[PATCH 0/2] ARM: davinci: fix eDMA probing for dm365

[PATCH 0/2] ARM: davinci: fix eDMA probing for dm365

[Alejandro Mery]

Hi, as an intermediate step toward migrating a davinci dm365 based product from 2.6.32.71
to 4.14(.3+) I'm trying to get the dm365 evm (evaluation board) to work

The very first problem is `edma: probe of edma.0 failed with error -5` caused by dm365 not
been included on some refactoring done by Peter Ujfalusi in 2015 which got solved by the
cherry-picking the following changes into dm365.c

* cef5b0da4019 2015-10-14 Peter Ujfalusi ARM: davinci: Add dma_mask to eDMA devices
* 7ab388e85faa 2015-10-14 Peter Ujfalusi ARM: davinci: Use platform_device_register_full() to create pdev for eDMA

[PATCH 1/2] ARM: davinci: Use platform_device_register_full() to create pdev for dm365's eDMA

[Alejandro Mery]

as it was done by 7ab388e85faa97a35d520720269e7c8e00ad54a0 for other davinci SoCs

Signed-off-by: Alejandro Mery <amery@hanoverdisplays.com>
---
 arch/arm/mach-davinci/dm365.c | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/arch/arm/mach-davinci/dm365.c b/arch/arm/mach-davinci/dm365.c
index 8be04ec95adf..855e0a57b219 100644
--- a/arch/arm/mach-davinci/dm365.c
+++ b/arch/arm/mach-davinci/dm365.c
@@ -925,12 +925,13 @@ static struct resource edma_resources[] = {
 	/* not using TC*_ERR */
 };
 
-static struct platform_device dm365_edma_device = {
-	.name			= "edma",
-	.id			= 0,
-	.dev.platform_data	= &dm365_edma_pdata,
-	.num_resources		= ARRAY_SIZE(edma_resources),
-	.resource		= edma_resources,
+static const struct platform_device_info dm365_edma_device __initconst = {
+	.name		= "edma",
+	.id		= 0,
+	.res		= edma_resources,
+	.num_res	= ARRAY_SIZE(edma_resources),
+	.data		= &dm365_edma_pdata,
+	.size_data	= sizeof(&dm365_edma_pdata),
 };
 
 static struct resource dm365_asp_resources[] = {
@@ -1428,13 +1429,18 @@ int __init dm365_init_video(struct vpfe_config *vpfe_cfg,
 
 static int __init dm365_init_devices(void)
 {
+	struct platform_device *edma_pdev;
 	int ret = 0;
 
 	if (!cpu_is_davinci_dm365())
 		return 0;
 
 	davinci_cfg_reg(DM365_INT_EDMA_CC);
-	platform_device_register(&dm365_edma_device);
+	edma_pdev = platform_device_register_full(&dm365_edma_device);
+	if (IS_ERR(edma_pdev)) {
+		pr_warn("%s: Failed to register eDMA\n", __func__);
+		return PTR_ERR(edma_pdev);
+	}
 
 	platform_device_register(&dm365_mdio_device);
 	platform_device_register(&dm365_emac_device);
-- 
2.15.0

[PATCH 2/2] ARM: davinci: Add dma_mask to dm365's eDMA device

[Alejandro Mery]

as previously done by cef5b0da4019358cb03c9b0a964d4d63cd7deaf6 for other davinci SoCs

Signed-off-by: Alejandro Mery <amery@hanoverdisplays.com>
---
 arch/arm/mach-davinci/dm365.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm/mach-davinci/dm365.c b/arch/arm/mach-davinci/dm365.c
index 855e0a57b219..90858bc9dd18 100644
--- a/arch/arm/mach-davinci/dm365.c
+++ b/arch/arm/mach-davinci/dm365.c
@@ -928,6 +928,7 @@ static struct resource edma_resources[] = {
 static const struct platform_device_info dm365_edma_device __initconst = {
 	.name		= "edma",
 	.id		= 0,
+	.dma_mask	= DMA_BIT_MASK(32),
 	.res		= edma_resources,
 	.num_res	= ARRAY_SIZE(edma_resources),
 	.data		= &dm365_edma_pdata,
-- 
2.15.0

