After commit f1ad5338a4d5("of: Fix "dma-ranges" handling for bus controllers"),
the dma-ranges is not allowed in decoder dts node. But the driver still need
to set dma mask, remove "dma-ranges" condition in prob function.
Signed-off-by: Yunfei Dong <[email protected]>
---
Reference series:
[1]: v5 of this series is present by Yong Wu.
[email protected]
---
.../mediatek/vcodec/mtk_vcodec_dec_drv.c | 24 +++++++++++++------
1 file changed, 17 insertions(+), 7 deletions(-)
diff --git a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
index bba7b932f4fa..2c3a4c2cdaee 100644
--- a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
+++ b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
@@ -137,6 +137,20 @@ static int mtk_vcodec_init_dec_resources(struct mtk_vcodec_dev *dev)
return 0;
}
+static int mtk_vcodec_dec_set_dma_mask(struct device *dev)
+{
+ int ret = 0;
+
+ return !(of_device_is_compatible(dev->of_node, "mediatek,mt8173-vcodec-dec") ||
+ of_device_is_compatible(dev->of_node, "mediatek,mt8183-vcodec-dec"));
+
+ ret = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(34));
+ if (ret)
+ mtk_v4l2_err("Failed to set dec mask");
+
+ return ret;
+}
+
static int fops_vcodec_open(struct file *file)
{
struct mtk_vcodec_dev *dev = video_drvdata(file);
@@ -323,13 +337,9 @@ static int mtk_vcodec_probe(struct platform_device *pdev)
}
}
- if (of_get_property(pdev->dev.of_node, "dma-ranges", NULL)) {
- ret = dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(34));
- if (ret) {
- mtk_v4l2_err("Failed to set mask");
- goto err_core_workq;
- }
- }
+ ret = mtk_vcodec_dec_set_dma_mask(&pdev->dev);
+ if (ret)
+ goto err_core_workq;
for (i = 0; i < MTK_VDEC_HW_MAX; i++)
mutex_init(&dev->dec_mutex[i]);
--
2.18.0
Il 23/03/23 09:38, Yunfei Dong ha scritto:
> After commit f1ad5338a4d5("of: Fix "dma-ranges" handling for bus controllers"),
> the dma-ranges is not allowed in decoder dts node. But the driver still need
> to set dma mask, remove "dma-ranges" condition in prob function.
>
> Signed-off-by: Yunfei Dong <[email protected]>
> ---
> Reference series:
> [1]: v5 of this series is present by Yong Wu.
> [email protected]
> ---
> .../mediatek/vcodec/mtk_vcodec_dec_drv.c | 24 +++++++++++++------
> 1 file changed, 17 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> index bba7b932f4fa..2c3a4c2cdaee 100644
> --- a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> +++ b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> @@ -137,6 +137,20 @@ static int mtk_vcodec_init_dec_resources(struct mtk_vcodec_dev *dev)
> return 0;
> }
>
> +static int mtk_vcodec_dec_set_dma_mask(struct device *dev)
> +{
> + int ret = 0;
> +
> + return !(of_device_is_compatible(dev->of_node, "mediatek,mt8173-vcodec-dec") ||
> + of_device_is_compatible(dev->of_node, "mediatek,mt8183-vcodec-dec"));
Like that, you're never reaching the code below...
Anyway, at a first glance, it looked like you were sending the same commit twice,
but then I noticed... so... I would propose to change the commit title(s) to follow
this format:
media: mediatek: vcodec: enc: Set DMA mask only for MT8173 and MT8183
(but then, why mt8173/83 and not the others? this deserves an explanation...)
Regards,
Angelo
On Thu, 2023-03-23 at 09:50 +0100, AngeloGioacchino Del Regno wrote:
> Il 23/03/23 09:38, Yunfei Dong ha scritto:
> > After commit f1ad5338a4d5("of: Fix "dma-ranges" handling for bus
> > controllers"),
> > the dma-ranges is not allowed in decoder dts node. But the driver
> > still need
> > to set dma mask, remove "dma-ranges" condition in prob function.
> >
> > Signed-off-by: Yunfei Dong <[email protected]>
> > ---
> > Reference series:
> > [1]: v5 of this series is present by Yong Wu.
