__kernel_suseconds_t is int on PA-RISC (and SPARC64)?

__kernel_suseconds_t is int on PA-RISC (and SPARC64)?

[Geert Uytterhoeven]

Apparently __kernel_suseconds_t is `int' on PA-RISC (and on SPARC64),
while it's `long' on all other platforms.

Is there a specific reason for that?

This causes compiler warnings in the HP SDC Real Time Clock driver,
which is shared between PA-RISC and m68k:

drivers/input/misc/hp_sdc_rtc.c:461: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
drivers/input/misc/hp_sdc_rtc.c:461: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
drivers/input/misc/hp_sdc_rtc.c:468: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
drivers/input/misc/hp_sdc_rtc.c:468: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
drivers/input/misc/hp_sdc_rtc.c:475: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
drivers/input/misc/hp_sdc_rtc.c:475: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
drivers/input/misc/hp_sdc_rtc.c:482: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
drivers/input/misc/hp_sdc_rtc.c:482: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
drivers/input/misc/hp_sdc_rtc.c:489: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
drivers/input/misc/hp_sdc_rtc.c:489: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds

Re: __kernel_suseconds_t is int on PA-RISC (and SPARC64)?

[Helge Deller]

Geert Uytterhoeven wrote:
> Apparently __kernel_suseconds_t is `int' on PA-RISC (and on SPARC64),
> while it's `long' on all other platforms.
> 
> Is there a specific reason for that?

I don't know for sure, but arch/parisc/kernel/syscall_table.S states:
/* even though manpage says struct timeval contains longs, ours has
  * time_t and suseconds_t -- both of which are safe wide/narrow */
    ENTRY_COMP(setitimer)

So it was probably chosen to be int so that it can be easily used in 32- 
and 64bit kernels (or compat code).

Helge


> This causes compiler warnings in the HP SDC Real Time Clock driver,
> which is shared between PA-RISC and m68k:
> 
> drivers/input/misc/hp_sdc_rtc.c:461: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
> drivers/input/misc/hp_sdc_rtc.c:461: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
> drivers/input/misc/hp_sdc_rtc.c:468: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
> drivers/input/misc/hp_sdc_rtc.c:468: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
> drivers/input/misc/hp_sdc_rtc.c:475: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
> drivers/input/misc/hp_sdc_rtc.c:475: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
> drivers/input/misc/hp_sdc_rtc.c:482: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
> drivers/input/misc/hp_sdc_rtc.c:482: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
> drivers/input/misc/hp_sdc_rtc.c:489: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
> drivers/input/misc/hp_sdc_rtc.c:489: warning: format '%02d' expects type 'int', but argument 4 has type 'long int'
> 
> Gr{oetje,eeting}s,
> 
> 						Geert

Re: __kernel_suseconds_t is int on PA-RISC (and SPARC64)?

[Geert Uytterhoeven]

On Tue, 26 Aug 2008, Helge Deller wrote:
> Geert Uytterhoeven wrote:
> > Apparently __kernel_suseconds_t is `int' on PA-RISC (and on SPARC64),
> > while it's `long' on all other platforms.
> > 
> > Is there a specific reason for that?
> 
> I don't know for sure, but arch/parisc/kernel/syscall_table.S states:
> /* even though manpage says struct timeval contains longs, ours has
>  * time_t and suseconds_t -- both of which are safe wide/narrow */
>    ENTRY_COMP(setitimer)
> 
> So it was probably chosen to be int so that it can be easily used in 32- and
> 64bit kernels (or compat code).
> 
> > This causes compiler warnings in the HP SDC Real Time Clock driver,
> > which is shared between PA-RISC and m68k:

Here's a patch to fix the compiler warnings. Is this OK for PA-RISC?