[PATCH 0/2] ARM: davinci: fix eDMA probing for dm365

[Peter Ujfalusi]

Hi,

On 2017-12-04 15:33, Alejandro Mery wrote:
> Hi, as an intermediate step toward migrating a davinci dm365 based product from 2.6.32.71
> to 4.14(.3+) I'm trying to get the dm365 evm (evaluation board) to work
> 
> The very first problem is `edma: probe of edma.0 failed with error -5` caused by dm365 not
> been included on some refactoring done by Peter Ujfalusi in 2015 which got solved by the
> cherry-picking the following changes into dm365.c
> 
> * cef5b0da4019 2015-10-14 Peter Ujfalusi ARM: davinci: Add dma_mask to eDMA devices
> * 7ab388e85faa 2015-10-14 Peter Ujfalusi ARM: davinci: Use platform_device_register_full() to create pdev for eDMA

Oh, how did I missed them???

Thanks for taking care of dm365!

To both:
Reviewed-by: Peter Ujfalusi <peter.ujfalusi@ti.com>

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

[PATCH v2 0/3] ARM: davinci: fix eDMA for DM365

[Alejandro Mery]

Hi, as an intermediate step toward migrating a davinci dm365 based product from 2.6.32.71
to 4.14(.3+) I'm trying to get the dm365 evm (evaluation board) to work

The very first problem is `edma: probe of edma.0 failed with error -5` caused by dm365 not
been included on some refactoring done by Peter Ujfalusi in 2015 which got solved by the
cherry-picking the following changes into dm365.c

* cef5b0da4019 2015-10-14 Peter Ujfalusi ARM: davinci: Add dma_mask to eDMA devices
* 7ab388e85faa 2015-10-14 Peter Ujfalusi ARM: davinci: Use platform_device_register_full() to create pdev for eDMA

Changes since v1:
* fix dumb typo on 1/2
* add extra change (3/3) correcting the name of mmc devices

[PATCH v2 1/3] ARM: davinci: Use platform_device_register_full() to create pdev for dm365's eDMA

[Alejandro Mery]

as it was done by 7ab388e85faa97a35d520720269e7c8e00ad54a0 for other davinci SoCs

changes since v1:
* fix typo

Signed-off-by: Alejandro Mery <amery@hanoverdisplays.com>
---
 arch/arm/mach-davinci/dm365.c | 20 +++++++++++++-------
 1 file changed, 13 insertions(+), 7 deletions(-)

diff --git a/arch/arm/mach-davinci/dm365.c b/arch/arm/mach-davinci/dm365.c
index 8be04ec95adf..9bd17bc77b5c 100644
--- a/arch/arm/mach-davinci/dm365.c
+++ b/arch/arm/mach-davinci/dm365.c
@@ -925,12 +925,13 @@ static struct resource edma_resources[] = {
 	/* not using TC*_ERR */
 };
 
-static struct platform_device dm365_edma_device = {
-	.name			= "edma",
-	.id			= 0,
-	.dev.platform_data	= &dm365_edma_pdata,
-	.num_resources		= ARRAY_SIZE(edma_resources),
-	.resource		= edma_resources,
+static const struct platform_device_info dm365_edma_device __initconst = {
+	.name		= "edma",
+	.id		= 0,
+	.res		= edma_resources,
+	.num_res	= ARRAY_SIZE(edma_resources),
+	.data		= &dm365_edma_pdata,
+	.size_data	= sizeof(dm365_edma_pdata),
 };
 
 static struct resource dm365_asp_resources[] = {
@@ -1428,13 +1429,18 @@ int __init dm365_init_video(struct vpfe_config *vpfe_cfg,
 
 static int __init dm365_init_devices(void)
 {
+	struct platform_device *edma_pdev;
 	int ret = 0;
 
 	if (!cpu_is_davinci_dm365())
 		return 0;
 
 	davinci_cfg_reg(DM365_INT_EDMA_CC);
-	platform_device_register(&dm365_edma_device);
+	edma_pdev = platform_device_register_full(&dm365_edma_device);
+	if (IS_ERR(edma_pdev)) {
+		pr_warn("%s: Failed to register eDMA\n", __func__);
+		return PTR_ERR(edma_pdev);
+	}
 
 	platform_device_register(&dm365_mdio_device);
 	platform_device_register(&dm365_emac_device);
-- 
2.15.0

