ARM: mxs: move to generic DMA device tree binding
Subsystems (13)
cover ARM: mxs: move to generic DMA device tree binding
The series converts mxs-dma and its clients to generic DMA device tree binding/helper.
Subsystem maintainers,
I need your ACKs to have the series go via arm-soc as a whole if the patches look good to you.
Mark,
Patch "ASoC: dmaengine_pcm: support use of generic DMA helper" is not mxs specific. It makes more sense for you to maintain a topic branch for that patch. It should be useful for any dmaengine_pcm users that want to move to generic DMA helper.
Regards,
Shawn
Shawn Guo (12):
ARM: dts: add generic DMA device tree binding for mxs-dma
dma: mxs-dma: use devm_* managed functions
dma: mxs-dma: move to generic device tree binding
mmc: mxs-mmc: move to use generic DMA helper
spi: mxs-spi: move to use generic DMA helper
i2c: i2c-mxs: move to use generic DMA helper
mtd: gpmi: move to use generic DMA helper
serial: mxs-auart: move to use generic DMA helper
ASoC: dmaengine_pcm: support use of generic DMA helper
ASoC: mxs: move to use generic DMA helper
dma: mxs-dma: remove code left from generic DMA binding conversion
ARM: dts: remove mxs-dma channel interrupt number from client nodes
.../devicetree/bindings/dma/fsl-mxs-dma.txt | 49 ++++++- Documentation/devicetree/bindings/i2c/i2c-mxs.txt | 12 +- Documentation/devicetree/bindings/mmc/mxs-mmc.txt | 12 +- .../devicetree/bindings/mtd/gpmi-nand.txt | 17 ++- .../devicetree/bindings/sound/mxs-saif.txt | 18 ++- Documentation/devicetree/bindings/spi/mxs-spi.txt | 12 +- .../bindings/tty/serial/fsl-mxs-auart.txt | 16 +-- arch/arm/boot/dts/imx23.dtsi | 54 ++++++-- arch/arm/boot/dts/imx28.dtsi | 94 +++++++++++--- arch/arm/boot/dts/imx6qdl.dtsi | 12 +- drivers/dma/mxs-dma.c | 135 ++++++++++++-------- drivers/i2c/busses/i2c-mxs.c | 40 +----- drivers/mmc/host/mxs-mmc.c | 48 +------ drivers/mtd/nand/gpmi-nand/gpmi-nand.c | 51 +------- drivers/mtd/nand/gpmi-nand/gpmi-nand.h | 3 +- drivers/spi/spi-mxs.c | 60 ++------- drivers/tty/serial/mxs-auart.c | 52 +------- include/linux/fsl/mxs-dma.h | 20 --- include/linux/spi/mxs-spi.h | 4 +- include/sound/dmaengine_pcm.h | 5 + sound/soc/mxs/mxs-pcm.c | 43 +------ sound/soc/mxs/mxs-pcm.h | 5 - sound/soc/mxs/mxs-saif.c | 28 +--- sound/soc/mxs/mxs-saif.h | 3 +- sound/soc/soc-dmaengine-pcm.c | 19 ++- 25 files changed, 357 insertions(+), 455 deletions(-) delete mode 100644 include/linux/fsl/mxs-dma.h
Signature · 1 line
1.7.9.5
PATCH 1/12 ARM: dts: add generic DMA device tree binding for mxs-dma
Add generic DMA device tree binding for mxs-dma. The changes include:
* Add channel interrupts into DMA controller nodes
* Add properties '#dma-cells' and 'dma-channels' for DMA controller nodes
* And properties 'dmas' and 'dma-names' for DMA client nodes
* Update mxs-dma device tree binding doc
--- .../devicetree/bindings/dma/fsl-mxs-dma.txt | 49 +++++++++++++++-- arch/arm/boot/dts/imx23.dtsi | 42 +++++++++++++- arch/arm/boot/dts/imx28.dtsi | 58 +++++++++++++++++++- arch/arm/boot/dts/imx6qdl.dtsi | 8 ++- 4 files changed, 148 insertions(+), 9 deletions(-)
Documentation/devicetree/bindings/dma/fsl-mxs-dma.txt +45 -4
@@ -3,17 +3,58 @@ Required properties: - compatible : Should be "fsl,<chip>-dma-apbh" or "fsl,<chip>-dma-apbx" - reg : Should contain registers location and length +- interrupts : Should contain the interrupt numbers of DMA channels. + If a channel is empty/reserved, 0 should be filled in place. +- #dma-cells : Must be <1>. The number cell specifies the channel ID. +- dma-channels : Number of channels supported by the DMA controller + +Optional properties: +- interrupt-names : Name of DMA channel interrupts Supported chips: imx23, imx28. Examples: -dma-apbh at 80004000 { + +dma_apbh: dma-apbh at 80004000 { compatible = "fsl,imx28-dma-apbh"; - reg = <0x80004000 2000>; + reg = <0x80004000 0x2000>; + interrupts = <82 83 84 85 + 88 88 88 88 + 88 88 88 88 + 87 86 0 0>; + interrupt-names = "ssp0", "ssp1", "ssp2", "ssp3", + "gpmi0", "gmpi1", "gpmi2", "gmpi3", + "gpmi4", "gmpi5", "gpmi6", "gmpi7", + "hsadc", "lcdif", "empty", "empty"; + #dma-cells = <1>; + dma-channels = <16>; }; -dma-apbx at 80024000 { +dma_apbx: dma-apbx at 80024000 { compatible = "fsl,imx28-dma-apbx"; - reg = <0x80024000 2000>; + reg = <0x80024000 0x2000>; + interrupts = <78 79 66 0 + 80 81 68 69 + 70 71 72 73 + 74 75 76 77>; + interrupt-names = "auart4-rx", "aurat4-tx", "spdif-tx", "empty", + "saif0", "saif1", "i2c0", "i2c1", + "auart0-rx", "auart0-tx", "auart1-rx", "auart1-tx", + "auart2-rx", "auart2-tx", "auart3-rx", "auart3-tx"; + #dma-cells = <1>; + dma-channels = <16>; +}; + +DMA clients connected to the MXS DMA controller must use the format +described in the dma.txt file. + +Examples: + +auart0: serial at 8006a000 { + compatible = "fsl,imx28-auart", "fsl,imx23-auart"; + reg = <0x8006a000 0x2000>; + interrupts = <112>; + dmas = <&dma_apbx 8>, <&dma_apbx 9>; + dma-names = "rx", "tx"; };
arch/arm/boot/dts/imx23.dtsi +40 -2
121 lines not on this page · raw message
arch/arm/boot/dts/imx28.dtsi +56 -2
173 lines not on this page · raw message
arch/arm/boot/dts/imx6qdl.dtsi +7 -1
@@ -65,9 +65,13 @@ interrupt-parent = <&intc>; ranges; - dma-apbh at 00110000 { + dma_apbh: dma-apbh at 00110000 { compatible = "fsl,imx6q-dma-apbh", "fsl,imx28-dma-apbh"; reg = <0x00110000 0x2000>; + interrupts = <0 13 0x04>, <0 13 0x04>, <0 13 0x04>, <0 13 0x04>; + interrupt-names = "gpmi0", "gpmi1", "gpmi2", "gpmi3"; + #dma-cells = <1>; + dma-channels = <4>; clocks = <&clks 106>; };
@@ -83,6 +87,8 @@ <&clks 150>, <&clks 149>; clock-names = "gpmi_io", "gpmi_apb", "gpmi_bch", "gpmi_bch_apb", "per1_bch"; + dmas = <&dma_apbh 0>; + dma-names = "gpmi"; fsl,gpmi-dma-channel = <0>; status = "disabled"; };
Signature · 1 line
1.7.9.5
PATCH 1/12 ARM: dts: add generic DMA device tree binding for mxs-dma
Quoted 21 lines · Shawn Guo
Add generic DMA device tree binding for mxs-dma. The changes include:
* Add channel interrupts into DMA controller nodes
* Add properties '#dma-cells' and 'dma-channels' for DMA controller nodes
* And properties 'dmas' and 'dma-names' for DMA client nodes
* Update mxs-dma device tree binding docSigned-off-by Shawn Guo <[email protected]>--- .../devicetree/bindings/dma/fsl-mxs-dma.txt | 49 +++++++++++++++-- arch/arm/boot/dts/imx23.dtsi | 42 +++++++++++++- arch/arm/boot/dts/imx28.dtsi | 58 +++++++++++++++++++- arch/arm/boot/dts/imx6qdl.dtsi | 8 ++- 4 files changed, 148 insertions(+), 9 deletions(-)Documentation/devicetree/bindings/dma/fsl-mxs-dma.txt +0
-0@@ -3,17 +3,58 @@ Required properties:
Documentation/devicetree/bindings/dma/fsl-mxs-dma.txt line 13
Quoted 10 lines
Documentation/devicetree/bindings/dma/fsl-mxs-dma.txt +7
-0- compatible : Should be "fsl,<chip>-dma-apbh" or "fsl,<chip>-dma-apbx" - reg : Should contain registers location and length +- interrupts : Should contain the interrupt numbers of DMA channels. + If a channel is empty/reserved, 0 should be filled in place. +- #dma-cells : Must be <1>. The number cell specifies the channel ID. +- dma-channels : Number of channels supported by the DMA controller + +Optional properties: +- interrupt-names : Name of DMA channel interrupts
Looks good overall, just one small comment: The generic binding requires both "dma-channels" and "dma-requests" to be listed here. My understanding of this driver is that each channel is hardwired to one request, so you can mandate here that the two values must be the same.
