| Message ID | 1347972540-29667-1-git-send-email-stigge@antcom.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, Sep 18, 2012 at 9:49 AM, Roland Stigge <stigge@antcom.de> wrote: > Even with devicetree support, we needed platform data to provide some data like > bus_id and enable_dma, leading to mixed device tree and platform data. This > patch makes it possible to provide all that information via device tree. Now, > the data must be provided via platform data _or_ device tree completely. > > Only in case of callback specification (dma_filter()), platform data is > necessary. > > Signed-off-by: Roland Stigge <stigge@antcom.de> > Acked-by: Linus Walleij <linus.walleij@linaro.org> Acked-by: Alexandre Pereira da Silva <aletes.xgr@gmail.com> > --- > Documentation/devicetree/bindings/spi/spi_pl022.txt | 9 ++++ > drivers/spi/spi-pl022.c | 41 +++++++++++++++++--- > 2 files changed, 45 insertions(+), 5 deletions(-) > > --- linux-2.6.orig/Documentation/devicetree/bindings/spi/spi_pl022.txt > +++ linux-2.6/Documentation/devicetree/bindings/spi/spi_pl022.txt > @@ -10,6 +10,15 @@ Optional properties: > - cs-gpios : should specify GPIOs used for chipselects. > The gpios will be referred to as reg = <index> in the SPI child nodes. > If unspecified, a single SPI device without a chip select can be used. > +- pl022,bus-id : Bus ID (0, 1, ...) > +- pl022,enable-dma : enables DMA driven transfers (boolean) > +- pl022,autosuspend-delay : delay in ms following transfer completion before > + the runtime power management system suspends the > + device. A setting of 0 indicates no delay and the > + device will be suspended immediately > +- pl022,rt : indicates the controller should run the message pump with realtime > + priority to minimise the transfer latency on the bus (boolean) > + > > SPI slave nodes must be children of the SPI master node and can > contain the following properties. > --- linux-2.6.orig/drivers/spi/spi-pl022.c > +++ linux-2.6/drivers/spi/spi-pl022.c > @@ -2024,6 +2024,36 @@ static void pl022_cleanup(struct spi_dev > kfree(chip); > } > > +static struct pl022_ssp_controller * > +pl022_platform_data_dt_get(struct device *dev) > +{ > + struct device_node *np = dev->of_node; > + struct pl022_ssp_controller *pd; > + u32 tmp; > + > + if (!np) { > + dev_err(dev, "no dt node defined\n"); > + return NULL; > + } > + > + pd = devm_kzalloc(dev, sizeof(struct pl022_ssp_controller), GFP_KERNEL); > + if (!pd) { > + dev_err(dev, "cannot allocate platform data memory\n"); > + return NULL; > + } > + > + of_property_read_u32(np, "pl022,bus-id", &tmp); > + pd->bus_id = tmp; > + of_property_read_u32(np, "num-cs", &tmp); > + pd->num_chipselect = tmp; > + pd->enable_dma = of_property_read_bool(np, "pl022,enable-dma") ? 1 : 0; > + of_property_read_u32(np, "pl022,autosuspend-delay", > + &pd->autosuspend_delay); > + pd->rt = of_property_read_bool(np, "pl022,rt"); > + > + return pd; > +} > + > static int __devinit > pl022_probe(struct amba_device *adev, const struct amba_id *id) > { > @@ -2036,18 +2066,19 @@ pl022_probe(struct amba_device *adev, co > > dev_info(&adev->dev, > "ARM PL022 driver, device ID: 0x%08x\n", adev->periphid); > - if (platform_info == NULL) { > - dev_err(&adev->dev, "probe - no platform data supplied\n"); > + if (!platform_info && IS_ENABLED(CONFIG_OF)) > + platform_info = pl022_platform_data_dt_get(dev); > + > + if (!platform_info) { > + dev_err(dev, "probe: no platform data defined\n"); > status = -ENODEV; > goto err_no_pdata; > } > > if (platform_info->num_chipselect) { > num_cs = platform_info->num_chipselect; > - } else if (IS_ENABLED(CONFIG_OF)) { > - of_property_read_u32(np, "num-cs", &num_cs); > } else { > - dev_err(&adev->dev, "probe: no chip select defined\n"); > + dev_err(dev, "probe: no chip select defined\n"); > status = -ENODEV; > goto err_no_pdata; > }
On Tuesday 18 September 2012, Roland Stigge wrote: > --- linux-2.6.orig/Documentation/devicetree/bindings/spi/spi_pl022.txt > +++ linux-2.6/Documentation/devicetree/bindings/spi/spi_pl022.txt > @@ -10,6 +10,15 @@ Optional properties: > - cs-gpios : should specify GPIOs used for chipselects. > The gpios will be referred to as reg = <index> in the SPI child nodes. > If unspecified, a single SPI device without a chip select can be used. > +- pl022,bus-id : Bus ID (0, 1, ...) I don't understand why we need the bus-id here. My understanding is that this is a linux-specific number that should not be necessary when all devices are described in the device tree rather than through spi_board_info. Can't you just set master->bus_num to -1 when using the DT for probing? > +- pl022,enable-dma : enables DMA driven transfers (boolean) Similar thing here: Can't you just set the enable_dma flag when a dma channel is provided in the device tree and unset it otherwise? It also seems a bit pointless to enable DMA mode when you don't provide a way to specify the channel at the same time. Arnd
