All of lore.kernel.org
 help / color / mirror / Atom feed
* [Patch 1/1] media: ti-vpe: cal: fix disable_irqs to only the intended target
@ 2020-03-02 13:56 Benoit Parrot
  2020-03-04  8:49 ` Tomi Valkeinen
  2020-03-04 15:35 ` Sasha Levin
  0 siblings, 2 replies; 4+ messages in thread
From: Benoit Parrot @ 2020-03-02 13:56 UTC (permalink / raw)
  To: Hans Verkuil
  Cc: linux-media, devicetree, linux-kernel, Tomi Valkeinen,
	Benoit Parrot, stable

disable_irqs() was mistakenly disabling all interrupts when called.
This cause all port stream to stop even if only stopping one of them.

Cc: stable <stable@vger.kernel.org>
Signed-off-by: Benoit Parrot <bparrot@ti.com>
---
 drivers/media/platform/ti-vpe/cal.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/drivers/media/platform/ti-vpe/cal.c b/drivers/media/platform/ti-vpe/cal.c
index 6e009e479be3..6d4cbb8782ed 100644
--- a/drivers/media/platform/ti-vpe/cal.c
+++ b/drivers/media/platform/ti-vpe/cal.c
@@ -722,16 +722,16 @@ static void enable_irqs(struct cal_ctx *ctx)
 
 static void disable_irqs(struct cal_ctx *ctx)
 {
+	u32 val;
+
 	/* Disable IRQ_WDMA_END 0/1 */
-	reg_write_field(ctx->dev,
-			CAL_HL_IRQENABLE_CLR(2),
-			CAL_HL_IRQ_CLEAR,
-			CAL_HL_IRQ_MASK(ctx->csi2_port));
+	val = 0;
+	set_field(&val, CAL_HL_IRQ_CLEAR, CAL_HL_IRQ_MASK(ctx->csi2_port));
+	reg_write(ctx->dev, CAL_HL_IRQENABLE_CLR(2), val);
 	/* Disable IRQ_WDMA_START 0/1 */
-	reg_write_field(ctx->dev,
-			CAL_HL_IRQENABLE_CLR(3),
-			CAL_HL_IRQ_CLEAR,
-			CAL_HL_IRQ_MASK(ctx->csi2_port));
+	val = 0;
+	set_field(&val, CAL_HL_IRQ_CLEAR, CAL_HL_IRQ_MASK(ctx->csi2_port));
+	reg_write(ctx->dev, CAL_HL_IRQENABLE_CLR(3), val);
 	/* Todo: Add VC_IRQ and CSI2_COMPLEXIO_IRQ handling */
 	reg_write(ctx->dev, CAL_CSI2_VC_IRQENABLE(1), 0);
 }
-- 
2.17.1


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [Patch 1/1] media: ti-vpe: cal: fix disable_irqs to only the intended target
  2020-03-02 13:56 [Patch 1/1] media: ti-vpe: cal: fix disable_irqs to only the intended target Benoit Parrot
@ 2020-03-04  8:49 ` Tomi Valkeinen
  2020-03-04 13:17   ` Benoit Parrot
  2020-03-04 15:35 ` Sasha Levin
  1 sibling, 1 reply; 4+ messages in thread
From: Tomi Valkeinen @ 2020-03-04  8:49 UTC (permalink / raw)
  To: Benoit Parrot, Hans Verkuil; +Cc: linux-media, devicetree, linux-kernel, stable