Arnd
PATCH 1/12 ARM: dts: add generic DMA device tree binding for mxs-dma
Quoted 2 lines · Arnd Bergmann
Looks good overall, just one small comment: The generic binding requires both "dma-channels" and "dma-requests" to be listed here.
Hmm, what I read from Documentation/devicetree/bindings/dma/dma.txt is that both dma-channels and dma-requests are optional properties.
Shawn
Quoted 3 lines
My understanding of this
driver is that each channel is hardwired to one request, so you can mandate here that the two values must be the same.
PATCH 1/12 ARM: dts: add generic DMA device tree binding for mxs-dma
Quoted 7 lines · Shawn Guo
Quoted 2 lines · Arnd Bergmann
Looks good overall, just one small comment: The generic binding requires both "dma-channels" and "dma-requests" to be listed here.
Hmm, what I read from
Documentation/devicetree/bindings/dma/dma.txtis that both dma-channels and dma-requests are optional properties.
Ah, you are right. Nevermind then.
Arnd
PATCH 2/12 dma: mxs-dma: use devm_* managed functions
In reply to Shawn GuoUse devm_* managed functions to simplify probe() error handling.
--- drivers/dma/mxs-dma.c | 39 +++++++++------------------------------ 1 file changed, 9 insertions(+), 30 deletions(-)
drivers/dma/mxs-dma.c +9 -30
@@ -674,7 +674,7 @@ static int __init mxs_dma_probe(struct platform_device *pdev)struct resource *iores;int ret, i; - mxs_dma =kzalloc (sizeof (*mxs_dma),GFP_KERNEL ); + mxs_dma =devm_kzalloc (&pdev->dev,sizeof (*mxs_dma),GFP_KERNEL );if (!mxs_dma)return -ENOMEM ;
@@ -689,24 +689,13 @@ static int __init mxs_dma_probe(struct platform_device *pdev) mxs_dma->dev_id = dma_type->id; iores =platform_get_resource (pdev,IORESOURCE_MEM ,0 ); + mxs_dma->base =devm_ioremap_resource (&pdev->dev, iores); +if (IS_ERR (mxs_dma->base)) +return PTR_ERR (mxs_dma->base); -if (!request_mem_region (iores->start,resource_size (iores), - pdev->name)) { - ret = -EBUSY ; - goto err_request_region; - } - - mxs_dma->base =ioremap (iores->start,resource_size (iores)); -if (!mxs_dma->base) { - ret = -ENOMEM ; - goto err_ioremap; - } - - mxs_dma->clk =clk_get (&pdev->dev,NULL ); -if (IS_ERR (mxs_dma->clk)) { - ret =PTR_ERR (mxs_dma->clk); - goto err_clk; - } + mxs_dma->clk =devm_clk_get (&pdev->dev,NULL ); +if (IS_ERR (mxs_dma->clk)) +return PTR_ERR (mxs_dma->clk);dma_cap_set (DMA_SLAVE , mxs_dma->dma_device.cap_mask);dma_cap_set (DMA_CYCLIC , mxs_dma->dma_device.cap_mask);
@@ -732,7 +721,7 @@ static int __init mxs_dma_probe(struct platform_device *pdev) ret =mxs_dma_init (mxs_dma);if (ret) - goto err_init; +return ret; mxs_dma->dma_device.dev = &pdev->dev;
@@ -751,22 +740,12 @@ static int __init mxs_dma_probe(struct platform_device *pdev) ret =dma_async_device_register (&mxs_dma->dma_device);if (ret) {dev_err (mxs_dma->dma_device.dev,"unable to register\n" ); - goto err_init; +return ret; }dev_info (mxs_dma->dma_device.dev,"initialized\n" );return 0 ; - -err_init: -clk_put (mxs_dma->clk); -err_clk: -iounmap (mxs_dma->base); -err_ioremap: -release_mem_region (iores->start,resource_size (iores)); -err_request_region: -kfree (mxs_dma); -return ret; }static struct platform_driver mxs_dma_driver = {
Signature · 1 line
1.7.9.5
PATCH 3/12 dma: mxs-dma: move to generic device tree binding
In reply to Shawn GuoUpdate mxs-dma driver to adopt generic DMA device tree binding. It calls of_dma_controller_register() with mxs specific of_dma_xlate to get the generic DMA device tree helper support. Then DMA clients only need to call dma_request_slave_channel() for requesting a DMA channel from dmaengine.
The existing way of requesting channel, clients directly call dma_request_channel(), still work there, and will be removed after all mxs-dma clients get converted to generic DMA device tree helper.
--- drivers/dma/mxs-dma.c | 77 ++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 73 insertions(+), 4 deletions(-)
drivers/dma/mxs-dma.c +73 -4
@@ -27,6 +27,7 @@#include <linux/stmp_device.h> #include <linux/of.h> #include <linux/of_device.h> +#include <linux/of_dma.h> #include <asm/irq.h>
@@ -139,6 +140,9 @@ struct mxs_dma_engine {struct dma_device dma_device;struct device_dma_parameters dma_parms;struct mxs_dma_chan mxs_chans[MXS_DMA_CHANNELS ]; +struct platform_device *pdev; +struct of_dma_filter_info filter_info; +unsigned int nr_channels; };struct mxs_dma_type {
@@ -350,10 +354,8 @@ static int mxs_dma_alloc_chan_resources(struct dma_chan *chan)struct mxs_dma_engine *mxs_dma = mxs_chan->mxs_dma;int ret; -if (!data) -return -EINVAL ; - - mxs_chan->chan_irq = data->chan_irq; +if (data) + mxs_chan->chan_irq = data->chan_irq; mxs_chan->ccw = dma_alloc_coherent(mxs_dma->dma_device.dev, CCW_BLOCK_SIZE, &mxs_chan->ccw_phys,
@@ -665,8 +667,59 @@ err_out:return ret; } +struct mxs_dma_filter_param { +struct device_node *of_node; +unsigned int chan_id; +}; + +struct dma_chan *mxs_dma_xlate (struct of_phandle_args *dma_spec, +struct of_dma *ofdma) +{ +struct mxs_dma_engine *mxs_dma = ofdma->of_dma_data; +struct of_dma_filter_info *info = &mxs_dma->filter_info; +struct mxs_dma_filter_param param; +int count = dma_spec->args_count; + +if (!info || !info->filter_fn) +return NULL ; + +if (count !=1 ) +return NULL ; + + param.of_node = ofdma->of_node; + param.chan_id = dma_spec->args[0 ]; + +if (param.chan_id >= mxs_dma->nr_channels) +return NULL ; + +return dma_request_channel (info->dma_cap, info->filter_fn, ¶m); +} + +static bool mxs_dma_filter_fn (struct dma_chan *chan,void *fn_param) +{ +struct mxs_dma_filter_param *param = fn_param; +struct mxs_dma_chan *mxs_chan =to_mxs_dma_chan (chan); +struct mxs_dma_engine *mxs_dma = mxs_chan->mxs_dma; +int chan_irq; + +if (mxs_dma->dma_device.dev->of_node != param->of_node) +return false; + +if (chan->chan_id != param->chan_id) +return false; + + chan_irq =platform_get_irq (mxs_dma->pdev, param->chan_id); +if (chan_irq <0 ) +return false; + + mxs_chan->chan_irq = chan_irq; + +return true; +} +static int __init mxs_dma_probe (struct platform_device *pdev) { +struct device_node *np = pdev->dev.of_node;const struct platform_device_id *id_entry;const struct of_device_id *of_id;const struct mxs_dma_type *dma_type;
@@ -678,6 +731,12 @@ static int __init mxs_dma_probe(struct platform_device *pdev)if (!mxs_dma)return -ENOMEM ; + ret =of_property_read_u32 (np,"dma-channels" , &mxs_dma->nr_channels); +if (ret) { +dev_err (&pdev->dev,"failed to read dma-channels\n" ); +return ret; + } + of_id =of_match_device (mxs_dma_dt_ids, &pdev->dev);if (of_id) id_entry = of_id->data;
@@ -723,6 +782,7 @@ static int __init mxs_dma_probe(struct platform_device *pdev)if (ret)return ret; + mxs_dma->pdev = pdev; mxs_dma->dma_device.dev = &pdev->dev;/* mxs_dma gets 65535 bytes maximum sg size */
@@ -743,6 +803,15 @@ static int __init mxs_dma_probe(struct platform_device *pdev)return ret; } + mxs_dma->filter_info.dma_cap = mxs_dma->dma_device.cap_mask; + mxs_dma->filter_info.filter_fn = mxs_dma_filter_fn; + ret =of_dma_controller_register (np, mxs_dma_xlate, mxs_dma); +if (ret) { +dev_err (mxs_dma->dma_device.dev, +"failed to register controller\n" ); +dma_async_device_unregister (&mxs_dma->dma_device); + } +dev_info (mxs_dma->dma_device.dev,"initialized\n" );return 0 ;
Signature · 1 line
1.7.9.5
PATCH 3/12 dma: mxs-dma: move to generic device tree binding
drivers/dma/mxs-dma.c struct mxs_dma_engine { · line 146
Quoted 8 lines · Shawn Guo
drivers/dma/mxs-dma.c +3
-0@@ -139,6 +140,9 @@ struct mxs_dma_engine {struct dma_device dma_device;struct device_dma_parameters dma_parms;struct mxs_dma_chan mxs_chans[MXS_DMA_CHANNELS ]; +struct platform_device *pdev; +struct of_dma_filter_info filter_info; +unsigned int nr_channels; };
I don't think using of_dma_filter_info makes sense when we are not using the generic xlate function:
Quoted 23 lines
drivers/dma/mxs-dma.c +19
-0@@ -665,8 +667,59 @@ err_out:return ret; } +struct mxs_dma_filter_param { +struct device_node *of_node; +unsigned int chan_id; +}; + +struct dma_chan *mxs_dma_xlate (struct of_phandle_args *dma_spec, +struct of_dma *ofdma) +{ +struct mxs_dma_engine *mxs_dma = ofdma->of_dma_data; +struct of_dma_filter_info *info = &mxs_dma->filter_info; +struct mxs_dma_filter_param param; +int count = dma_spec->args_count; + +if (!info || !info->filter_fn) +return NULL ; + +if (count !=1 ) +return NULL ; +
drivers/dma/mxs-dma.c err_out: · line 698
Quoted 10 lines
drivers/dma/mxs-dma.c +10
-0+ param.of_node = ofdma->of_node; + param.chan_id = dma_spec->args[0 ]; + +if (param.chan_id >= mxs_dma->nr_channels) +return NULL ; + +return dma_request_channel (info->dma_cap, info->filter_fn, ¶m); +} + +static bool mxs_dma_filter_fn (struct dma_chan *chan,void *fn_param)
You already know the value of info->filter_fn, it's always &mxs_dma_filter_fn, so if you reorder these two functions, you can pass it directly into dma_request_channel.