[PATCH v2 1/3] ARM: davinci: Use platform_device_register_full() to create pdev for dm365's eDMA

[Sekhar Nori]

Hi Alejandro,

On Tuesday 05 December 2017 06:04 PM, Alejandro Mery wrote:
> as it was done by 7ab388e85faa97a35d520720269e7c8e00ad54a0 for other davinci SoCs
> 
> changes since v1:
> * fix typo
> 
> Signed-off-by: Alejandro Mery <amery@hanoverdisplays.com>

The patch itself looks good (and apologies for not catching the missing 
DM365 conversion). But:

The way you refer to the commit in patch description will lead to 
checkpatch error. And, we need more explanation in the patch description 
than just referencing the commit which is being fixed. When fixing a 
previous commit, its best to use the Fixes: tag (see below). Thats an 
indication for stable tree maintainers to backport this patch all the 
way till the time the offending commit was introduced into the kernel. 
This way you don't have to do the backporting yourself.

Please keep the changelog below the tearline "---" so it will not get 
included in commit description. And, when submitting another version, 
please carry forward the Reviewed-by: and and Acked-by: tags you 
received for previous version (unless the patch has changed so 
significantly that it does not make sense to do so anymore).

For this time, I fixed all of these up locally. Here is the final commit 
text.

--8<--
ARM: davinci: Use platform_device_register_full() to create pdev for dm365's eDMA

Convert the DM365 EDMA platform device creation to use
struct platform_device_info XXXXXX __initconst and
platform_device_register_full()

This will allow us to specify the dma_mask for the device
in an upcoming patch. Without this, EDMA on DM365 refuses
to probe.

Fixes: 7ab388e85faa ("ARM: davinci: Use platform_device_register_full() to create pdev for eDMA")
Reviewed-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Alejandro Mery <amery@hanoverdisplays.com>
Signed-off-by: Sekhar Nori <nsekhar@ti.com>
--8<--

Thanks,
Sekhar


> ---
>  arch/arm/mach-davinci/dm365.c | 20 +++++++++++++-------
>  1 file changed, 13 insertions(+), 7 deletions(-)
> 
> diff --git a/arch/arm/mach-davinci/dm365.c b/arch/arm/mach-davinci/dm365.c
> index 8be04ec95adf..9bd17bc77b5c 100644
> --- a/arch/arm/mach-davinci/dm365.c
> +++ b/arch/arm/mach-davinci/dm365.c
> @@ -925,12 +925,13 @@ static struct resource edma_resources[] = {
>  	/* not using TC*_ERR */
>  };
>  
> -static struct platform_device dm365_edma_device = {
> -	.name			= "edma",
> -	.id			= 0,
> -	.dev.platform_data	= &dm365_edma_pdata,
> -	.num_resources		= ARRAY_SIZE(edma_resources),
> -	.resource		= edma_resources,
> +static const struct platform_device_info dm365_edma_device __initconst = {
> +	.name		= "edma",
> +	.id		= 0,
> +	.res		= edma_resources,
> +	.num_res	= ARRAY_SIZE(edma_resources),
> +	.data		= &dm365_edma_pdata,
> +	.size_data	= sizeof(dm365_edma_pdata),
>  };
>  
>  static struct resource dm365_asp_resources[] = {
> @@ -1428,13 +1429,18 @@ int __init dm365_init_video(struct vpfe_config *vpfe_cfg,
>  
>  static int __init dm365_init_devices(void)
>  {
> +	struct platform_device *edma_pdev;
>  	int ret = 0;
>  
>  	if (!cpu_is_davinci_dm365())
>  		return 0;
>  
>  	davinci_cfg_reg(DM365_INT_EDMA_CC);
> -	platform_device_register(&dm365_edma_device);
> +	edma_pdev = platform_device_register_full(&dm365_edma_device);
> +	if (IS_ERR(edma_pdev)) {
> +		pr_warn("%s: Failed to register eDMA\n", __func__);
> +		return PTR_ERR(edma_pdev);
> +	}
>  
>  	platform_device_register(&dm365_mdio_device);
>  	platform_device_register(&dm365_emac_device);
> -- 
> 2.15.0
> 