On 02/03/2020 15:56, Benoit Parrot wrote:
> disable_irqs() was mistakenly disabling all interrupts when called.
> This cause all port stream to stop even if only stopping one of them.
> 
> Cc: stable <stable@vger.kernel.org>
> Signed-off-by: Benoit Parrot <bparrot@ti.com>
> ---
>   drivers/media/platform/ti-vpe/cal.c | 16 ++++++++--------
>   1 file changed, 8 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/media/platform/ti-vpe/cal.c b/drivers/media/platform/ti-vpe/cal.c
> index 6e009e479be3..6d4cbb8782ed 100644
> --- a/drivers/media/platform/ti-vpe/cal.c
> +++ b/drivers/media/platform/ti-vpe/cal.c
> @@ -722,16 +722,16 @@ static void enable_irqs(struct cal_ctx *ctx)
>   
>   static void disable_irqs(struct cal_ctx *ctx)
>   {
> +	u32 val;
> +
>   	/* Disable IRQ_WDMA_END 0/1 */
> -	reg_write_field(ctx->dev,
> -			CAL_HL_IRQENABLE_CLR(2),
> -			CAL_HL_IRQ_CLEAR,
> -			CAL_HL_IRQ_MASK(ctx->csi2_port));
> +	val = 0;
> +	set_field(&val, CAL_HL_IRQ_CLEAR, CAL_HL_IRQ_MASK(ctx->csi2_port));
> +	reg_write(ctx->dev, CAL_HL_IRQENABLE_CLR(2), val);
>   	/* Disable IRQ_WDMA_START 0/1 */
> -	reg_write_field(ctx->dev,
> -			CAL_HL_IRQENABLE_CLR(3),
> -			CAL_HL_IRQ_CLEAR,
> -			CAL_HL_IRQ_MASK(ctx->csi2_port));
> +	val = 0;
> +	set_field(&val, CAL_HL_IRQ_CLEAR, CAL_HL_IRQ_MASK(ctx->csi2_port));
> +	reg_write(ctx->dev, CAL_HL_IRQENABLE_CLR(3), val);
>   	/* Todo: Add VC_IRQ and CSI2_COMPLEXIO_IRQ handling */
>   	reg_write(ctx->dev, CAL_CSI2_VC_IRQENABLE(1), 0);
>   }
> 

I think the above works. But the enable_irqs is broken too, even if it doesn't cause any issues. Both IRQ_SET and IRQ_CLR are not supposed to be "modified" by a read-mod-write operation, but just written to.

The macros used also make the code very difficult to read. Something like this fixes both irq enable and disable, and makes it readable:

diff --git a/drivers/media/platform/ti-vpe/cal.c b/drivers/media/platform/ti-vpe/cal.c
index c8b1290c9e2b..660653031a0b 100644
--- a/drivers/media/platform/ti-vpe/cal.c
+++ b/drivers/media/platform/ti-vpe/cal.c
@@ -707,15 +707,9 @@ static void cal_quickdump_regs(struct cal_dev *dev)
 static void enable_irqs(struct cal_ctx *ctx)
 {
 	/* Enable IRQ_WDMA_END 0/1 */
-	reg_write_field(ctx->dev,
-			CAL_HL_IRQENABLE_SET(2),
-			CAL_HL_IRQ_ENABLE,
-			CAL_HL_IRQ_MASK(ctx->csi2_port));
+	reg_write(ctx->dev, CAL_HL_IRQENABLE_SET(2), 1 << (ctx->csi2_port - 1));
 	/* Enable IRQ_WDMA_START 0/1 */
-	reg_write_field(ctx->dev,
-			CAL_HL_IRQENABLE_SET(3),
-			CAL_HL_IRQ_ENABLE,
-			CAL_HL_IRQ_MASK(ctx->csi2_port));
+	reg_write(ctx->dev, CAL_HL_IRQENABLE_SET(3), 1 << (ctx->csi2_port - 1));
 	/* Todo: Add VC_IRQ and CSI2_COMPLEXIO_IRQ handling */
 	reg_write(ctx->dev, CAL_CSI2_VC_IRQENABLE(1), 0xFF000000);
 }
@@ -723,15 +717,9 @@ static void enable_irqs(struct cal_ctx *ctx)
 static void disable_irqs(struct cal_ctx *ctx)
 {
 	/* Disable IRQ_WDMA_END 0/1 */
-	reg_write_field(ctx->dev,
-			CAL_HL_IRQENABLE_CLR(2),
-			CAL_HL_IRQ_CLEAR,
-			CAL_HL_IRQ_MASK(ctx->csi2_port));
+	reg_write(ctx->dev, CAL_HL_IRQENABLE_CLR(2), 1 << (ctx->csi2_port - 1));
 	/* Disable IRQ_WDMA_START 0/1 */
-	reg_write_field(ctx->dev,
-			CAL_HL_IRQENABLE_CLR(3),
-			CAL_HL_IRQ_CLEAR,
-			CAL_HL_IRQ_MASK(ctx->csi2_port));
+	reg_write(ctx->dev, CAL_HL_IRQENABLE_CLR(3), 1 << (ctx->csi2_port - 1));
 	/* Todo: Add VC_IRQ and CSI2_COMPLEXIO_IRQ handling */
 	reg_write(ctx->dev, CAL_CSI2_VC_IRQENABLE(1), 0);
 }