drivers/dma/mxs-dma.c err_out: · line 709
Quoted 10 lines
drivers/dma/mxs-dma.c +10
-0+struct mxs_dma_filter_param *param = fn_param; +struct mxs_dma_chan *mxs_chan =to_mxs_dma_chan (chan); +struct mxs_dma_engine *mxs_dma = mxs_chan->mxs_dma; +int chan_irq; + +if (mxs_dma->dma_device.dev->of_node != param->of_node) +return false; + +if (chan->chan_id != param->chan_id) +return false;
It would be nice if we could just get a pointer to the right channel structure from the xlate() function, rather than having to walk the list of channels and trying each one. Your method is what everyone else is doing as well, so we can stick with it for now, but at some point I'd like to open up this discussion again.
Arnd
PATCH 3/12 dma: mxs-dma: move to generic device tree binding
Quoted 12 lines · Arnd Bergmann
drivers/dma/mxs-dma.c struct mxs_dma_engine { · line 146
Quoted 8 lines · Shawn Guo
drivers/dma/mxs-dma.c +3
-0@@ -139,6 +140,9 @@ struct mxs_dma_engine {struct dma_device dma_device;struct device_dma_parameters dma_parms;struct mxs_dma_chan mxs_chans[MXS_DMA_CHANNELS ]; +struct platform_device *pdev; +struct of_dma_filter_info filter_info; +unsigned int nr_channels; };I don't think using
of_dma_filter_infomakes sense when we are not using the generic xlate function:
...
Quoted 5 lines
Quoted 1 line
+static bool
mxs_dma_filter_fn(struct dma_chan *chan, void *fn_param)You already know the value of
info->filter_fn, it's always&mxs_dma_filter_fn, so if you reorder these two functions, you can pass it directly intodma_request_channel.
Yeah, you're right. I was blindly following the generic xlate example. Thanks for spotting this. Will fix it.
Shawn
PATCH 4/12 mmc: mxs-mmc: move to use generic DMA helper
In reply to Shawn GuoWith the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call dma_request_slave_channel() for requesting a DMA channel from dmaengine.
Since mxs is a DT only platform now, along with the changes, the non-DT case checking in probe function also gets cleaned up.
--- Documentation/devicetree/bindings/mmc/mxs-mmc.txt | 12 ++++-- drivers/mmc/host/mxs-mmc.c | 48 +++------------------ 2 files changed, 13 insertions(+), 47 deletions(-)
Documentation/devicetree/bindings/mmc/mxs-mmc.txt +8 -4
@@ -9,15 +9,19 @@ and the properties used by the mxsmmc driver. Required properties: - compatible: Should be "fsl,<chip>-mmc". The supported chips include imx23 and imx28. -- interrupts: Should contain ERROR and DMA interrupts -- fsl,ssp-dma-channel: APBH DMA channel for the SSP +- interrupts: Should contain ERROR interrupt number +- dmas: DMA specifier, consisting of a phandle to DMA controller node + and SSP DMA channel ID. + Refer to dma.txt and fsl-mxs-dma.txt for details. +- dma-names: Must be "ssp". Examples: ssp0: ssp at 80010000 { compatible = "fsl,imx28-mmc"; reg = <0x80010000 2000>; - interrupts = <96 82>; - fsl,ssp-dma-channel = <0>; + interrupts = <96>; + dmas = <&dma_apbh 0>; + dma-names = "ssp"; bus-width = <8>; };
drivers/mmc/host/mxs-mmc.c +5 -43
80 lines not on this page · raw message
Signature · 1 line
1.7.9.5
PATCH 4/12 mmc: mxs-mmc: move to use generic DMA helper
Documentation/devicetree/bindings/mmc/mxs-mmc.txt and the properties used by the mxsmmc driver. · line 15
Quoted 9 lines · Shawn Guo
Required properties:
- compatible: Should be "fsl,<chip>-mmc". The supported chips include imx23 and imx28.
-- interrupts: Should contain ERROR and DMA interrupts
-- fsl,ssp-dma-channel: APBH DMA channel for the SSP+- interrupts: Should contain ERROR interrupt number +- dmas: DMA specifier, consisting of a phandle to DMA controller nodeDocumentation/devicetree/bindings/mmc/mxs-mmc.txt +2
-0+ and SSP DMA channel ID. + Refer to dma.txt and fsl-mxs-dma.txt for details.
I wonder if we should leave support for old device trees files around, at least for a while. Your patch removes a lot of unnecessary code if we decide not to worry about backwards compatibility here, but I could imagine that we see a few surprises here.
Quoted 1 line
+- dma-names: Must be "ssp".
I would prefer calling this "data" rather than "ssp". The name only has local significance in the ssp device, so calling the channel "ssp" seems wrong.
Arnd
PATCH 4/12 mmc: mxs-mmc: move to use generic DMA helper
Quoted 11 lines · Arnd Bergmann
Documentation/devicetree/bindings/mmc/mxs-mmc.txt and the properties used by the mxsmmc driver. · line 15
Quoted 9 lines · Shawn Guo
Required properties:
- compatible: Should be "fsl,<chip>-mmc". The supported chips include imx23 and imx28.
-- interrupts: Should contain ERROR and DMA interrupts
-- fsl,ssp-dma-channel: APBH DMA channel for the SSP+- interrupts: Should contain ERROR interrupt number +- dmas: DMA specifier, consisting of a phandle to DMA controller nodeDocumentation/devicetree/bindings/mmc/mxs-mmc.txt +2
-0+ and SSP DMA channel ID. + Refer to dma.txt and fsl-mxs-dma.txt for details.
Quoted 5 lines
I wonder if we should leave support for old device trees files around, at least for a while. Your patch removes a lot of unnecessary code if we decide not to worry about backwards compatibility here, but I could imagine that we see a few surprises here.
I'm less concerned by that at this point. Though platform mxs has been fully converted to device tree, device tree is still quite new for the platform. The incompatible device tree should be acceptable, considering we have noticed that when adding the temporary DMA binding for client devices.
/*
* TODO: This is a temporary solution and should be changed
* to use generic DMA binding later when the helper get in. */
Unless we have these TODO remarks leave in the tree forever, we will break old device tree sooner or later anyway. And I would have it happen sooner than later.
And I would start thinking about maintaining the device tree compatibility for mxs after generic DMA binding adoption.
Quoted 6 lines
Quoted 1 line
+- dma-names: Must be "ssp".
I would prefer calling this "data" rather than "ssp". The name only has local significance in the ssp device, so calling the channel "ssp" seems wrong.
Ok, will rename it per the use of the channel in the controller.
