| Message ID | 1459232852-15697-1-git-send-email-sdliyong@gmail.com |
|---|---|
| State | New |
| Headers | show |
G'day Yong, One comment below. On 29/03/2016 2:27 PM, Yong Li wrote: > The current implementation only uses the first byte in *val, > the second data is always 0. Change it to *(u16 *)val > to write the two bytes into the register > > Signed-off-by: Yong Li <sdliyong@gmail.com> > --- > drivers/gpio/gpio-pca953x.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/gpio/gpio-pca953x.c b/drivers/gpio/gpio-pca953x.c > index d0d3065..cf3d410 100644 > --- a/drivers/gpio/gpio-pca953x.c > +++ b/drivers/gpio/gpio-pca953x.c > @@ -159,7 +159,7 @@ static int pca953x_write_regs(struct pca953x_chip *chip, int reg, u8 *val) > switch (chip->chip_type) { > case PCA953X_TYPE: > ret = i2c_smbus_write_word_data(chip->client, > - reg << 1, (u16) *val); > + reg << 1, *(u16 *)val); I don't think this is safe for systems that don't support unaligned memory access. > break; > case PCA957X_TYPE: > ret = i2c_smbus_write_byte_data(chip->client, reg << 1, >
Thanks for your comment, I think I can change it to val[0] | (val[1] << 8), is it okay ? 2016-03-29 20:06 GMT+08:00 Phil Reid <preid@electromag.com.au>: > G'day Yong, > > One comment below. > > On 29/03/2016 2:27 PM, Yong Li wrote: >> >> The current implementation only uses the first byte in *val, >> the second data is always 0. Change it to *(u16 *)val >> to write the two bytes into the register >> >> Signed-off-by: Yong Li <sdliyong@gmail.com> >> --- >> drivers/gpio/gpio-pca953x.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/drivers/gpio/gpio-pca953x.c b/drivers/gpio/gpio-pca953x.c >> index d0d3065..cf3d410 100644 >> --- a/drivers/gpio/gpio-pca953x.c >> +++ b/drivers/gpio/gpio-pca953x.c >> @@ -159,7 +159,7 @@ static int pca953x_write_regs(struct pca953x_chip >> *chip, int reg, u8 *val) >> switch (chip->chip_type) { >> case PCA953X_TYPE: >> ret = i2c_smbus_write_word_data(chip->client, >> - reg << 1, (u16) >> *val); >> + reg << 1, *(u16 >> *)val); > > I don't think this is safe for systems that don't support unaligned memory > access. > > >> break; >> case PCA957X_TYPE: >> ret = i2c_smbus_write_byte_data(chip->client, reg >> << 1, >> > > > -- > Regards > Phil Reid > -- To unsubscribe from this list: send the line "unsubscribe linux-gpio" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
You missed CC'ing Phil (Added for this post) On Tuesday 29 March 2016 20:53:58, Yong Li wrote: > Thanks for your comment, I think I can change it to val[0] | (val[1] > << 8), is it okay ? Mh, currently there is only one caller (device_pca953x_init) which passes only 0, 0 or 0xff, 0xff, so endianess is irrelevant. But to be future proof this should be done in an endian-safe manner. Though cpu_to_le16p does not work, due to same alignment problem as casting to u16*. Best regards, Alexander -- To unsubscribe from this list: send the line "unsubscribe linux-gpio" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 29/03/2016 10:39 PM, Alexander Stein wrote: > You missed CC'ing Phil (Added for this post) > > On Tuesday 29 March 2016 20:53:58, Yong Li wrote: >> Thanks for your comment, I think I can change it to val[0] | (val[1] >> << 8), is it okay ? > > Mh, currently there is only one caller (device_pca953x_init) which passes only > 0, 0 or 0xff, 0xff, so endianess is irrelevant. But to be future proof this > should be done in an endian-safe manner. Though cpu_to_le16p does not work, > due to same alignment problem as casting to u16*. > I think get_unaligned((u16 *) val) should do the job. There's also get_unaligned_le* get_unaligned_be*