 Tomi

-- 
Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki.
Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki

^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [Patch 1/1] media: ti-vpe: cal: fix disable_irqs to only the intended target
  2020-03-04  8:49 ` Tomi Valkeinen
@ 2020-03-04 13:17   ` Benoit Parrot
  0 siblings, 0 replies; 4+ messages in thread
From: Benoit Parrot @ 2020-03-04 13:17 UTC (permalink / raw)
  To: Tomi Valkeinen
  Cc: Hans Verkuil, linux-media, devicetree, linux-kernel, stable

Tomi,

Thanks for the review.

Tomi Valkeinen <tomi.valkeinen@ti.com> wrote on Wed [2020-Mar-04 10:49:55 +0200]:
> On 02/03/2020 15:56, Benoit Parrot wrote:
> > disable_irqs() was mistakenly disabling all interrupts when called.
> > This cause all port stream to stop even if only stopping one of them.
> > 
> > Cc: stable <stable@vger.kernel.org>
> > Signed-off-by: Benoit Parrot <bparrot@ti.com>
> > ---
> >   drivers/media/platform/ti-vpe/cal.c | 16 ++++++++--------
> >   1 file changed, 8 insertions(+), 8 deletions(-)
> > 
> > diff --git a/drivers/media/platform/ti-vpe/cal.c b/drivers/media/platform/ti-vpe/cal.c
> > index 6e009e479be3..6d4cbb8782ed 100644
> > --- a/drivers/media/platform/ti-vpe/cal.c
> > +++ b/drivers/media/platform/ti-vpe/cal.c
> > @@ -722,16 +722,16 @@ static void enable_irqs(struct cal_ctx *ctx)
> >   
> >   static void disable_irqs(struct cal_ctx *ctx)
> >   {
> > +	u32 val;
> > +
> >   	/* Disable IRQ_WDMA_END 0/1 */
> > -	reg_write_field(ctx->dev,
> > -			CAL_HL_IRQENABLE_CLR(2),
> > -			CAL_HL_IRQ_CLEAR,
> > -			CAL_HL_IRQ_MASK(ctx->csi2_port));
> > +	val = 0;
> > +	set_field(&val, CAL_HL_IRQ_CLEAR, CAL_HL_IRQ_MASK(ctx->csi2_port));
> > +	reg_write(ctx->dev, CAL_HL_IRQENABLE_CLR(2), val);
> >   	/* Disable IRQ_WDMA_START 0/1 */
> > -	reg_write_field(ctx->dev,
> > -			CAL_HL_IRQENABLE_CLR(3),
> > -			CAL_HL_IRQ_CLEAR,
> > -			CAL_HL_IRQ_MASK(ctx->csi2_port));
> > +	val = 0;
> > +	set_field(&val, CAL_HL_IRQ_CLEAR, CAL_HL_IRQ_MASK(ctx->csi2_port));
> > +	reg_write(ctx->dev, CAL_HL_IRQENABLE_CLR(3), val);
> >   	/* Todo: Add VC_IRQ and CSI2_COMPLEXIO_IRQ handling */
> >   	reg_write(ctx->dev, CAL_CSI2_VC_IRQENABLE(1), 0);
> >   }
> > 
> 
> I think the above works. But the enable_irqs is broken too, even if it doesn't cause any issues. Both IRQ_SET and IRQ_CLR are not supposed to be "modified" by a read-mod-write operation, but just written to.
>

Well maybe not consistent now, that disable_irqs has been modified but not
broken per se.

> The macros used also make the code very difficult to read. Something like this fixes both irq enable and disable, and makes it readable:

Ah but the mask macro are all consistent with each other, whether that
create one too many level of abstraction is a different subject. This setup
was oroginally requested by the maintainer. So I'll "fix" enable_irqs also
but I'll keep the macro as is.