Shawn
PATCH 4/12 mmc: mxs-mmc: move to use generic DMA helper
Quoted 19 lines · Shawn Guo
Quoted 7 lines · Arnd Bergmann
On Wednesday 27 February 2013, Shawn Guo wrote:
I wonder if we should leave support for old device trees files around, at least for a while. Your patch removes a lot of unnecessary code if we decide not to worry about backwards compatibility here, but I could imagine that we see a few surprises here.
I'm less concerned by that at this point. Though platform mxs has been fully converted to device tree, device tree is still quite new for the platform. The incompatible device tree should be acceptable, considering we have noticed that when adding the temporary DMA binding for client devices.
/*
* TODO: This is a temporary solution and should be changed
* to use generic DMA binding later when the helper get in. */
Quoted 6 lines
Unless we have these TODO remarks leave in the tree forever, we will break old device tree sooner or later anyway. And I would have it happen sooner than later.
And I would start thinking about maintaining the device tree compatibility for mxs after generic DMA binding adoption.
Right. It's your decision anyway, and removing the code as early as possible has some advantages as well.
Arnd
PATCH 5/12 spi: mxs-spi: move to use generic DMA helper
In reply to Shawn GuoWith the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call dma_request_slave_channel() for requesting a DMA channel from dmaengine.
Since mxs is a DT only platform now, along with the changes, the non-DT case handling in probe function also gets removed.
--- Documentation/devicetree/bindings/spi/mxs-spi.txt | 12 +++-- drivers/spi/spi-mxs.c | 60 ++++----------------- include/linux/spi/mxs-spi.h | 4 +- 3 files changed, 18 insertions(+), 58 deletions(-)
Documentation/devicetree/bindings/spi/mxs-spi.txt +8 -4
@@ -3,8 +3,11 @@ Required properties: - compatible: Should be "fsl,<soc>-spi", where soc is "imx23" or "imx28" - reg: Offset and length of the register set for the device -- interrupts: Should contain SSP interrupts (error irq first, dma irq second) -- fsl,ssp-dma-channel: APBX DMA channel for the SSP +- interrupts: Should contain SSP ERROR interrupt +- dmas: DMA specifier, consisting of a phandle to DMA controller node + and SSP DMA channel ID. + Refer to dma.txt and fsl-mxs-dma.txt for details. +- dma-names: Must be "ssp". Optional properties: - clock-frequency : Input clock frequency to the SPI block in Hz.
@@ -17,6 +20,7 @@ ssp0: ssp at 80010000 { #size-cells = <0>; compatible = "fsl,imx28-spi"; reg = <0x80010000 0x2000>; - interrupts = <96 82>; - fsl,ssp-dma-channel = <0>; + interrupts = <96>; + dmas = <&dma_apbh 0>; + dma-names = "ssp"; };
drivers/spi/spi-mxs.c +9 -51
100 lines not on this page · raw message
include/linux/spi/mxs-spi.h +1 -3
@@ -24,7 +24,7 @@#ifndef __LINUX_SPI_MXS_SPI_H__#define __LINUX_SPI_MXS_SPI_H__ -#include <linux/fsl/mxs-dma.h> +#include <linux/dmaengine.h> #define ssp_is_old (host) ((host)->devid == IMX23_SSP)
@@ -137,9 +137,7 @@ struct mxs_ssp {unsigned int clk_rate;enum mxs_ssp_id devid; -int dma_channel;struct dma_chan *dmach; -struct mxs_dma_data dma_data;unsigned int dma_dir;enum dma_transfer_direction slave_dirn;u32 ssp_pio_words[SSP_PIO_NUM ];
Signature · 1 line
1.7.9.5
PATCH 5/12 spi: mxs-spi: move to use generic DMA helper
Quoted 10 lines · Shawn Guo
With the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call
dma_request_slave_channel()for requesting a DMA channel from dmaengine.Since mxs is a DT only platform now, along with the changes, the non-DT case handling in probe function also gets removed.
Signed-off-by Shawn Guo <[email protected]>Cc Grant Likely <[email protected]>Cc spi-devel-general at lists.sourceforge.net
PATCH 6/12 i2c: i2c-mxs: move to use generic DMA helper
In reply to Shawn GuoWith the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call dma_request_slave_channel() for requesting a DMA channel from dmaengine.
--- Documentation/devicetree/bindings/i2c/i2c-mxs.txt | 12 ++++--- drivers/i2c/busses/i2c-mxs.c | 40 ++------------------- 2 files changed, 11 insertions(+), 41 deletions(-)
Documentation/devicetree/bindings/i2c/i2c-mxs.txt +8 -4
@@ -3,10 +3,13 @@ Required properties: - compatible: Should be "fsl,<chip>-i2c" - reg: Should contain registers location and length -- interrupts: Should contain ERROR and DMA interrupts +- interrupts: Should contain ERROR interrupt number - clock-frequency: Desired I2C bus clock frequency in Hz. Only 100000Hz and 400000Hz modes are supported. -- fsl,i2c-dma-channel: APBX DMA channel for the I2C +- dmas: DMA specifier, consisting of a phandle to DMA controller node + and I2C DMA channel ID. + Refer to dma.txt and fsl-mxs-dma.txt for details. +- dma-names: Must be "i2c". Examples:
@@ -15,7 +18,8 @@ i2c0: i2c at 80058000 { #size-cells = <0>; compatible = "fsl,imx28-i2c"; reg = <0x80058000 2000>; - interrupts = <111 68>; + interrupts = <111>; clock-frequency = <100000>; - fsl,i2c-dma-channel = <6>; + dmas = <&dma_apbx 6>; + dma-names = "i2c"; };
drivers/i2c/busses/i2c-mxs.c +3 -37
@@ -31,7 +31,6 @@#include <linux/of_i2c.h> #include <linux/dma-mapping.h> #include <linux/dmaengine.h> -#include <linux/fsl/mxs-dma.h> #define DRIVER_NAME "mxs-i2c"
@@ -113,9 +112,7 @@ struct mxs_i2c_dev {uint32_t timing1;/* DMA support components */ -int dma_channel;struct dma_chan *dmach; -struct mxs_dma_data dma_data;uint32_t pio_data[2 ];uint32_t addr_data;struct scatterlist sg_io[2 ];
@@ -518,21 +515,6 @@ static const struct i2c_algorithm mxs_i2c_algo = { .functionality = mxs_i2c_func, }; -static bool mxs_i2c_dma_filter (struct dma_chan *chan,void *param) -{ -struct mxs_i2c_dev *i2c = param; - -if (!mxs_dma_is_apbx (chan)) -return false; - -if (chan->chan_id != i2c->dma_channel) -return false; - - chan->private = &i2c->dma_data; - -return true; -} -static void mxs_i2c_derive_timing (struct mxs_i2c_dev *i2c,int speed) {/* The I2C block clock run at 24MHz */
@@ -577,17 +559,6 @@ static int mxs_i2c_get_ofdata(struct mxs_i2c_dev *i2c)struct device_node *node = dev->of_node;int ret; -/* -* TODO: This is a temporary solution and should be changed -* to use generic DMA binding later when the helpers get in. -*/ - ret =of_property_read_u32 (node,"fsl,i2c-dma-channel" , - &i2c->dma_channel); -if (ret) { -dev_err (dev,"Failed to get DMA channel!\n" ); -return -ENODEV ; - } - ret =of_property_read_u32 (node,"clock-frequency" , &speed);if (ret) {dev_warn (dev,"No I2C speed selected, using 100kHz\n" );
@@ -607,8 +578,7 @@ static int mxs_i2c_probe(struct platform_device *pdev)struct pinctrl *pinctrl;struct resource *res;resource_size_t res_size; -int err, irq, dmairq; -dma_cap_mask_t mask; +int err, irq; pinctrl =devm_pinctrl_get_select_default (dev);if (IS_ERR (pinctrl))
@@ -620,9 +590,8 @@ static int mxs_i2c_probe(struct platform_device *pdev) res =platform_get_resource (pdev,IORESOURCE_MEM ,0 ); irq =platform_get_irq (pdev,0 ); - dmairq =platform_get_irq (pdev,1 ); -if (!res || irq <0 || dmairq <0 ) +if (!res || irq <0 )return -ENOENT ; res_size =resource_size (res);
@@ -648,10 +617,7 @@ static int mxs_i2c_probe(struct platform_device *pdev) }/* Setup the DMA */ -dma_cap_zero (mask); -dma_cap_set (DMA_SLAVE , mask); - i2c->dma_data.chan_irq = dmairq; - i2c->dmach =dma_request_channel (mask, mxs_i2c_dma_filter, i2c); + i2c->dmach =dma_request_slave_channel (dev,"i2c" );if (!i2c->dmach) {dev_err (dev,"Failed to request dma\n" );return -ENODEV ;
Signature · 1 line
1.7.9.5
PATCH 7/12 mtd: gpmi: move to use generic DMA helper
In reply to Shawn GuoWith the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call dma_request_slave_channel() for requesting a DMA channel from dmaengine.
