7:50 a.m.
Hi,
On 1/12/22 21:44, Andy Shevchenko wrote:
On Wed, Jan 12, 2022 at 08:58:00PM +0100, Hans de Goede wrote:
> On 1/8/22 19:54, kernel test robot wrote:
>>>> drivers/pinctrl/intel/pinctrl-baytrail.c:1483:58: warning: format
specifies type 'long' but the argument has type 'int' [-Wformat]
>> dev_dbg(vg->dev, "Pin %i: uses direct IRQ %ld\n",
pin, match - direct_irq);
>> ~~~
^~~~~~~~~~~~~~~~~~
>> %d
>> include/linux/dev_printk.h:163:47: note: expanded from macro
'dev_dbg'
>> dev_printk(KERN_DEBUG, dev, dev_fmt(fmt), ##__VA_ARGS__); \
>> ~~~ ^~~~~~~~~~~
>> include/linux/dev_printk.h:129:34: note: expanded from macro
'dev_printk'
>> _dev_printk(level, dev, fmt, ##__VA_ARGS__); \
>> ~~~ ^~~~~~~~~~~
>> 1 warning generated.
>
> Hmm, ok. so x86_64 needs a %ld for the pointer arithmic result on i386 needs a %d
> without the 'l' what fun. I'll just store it in a temp int variable in
the next
> version.
Why not to use uintptr_t and corresponding specifier (or ptrdiff_t)?
those or long resp. int depending on the arch to, but I could just
cast the result of the "match - direct_irq" pointer arithmetic to
an int, however due to the other changes in v4 using a local variable
seems cleaner, see the v4 which I'm about to send out in a couple of
minutes.
I think you will find v4 interesting because I've done a detailed
analysis of how the pinctrl conf0 pad register trigger bits and the IO-APIC
trigger bits work together, which is quite enlightening to how this all
works (and AFAICT not documented).
Regards,
Hans