Thread (33 messages) 33 messages, 2 authors, 2026-01-05

Re: [PATCH v11 02/12] media: mediatek: jpeg: fix jpeg buffer payload setting

From: Kyrie Wu (吴晗) <hidden>
Date: 2025-12-25 05:48:52
Also in: linux-devicetree, linux-media, linux-mediatek, lkml

On Tue, 2025-12-16 at 16:15 -0500, Nicolas Dufresne wrote:
Hi.

Le mardi 02 décembre 2025 à 17:47 +0800, Kyrie Wu a écrit :
quoted
For multi-core jpegdec, if one of hws gets the event of resolution
changing, the payload size, representing the size of Y/C data,
-> "gets the resolution change event, ..."
quoted
needed to change. But others hws are decoding at the same time and
needed -> needs,               hws -> cores ?

It this specific to decoders or any type of cores ?
quoted
it can not be changed immediately, which results that the payload
                                                    in the playload
quoted
size is not equal to the real buffer length of the hw's, which
occurred
size to not match the real buffer lenght for the ... hw ?
quoted
resolution changing and a warnning call trace will print.
Can't parse. You can probably split that large sentence, and it needs
to be
rework.
quoted
So the setting of payload size must less than the real buffer
length
to remove the warnning logs.

Fixes: 0fa49df4222f ("media: mtk-jpegdec: support jpegdec multi-
hardware")
Don't add blank line in tags please.
quoted
Signed-off-by: Kyrie Wu <redacted>
---
 .../platform/mediatek/jpeg/mtk_jpeg_core.c    | 19 ++++++++++++++-
----
 1 file changed, 14 insertions(+), 5 deletions(-)
diff --git a/drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c
b/drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c
index 6a7e01130f1c..0cf3dc5407e4 100644
--- a/drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c
+++ b/drivers/media/platform/mediatek/jpeg/mtk_jpeg_core.c
@@ -709,6 +709,7 @@ static int mtk_jpeg_buf_prepare(struct
vb2_buffer *vb)
 	struct mtk_jpeg_ctx *ctx = vb2_get_drv_priv(vb->vb2_queue);
 	struct mtk_jpeg_q_data *q_data = NULL;
 	struct v4l2_plane_pix_format plane_fmt = {};
+	unsigned long max_size;
size_t ?
quoted
 	int i;
 
 	q_data = mtk_jpeg_get_q_data(ctx, vb->vb2_queue->type);
@@ -717,12 +718,20 @@ static int mtk_jpeg_buf_prepare(struct
vb2_buffer *vb)
 
 	for (i = 0; i < q_data->fmt->colplanes; i++) {
 		plane_fmt = q_data->pix_mp.plane_fmt[i];
+		max_size = plane_fmt.sizeimage;
+
 		if (ctx->enable_exif &&
-		    q_data->fmt->fourcc == V4L2_PIX_FMT_JPEG)
-			vb2_set_plane_payload(vb, i,
plane_fmt.sizeimage +
-					      MTK_JPEG_MAX_EXIF_SIZE);
-		else
-			vb2_set_plane_payload(vb, i, 
plane_fmt.sizeimage);
+			q_data->fmt->fourcc == V4L2_PIX_FMT_JPEG) {
+			max_size += MTK_JPEG_MAX_EXIF_SIZE;
+
+			vb2_set_plane_payload(vb, i,
+					      MIN(vb->planes[i].length,
+						  max_size));
This is still not quite right. sizeimage, straight from s_fmt should
already
account for the EXIF headers. If enable_exif is unknown at the
moment, then you
should just always include that space. This way, the buffer length
will only be
bigger if userspace asked for more at allocation time, and wil never
be smaller
then this.

The point of sizeimage it to allow userspace allocate externally the
right size,
this driver failed at that task it seems.

Stepping back a little, I don't even understand why you set the
payload size
like this here. Why isn't the true payload size written once, when
the buffer is
encoded ? Are you using bytesused to tell the HW how much space can
be written
too ? Just give it the full space, wich is the allocated size
(length).

Nicolas
Dear Nicolas,

The value, max_size = plane_fmt.sizeimage, recorded the current size,
but if the resolution changed to a smaller one, the sizeimage will
bigger
than the vb->planes[i].length, in the former sw flow, the bigger value
will set to payload size and caused a warnning call trace printed.

Here are the answers of your questions:
1. The enable_exif is a
determined value, which is set in s_ctrl before 
jpeg decoder/encoder starting.

2. If enable_exif is set as true, the buffer planes length is equal to
sizeimage + MTK_JPEG_MAX_EXIF_SIZE.

3. The reason, Why isn't the true payload size written once,
is due to this setting is used to avoid the condition descripted in
commit message. And it will set the true value in irq handler.

Thanks.

Regards,
Kyrie.
quoted
+		} else {
+			vb2_set_plane_payload(vb, i,
+					      MIN(plane_fmt.sizeimage,
+						  vb-
quoted
planes[i].length));
+		}
 	}
 
 	return 0;
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help