--- .../devicetree/bindings/mtd/gpmi-nand.txt | 17 ++++--- drivers/mtd/nand/gpmi-nand/gpmi-nand.c | 51 +------------------- drivers/mtd/nand/gpmi-nand/gpmi-nand.h | 3 +- 3 files changed, 12 insertions(+), 59 deletions(-)
Documentation/devicetree/bindings/mtd/gpmi-nand.txt +10 -7
@@ -7,10 +7,12 @@ Required properties: - compatible : should be "fsl,<chip>-gpmi-nand" - reg : should contain registers location and length for gpmi and bch. - reg-names: Should contain the reg names "gpmi-nand" and "bch" - - interrupts : The first is the DMA interrupt number for GPMI. - The second is the BCH interrupt number. - - interrupt-names : The interrupt names "gpmi-dma", "bch"; - - fsl,gpmi-dma-channel : Should contain the dma channel it uses. + - interrupts : BCH interrupt number. + - interrupt-names : Should be "bch". + - dmas: DMA specifier, consisting of a phandle to DMA controller node + and GPMI DMA channel ID. + Refer to dma.txt and fsl-mxs-dma.txt for details. + - dma-names: Must be "gpmi". Optional properties: - nand-on-flash-bbt: boolean to enable on flash bbt option if not
@@ -27,9 +29,10 @@ gpmi-nand at 8000c000 { #size-cells = <1>; reg = <0x8000c000 2000>, <0x8000a000 2000>; reg-names = "gpmi-nand", "bch"; - interrupts = <88>, <41>; - interrupt-names = "gpmi-dma", "bch"; - fsl,gpmi-dma-channel = <4>; + interrupts = <41>; + interrupt-names = "bch"; + dmas = <&dma_apbh 4>; + dma-names = "gpmi"; partition at 0 { ...
drivers/mtd/nand/gpmi-nand/gpmi-nand.c +1 -50
72 lines not on this page · raw message
drivers/mtd/nand/gpmi-nand/gpmi-nand.h +1 -2
@@ -20,7 +20,7 @@#include <linux/mtd/nand.h> #include <linux/platform_device.h> #include <linux/dma-mapping.h> -#include <linux/fsl/mxs-dma.h> +#include <linux/dmaengine.h> #define GPMI_CLK_MAX 5/* MX6Q needs five clocks */ struct resources {
@@ -180,7 +180,6 @@ struct gpmi_nand_data {/* DMA channels */ #define DMA_CHANS 8struct dma_chan *dma_chans[DMA_CHANS ]; -struct mxs_dma_data dma_data;enum dma_ops_type last_dma_type;enum dma_ops_type dma_type;struct completion dma_done;
Signature · 1 line
1.7.9.5
PATCH 7/12 mtd: gpmi: move to use generic DMA helper
? 2013?02?27? 23:25, Shawn Guo ??:
Documentation/devicetree/bindings/mtd/gpmi-nand.txt Required properties: · line 11
Quoted 11 lines
+++
b/Documentation/devicetree/bindings/mtd/gpmi-nand.txtDocumentation/devicetree/bindings/mtd/gpmi-nand.txt +2
-4@@ -7,10 +7,12 @@ Required properties: - compatible : should be "fsl,<chip>-gpmi-nand" - reg : should contain registers location and length for gpmi and bch. - reg-names: Should contain the reg names "gpmi-nand" and "bch" - - interrupts : The first is the DMA interrupt number for GPMI. - The second is the BCH interrupt number. - - interrupt-names : The interrupt names "gpmi-dma", "bch"; - - fsl,gpmi-dma-channel : Should contain the dma channel it uses. + - interrupts : BCH interrupt number. + - interrupt-names : Should be "bch".
why split out the bch interrupt? It seems no use to do it.
thanks
Huang Shijie
PATCH 7/12 mtd: gpmi: move to use generic DMA helper
Quoted 14 lines · Huang Shijie
? 2013?02?27? 23:25, Shawn Guo ??:
Documentation/devicetree/bindings/mtd/gpmi-nand.txt Required properties: · line 11
Quoted 11 lines
+++
b/Documentation/devicetree/bindings/mtd/gpmi-nand.txtDocumentation/devicetree/bindings/mtd/gpmi-nand.txt +2
-4@@ -7,10 +7,12 @@ Required properties: - compatible : should be "fsl,<chip>-gpmi-nand" - reg : should contain registers location and length for gpmi and bch. - reg-names: Should contain the reg names "gpmi-nand" and "bch" - - interrupts : The first is the DMA interrupt number for GPMI. - The second is the BCH interrupt number. - - interrupt-names : The interrupt names "gpmi-dma", "bch"; - - fsl,gpmi-dma-channel : Should contain the dma channel it uses. + - interrupts : BCH interrupt number. + - interrupt-names : Should be "bch".why split out the bch interrupt? It seems no use to do it.
The real change here is to remove gpmi-dma interrupt, since we have all dma interrupts defined in dma controller node.
Shawn
PATCH 7/12 mtd: gpmi: move to use generic DMA helper
In reply to Shawn GuoQuoted 7 lines · Shawn Guo
With the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call
dma_request_slave_channel()for requesting a DMA channel from dmaengine.Signed-off-by Shawn Guo <[email protected]>Cc Artem Bityutskiy <[email protected]>Cc linux-mtd at lists.infradead.org
It looks like this ne is rather independent and can go via the MTD tree. So pushed this one to l2-mtd.git - let me know if I am confused. Thanks!
Signature · 2 lines
Best Regards, Artem Bityutskiy
PATCH 7/12 mtd: gpmi: move to use generic DMA helper
Quoted 11 lines · Artem Bityutskiy
Quoted 7 lines · Shawn Guo
With the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call
dma_request_slave_channel()for requesting a DMA channel from dmaengine.Signed-off-by Shawn Guo <[email protected]>Cc Artem Bityutskiy <[email protected]>Cc linux-mtd at lists.infradead.orgIt looks like this ne is rather independent and can go via the MTD tree. So pushed this one to l2-mtd.git - let me know if I am confused. Thanks!
I think you should not apply this patch on a tree that does not already have the first three patches in the series. The easiest way would be to keep the series together. As an alternative, Shawn could create a git branch with those three patches and let you (and the other subsystem maintainers) pull that branch in before applying the subsystem specific patch.
Arnd
PATCH 7/12 mtd: gpmi: move to use generic DMA helper
Quoted 13 lines · Arnd Bergmann
Quoted 11 lines · Artem Bityutskiy
Quoted 7 lines · Shawn Guo
With the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call
dma_request_slave_channel()for requesting a DMA channel from dmaengine.Signed-off-by Shawn Guo <[email protected]>Cc Artem Bityutskiy <[email protected]>Cc linux-mtd at lists.infradead.orgIt looks like this ne is rather independent and can go via the MTD tree. So pushed this one to l2-mtd.git - let me know if I am confused. Thanks!
Quoted 6 lines
I think you should not apply this patch on a tree that does not already have the first three patches in the series. The easiest way would be to keep the series together. As an alternative, Shawn could create a git branch with those three patches and let you (and the other subsystem maintainers) pull that branch in before applying the subsystem specific patch.
OK, this one looked independent to me. Anyway, dropped it and for this one:
Signature · 2 lines
Best Regards, Artem Bityutskiy
PATCH 7/12 mtd: gpmi: move to use generic DMA helper
In reply to Artem BityutskiyQuoted 11 lines · Artem Bityutskiy
Quoted 7 lines · Shawn Guo
With the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call
dma_request_slave_channel()for requesting a DMA channel from dmaengine.Signed-off-by Shawn Guo <[email protected]>Cc Artem Bityutskiy <[email protected]>Cc linux-mtd at lists.infradead.orgIt looks like this ne is rather independent and can go via the MTD tree. So pushed this one to l2-mtd.git - let me know if I am confused. Thanks!
I suggest do not apply this patch. the whole patch set is still under review now.
Best Regards
Huang Shijie
Quoted 10 lines
--
Best Regards,
Artem Bityutskiy_______________________________________________
linux-arm-kernel mailing list
linux-arm-kernel at lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
PATCH 7/12 mtd: gpmi: move to use generic DMA helper
In reply to Artem BityutskiyArtem,
Quoted 11 lines · Artem Bityutskiy
Quoted 7 lines · Shawn Guo
With the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call
dma_request_slave_channel()for requesting a DMA channel from dmaengine.Signed-off-by Shawn Guo <[email protected]>Cc Artem Bityutskiy <[email protected]>Cc linux-mtd at lists.infradead.orgIt looks like this ne is rather independent and can go via the MTD tree. So pushed this one to l2-mtd.git - let me know if I am confused. Thanks!
First of all, there is a new version (v2) of the patch (series) to address review comments.
Secondly, as Arnd pointed it out, you shouldn't apply this single patch, because it depends on the first 3 patches in the series to work. Applying this single patch will result in a non-functional gpmi driver.
As I mentioned in the cover letter, I prefer to have your ACK on the patch to merge the series through arm-soc tree as a whole. If you insist that you merge the patch through your tree, I can also create a topic branch containing the first 3 patches, as Arnd suggests. Please just let me know.
