linux-media.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Tomi Valkeinen <tomi.valkeinen@ti.com>
To: <linux-media@vger.kernel.org>, Benoit Parrot <bparrot@ti.com>,
	Laurent Pinchart <laurent.pinchart@ideasonboard.com>,
	Hans Verkuil <hverkuil@xs4all.nl>
Cc: Mauro Carvalho Chehab <mchehab@kernel.org>,
	Tomi Valkeinen <tomi.valkeinen@ti.com>
Subject: [PATCH v3 01/19] media: ti-vpe: cal: fix DMA memory corruption
Date: Wed, 25 Mar 2020 14:14:52 +0200	[thread overview]
Message-ID: <20200325121510.25923-2-tomi.valkeinen@ti.com> (raw)
In-Reply-To: <20200325121510.25923-1-tomi.valkeinen@ti.com>

When the CAL driver stops streaming, it will shut everything down
without waiting for the current frame to finish. This leaves the CAL DMA
in a slightly undefined state, and when CAL DMA is enabled when the
stream is started the next time, the old DMA transfer will continue.

It is not clear if the old DMA transfer continues with the exact
settings of the original transfer, or is it a mix of old and new
settings, but in any case the end result is memory corruption as the
destination memory address is no longer valid.

I could not find any way to ensure that any old DMA transfer would be
discarded, except perhaps full CAL reset. But we cannot do a full reset
when one port is getting enabled, as that would reset both ports.

This patch tries to make sure that the DMA transfer is finished properly
when the stream is being stopped. I say "tries", as, as mentioned above,
I don't see a way to force the DMA transfer to finish. I believe this
fixes the corruptions for normal cases, but if for some reason the DMA
of the final frame would stall a lot, resulting in timeout in the code
waiting for the DMA to finish, we'll again end up with unfinished DMA
transfer. However, I don't know what could cause such a timeout.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Reviewed-by: Benoit Parrot <bparrot@ti.com>
---
 drivers/media/platform/ti-vpe/cal.c | 32 +++++++++++++++++++++++++++++
 1 file changed, 32 insertions(+)

diff --git a/drivers/media/platform/ti-vpe/cal.c b/drivers/media/platform/ti-vpe/cal.c
index 6c8f3702eac0..9dd6de14189b 100644
--- a/drivers/media/platform/ti-vpe/cal.c
+++ b/drivers/media/platform/ti-vpe/cal.c
@@ -412,6 +412,8 @@ struct cal_ctx {
 	struct cal_buffer	*cur_frm;
 	/* Pointer pointing to next v4l2_buffer */
 	struct cal_buffer	*next_frm;
+
+	bool dma_act;
 };
 
 static const struct cal_fmt *find_format_by_pix(struct cal_ctx *ctx,
@@ -942,6 +944,7 @@ static void csi2_lane_config(struct cal_ctx *ctx)
 
 static void csi2_ppi_enable(struct cal_ctx *ctx)
 {
+	reg_write(ctx->dev, CAL_CSI2_PPI_CTRL(ctx->csi2_port), BIT(3));
 	reg_write_field(ctx->dev, CAL_CSI2_PPI_CTRL(ctx->csi2_port),
 			CAL_GEN_ENABLE, CAL_CSI2_PPI_CTRL_IF_EN_MASK);
 }
@@ -1204,15 +1207,25 @@ static irqreturn_t cal_irq(int irq_cal, void *data)
 		if (isportirqset(irqst2, 1)) {
 			ctx = dev->ctx[0];
 
+			spin_lock(&ctx->slock);
+			ctx->dma_act = false;
+
 			if (ctx->cur_frm != ctx->next_frm)
 				cal_process_buffer_complete(ctx);
+
+			spin_unlock(&ctx->slock);
 		}
 
 		if (isportirqset(irqst2, 2)) {
 			ctx = dev->ctx[1];
 
+			spin_lock(&ctx->slock);
+			ctx->dma_act = false;
+
 			if (ctx->cur_frm != ctx->next_frm)
 				cal_process_buffer_complete(ctx);
+
+			spin_unlock(&ctx->slock);
 		}
 	}
 
