[i2c] [PATCH] i2c-s3c2410.c: Fix build warning

[Arnaud Patard (Rtp)]

Jean Delvare <khali at linux-fr.org> writes:

Hi Jean and Ben !

> Hi Ben,
>
> On Sun, 6 May 2007 14:51:24 +0000, Ben Dooks wrote:
>> On Sat, May 05, 2007 at 09:58:37PM +0200, Jean Delvare wrote:
>> > On Sat, 05 May 2007 16:02:20 +0200, Arnaud Patard (Rtp) wrote:
>> > > 
>> > > Fix for the following build warning :
>> > >   CC      drivers/i2c/busses/i2c-s3c2410.o drivers/i2c/busses/i2c-s3c2410.c: In function 's3c24xx_i2c_probe':
>> > >   drivers/i2c/busses/i2c-s3c2410.c:839: warning: format '%ld' expects type 'long int', but argument 4 has type 'resource_size_t'
>> > > 
>> > > Signed-off-by: Arnaud Patard <arnaud.patard at rtp-net.org>
>> > 
>> > > ---
>> > > diff --git a/drivers/i2c/busses/i2c-s3c2410.c b/drivers/i2c/busses/i2c-s3c2410.c
>> > > index 556f244..674ca30 100644
>> > > --- a/drivers/i2c/busses/i2c-s3c2410.c
>> > > +++ b/drivers/i2c/busses/i2c-s3c2410.c
>> > > @@ -836,7 +836,7 @@ static int s3c24xx_i2c_probe(struct platform_device *pdev)
>> > >  
>> > >  	i2c->irq = res;
>> > >  		
>> > > -	dev_dbg(&pdev->dev, "irq resource %p (%ld)\n", res, res->start);
>> > > +	dev_dbg(&pdev->dev, "irq resource %p (%zd)\n", res, res->start);
>> > >  
>> > >  	ret = i2c_add_adapter(&i2c->adap);
>> > >  	if (ret < 0) {
>> > 
>> > I'm not sure this is the right fix. resource_size_t might be either u32
>> > or u64 depending on CONFIG_RESOURCES_64BIT, so I suspect that you are
>> > clearing the warning in one case but adding it in the other. As a
>> > matter of fact, I couldn't find any other example in the kernel tree of
>> > %zd, %zu or %zx being used to print a resource boundary.
>> > 
>> > So I think that it would be better to keep %ld and cast res->start to
>> > an unsigned long. Ben, what do you think?
>> 
>> I think the best fix is to cast to (unsigned long long) and use %llu
>> as this means you are not truncating 64bit resources... although why
>> anyone would want to use 64bit resources on an ARM chip that is 32bit
>> only is a little strange.
>
> long longs are costly to handle so I'd rather avoid that. This is an
> IRQ we're talking about, so I'd be very surprised if it didn't fit in
> 32 bits.
>
> I can't tell you why anyone would want to use 64-bit resources on ARM.
> In fact I simply don't get why CONFIG_RESOURCES_64BIT exists for 32-bit
> architectures.

Did you find an agreement on this ? I should use a cast on res->start ?


Thanks,
Arnaud