Shawn
PATCH 8/12 serial: mxs-auart: move to use generic DMA helper
In reply to Shawn GuoWith the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call dma_request_slave_channel() for requesting a DMA channel from dmaengine.
--- .../bindings/tty/serial/fsl-mxs-auart.txt | 16 +++--- drivers/tty/serial/mxs-auart.c | 52 ++------------------ 2 files changed, 11 insertions(+), 57 deletions(-)
Documentation/devicetree/bindings/tty/serial/fsl-mxs-auart.txt +7 -9
@@ -5,20 +5,18 @@ Required properties: imx23 and imx28. - reg : Address and length of the register set for the device - interrupts : Should contain the auart interrupt numbers - -Optional properties: -- fsl,auart-dma-channel : The DMA channels, the first is for RX, the other - is for TX. If you add this property, it also means that you - will enable the DMA support for the auart. - Note: due to the hardware bug in imx23(see errata : 2836), - only the imx28 can enable the DMA support for the auart. +- dmas: DMA specifier, consisting of a phandle to DMA controller node + and AUART DMA channel ID. + Refer to dma.txt and fsl-mxs-dma.txt for details. +- dma-names: "rx" for RX channel, "tx" for TX channel. Example: auart0: serial at 8006a000 { compatible = "fsl,imx28-auart", "fsl,imx23-auart"; reg = <0x8006a000 0x2000>; - interrupts = <112 70 71>; - fsl,auart-dma-channel = <8 9>; + interrupts = <112>; + dmas = <&dma_apbx 8>, <&dma_apbx 9>; + dma-names = "rx", "tx"; }; Note: Each auart port should have an alias correctly numbered in "aliases"
drivers/tty/serial/mxs-auart.c +4 -48
99 lines not on this page · raw message
Signature · 1 line
1.7.9.5
PATCH 9/12 ASoC: dmaengine_pcm: support use of generic DMA helper
In reply to Shawn GuoWith generic DMA device tree binding and helper function dma_request_slave_channel() in place, dmaengine_pcm should support that in requesting DMA channel for users that support generic DMA device tree binding.
Instead of inventing a new API, it defines the parameters needed by dma_request_slave_channel() into struct snd_dma_channel_params, interprets filter_data into snd_dma_channel_params, and calls the helper in case that dmaengine_pcm users pass in a NULL filter_fn.
Then, dmaengine_pcm users can call snd_dmaengine_pcm_open() with NULL filter_fn and snd_dma_channel_params being filter_data to direct the API to request DMA channel using generic DMA helper.
--- include/sound/dmaengine_pcm.h | 5 +++++ sound/soc/soc-dmaengine-pcm.c | 19 +++++++++++++------ 2 files changed, 18 insertions(+), 6 deletions(-)
include/sound/dmaengine_pcm.h +5 -0
@@ -18,6 +18,11 @@#include <sound/pcm.h> #include <linux/dmaengine.h> +struct snd_dma_channel_params { +struct device *dev; +char *name; +}; +/** * snd_pcm_substream_to_dma_direction - Get dma_transfer_direction for a PCM * substream
sound/soc/soc-dmaengine-pcm.c +13 -6
@@ -247,12 +247,19 @@ EXPORT_SYMBOL_GPL(snd_dmaengine_pcm_pointer);static int dmaengine_pcm_request_channel (struct dmaengine_pcm_runtime_data *prtd,dma_filter_fn filter_fn,void *filter_data) { -dma_cap_mask_t mask; - -dma_cap_zero (mask); -dma_cap_set (DMA_SLAVE , mask); -dma_cap_set (DMA_CYCLIC , mask); - prtd->dma_chan =dma_request_channel (mask, filter_fn, filter_data); +if (filter_fn) { +dma_cap_mask_t mask; + +dma_cap_zero (mask); +dma_cap_set (DMA_SLAVE , mask); +dma_cap_set (DMA_CYCLIC , mask); + prtd->dma_chan =dma_request_channel (mask, filter_fn, + filter_data); + }else { +struct snd_dma_channel_params *params = filter_data; + prtd->dma_chan =dma_request_slave_channel (params->dev, + params->name); + }if (!prtd->dma_chan)return -ENXIO ;
Signature · 1 line
1.7.9.5
Re: [PATCH 09/12] ASoC: dmaengine_pcm: support use of generic DMA helper
Quoted 10 lines · Shawn Guo
With generic DMA device tree binding and helper function
dma_request_slave_channel()in place,dmaengine_pcmshould support that in requesting DMA channel for users that support generic DMA device tree binding.Instead of inventing a new API, it defines the parameters needed by
dma_request_slave_channel()into structsnd_dma_channel_params, interpretsfilter_dataintosnd_dma_channel_params, and calls the helper in case thatdmaengine_pcmusers pass in a NULLfilter_fn.
Quoted 7 lines
Then,
dmaengine_pcmusers can callsnd_dmaengine_pcm_open()with NULLfilter_fnandsnd_dma_channel_paramsbeingfilter_datato direct the API to request DMA channel using generic DMA helper.Signed-off-by Shawn Guo <[email protected]>Cc Mark Brown <[email protected]>Cc alsa-devel at alsa-project.org
I would actually prefer having a new API in the soc-dmaengine-pcm module, like
static int dmaengine_pcm_request_slave_channel(struct dmaengine_pcm_runtime_data *prtd,
struct device *dev, const char *id);
For sound drivers that are fully converted to using DT, it would be a more natural interface to use IMHO.
Your patch looks technically correct though, so it's up to Mark to decide what he prefers.
Arnd
Re: [PATCH 09/12] ASoC: dmaengine_pcm: support use of generic DMA helper
Quoted 19 lines · Arnd Bergmann
Quoted 10 lines · Shawn Guo
With generic DMA device tree binding and helper function
dma_request_slave_channel()in place,dmaengine_pcmshould support that in requesting DMA channel for users that support generic DMA device tree binding.Instead of inventing a new API, it defines the parameters needed by
dma_request_slave_channel()into structsnd_dma_channel_params, interpretsfilter_dataintosnd_dma_channel_params, and calls the helper in case thatdmaengine_pcmusers pass in a NULLfilter_fn.Quoted 7 lines
Then,
dmaengine_pcmusers can callsnd_dmaengine_pcm_open()with NULLfilter_fnandsnd_dma_channel_paramsbeingfilter_datato direct the API to request DMA channel using generic DMA helper.
Quoted 9 lines
I would actually prefer having a new API in the soc-dmaengine-pcm module, like
static int dmaengine_pcm_request_slave_channel(struct
dmaengine_pcm_runtime_data*prtd,struct device *dev, const char *id);For sound drivers that are fully converted to using DT, it would be a more natural interface to use IMHO.
Ok, let's do this. While my patch saves a new API, it adds a new struct anyway, which I'm not happy with.
Shawn
Re: [PATCH 09/12] ASoC: dmaengine_pcm: support use of generic DMA helper
In reply to Arnd BergmannQuoted 8 lines · Arnd Bergmann
I would actually prefer having a new API in the soc-dmaengine-pcm module, like
static int dmaengine_pcm_request_slave_channel(struct
dmaengine_pcm_runtime_data*prtd,struct device *dev, const char *id);For sound drivers that are fully converted to using DT, it would be a more natural interface to use IMHO.
Yes, please.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.infradead.org/pipermail/linux-arm-kernel/attachments/20130301/ddab8998/attachment.sig>
Re: [PATCH 09/12] ASoC: dmaengine_pcm: support use of generic DMA helper
In reply to Arnd BergmannQuoted 19 lines · Arnd Bergmann
Quoted 10 lines · Shawn Guo
With generic DMA device tree binding and helper function
dma_request_slave_channel()in place,dmaengine_pcmshould support that in requesting DMA channel for users that support generic DMA device tree binding.Instead of inventing a new API, it defines the parameters needed by
dma_request_slave_channel()into structsnd_dma_channel_params, interpretsfilter_dataintosnd_dma_channel_params, and calls the helper in case thatdmaengine_pcmusers pass in a NULLfilter_fn.Quoted 7 lines
Then,
dmaengine_pcmusers can callsnd_dmaengine_pcm_open()with NULLfilter_fnandsnd_dma_channel_paramsbeingfilter_datato direct the API to request DMA channel using generic DMA helper.
Quoted 8 lines
I would actually prefer having a new API in the soc-dmaengine-pcm module, like
static int dmaengine_pcm_request_slave_channel(struct
dmaengine_pcm_runtime_data*prtd,struct device *dev, const char *id);For sound drivers that are fully converted to using DT, it would be a more natural interface to use IMHO.
Just to be clear, if we choose to have a new API, it's not the above one but something like
int snd_dmaengine_generic_pcm_open(struct snd_pcm_substream *substream,
struct device *dev, const char *id)
as snd_dmaengine_pcm_open() is the interface to soc-dmaengine-pcm clients, not dmaengine_pcm_request_channel().
Shawn
PATCH 10/12 ASoC: mxs: move to use generic DMA helper
In reply to Shawn GuoWith the generic DMA device tree helper supported by mxs-dma driver, client devices only need to call dma_request_slave_channel() for requesting a DMA channel from dmaengine.
