| Message ID | 1410533640-21469-1-git-send-email-grygorii.strashko@ti.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Fri, Sep 12, 2014 at 05:54:00PM +0300, Grygorii Strashko wrote: > static void davinci_spi_cleanup(struct spi_device *spi) > { > - if (spi->cs_gpio >= 0) > - gpio_free(spi->cs_gpio); > } This function is now empty so should be removed. I've applied for now but please send a followup fixing this.
On 09/13/2014 07:28 PM, Mark Brown wrote: > On Fri, Sep 12, 2014 at 05:54:00PM +0300, Grygorii Strashko wrote: > >> static void davinci_spi_cleanup(struct spi_device *spi) >> { >> - if (spi->cs_gpio >= 0) >> - gpio_free(spi->cs_gpio); >> } > > This function is now empty so should be removed. I've applied for now > but please send a followup fixing this. > The function davinci_spi_cleanup() will be reused by following patch "[PATCH v3] spi: davinci: add support for adding delay between word's transmissions" http://www.spinics.net/lists/devicetree/msg49141.html So, in broonie/spi.git/for-next it will not be empty. Again, I've missed description of this dependency, sorry for that. Regards, -grygorii
On Mon, Sep 15, 2014 at 04:06:42PM +0300, Grygorii Strashko wrote: > The function davinci_spi_cleanup() will be reused by following patch > "[PATCH v3] spi: davinci: add support for adding delay between word's transmissions" > http://www.spinics.net/lists/devicetree/msg49141.html > So, in broonie/spi.git/for-next it will not be empty. > Again, I've missed description of this dependency, sorry for that. Don't do things like this, delete the function and then re-add it - it makes the review simpler and it means that if the second patch doesn't get applied (as will happen here for v3.17) then the code still looks good.
diff --git a/drivers/spi/spi-davinci.c b/drivers/spi/spi-davinci.c index 48f1d26..d493507 100644 --- a/drivers/spi/spi-davinci.c +++ b/drivers/spi/spi-davinci.c @@ -397,24 +397,21 @@ static int davinci_spi_setup(struct spi_device *spi) struct spi_master *master = spi->master; struct device_node *np = spi->dev.of_node; bool internal_cs = true; - unsigned long flags = GPIOF_DIR_OUT; dspi = spi_master_get_devdata(spi->master); pdata = &dspi->pdata; - flags |= (spi->mode & SPI_CS_HIGH) ? GPIOF_INIT_LOW : GPIOF_INIT_HIGH; - if (!(spi->mode & SPI_NO_CS)) { if (np && (master->cs_gpios != NULL) && (spi->cs_gpio >= 0)) { - retval = gpio_request_one(spi->cs_gpio, - flags, dev_name(&spi->dev)); + retval = gpio_direction_output( + spi->cs_gpio, !(spi->mode & SPI_CS_HIGH)); internal_cs = false; } else if (pdata->chip_sel && spi->chip_select < pdata->num_chipselect && pdata->chip_sel[spi->chip_select] != SPI_INTERN_CS) { spi->cs_gpio = pdata->chip_sel[spi->chip_select]; - retval = gpio_request_one(spi->cs_gpio, - flags, dev_name(&spi->dev)); + retval = gpio_direction_output( + spi->cs_gpio, !(spi->mode & SPI_CS_HIGH)); internal_cs = false; } @@ -441,8 +438,6 @@ static int davinci_spi_setup(struct spi_device *spi) static void davinci_spi_cleanup(struct spi_device *spi) { - if (spi->cs_gpio >= 0) - gpio_free(spi->cs_gpio); } static int davinci_spi_check_error(struct davinci_spi *dspi, int int_status) @@ -967,6 +962,27 @@ static int davinci_spi_probe(struct platform_device *pdev) if (dspi->version == SPI_VERSION_2) dspi->bitbang.flags |= SPI_READY; + if (pdev->dev.of_node) { + int i; + + for (i = 0; i < pdata->num_chipselect; i++) { + int cs_gpio = of_get_named_gpio(pdev->dev.of_node, + "cs-gpios", i); + + if (cs_gpio == -EPROBE_DEFER) { + ret = cs_gpio; + goto free_clk; + } + + if (gpio_is_valid(cs_gpio)) { + ret = devm_gpio_request(&pdev->dev, cs_gpio, + dev_name(&pdev->dev)); + if (ret) + goto free_clk; + } + } + } + r = platform_get_resource(pdev, IORESOURCE_DMA, 0); if (r) dma_rx_chan = r->start;
Now CS GPIOs are requested from struct spi_master.setup() callback and that causes failures when Client SPI device is getting accessed through SPIDEV driver. The failure happens, because .setup() callback may be called many times from IOCTL handler and when it's called second time gpio_request() will fail and return -EBUSY. Hence, fix it by moving CS GPIOs requesting code in .probe(). Reported-by: Murali Karicheri <m-karicheri2@ti.com> Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com> --- Tested using spidev_test utility. drivers/spi/spi-davinci.c | 34 +++++++++++++++++++++++++--------- 1 file changed, 25 insertions(+), 9 deletions(-)