On 06/06/2024 17:56, Guenter Roeck wrote: > On 6/6/24 07:25, Krzysztof Kozlowski wrote: >> Use spi_get_device_match_data() helper to simplify a bit the driver. >> Also kernel_ulong_t type is preferred for kernel code over uintptr_t >> (needed for the cast). >> >> Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@xxxxxxxxxx> >> --- >> drivers/hwmon/lm70.c | 6 +----- >> 1 file changed, 1 insertion(+), 5 deletions(-) >> >> diff --git a/drivers/hwmon/lm70.c b/drivers/hwmon/lm70.c >> index 481e4e1f8f4f..0d5a250cb672 100644 >> --- a/drivers/hwmon/lm70.c >> +++ b/drivers/hwmon/lm70.c >> @@ -169,11 +169,7 @@ static int lm70_probe(struct spi_device *spi) >> struct lm70 *p_lm70; >> int chip; >> >> - if (dev_fwnode(&spi->dev)) >> - chip = (int)(uintptr_t)device_get_match_data(&spi->dev); >> - else >> - chip = spi_get_device_id(spi)->driver_data; >> - >> + chip = (kernel_ulong_t)spi_get_device_match_data(spi); >> >> /* signaling is SPI_MODE_0 */ >> if ((spi->mode & SPI_MODE_X_MASK) != SPI_MODE_0) > > I'll apply this patch, but ... > > I take it that the earlier claims that driver_data must not be > something like "(void *) LM70_CHIP_LM70", where LM70_CHIP_LM70 is 0, > was wrong. I'll start reverting all patches which changed starting > enum values from 0 to 1 based on that wrong claim. Yeah, they should not be 0. I think Jonathan also brought it some time ago for iio drivers: https://lore.kernel.org/linux-iio/20240226192555.14aa178e@jic23-huawei/ but if all the ID tables are complete, the 0 will work fine: https://lore.kernel.org/all/20240413154511.52576-1-krzysztof.kozlowski@xxxxxxxxxx/ Best regards, Krzysztof