If use the get_unaligned, below is the code example, but we cannot detect if it is big endian or little endian. I would like to use the same write logic as PCA957X_TYPE: use the i2c_smbus_write_byte_data API to write two times. How do you think about it? if (big_endian) value = get_unaligned_be16(buf); else value = get_unaligned_le16(buf); Thanks, Yong Li 2016-03-30 0:33 GMT+08:00 Phil Reid <preid@electromag.com.au>: > On 29/03/2016 10:39 PM, Alexander Stein wrote: >> >> You missed CC'ing Phil (Added for this post) >> >> On Tuesday 29 March 2016 20:53:58, Yong Li wrote: >>> >>> Thanks for your comment, I think I can change it to val[0] | (val[1] >>> << 8), is it okay ? >> >> >> Mh, currently there is only one caller (device_pca953x_init) which passes >> only >> 0, 0 or 0xff, 0xff, so endianess is irrelevant. But to be future proof >> this >> should be done in an endian-safe manner. Though cpu_to_le16p does not >> work, >> due to same alignment problem as casting to u16*. >> > > I think get_unaligned((u16 *) val) should do the job. > There's also get_unaligned_le* get_unaligned_be* > > -- > Regards > Phil Reid > -- To unsubscribe from this list: send the line "unsubscribe linux-gpio" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Or another method is using the below to convert the u8 to u16: cpu_to_le16(get_unaligned((u16 *) val)), compared with the i2c_smbus_write_byte_data method, which one is better? Thanks, Yong 2016-03-30 10:43 GMT+08:00 Yong Li <sdliyong@gmail.com>: > If use the get_unaligned, below is the code example, but we cannot detect if > it is big endian or little endian. I would like to use the same write logic > as PCA957X_TYPE: use the i2c_smbus_write_byte_data API to write two times. > How do you think about it? > > if (big_endian) > value = get_unaligned_be16(buf); > else > value = get_unaligned_le16(buf); > > Thanks, > Yong Li > 2016-03-30 0:33 GMT+08:00 Phil Reid <preid@electromag.com.au>: >> >> On 29/03/2016 10:39 PM, Alexander Stein wrote: >>> >>> You missed CC'ing Phil (Added for this post) >>> >>> On Tuesday 29 March 2016 20:53:58, Yong Li wrote: >>>> >>>> Thanks for your comment, I think I can change it to val[0] | (val[1] >>>> << 8), is it okay ? >>> >>> >>> Mh, currently there is only one caller (device_pca953x_init) which passes >>> only >>> 0, 0 or 0xff, 0xff, so endianess is irrelevant. But to be future proof >>> this >>> should be done in an endian-safe manner. Though cpu_to_le16p does not >>> work, >>> due to same alignment problem as casting to u16*. >>> >> >> I think get_unaligned((u16 *) val) should do the job. >> There's also get_unaligned_le* get_unaligned_be* >> >> -- >> Regards >> Phil Reid >> -- To unsubscribe from this list: send the line "unsubscribe linux-gpio" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On 30/03/2016 1:01 PM, Yong Li wrote: > Or another method is using the below to convert the u8 to u16: > cpu_to_le16(get_unaligned((u16 *) val)), compared with the > i2c_smbus_write_byte_data method, which one is better? > > G'day Yong, I'd go with the cpu_to_le16(get_unaligned((u16 *) val))
diff --git a/drivers/gpio/gpio-pca953x.c b/drivers/gpio/gpio-pca953x.c index d0d3065..cf3d410 100644 --- a/drivers/gpio/gpio-pca953x.c +++ b/drivers/gpio/gpio-pca953x.c @@ -159,7 +159,7 @@ static int pca953x_write_regs(struct pca953x_chip *chip, int reg, u8 *val) switch (chip->chip_type) { case PCA953X_TYPE: ret = i2c_smbus_write_word_data(chip->client, - reg << 1, (u16) *val); + reg << 1, *(u16 *)val); break; case PCA957X_TYPE: ret = i2c_smbus_write_byte_data(chip->client, reg << 1,
The current implementation only uses the first byte in *val, the second data is always 0. Change it to *(u16 *)val to write the two bytes into the register Signed-off-by: Yong Li <sdliyong@gmail.com> --- drivers/gpio/gpio-pca953x.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)