@@ -1228,6 +1241,7 @@ static irqreturn_t cal_irq(int irq_cal, void *data)
 			dma_q = &ctx->vidq;
 
 			spin_lock(&ctx->slock);
+			ctx->dma_act = true;
 			if (!list_empty(&dma_q->active) &&
 			    ctx->cur_frm == ctx->next_frm)
 				cal_schedule_next_buffer(ctx);
@@ -1239,6 +1253,7 @@ static irqreturn_t cal_irq(int irq_cal, void *data)
 			dma_q = &ctx->vidq;
 
 			spin_lock(&ctx->slock);
+			ctx->dma_act = true;
 			if (!list_empty(&dma_q->active) &&
 			    ctx->cur_frm == ctx->next_frm)
 				cal_schedule_next_buffer(ctx);
@@ -1711,10 +1726,27 @@ static void cal_stop_streaming(struct vb2_queue *vq)
 	struct cal_ctx *ctx = vb2_get_drv_priv(vq);
 	struct cal_dmaqueue *dma_q = &ctx->vidq;
 	struct cal_buffer *buf, *tmp;
+	unsigned long timeout;
 	unsigned long flags;
 	int ret;
+	bool dma_act;
 
 	csi2_ppi_disable(ctx);
+
+	/* wait for stream and dma to finish */
+	dma_act = true;
+	timeout = jiffies + msecs_to_jiffies(500);
+	while (dma_act && time_before(jiffies, timeout)) {
+		msleep(50);
+
+		spin_lock_irqsave(&ctx->slock, flags);
+		dma_act = ctx->dma_act;
+		spin_unlock_irqrestore(&ctx->slock, flags);
+	}
+
+	if (dma_act)
+		ctx_err(ctx, "failed to disable dma cleanly\n");
+
 	disable_irqs(ctx);
 	csi2_phy_deinit(ctx);
 
-- 
Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki.
Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki


  reply	other threads:[~2020-03-25 12:15 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-03-25 12:14 [PATCH v3 00/19] CAL fixes and improvements Tomi Valkeinen
2020-03-25 12:14 ` Tomi Valkeinen [this message]
2020-03-25 12:14 ` [PATCH v3 02/19] media: ti-vpe: cal: improve enable_irqs Tomi Valkeinen
2020-03-25 12:14 ` [PATCH v3 03/19] media: ti-vpe: cal: fix use of wrong macro Tomi Valkeinen
2020-03-25 12:14 ` [PATCH v3 04/19] media: ti-vpe: cal: use runtime_resume for errata handling Tomi Valkeinen
2020-03-25 12:14 ` [PATCH v3 05/19] media: ti-vpe: cal: drop cal_runtime_get/put Tomi Valkeinen
2020-03-25 12:14 ` [PATCH v3 06/19] media: ti-vpe: cal: catch error irqs and print errors Tomi Valkeinen
2020-03-25 12:14 ` [PATCH v3 07/19] media: ti-vpe: cal: print errors on timeouts Tomi Valkeinen
2020-03-25 12:14 ` [PATCH v3 08/19] media: ti-vpe: cal: simplify irq handling Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 09/19] media: ti-vpe: cal: remove useless CAL_GEN_* macros Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 10/19] media: ti-vpe: cal: remove useless IRQ defines Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 11/19] media: ti-vpe: cal: use reg_write_field Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 12/19] media: ti-vpe: cal: cleanup CIO power enable/disable Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 13/19] media: ti-vpe: cal: fix dummy read to phy Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 14/19] media: ti-vpe: cal: program number of lines properly Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 15/19] media: ti-vpe: cal: set DMA max seg size Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 16/19] media: ti-vpe: cal: move code to separate functions Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 17/19] media: ti-vpe: cal: improve wait for CIO resetdone Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 18/19] media: ti-vpe: cal: improve wait for stop-state Tomi Valkeinen
2020-03-25 12:15 ` [PATCH v3 19/19] media: ti-vpe: cal: fix stop state timeout Tomi Valkeinen

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=20200325121510.25923-2-tomi.valkeinen@ti.com \
    --to=tomi.valkeinen@ti.com \
    --cc=bparrot@ti.com \
    --cc=hverkuil@xs4all.nl \
    --cc=laurent.pinchart@ideasonboard.com \
    --cc=linux-media@vger.kernel.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;
as well as URLs for NNTP newsgroup(s).