Re: [PATCH 1/1] mfd: twl4030: clarify the return value for readand write

[Eduardo Valentin]

On Fri, Dec 11, 2009 at 11:36:10AM +0100, ext Samuel Ortiz wrote:
> Hi Amit,
> 
> On Mon, Dec 07, 2009 at 03:53:09PM +0200, Eduardo Valentin wrote:
> > On Mon, Dec 07, 2009 at 01:17:29PM +0100, ext Amit Kucheria wrote:
> > > Infact, we can just return -EIO so that caller knows for sure that all
> > > messages were not tranferred. Please consider fixed patch instead.
> > > 
> > > We should be checking if all the messages were tranferred or not. And return
> > > -1 for failure. Currently we return success (0) even if none of messages were
> > > transferred successfully.
> > > 
> > > Signed-off-by: Amit Kucheria <amit.kucheria@xxxxxxxxxxxxx>
> > > ---
> > >  drivers/mfd/twl4030-core.c |   24 ++++++++++++++++--------
> > >  1 files changed, 16 insertions(+), 8 deletions(-)
> > > 
> > > diff --git a/drivers/mfd/twl4030-core.c b/drivers/mfd/twl4030-core.c
> > > index 56f1de5..3d2c413 100644
> > > --- a/drivers/mfd/twl4030-core.c
> > > +++ b/drivers/mfd/twl4030-core.c
> > > @@ -292,10 +292,14 @@ int twl4030_i2c_write(u8 mod_no, u8 *value, u8 reg, unsigned num_bytes)
> > >  	ret = i2c_transfer(twl->client->adapter, twl->xfer_msg, 1);
> > >  	mutex_unlock(&twl->xfer_lock);
> > >  
> > > -	/* i2cTransfer returns num messages.translate it pls.. */
> > > -	if (ret >= 0)
> > > -		ret = 0;
> > > -	return ret;
> > > +	/* i2c_transfer returns number of messages transferred */
> > > +	if (ret != 1) {
> > > +		pr_err("%s: i2c_write failed to transfer all messages\n",
> > > +			DRIVER_NAME);
> > > +		return -EIO;
> > 
> > How about reporting the actual error that has occurred and reported by i2c_transfer?
> > Instead of just masking it as EIO? If i2c_transfer returns something > 0 then EIO should be right.
> > But if returns and error code, then that error code must be reported to upper layers.
> >
> So, I applied a modified version of this patch, see below. If you disagree
> with it, please let me know and I wont push it upstream.
> 

OK.

> 
> commit 8aa7cfd5732dee80aaaf3caa0a1d2f9621126315
> Author: Amit Kucheria <amit.kucheria@xxxxxxxxxxxxx>
> Date:   Fri Dec 11 11:31:11 2009 +0100
> 
>     mfd: Clarify twl4030 return value for read and write
>     
>     We should be checking if all the messages were tranferred. If not, then we
>     should propagate the i2c core error code.
>     Currently we return success (0) even if none of messages were transferred
>     successfully.
>     
>     Signed-off-by: Amit Kucheria <amit.kucheria@xxxxxxxxxxxxx>
>     Signed-off-by: Samuel Ortiz <sameo@xxxxxxxxxxxxxxx>
> 
> diff --git a/drivers/mfd/twl4030-core.c b/drivers/mfd/twl4030-core.c
> index e4a5d48..8add16c 100644
> --- a/drivers/mfd/twl4030-core.c
> +++ b/drivers/mfd/twl4030-core.c
> @@ -306,10 +306,14 @@ int twl4030_i2c_write(u8 mod_no, u8 *value, u8 reg, unsigned num_bytes)
>  	ret = i2c_transfer(twl->client->adapter, twl->xfer_msg, 1);
>  	mutex_unlock(&twl->xfer_lock);
>  
> -	/* i2cTransfer returns num messages.translate it pls.. */
> -	if (ret >= 0)
> -		ret = 0;
> -	return ret;
> +	/* i2c_transfer returns number of messages transferred */
> +	if (ret != 1) {
> +		pr_err("%s: i2c_write failed to transfer all messages\n",
> +			DRIVER_NAME);
> +		return ret;

For this case we will be returning success if i2c_transfer fails to transfer the one message.
I guess we should return ret only if ret < 0 ? otherwise -EIO.

> +	} else {
> +		return 0;
> +	}
>  }
>  EXPORT_SYMBOL(twl4030_i2c_write);
>  
> @@ -358,10 +362,14 @@ int twl4030_i2c_read(u8 mod_no, u8 *value, u8 reg, unsigned num_bytes)
>  	ret = i2c_transfer(twl->client->adapter, twl->xfer_msg, 2);
>  	mutex_unlock(&twl->xfer_lock);
>  
> -	/* i2cTransfer returns num messages.translate it pls.. */
> -	if (ret >= 0)
> -		ret = 0;
> -	return ret;
> +	/* i2c_transfer returns number of messages transferred */
> +	if (ret != 2) {
> +		pr_err("%s: i2c_read failed to transfer all messages\n",
> +			DRIVER_NAME);
> +		return ret;

Same case here, we will be returning success if i2c_transfer fails to transfer any of those messages.
I guess we should return ret only if ret < 0 ? otherwise -EIO.


> +	} else {
> +		return 0;
> +	}
>  }
>  EXPORT_SYMBOL(twl4030_i2c_read);
>  
> -- 
> Intel Open Source Technology Centre
> http://oss.intel.com/

-- 
Eduardo Valentin
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/