From: Stefan Klug <stefan.klug@ideasonboard.com>
To: linux-media@vger.kernel.org,
Jacopo Mondi <jacopo.mondi@ideasonboard.com>,
Dafna Hirschfeld <dafna@fastmail.com>,
Laurent Pinchart <laurent.pinchart@ideasonboard.com>,
Mauro Carvalho Chehab <mchehab@kernel.org>,
Heiko Stuebner <heiko@sntech.de>
Cc: Stefan Klug <stefan.klug@ideasonboard.com>,
linux-rockchip@lists.infradead.org,
linux-arm-kernel@lists.infradead.org,
linux-kernel@vger.kernel.org
Subject: [PATCH] media: rkisp1: Improve frame sequence correctness on stats and params buffers
Date: Mon, 8 Sep 2025 11:41:48 +0200 [thread overview]
Message-ID: <20250908094200.48002-1-stefan.klug@ideasonboard.com> (raw)
On the rkisp1 (in my case on a NXP i.MX8 M Plus) the ISP interrupt
handler is sometimes called with RKISP1_CIF_ISP_V_START (start of frame)
and RKISP1_CIF_ISP_FRAME (end of frame) being set at the same time. In
commit 8524fa22fd2f ("media: staging: rkisp1: isp: add a warning and
debugfs var for irq delay") a warning was added for that. There are two
cases where this condition can occur:
1. The v-sync and the frame-end belong to the same frame. This means,
the irq was heavily delayed and the warning is likely appropriate.
2. The v-sync belongs to the next frame. This can happen if the vertical
blanking between two frames is very short.
The current code always handles case 1 although case 2 is in my
experience the more common case and happens in regular usage. This leads
to incorrect sequence numbers on stats and params buffers which in turn
breaks the regulation in user space. Fix that by adding a frame_active
flag to distinguish between these cases and handle the start of frame
either at the beginning or at the end of the rkisp1_isp_isr().
Signed-off-by: Stefan Klug <stefan.klug@ideasonboard.com>
---
.../platform/rockchip/rkisp1/rkisp1-common.h | 1 +
.../media/platform/rockchip/rkisp1/rkisp1-isp.c | 17 +++++++++++++----
2 files changed, 14 insertions(+), 4 deletions(-)
diff --git a/drivers/media/platform/rockchip/rkisp1/rkisp1-common.h b/drivers/media/platform/rockchip/rkisp1/rkisp1-common.h
index ca952fd0829b..adf23416de9a 100644
--- a/drivers/media/platform/rockchip/rkisp1/rkisp1-common.h
+++ b/drivers/media/platform/rockchip/rkisp1/rkisp1-common.h
@@ -222,6 +222,7 @@ struct rkisp1_isp {
struct media_pad pads[RKISP1_ISP_PAD_MAX];
const struct rkisp1_mbus_info *sink_fmt;
__u32 frame_sequence;
+ bool frame_active;
};
/*
diff --git a/drivers/media/platform/rockchip/rkisp1/rkisp1-isp.c b/drivers/media/platform/rockchip/rkisp1/rkisp1-isp.c
index 8c29a1c9309a..1469075b2d45 100644
--- a/drivers/media/platform/rockchip/rkisp1/rkisp1-isp.c
+++ b/drivers/media/platform/rockchip/rkisp1/rkisp1-isp.c
@@ -965,6 +965,7 @@ static int rkisp1_isp_s_stream(struct v4l2_subdev *sd, int enable)
}
isp->frame_sequence = -1;
+ isp->frame_active = false;
sd_state = v4l2_subdev_lock_and_get_active_state(sd);
@@ -1086,12 +1087,15 @@ void rkisp1_isp_unregister(struct rkisp1_device *rkisp1)
* Interrupt handlers
*/
-static void rkisp1_isp_queue_event_sof(struct rkisp1_isp *isp)
+static void rkisp1_isp_sof(struct rkisp1_isp *isp)
{
struct v4l2_event event = {
.type = V4L2_EVENT_FRAME_SYNC,
};
+ isp->frame_sequence++;
+ isp->frame_active = true;
+
event.u.frame_sync.frame_sequence = isp->frame_sequence;
v4l2_event_queue(isp->sd.devnode, &event);
}
@@ -1112,14 +1116,15 @@ irqreturn_t rkisp1_isp_isr(int irq, void *ctx)
rkisp1_write(rkisp1, RKISP1_CIF_ISP_ICR, status);
/* Vertical sync signal, starting generating new frame */
- if (status & RKISP1_CIF_ISP_V_START) {
- rkisp1->isp.frame_sequence++;
- rkisp1_isp_queue_event_sof(&rkisp1->isp);
+ if (status & RKISP1_CIF_ISP_V_START && !rkisp1->isp.frame_active) {
+ status &= ~RKISP1_CIF_ISP_V_START;
+ rkisp1_isp_sof(&rkisp1->isp);
if (status & RKISP1_CIF_ISP_FRAME) {
WARN_ONCE(1, "irq delay is too long, buffers might not be in sync\n");
rkisp1->debug.irq_delay++;
}
}
+
if (status & RKISP1_CIF_ISP_PIC_SIZE_ERROR) {
/* Clear pic_size_error */
isp_err = rkisp1_read(rkisp1, RKISP1_CIF_ISP_ERR);
@@ -1138,6 +1143,7 @@ irqreturn_t rkisp1_isp_isr(int irq, void *ctx)
if (status & RKISP1_CIF_ISP_FRAME) {
u32 isp_ris;
+ rkisp1->isp.frame_active = false;
rkisp1->debug.complete_frames++;
/* New frame from the sensor received */
@@ -1152,5 +1158,8 @@ irqreturn_t rkisp1_isp_isr(int irq, void *ctx)
rkisp1_params_isr(rkisp1);
}
+ if (status & RKISP1_CIF_ISP_V_START && !rkisp1->isp.frame_active)
+ rkisp1_isp_sof(&rkisp1->isp);
+
return IRQ_HANDLED;
}
--
2.48.1
next reply other threads:[~2025-09-08 12:04 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-09-08 9:41 Stefan Klug [this message]
2025-09-15 16:55 ` [PATCH] media: rkisp1: Improve frame sequence correctness on stats and params buffers Jacopo Mondi
2025-09-16 7:49 ` Stefan Klug
2025-09-16 8:11 ` Jacopo Mondi
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=20250908094200.48002-1-stefan.klug@ideasonboard.com \
--to=stefan.klug@ideasonboard.com \
--cc=dafna@fastmail.com \
--cc=heiko@sntech.de \
--cc=jacopo.mondi@ideasonboard.com \
--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=linux-rockchip@lists.infradead.org \
--cc=mchehab@kernel.org \
/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