Benoit

> 
> diff --git a/drivers/media/platform/ti-vpe/cal.c b/drivers/media/platform/ti-vpe/cal.c
> index c8b1290c9e2b..660653031a0b 100644
> --- a/drivers/media/platform/ti-vpe/cal.c
> +++ b/drivers/media/platform/ti-vpe/cal.c
> @@ -707,15 +707,9 @@ static void cal_quickdump_regs(struct cal_dev *dev)
>  static void enable_irqs(struct cal_ctx *ctx)
>  {
>  	/* Enable IRQ_WDMA_END 0/1 */
> -	reg_write_field(ctx->dev,
> -			CAL_HL_IRQENABLE_SET(2),
> -			CAL_HL_IRQ_ENABLE,
> -			CAL_HL_IRQ_MASK(ctx->csi2_port));
> +	reg_write(ctx->dev, CAL_HL_IRQENABLE_SET(2), 1 << (ctx->csi2_port - 1));
>  	/* Enable IRQ_WDMA_START 0/1 */
> -	reg_write_field(ctx->dev,
> -			CAL_HL_IRQENABLE_SET(3),
> -			CAL_HL_IRQ_ENABLE,
> -			CAL_HL_IRQ_MASK(ctx->csi2_port));
> +	reg_write(ctx->dev, CAL_HL_IRQENABLE_SET(3), 1 << (ctx->csi2_port - 1));
>  	/* Todo: Add VC_IRQ and CSI2_COMPLEXIO_IRQ handling */
>  	reg_write(ctx->dev, CAL_CSI2_VC_IRQENABLE(1), 0xFF000000);
>  }
> @@ -723,15 +717,9 @@ static void enable_irqs(struct cal_ctx *ctx)
>  static void disable_irqs(struct cal_ctx *ctx)
>  {
>  	/* Disable IRQ_WDMA_END 0/1 */
> -	reg_write_field(ctx->dev,
> -			CAL_HL_IRQENABLE_CLR(2),
> -			CAL_HL_IRQ_CLEAR,
> -			CAL_HL_IRQ_MASK(ctx->csi2_port));
> +	reg_write(ctx->dev, CAL_HL_IRQENABLE_CLR(2), 1 << (ctx->csi2_port - 1));
>  	/* Disable IRQ_WDMA_START 0/1 */
> -	reg_write_field(ctx->dev,
> -			CAL_HL_IRQENABLE_CLR(3),
> -			CAL_HL_IRQ_CLEAR,
> -			CAL_HL_IRQ_MASK(ctx->csi2_port));
> +	reg_write(ctx->dev, CAL_HL_IRQENABLE_CLR(3), 1 << (ctx->csi2_port - 1));
>  	/* Todo: Add VC_IRQ and CSI2_COMPLEXIO_IRQ handling */
>  	reg_write(ctx->dev, CAL_CSI2_VC_IRQENABLE(1), 0);
>  }
> 
>  Tomi
> 
> -- 
> Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki.
> Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [Patch 1/1] media: ti-vpe: cal: fix disable_irqs to only the intended target
  2020-03-02 13:56 [Patch 1/1] media: ti-vpe: cal: fix disable_irqs to only the intended target Benoit Parrot
  2020-03-04  8:49 ` Tomi Valkeinen
@ 2020-03-04 15:35 ` Sasha Levin
  1 sibling, 0 replies; 4+ messages in thread
From: Sasha Levin @ 2020-03-04 15:35 UTC (permalink / raw)
  To: Sasha Levin, Benoit Parrot, Hans Verkuil
  Cc: linux-media, devicetree, stable, stable

Hi

[This is an automated email]

This commit has been processed because it contains a -stable tag.
The stable tag indicates that it's relevant for the following trees: all

The bot has tested the following trees: v5.5.7, v5.4.23, v4.19.107, v4.14.172, v4.9.215, v4.4.215.

v5.5.7: Build OK!
v5.4.23: Build OK!
v4.19.107: Build OK!
v4.14.172: Build OK!
v4.9.215: Build OK!
v4.4.215: Failed to apply! Possible dependencies:
    343e89a792a5 ("[media] media: ti-vpe: Add CAL v4l2 camera capture driver")


NOTE: The patch will not be queued to stable trees until it is upstream.

How should we proceed with this patch?

-- 
Thanks
Sasha

^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2020-03-04 15:36 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-03-02 13:56 [Patch 1/1] media: ti-vpe: cal: fix disable_irqs to only the intended target Benoit Parrot
2020-03-04  8:49 ` Tomi Valkeinen
2020-03-04 13:17   ` Benoit Parrot
2020-03-04 15:35 ` Sasha Levin

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.