[PATCH v2 2/3] ARM: davinci: Add dma_mask to dm365's eDMA device

[Alejandro Mery]

as previously done by cef5b0da4019358cb03c9b0a964d4d63cd7deaf6 for other davinci SoCs

Signed-off-by: Alejandro Mery <amery@hanoverdisplays.com>
---
 arch/arm/mach-davinci/dm365.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/arch/arm/mach-davinci/dm365.c b/arch/arm/mach-davinci/dm365.c
index 9bd17bc77b5c..103316f01a22 100644
--- a/arch/arm/mach-davinci/dm365.c
+++ b/arch/arm/mach-davinci/dm365.c
@@ -928,6 +928,7 @@ static struct resource edma_resources[] = {
 static const struct platform_device_info dm365_edma_device __initconst = {
 	.name		= "edma",
 	.id		= 0,
+	.dma_mask	= DMA_BIT_MASK(32),
 	.res		= edma_resources,
 	.num_res	= ARRAY_SIZE(edma_resources),
 	.data		= &dm365_edma_pdata,
-- 
2.15.0

[PATCH v2 2/3] ARM: davinci: Add dma_mask to dm365's eDMA device

[Sekhar Nori]

On Tuesday 05 December 2017 06:04 PM, Alejandro Mery wrote:
> as previously done by cef5b0da4019358cb03c9b0a964d4d63cd7deaf6 for other davinci SoCs
> 
> Signed-off-by: Alejandro Mery <amery@hanoverdisplays.com>

Most of the comments on 1/2 apply here. Here is the updated description:

--8<--
ARM: davinci: Add dma_mask to dm365's eDMA device

Add dma_mask to dm365's EDMA device.

Without a valid dma_mask, EDMA on DM365 refuses to
probe.

Fixes: cef5b0da4019 ("ARM: davinci: Add dma_mask to eDMA devices")
Reviewed-by: Peter Ujfalusi <peter.ujfalusi@ti.com>
Signed-off-by: Alejandro Mery <amery@hanoverdisplays.com>
Signed-off-by: Sekhar Nori <nsekhar@ti.com>
--8<--

Thanks,
Sekhar

> ---
>  arch/arm/mach-davinci/dm365.c | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/arch/arm/mach-davinci/dm365.c b/arch/arm/mach-davinci/dm365.c
> index 9bd17bc77b5c..103316f01a22 100644
> --- a/arch/arm/mach-davinci/dm365.c
> +++ b/arch/arm/mach-davinci/dm365.c
> @@ -928,6 +928,7 @@ static struct resource edma_resources[] = {
>  static const struct platform_device_info dm365_edma_device __initconst = {
>  	.name		= "edma",
>  	.id		= 0,
> +	.dma_mask	= DMA_BIT_MASK(32),
>  	.res		= edma_resources,
>  	.num_res	= ARRAY_SIZE(edma_resources),
>  	.data		= &dm365_edma_pdata,
> -- 
> 2.15.0
> 

[PATCH v2 3/3] ARM: davinci: fix mmc entries in DM365's eDMA slaves table

[Alejandro Mery]

