This patch serie adds JZ4755 SoC DMA support.
Diffs from 1-st patchset:
- DMA patches splitted into its own patchset
- acks collected
Siarhei Volkau (2):
dt-bindings: ingenic: Add support for the JZ4755 dmaengine
dmaengine: JZ4780: Add support for the JZ4755.
Documentation/devicetree/bindings/dma/ingenic,dma.yaml | 1 +
drivers/dma/dma-jz4780.c | 8 ++++++++
2 files changed, 9 insertions(+)
--
2.36.1
The JZ4755 has 4 DMA channels per DMA unit, two idential DMA units.
The JZ4755 has the similar DMA engine to JZ4725b, so I assume it has the
same bug as JZ4725b, see commit a40c94be2336.
Signed-off-by: Siarhei Volkau <[email protected]>
---
drivers/dma/dma-jz4780.c | 8 ++++++++
1 file changed, 8 insertions(+)
diff --git a/drivers/dma/dma-jz4780.c b/drivers/dma/dma-jz4780.c
index 2a483802d..9c1a6e9a9 100644
--- a/drivers/dma/dma-jz4780.c
+++ b/drivers/dma/dma-jz4780.c
@@ -1038,6 +1038,13 @@ static const struct jz4780_dma_soc_data jz4725b_dma_soc_data = {
JZ_SOC_DATA_BREAK_LINKS,
};
+static const struct jz4780_dma_soc_data jz4755_dma_soc_data = {
+ .nb_channels = 4,
+ .transfer_ord_max = 5,
+ .flags = JZ_SOC_DATA_PER_CHAN_PM | JZ_SOC_DATA_NO_DCKES_DCKEC |
+ JZ_SOC_DATA_BREAK_LINKS,
+};
+
static const struct jz4780_dma_soc_data jz4760_dma_soc_data = {
.nb_channels = 5,
.transfer_ord_max = 6,
@@ -1101,6 +1108,7 @@ static const struct jz4780_dma_soc_data x1830_dma_soc_data = {
static const struct of_device_id jz4780_dma_dt_match[] = {
{ .compatible = "ingenic,jz4740-dma", .data = &jz4740_dma_soc_data },
{ .compatible = "ingenic,jz4725b-dma", .data = &jz4725b_dma_soc_data },
+ { .compatible = "ingenic,jz4755-dma", .data = &jz4755_dma_soc_data },
{ .compatible = "ingenic,jz4760-dma", .data = &jz4760_dma_soc_data },
{ .compatible = "ingenic,jz4760-mdma", .data = &jz4760_mdma_soc_data },
{ .compatible = "ingenic,jz4760-bdma", .data = &jz4760_bdma_soc_data },
--
2.36.1
Acked-by: Krzysztof Kozlowski <[email protected]>
Signed-off-by: Siarhei Volkau <[email protected]>
---
Documentation/devicetree/bindings/dma/ingenic,dma.yaml | 1 +
1 file changed, 1 insertion(+)
diff --git a/Documentation/devicetree/bindings/dma/ingenic,dma.yaml b/Documentation/devicetree/bindings/dma/ingenic,dma.yaml
index 3b0b3b919..e42b8ce94 100644
--- a/Documentation/devicetree/bindings/dma/ingenic,dma.yaml
+++ b/Documentation/devicetree/bindings/dma/ingenic,dma.yaml
@@ -18,6 +18,7 @@ properties:
- enum:
- ingenic,jz4740-dma
- ingenic,jz4725b-dma
+ - ingenic,jz4755-dma
- ingenic,jz4760-dma
- ingenic,jz4760-bdma
- ingenic,jz4760-mdma
--
2.36.1
Hi Siarhei,
Le dim., oct. 16 2022 at 18:12:56 +0300, Siarhei Volkau
<[email protected]> a ?crit :
> The JZ4755 has 4 DMA channels per DMA unit, two idential DMA units.
>
> The JZ4755 has the similar DMA engine to JZ4725b, so I assume it has
> the
> same bug as JZ4725b, see commit a40c94be2336.
Can you verify that?
It should be pretty simple, if it has the bug you'll see I/O errors on
the SD card.
Cheers,
-Paul
> Signed-off-by: Siarhei Volkau <[email protected]>
> ---
> drivers/dma/dma-jz4780.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/drivers/dma/dma-jz4780.c b/drivers/dma/dma-jz4780.c
> index 2a483802d..9c1a6e9a9 100644
> --- a/drivers/dma/dma-jz4780.c
> +++ b/drivers/dma/dma-jz4780.c
> @@ -1038,6 +1038,13 @@ static const struct jz4780_dma_soc_data
> jz4725b_dma_soc_data = {
> JZ_SOC_DATA_BREAK_LINKS,
> };
>
> +static const struct jz4780_dma_soc_data jz4755_dma_soc_data = {
> + .nb_channels = 4,
> + .transfer_ord_max = 5,
> + .flags = JZ_SOC_DATA_PER_CHAN_PM | JZ_SOC_DATA_NO_DCKES_DCKEC |
> + JZ_SOC_DATA_BREAK_LINKS,
> +};
> +
> static const struct jz4780_dma_soc_data jz4760_dma_soc_data = {
> .nb_channels = 5,
> .transfer_ord_max = 6,
> @@ -1101,6 +1108,7 @@ static const struct jz4780_dma_soc_data
> x1830_dma_soc_data = {
> static const struct of_device_id jz4780_dma_dt_match[] = {
> { .compatible = "ingenic,jz4740-dma", .data = &jz4740_dma_soc_data
> },
> { .compatible = "ingenic,jz4725b-dma", .data =
> &jz4725b_dma_soc_data },
> + { .compatible = "ingenic,jz4755-dma", .data = &jz4755_dma_soc_data
> },
> { .compatible = "ingenic,jz4760-dma", .data = &jz4760_dma_soc_data
> },
> { .compatible = "ingenic,jz4760-mdma", .data =
> &jz4760_mdma_soc_data },
> { .compatible = "ingenic,jz4760-bdma", .data =
> &jz4760_bdma_soc_data },
> --
> 2.36.1
>
пн, 17 окт. 2022 г. в 12:34, Paul Cercueil <[email protected]>:
> Can you verify that?
