpeg/mxc-jpeg.c
@@ -817,6 +817,7 @@ static void mxc_jpeg_config_dec_desc(struct vb2_buffer *out_buf,
 	img_fmt = mxc_jpeg_fourcc_to_imgfmt(q_data_cap->fmt->fourcc);
 	desc->stm_ctrl &= ~STM_CTRL_IMAGE_FORMAT(0xF); /* clear image format */
 	desc->stm_ctrl |= STM_CTRL_IMAGE_FORMAT(img_fmt);
+	desc->stm_ctrl |= STM_CTRL_BITBUF_PTR_CLR(1);
 	desc->line_pitch = q_data_cap->bytesperline[0];
 	mxc_jpeg_addrs(desc, dst_buf, src_buf, 0);
 	mxc_jpeg_set_bufsize(desc, ALIGN(vb2_plane_size(src_buf, 0), 1024));
@@ -843,6 +844,7 @@ static void mxc_jpeg_config_dec_desc(struct vb2_buffer *out_buf,
 	cfg_desc->imgsize |= MXC_JPEG_MIN_HEIGHT;
 	cfg_desc->line_pitch = MXC_JPEG_MIN_WIDTH * 2;
 	cfg_desc->stm_ctrl = STM_CTRL_IMAGE_FORMAT(MXC_JPEG_YUV422);
+	cfg_desc->stm_ctrl |= STM_CTRL_BITBUF_PTR_CLR(1);
 	cfg_desc->stm_bufbase = cfg_stream_handle;
 	cfg_desc->stm_bufsize = ALIGN(*cfg_size, 1024);
 	print_descriptor_info(jpeg->dev, cfg_desc);
@@ -886,6 +888,7 @@ static void mxc_jpeg_config_enc_desc(struct vb2_buffer *out_buf,
 	cfg_desc->stm_bufsize = 0x0;
 	cfg_desc->imgsize = 0;
 	cfg_desc->stm_ctrl = STM_CTRL_CONFIG_MOD(1);
+	cfg_desc->stm_ctrl |= STM_CTRL_BITBUF_PTR_CLR(1);
 
 	desc->next_descpt_ptr = 0; /* end of chain */
 
@@ -900,6 +903,7 @@ static void mxc_jpeg_config_enc_desc(struct vb2_buffer *out_buf,
 		dev_err(jpeg->dev, "No valid image format detected\n");
 	desc->stm_ctrl = STM_CTRL_CONFIG_MOD(0) |
 			 STM_CTRL_IMAGE_FORMAT(img_fmt);
+	desc->stm_ctrl |= STM_CTRL_BITBUF_PTR_CLR(1);
 	mxc_jpeg_addrs(desc, src_buf, dst_buf, 0);
 	dev_dbg(jpeg->dev, "cfg_desc:\n");
 	print_descriptor_info(jpeg->dev, cfg_desc);
@@ -955,11 +959,6 @@ static void mxc_jpeg_device_run(void *priv)
 		return;
 	}
 
-	/*
-	 * TODO: this reset should be removed, once we figure out
-	 * how to overcome hardware issues both on encoder and decoder
-	 */
-	mxc_jpeg_sw_reset(reg);
 	mxc_jpeg_enable(reg);
 	mxc_jpeg_set_l_endian(reg, 1);
 
-- 
2.25.1