Signed-off-by: Alejandro Mery <amery@hanoverdisplays.com>
---
 arch/arm/mach-davinci/dm365.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/arch/arm/mach-davinci/dm365.c b/arch/arm/mach-davinci/dm365.c
index 103316f01a22..5ace9380626a 100644
--- a/arch/arm/mach-davinci/dm365.c
+++ b/arch/arm/mach-davinci/dm365.c
@@ -868,10 +868,10 @@ static const struct dma_slave_map dm365_edma_map[] = {
 	{ "spi_davinci.0", "rx", EDMA_FILTER_PARAM(0, 17) },
 	{ "spi_davinci.3", "tx", EDMA_FILTER_PARAM(0, 18) },
 	{ "spi_davinci.3", "rx", EDMA_FILTER_PARAM(0, 19) },
-	{ "dm6441-mmc.0", "rx", EDMA_FILTER_PARAM(0, 26) },
-	{ "dm6441-mmc.0", "tx", EDMA_FILTER_PARAM(0, 27) },
-	{ "dm6441-mmc.1", "rx", EDMA_FILTER_PARAM(0, 30) },
-	{ "dm6441-mmc.1", "tx", EDMA_FILTER_PARAM(0, 31) },
+	{ "da830-mmc.0", "rx", EDMA_FILTER_PARAM(0, 26) },
+	{ "da830-mmc.0", "tx", EDMA_FILTER_PARAM(0, 27) },
+	{ "da830-mmc.1", "rx", EDMA_FILTER_PARAM(0, 30) },
+	{ "da830-mmc.1", "tx", EDMA_FILTER_PARAM(0, 31) },
 };
 
 static struct edma_soc_info dm365_edma_pdata = {
-- 
2.15.0

[PATCH v2 3/3] ARM: davinci: fix mmc entries in DM365's eDMA slaves table

[Sekhar Nori]

Hi,

On Tuesday 05 December 2017 06:04 PM, Alejandro Mery wrote:
> Signed-off-by: Alejandro Mery <amery@hanoverdisplays.com>

Patch looks good, but please resubmit with patch description. Patch
description should be readable independent of headline so please add it
even if its just restating what is present in the headline.

Also, please add a Fixes tag.

Fixes: 0c750e1fe481 ("ARM: davinci: dm365: Add dma_slave_map to edma")

You can base the next version on fixes branch of my tree[1].

Thanks,
Sekhar

[1] git://git.kernel.org/pub/scm/linux/kernel/git/nsekhar/linux-davinci.git

> ---
>  arch/arm/mach-davinci/dm365.c | 8 ++++----
>  1 file changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/arch/arm/mach-davinci/dm365.c b/arch/arm/mach-davinci/dm365.c
> index 103316f01a22..5ace9380626a 100644
> --- a/arch/arm/mach-davinci/dm365.c
> +++ b/arch/arm/mach-davinci/dm365.c
> @@ -868,10 +868,10 @@ static const struct dma_slave_map dm365_edma_map[] = {
>  	{ "spi_davinci.0", "rx", EDMA_FILTER_PARAM(0, 17) },
>  	{ "spi_davinci.3", "tx", EDMA_FILTER_PARAM(0, 18) },
>  	{ "spi_davinci.3", "rx", EDMA_FILTER_PARAM(0, 19) },
> -	{ "dm6441-mmc.0", "rx", EDMA_FILTER_PARAM(0, 26) },
> -	{ "dm6441-mmc.0", "tx", EDMA_FILTER_PARAM(0, 27) },
> -	{ "dm6441-mmc.1", "rx", EDMA_FILTER_PARAM(0, 30) },
> -	{ "dm6441-mmc.1", "tx", EDMA_FILTER_PARAM(0, 31) },
> +	{ "da830-mmc.0", "rx", EDMA_FILTER_PARAM(0, 26) },
> +	{ "da830-mmc.0", "tx", EDMA_FILTER_PARAM(0, 27) },
> +	{ "da830-mmc.1", "rx", EDMA_FILTER_PARAM(0, 30) },
> +	{ "da830-mmc.1", "tx", EDMA_FILTER_PARAM(0, 31) },
>  };
>  
>  static struct edma_soc_info dm365_edma_pdata = {
> -- 
> 2.15.0
> 

[PATCH v2 0/3] ARM: davinci: fix eDMA for DM365

[Sekhar Nori]

Hi Alejandro,

On Tuesday 05 December 2017 06:04 PM, Alejandro Mery wrote:
> Hi, as an intermediate step toward migrating a davinci dm365 based product from 2.6.32.71
> to 4.14(.3+) I'm trying to get the dm365 evm (evaluation board) to work

Thanks for trying the mainline kernel on this device!

For the next time, when submitting another version, please send as an
independent series, not as reply to the the previous post.

Many maintainers will ignore old threads where comments have already
been provided. Sending fresh patches in reply to old thread just
increases the chance of patches getting lost.

Thanks,
Sekhar