Hi! On 09/18/2012 03:13 PM, Arnd Bergmann wrote: > On Tuesday 18 September 2012, Roland Stigge wrote: >> --- linux-2.6.orig/Documentation/devicetree/bindings/spi/spi_pl022.txt >> +++ linux-2.6/Documentation/devicetree/bindings/spi/spi_pl022.txt >> @@ -10,6 +10,15 @@ Optional properties: >> - cs-gpios : should specify GPIOs used for chipselects. >> The gpios will be referred to as reg = <index> in the SPI child nodes. >> If unspecified, a single SPI device without a chip select can be used. >> +- pl022,bus-id : Bus ID (0, 1, ...) > > I don't understand why we need the bus-id here. My understanding is that > this is a linux-specific number that should not be necessary when all > devices are described in the device tree rather than through spi_board_info. > > Can't you just set master->bus_num to -1 when using the DT for probing? > >> +- pl022,enable-dma : enables DMA driven transfers (boolean) > > Similar thing here: Can't you just set the enable_dma flag when > a dma channel is provided in the device tree and unset it otherwise? > > It also seems a bit pointless to enable DMA mode when you don't provide > a way to specify the channel at the same time. Completely right! I'm removing pl022,bus-id, setting bus_id to -1 in the dt case. And removing pl022,enable-dma since dma is anyway only useful in the platform data case where the dma_filter() callback is specified. Thanks for pointing this out, Roland
--- linux-2.6.orig/Documentation/devicetree/bindings/spi/spi_pl022.txt +++ linux-2.6/Documentation/devicetree/bindings/spi/spi_pl022.txt @@ -10,6 +10,15 @@ Optional properties: - cs-gpios : should specify GPIOs used for chipselects. The gpios will be referred to as reg = <index> in the SPI child nodes. If unspecified, a single SPI device without a chip select can be used. +- pl022,bus-id : Bus ID (0, 1, ...) +- pl022,enable-dma : enables DMA driven transfers (boolean) +- pl022,autosuspend-delay : delay in ms following transfer completion before + the runtime power management system suspends the + device. A setting of 0 indicates no delay and the + device will be suspended immediately +- pl022,rt : indicates the controller should run the message pump with realtime + priority to minimise the transfer latency on the bus (boolean) + SPI slave nodes must be children of the SPI master node and can contain the following properties. --- linux-2.6.orig/drivers/spi/spi-pl022.c +++ linux-2.6/drivers/spi/spi-pl022.c @@ -2024,6 +2024,36 @@ static void pl022_cleanup(struct spi_dev kfree(chip); } +static struct pl022_ssp_controller * +pl022_platform_data_dt_get(struct device *dev) +{ + struct device_node *np = dev->of_node; + struct pl022_ssp_controller *pd; + u32 tmp; + + if (!np) { + dev_err(dev, "no dt node defined\n"); + return NULL; + } + + pd = devm_kzalloc(dev, sizeof(struct pl022_ssp_controller), GFP_KERNEL); + if (!pd) { + dev_err(dev, "cannot allocate platform data memory\n"); + return NULL; + } + + of_property_read_u32(np, "pl022,bus-id", &tmp); + pd->bus_id = tmp; + of_property_read_u32(np, "num-cs", &tmp); + pd->num_chipselect = tmp; + pd->enable_dma = of_property_read_bool(np, "pl022,enable-dma") ? 1 : 0; + of_property_read_u32(np, "pl022,autosuspend-delay", + &pd->autosuspend_delay); + pd->rt = of_property_read_bool(np, "pl022,rt"); + + return pd; +} + static int __devinit pl022_probe(struct amba_device *adev, const struct amba_id *id) { @@ -2036,18 +2066,19 @@ pl022_probe(struct amba_device *adev, co dev_info(&adev->dev, "ARM PL022 driver, device ID: 0x%08x\n", adev->periphid); - if (platform_info == NULL) { - dev_err(&adev->dev, "probe - no platform data supplied\n"); + if (!platform_info && IS_ENABLED(CONFIG_OF)) + platform_info = pl022_platform_data_dt_get(dev); + + if (!platform_info) { + dev_err(dev, "probe: no platform data defined\n"); status = -ENODEV; goto err_no_pdata; } if (platform_info->num_chipselect) { num_cs = platform_info->num_chipselect; - } else if (IS_ENABLED(CONFIG_OF)) { - of_property_read_u32(np, "num-cs", &num_cs); } else { - dev_err(&adev->dev, "probe: no chip select defined\n"); + dev_err(dev, "probe: no chip select defined\n"); status = -ENODEV; goto err_no_pdata; }