> > [email protected]
> > ---
> > .../mediatek/vcodec/mtk_vcodec_dec_drv.c | 24 +++++++++++++-
> > -----
> > 1 file changed, 17 insertions(+), 7 deletions(-)
> >
> > diff --git
> > a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> > b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> > index bba7b932f4fa..2c3a4c2cdaee 100644
> > --- a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> > +++ b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> > @@ -137,6 +137,20 @@ static int
> > mtk_vcodec_init_dec_resources(struct mtk_vcodec_dev *dev)
> > return 0;
> > }
> >
> > +static int mtk_vcodec_dec_set_dma_mask(struct device *dev)
> > +{
> > + int ret = 0;
> > +
> > + return !(of_device_is_compatible(dev->of_node,
> > "mediatek,mt8173-vcodec-dec") ||
> > + of_device_is_compatible(dev->of_node,
> > "mediatek,mt8183-vcodec-dec"));
>
> Like that, you're never reaching the code below...
>
> Anyway, at a first glance, it looked like you were sending the same
> commit twice,
> but then I noticed... so... I would propose to change the commit
> title(s) to follow
> this format:
>
> media: mediatek: vcodec: enc: Set DMA mask only for MT8173 and MT8183
>
> (but then, why mt8173/83 and not the others? this deserves an
> explanation...)
This should be mt8173/83 don't need this. After confirming with yunfei,
I will fix the "return" above and contain this two patches and a jpeg
fix patch into the patchset of "Adjust the dma-ranges for MTK IOMMU".
This issue was introduced by that one since I removed "dma-ranges" for
them. they should be a topic.
>
>
> Regards,
> Angelo
>
>
On Thu, 2023-03-23 at 17:13 +0800, YongWu wrote:
> On Thu, 2023-03-23 at 09:50 +0100, AngeloGioacchino Del Regno wrote:
> > Il 23/03/23 09:38, Yunfei Dong ha scritto:
> > > After commit f1ad5338a4d5("of: Fix "dma-ranges" handling for bus
> > > controllers"),
> > > the dma-ranges is not allowed in decoder dts node. But the driver
> > > still need
> > > to set dma mask, remove "dma-ranges" condition in prob function.
> > >
> > > Signed-off-by: Yunfei Dong <[email protected]>
> > > ---
> > > Reference series:
> > > [1]: v5 of this series is present by Yong Wu.
> > > [email protected]
> > > ---
> > > .../mediatek/vcodec/mtk_vcodec_dec_drv.c | 24
> > > +++++++++++++-
> > > -----
> > > 1 file changed, 17 insertions(+), 7 deletions(-)
> > >
> > > diff --git
> > > a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> > > b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> > > index bba7b932f4fa..2c3a4c2cdaee 100644
> > > --- a/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> > > +++ b/drivers/media/platform/mediatek/vcodec/mtk_vcodec_dec_drv.c
> > > @@ -137,6 +137,20 @@ static int
> > > mtk_vcodec_init_dec_resources(struct mtk_vcodec_dev *dev)
> > > return 0;
> > > }
> > >
> > > +static int mtk_vcodec_dec_set_dma_mask(struct device *dev)
> > > +{
> > > + int ret = 0;
> > > +
> > > + return !(of_device_is_compatible(dev->of_node,
> > > "mediatek,mt8173-vcodec-dec") ||
> > > + of_device_is_compatible(dev->of_node,
> > > "mediatek,mt8183-vcodec-dec"));
> >
> > Like that, you're never reaching the code below...
> >
> > Anyway, at a first glance, it looked like you were sending the same
> > commit twice,
> > but then I noticed... so... I would propose to change the commit
> > title(s) to follow
> > this format:
> >
> > media: mediatek: vcodec: enc: Set DMA mask only for MT8173 and
> > MT8183
> >
> > (but then, why mt8173/83 and not the others? this deserves an
> > explanation...)
>
> This should be mt8173/83 don't need this. After confirming with
> yunfei,
> I will fix the "return" above and contain this two patches and a jpeg
> fix patch into the patchset of "Adjust the dma-ranges for MTK IOMMU".
> This issue was introduced by that one since I removed "dma-ranges"
> for
> them. they should be a topic.
Hi Angelo,
We meet a problem when changing this for jpeg.
JPEG[1] uses "mediatek,mtk-jpgenc" for mt2701, mt8183, mt8186 and
mt8188. But only mt8186/mt8188 need this
"dma_set_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(34))".
It can not distinguish the different SoC. Then does it need a different
compatible string for mt8186/mt8188?
Alternatively, A more simple way, MTK iommu knows whether the jpeg need
set dma_mask, it could help this. But I'm not sure if this is a right
way. The master devices don't know if it need set dma_mask and its
dma_mask is set by the other drivers, is this right?
Thanks very much:)
[1]
https://elixir.bootlin.com/linux/latest/source/Documentation/devicetree/bindings/media/mediatek-jpeg-encoder.yaml#L23
>
> >
> >
> > Regards,
> > Angelo
> >
> >