Call snd_dmaengine_pcm_open() with NULL filter_fn, and having parameters needed by dma_request_slave_channel() wrap into snd_dma_channel_params and pass it in as filter_data, so that dmaengine_pcm will call the helper rather than dma_request_channel() to request DMA channel.
--- .../devicetree/bindings/sound/mxs-saif.txt | 18 ++++---- sound/soc/mxs/mxs-pcm.c | 43 ++------------------ sound/soc/mxs/mxs-pcm.h | 5 --- sound/soc/mxs/mxs-saif.c | 28 ++----------- sound/soc/mxs/mxs-saif.h | 3 +- 5 files changed, 20 insertions(+), 77 deletions(-)
Documentation/devicetree/bindings/sound/mxs-saif.txt +11 -7
@@ -3,8 +3,11 @@ Required properties: - compatible: Should be "fsl,<chip>-saif" - reg: Should contain registers location and length -- interrupts: Should contain ERROR and DMA interrupts -- fsl,saif-dma-channel: APBX DMA channel for the SAIF +- interrupts: Should contain ERROR interrupt number +- dmas: DMA specifier, consisting of a phandle to DMA controller node + and SAIF DMA channel ID. + Refer to dma.txt and fsl-mxs-dma.txt for details. +- dma-names: Must be "saif". Optional properties: - fsl,saif-master: phandle to the master SAIF. It's only required for
@@ -23,14 +26,15 @@ aliases { saif0: saif at 80042000 { compatible = "fsl,imx28-saif"; reg = <0x80042000 2000>; - interrupts = <59 80>; - fsl,saif-dma-channel = <4>; + interrupts = <59>; + dmas = <&dma_apbx 4>; + dma-names = "saif"; }; saif1: saif at 80046000 { compatible = "fsl,imx28-saif"; reg = <0x80046000 2000>; - interrupts = <58 81>; - fsl,saif-dma-channel = <5>; - fsl,saif-master = <&saif0>; + interrupts = <58>; + dmas = <&dma_apbx 5>; + dma-names = "saif"; };
sound/soc/mxs/mxs-pcm.c +4 -39
79 lines not on this page · raw message
sound/soc/mxs/mxs-pcm.h +0 -5
@@ -19,11 +19,6 @@#ifndef _MXS_PCM_H#define _MXS_PCM_H -struct mxs_pcm_dma_params { -int chan_irq; -int chan_num; -}; -int mxs_pcm_platform_register (struct device *dev);void mxs_pcm_platform_unregister (struct device *dev);
sound/soc/mxs/mxs-saif.c +3 -25
@@ -26,7 +26,6 @@#include <linux/clk.h> #include <linux/delay.h> #include <linux/time.h> -#include <linux/fsl/mxs-dma.h> #include <linux/pinctrl/consumer.h> #include <sound/core.h> #include <sound/pcm.h>
@@ -659,7 +658,7 @@ static irqreturn_t mxs_saif_irq(int irq, void *dev_id)static int mxs_saif_probe (struct platform_device *pdev) {struct device_node *np = pdev->dev.of_node; -struct resource *iores, *dmares; +struct resource *iores;struct mxs_saif *saif;struct pinctrl *pinctrl;int ret =0 ;
@@ -721,22 +720,6 @@ static int mxs_saif_probe(struct platform_device *pdev)if (IS_ERR (saif->base))return PTR_ERR (saif->base); - dmares =platform_get_resource (pdev,IORESOURCE_DMA ,0 ); -if (!dmares) { -/* -* TODO: This is a temporary solution and should be changed -* to use generic DMA binding later when the helplers get in. -*/ - ret =of_property_read_u32 (np,"fsl,saif-dma-channel" , - &saif->dma_param.chan_num); -if (ret) { -dev_err (&pdev->dev,"failed to get dma channel\n" ); -return ret; - } - }else { - saif->dma_param.chan_num = dmares->start; - } - saif->irq =platform_get_irq (pdev,0 );if (saif->irq <0 ) { ret = saif->irq;
@@ -753,13 +736,8 @@ static int mxs_saif_probe(struct platform_device *pdev)return ret; } - saif->dma_param.chan_irq =platform_get_irq (pdev,1 ); -if (saif->dma_param.chan_irq <0 ) { - ret = saif->dma_param.chan_irq; -dev_err (&pdev->dev,"failed to get dma irq resource: %d\n" , - ret); -return ret; - } + saif->dma_param.dev = saif->dev; + saif->dma_param.name ="saif" ;platform_set_drvdata (pdev, saif);
sound/soc/mxs/mxs-saif.h +2 -1
@@ -108,6 +108,7 @@#define MXS_SAIF_MCLK 0 +#include <sound/dmaengine_pcm.h> #include "mxs-pcm.h" struct mxs_saif {
@@ -117,7 +118,7 @@ struct mxs_saif {unsigned int mclk_in_use;void __iomem *base;int irq; -struct mxs_pcm_dma_params dma_param; +struct snd_dma_channel_params dma_param;unsigned int id;unsigned int master_id;unsigned int cur_rate;
Signature · 1 line
1.7.9.5
PATCH 11/12 dma: mxs-dma: remove code left from generic DMA binding conversion
In reply to Shawn GuoWith all mxs-dma clients moved to use generic DMA helper, the code left from generic DMA binding conversion can be removed now.
--- drivers/dma/mxs-dma.c | 23 ----------------------- include/linux/fsl/mxs-dma.h | 20 -------------------- 2 files changed, 43 deletions(-) delete mode 100644 include/linux/fsl/mxs-dma.h
drivers/dma/mxs-dma.c +0 -23
@@ -23,7 +23,6 @@#include <linux/dmaengine.h> #include <linux/delay.h> #include <linux/module.h> -#include <linux/fsl/mxs-dma.h> #include <linux/stmp_device.h> #include <linux/of.h> #include <linux/of_device.h>
@@ -198,24 +197,6 @@ static struct mxs_dma_chan *to_mxs_dma_chan(struct dma_chan *chan) return container_of(chan, struct mxs_dma_chan, chan); } -int mxs_dma_is_apbh (struct dma_chan *chan) -{ -struct mxs_dma_chan *mxs_chan =to_mxs_dma_chan (chan); -struct mxs_dma_engine *mxs_dma = mxs_chan->mxs_dma; - -return dma_is_apbh (mxs_dma); -} -EXPORT_SYMBOL_GPL (mxs_dma_is_apbh); - -int mxs_dma_is_apbx (struct dma_chan *chan) -{ -struct mxs_dma_chan *mxs_chan =to_mxs_dma_chan (chan); -struct mxs_dma_engine *mxs_dma = mxs_chan->mxs_dma; - -return !dma_is_apbh (mxs_dma); -} -EXPORT_SYMBOL_GPL (mxs_dma_is_apbx); -static void mxs_dma_reset_chan (struct mxs_dma_chan *mxs_chan) {struct mxs_dma_engine *mxs_dma = mxs_chan->mxs_dma;
@@ -350,13 +331,9 @@ static irqreturn_t mxs_dma_int_handler(int irq, void *dev_id)static int mxs_dma_alloc_chan_resources (struct dma_chan *chan) {struct mxs_dma_chan *mxs_chan =to_mxs_dma_chan (chan); -struct mxs_dma_data *data = chan->private;struct mxs_dma_engine *mxs_dma = mxs_chan->mxs_dma;int ret; -if (data) - mxs_chan->chan_irq = data->chan_irq; - mxs_chan->ccw =dma_alloc_coherent (mxs_dma->dma_device.dev,CCW_BLOCK_SIZE , &mxs_chan->ccw_phys,GFP_KERNEL );
include/linux/fsl/mxs-dma.h deleted +0 -20
@@ -1,20 +0,0 @@ -/* -* Copyright 2011 Freescale Semiconductor, Inc. All Rights Reserved. -* -* This program is free software; you can redistribute it and/or modify -* it under the terms of the GNU General Public License version 2 as -* published by the Free Software Foundation. -*/ - -#ifndef __MACH_MXS_DMA_H__ -#define __MACH_MXS_DMA_H__ - -#include <linux/dmaengine.h> - -struct mxs_dma_data { -int chan_irq; -}; - -extern int mxs_dma_is_apbh (struct dma_chan *chan); -extern int mxs_dma_is_apbx (struct dma_chan *chan); -#endif /* __MACH_MXS_DMA_H__ */
Signature · 1 line
1.7.9.5
PATCH 12/12 ARM: dts: remove mxs-dma channel interrupt number from client nodes
In reply to Shawn GuoWith adopting generic DMA device tree binding, all mxs-dma channel interrupt numbers are defined in DMA controller nodes, so they can be removed from client nodes.