>
> It should be pretty simple, if it has the bug you'll see I/O errors on
> the SD card.
Well, the result is ambiguous:
Without that 'broken' flag: mmc works poorly, but becomes more or less
stable when MMC clock downs to 6MHz (90% boots without errors).
On the 12MHz MMC clock the issue doesn't appear in 50-70% cases.
On the 24MHz MMC clock the device never boots up.
However with the flag: MMC works stable on 24MHz MMC clock
(boot issue observed only once), but if I increase MMC clock speed even
a bit the problem appears oftenly ( >70% of cases).
So, that flag definitely helps a lot, but the nature of the problem might be
different.
BR,
Siarhei
Hi Siarhei,
Le dim., oct. 16 2022 at 18:12:56 +0300, Siarhei Volkau
<[email protected]> a ?crit :
> The JZ4755 has 4 DMA channels per DMA unit, two idential DMA units.
>
> The JZ4755 has the similar DMA engine to JZ4725b, so I assume it has
> the
> same bug as JZ4725b, see commit a40c94be2336.
>
> Signed-off-by: Siarhei Volkau <[email protected]>
Reviewed-by: Paul Cercueil <[email protected]>
Cheers,
-Paul
> ---
> drivers/dma/dma-jz4780.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/drivers/dma/dma-jz4780.c b/drivers/dma/dma-jz4780.c
> index 2a483802d..9c1a6e9a9 100644
> --- a/drivers/dma/dma-jz4780.c
> +++ b/drivers/dma/dma-jz4780.c
> @@ -1038,6 +1038,13 @@ static const struct jz4780_dma_soc_data
> jz4725b_dma_soc_data = {
> JZ_SOC_DATA_BREAK_LINKS,
> };
>
> +static const struct jz4780_dma_soc_data jz4755_dma_soc_data = {
> + .nb_channels = 4,
> + .transfer_ord_max = 5,
> + .flags = JZ_SOC_DATA_PER_CHAN_PM | JZ_SOC_DATA_NO_DCKES_DCKEC |
> + JZ_SOC_DATA_BREAK_LINKS,
> +};
> +
> static const struct jz4780_dma_soc_data jz4760_dma_soc_data = {
> .nb_channels = 5,
> .transfer_ord_max = 6,
> @@ -1101,6 +1108,7 @@ static const struct jz4780_dma_soc_data
> x1830_dma_soc_data = {
> static const struct of_device_id jz4780_dma_dt_match[] = {
> { .compatible = "ingenic,jz4740-dma", .data = &jz4740_dma_soc_data
> },
> { .compatible = "ingenic,jz4725b-dma", .data =
> &jz4725b_dma_soc_data },
> + { .compatible = "ingenic,jz4755-dma", .data = &jz4755_dma_soc_data
> },
> { .compatible = "ingenic,jz4760-dma", .data = &jz4760_dma_soc_data
> },
> { .compatible = "ingenic,jz4760-mdma", .data =
> &jz4760_mdma_soc_data },
> { .compatible = "ingenic,jz4760-bdma", .data =
> &jz4760_bdma_soc_data },
> --
> 2.36.1
>
Hi,
Le lun., oct. 17 2022 at 20:29:27 +0300, Siarhei Volkau
<[email protected]> a écrit :
> пн, 17 окт. 2022 г. в 12:34, Paul Cercueil
> <[email protected]>:
>> Can you verify that?
>>
>> It should be pretty simple, if it has the bug you'll see I/O errors
>> on
>> the SD card.
>
> Well, the result is ambiguous:
>
> Without that 'broken' flag: mmc works poorly, but becomes more or less
> stable when MMC clock downs to 6MHz (90% boots without errors).
> On the 12MHz MMC clock the issue doesn't appear in 50-70% cases.
> On the 24MHz MMC clock the device never boots up.
>
> However with the flag: MMC works stable on 24MHz MMC clock
> (boot issue observed only once), but if I increase MMC clock speed
> even
> a bit the problem appears oftenly ( >70% of cases).
>
> So, that flag definitely helps a lot, but the nature of the problem
> might be
> different.
If the flag makes any difference then it's needed.
The controller should support up to 50 MHz, make sure that your "mmc"
clock is at 50 MHz otherwise changing the max-frequency DT property
won't do much.
Cheers,
-Paul