--- arch/arm/boot/dts/imx23.dtsi | 12 ++++++------ arch/arm/boot/dts/imx28.dtsi | 36 ++++++++++++++++++------------------ arch/arm/boot/dts/imx6qdl.dtsi | 4 ++-- 3 files changed, 26 insertions(+), 26 deletions(-)
arch/arm/boot/dts/imx23.dtsi +6 -6
@@ -72,8 +72,8 @@ #size-cells = <1>; reg = <0x8000c000 0x2000>, <0x8000a000 0x2000>; reg-names = "gpmi-nand", "bch"; - interrupts = <13>, <56>; - interrupt-names = "gpmi-dma", "bch"; + interrupts = <56>; + interrupt-names = "bch"; clocks = <&clks 34>; clock-names = "gpmi_io"; dmas = <&dma_apbh 4>;
@@ -84,7 +84,7 @@ ssp0: ssp at 80010000 { reg = <0x80010000 0x2000>; - interrupts = <15 14>; + interrupts = <15>; clocks = <&clks 33>; dmas = <&dma_apbh 1>; dma-names = "ssp";
@@ -360,7 +360,7 @@ ssp1: ssp at 80034000 { reg = <0x80034000 0x2000>; - interrupts = <2 20>; + interrupts = <2>; clocks = <&clks 33>; dmas = <&dma_apbh 2>; dma-names = "ssp";
@@ -465,7 +465,7 @@ auart0: serial at 8006c000 { compatible = "fsl,imx23-auart"; reg = <0x8006c000 0x2000>; - interrupts = <24 25 23>; + interrupts = <24>; clocks = <&clks 32>; dmas = <&dma_apbx 6>, <&dma_apbx 7>; dma-names = "rx", "tx";
@@ -475,7 +475,7 @@ auart1: serial at 8006e000 { compatible = "fsl,imx23-auart"; reg = <0x8006e000 0x2000>; - interrupts = <59 60 58>; + interrupts = <59>; clocks = <&clks 32>; dmas = <&dma_apbx 8>, <&dma_apbx 9>; dma-names = "rx", "tx";
arch/arm/boot/dts/imx28.dtsi +18 -18
@@ -60,7 +60,7 @@ hsadc at 80002000 { reg = <0x80002000 0x2000>; - interrupts = <13 87>; + interrupts = <13>; dmas = <&dma_apbh 12>; dma-names = "hsadc"; status = "disabled";
@@ -94,8 +94,8 @@ #size-cells = <1>; reg = <0x8000c000 0x2000>, <0x8000a000 0x2000>; reg-names = "gpmi-nand", "bch"; - interrupts = <88>, <41>; - interrupt-names = "gpmi-dma", "bch"; + interrupts = <41>; + interrupt-names = "bch"; clocks = <&clks 50>; clock-names = "gpmi_io"; dmas = <&dma_apbh 4>;
@@ -108,7 +108,7 @@ #address-cells = <1>; #size-cells = <0>; reg = <0x80010000 0x2000>; - interrupts = <96 82>; + interrupts = <96>; clocks = <&clks 46>; dmas = <&dma_apbh 0>; dma-names = "ssp";
@@ -120,7 +120,7 @@ #address-cells = <1>; #size-cells = <0>; reg = <0x80012000 0x2000>; - interrupts = <97 83>; + interrupts = <97>; clocks = <&clks 47>; dmas = <&dma_apbh 1>; dma-names = "ssp";
@@ -132,7 +132,7 @@ #address-cells = <1>; #size-cells = <0>; reg = <0x80014000 0x2000>; - interrupts = <98 84>; + interrupts = <98>; clocks = <&clks 48>; dmas = <&dma_apbh 2>; dma-names = "ssp";
@@ -144,7 +144,7 @@ #address-cells = <1>; #size-cells = <0>; reg = <0x80016000 0x2000>; - interrupts = <99 85>; + interrupts = <99>; clocks = <&clks 49>; dmas = <&dma_apbh 3>; dma-names = "ssp";
@@ -720,7 +720,7 @@ lcdif at 80030000 { compatible = "fsl,imx28-lcdif"; reg = <0x80030000 0x2000>; - interrupts = <38 86>; + interrupts = <38>; clocks = <&clks 55>; dmas = <&dma_apbh 13>; dma-names = "lcdif";
@@ -797,7 +797,7 @@ saif0: saif at 80042000 { compatible = "fsl,imx28-saif"; reg = <0x80042000 0x2000>; - interrupts = <59 80>; + interrupts = <59>; clocks = <&clks 53>; dmas = <&dma_apbx 4>; dma-names = "saif";
@@ -813,7 +813,7 @@ saif1: saif at 80046000 { compatible = "fsl,imx28-saif"; reg = <0x80046000 0x2000>; - interrupts = <58 81>; + interrupts = <58>; clocks = <&clks 54>; dmas = <&dma_apbx 5>; dma-names = "saif";
@@ -831,7 +831,7 @@ spdif at 80054000 { reg = <0x80054000 0x2000>; - interrupts = <45 66>; + interrupts = <45>; dmas = <&dma_apbx 2>; dma-names = "spdif"; status = "disabled";
@@ -848,7 +848,7 @@ #size-cells = <0>; compatible = "fsl,imx28-i2c"; reg = <0x80058000 0x2000>; - interrupts = <111 68>; + interrupts = <111>; clock-frequency = <100000>; dmas = <&dma_apbx 6>; dma-names = "i2c";
@@ -861,7 +861,7 @@ #size-cells = <0>; compatible = "fsl,imx28-i2c"; reg = <0x8005a000 0x2000>; - interrupts = <110 69>; + interrupts = <110>; clock-frequency = <100000>; dmas = <&dma_apbx 7>; dma-names = "i2c";
@@ -887,7 +887,7 @@ auart0: serial at 8006a000 { compatible = "fsl,imx28-auart", "fsl,imx23-auart"; reg = <0x8006a000 0x2000>; - interrupts = <112 70 71>; + interrupts = <112>; dmas = <&dma_apbx 8>, <&dma_apbx 9>; dma-names = "rx", "tx"; fsl,auart-dma-channel = <8 9>;
@@ -898,7 +898,7 @@ auart1: serial at 8006c000 { compatible = "fsl,imx28-auart", "fsl,imx23-auart"; reg = <0x8006c000 0x2000>; - interrupts = <113 72 73>; + interrupts = <113>; dmas = <&dma_apbx 10>, <&dma_apbx 11>; dma-names = "rx", "tx"; clocks = <&clks 45>;
@@ -908,7 +908,7 @@ auart2: serial at 8006e000 { compatible = "fsl,imx28-auart", "fsl,imx23-auart"; reg = <0x8006e000 0x2000>; - interrupts = <114 74 75>; + interrupts = <114>; dmas = <&dma_apbx 12>, <&dma_apbx 13>; dma-names = "rx", "tx"; clocks = <&clks 45>;
@@ -918,7 +918,7 @@ auart3: serial at 80070000 { compatible = "fsl,imx28-auart", "fsl,imx23-auart"; reg = <0x80070000 0x2000>; - interrupts = <115 76 77>; + interrupts = <115>; dmas = <&dma_apbx 14>, <&dma_apbx 15>; dma-names = "rx", "tx"; clocks = <&clks 45>;
@@ -928,7 +928,7 @@ auart4: serial at 80072000 { compatible = "fsl,imx28-auart", "fsl,imx23-auart"; reg = <0x80072000 0x2000>; - interrupts = <116 78 79>; + interrupts = <116>; dmas = <&dma_apbx 0>, <&dma_apbx 1>; dma-names = "rx", "tx"; clocks = <&clks 45>;
arch/arm/boot/dts/imx6qdl.dtsi +2 -2
@@ -81,8 +81,8 @@ #size-cells = <1>; reg = <0x00112000 0x2000>, <0x00114000 0x2000>; reg-names = "gpmi-nand", "bch"; - interrupts = <0 13 0x04>, <0 15 0x04>; - interrupt-names = "gpmi-dma", "bch"; + interrupts = <0 15 0x04>; + interrupt-names = "bch"; clocks = <&clks 152>, <&clks 153>, <&clks 151>, <&clks 150>, <&clks 149>; clock-names = "gpmi_io", "gpmi_apb", "gpmi_bch",
Signature · 1 line
1.7.9.5
cover ARM: mxs: move to generic DMA device tree binding
In reply to Shawn GuoQuoted 7 lines · Shawn Guo
The series converts mxs-dma and its clients to generic DMA device tree binding/helper.
Subsystem maintainers,
I need your ACKs to have the series go via arm-soc as a whole if the patches look good to you.
Great stuff!
I have a few smaller comments (my comments on patch 4 also apply to a lot of the other ones), but I'm very happy with the conversion overall.
Arnd
cover ARM: mxs: move to generic DMA device tree binding
Dear Arnd Bergmann,
Quoted 9 lines · Shawn Guo
Quoted 7 lines
The series converts mxs-dma and its clients to generic DMA device tree binding/helper.
Subsystem maintainers,
I need your ACKs to have the series go via arm-soc as a whole if the patches look good to you.
Quoted 7 lines
Great stuff!
I have a few smaller comments (my comments on patch 4 also apply to a lot of the other ones), but I'm very happy with the conversion overall.
Arnd
So true, amazing stuff. I'm currently snowed by email, so I'll try reviewing only from version 2 on.
Thanks!
Best regards,
Marek Vasut