SUbject: [PATCH] HP input: kill warnings due to suseconds_t differences

Kill compiler warnings related to printf() formats in the input drivers for
various HP9000 machines, which are shared between PA-RISC (suseconds_t is int)
and m68k (suseconds_t is long). As both are 32-bit, it's safe to cast to int.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
---
 drivers/input/misc/hp_sdc_rtc.c |   10 +++++-----
 drivers/input/serio/hp_sdc.c    |    2 +-
 2 files changed, 6 insertions(+), 6 deletions(-)

--- a/drivers/input/misc/hp_sdc_rtc.c
+++ b/drivers/input/misc/hp_sdc_rtc.c
@@ -458,35 +458,35 @@ static int hp_sdc_rtc_proc_output (char 
 		p += sprintf(p, "i8042 rtc\t: READ FAILED!\n");
 	} else {
 		p += sprintf(p, "i8042 rtc\t: %ld.%02d seconds\n", 
-			     tv.tv_sec, tv.tv_usec/1000);
+			     tv.tv_sec, (int)tv.tv_usec/1000);
 	}
 
 	if (hp_sdc_rtc_read_fhs(&tv)) {
 		p += sprintf(p, "handshake\t: READ FAILED!\n");
 	} else {
         	p += sprintf(p, "handshake\t: %ld.%02d seconds\n", 
-			     tv.tv_sec, tv.tv_usec/1000);
+			     tv.tv_sec, (int)tv.tv_usec/1000);
 	}
 
 	if (hp_sdc_rtc_read_mt(&tv)) {
 		p += sprintf(p, "alarm\t\t: READ FAILED!\n");
 	} else {
 		p += sprintf(p, "alarm\t\t: %ld.%02d seconds\n", 
-			     tv.tv_sec, tv.tv_usec/1000);
+			     tv.tv_sec, (int)tv.tv_usec/1000);
 	}
 
 	if (hp_sdc_rtc_read_dt(&tv)) {
 		p += sprintf(p, "delay\t\t: READ FAILED!\n");
 	} else {
 		p += sprintf(p, "delay\t\t: %ld.%02d seconds\n", 
-			     tv.tv_sec, tv.tv_usec/1000);
+			     tv.tv_sec, (int)tv.tv_usec/1000);
 	}
 
 	if (hp_sdc_rtc_read_ct(&tv)) {
 		p += sprintf(p, "periodic\t: READ FAILED!\n");
 	} else {
 		p += sprintf(p, "periodic\t: %ld.%02d seconds\n", 
-			     tv.tv_sec, tv.tv_usec/1000);
+			     tv.tv_sec, (int)tv.tv_usec/1000);
 	}
 
         p += sprintf(p,
--- a/drivers/input/serio/hp_sdc.c
+++ b/drivers/input/serio/hp_sdc.c
@@ -323,7 +323,7 @@ static void hp_sdc_tasklet(unsigned long
 			 * it back to the application. and be less verbose.
 			 */
 			printk(KERN_WARNING PREFIX "read timeout (%ius)!\n",
-			       tv.tv_usec - hp_sdc.rtv.tv_usec);
+			       (int)(tv.tv_usec - hp_sdc.rtv.tv_usec));
 			curr->idx += hp_sdc.rqty;
 			hp_sdc.rqty = 0;
 			tmp = curr->seq[curr->actidx];

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
							    -- Linus Torvalds

Re: __kernel_suseconds_t is int on PA-RISC (and SPARC64)?

[Helge Deller]

Geert Uytterhoeven wrote:
> On Tue, 26 Aug 2008, Helge Deller wrote:
>> Geert Uytterhoeven wrote:
>>> Apparently __kernel_suseconds_t is `int' on PA-RISC (and on SPARC64),
>>> while it's `long' on all other platforms.
>>>
>>> Is there a specific reason for that?
>> I don't know for sure, but arch/parisc/kernel/syscall_table.S states:
>> /* even though manpage says struct timeval contains longs, ours has
>>  * time_t and suseconds_t -- both of which are safe wide/narrow */
>>    ENTRY_COMP(setitimer)
>>
>> So it was probably chosen to be int so that it can be easily used in 32- and
>> 64bit kernels (or compat code).
>>
>>> This causes compiler warnings in the HP SDC Real Time Clock driver,
>>> which is shared between PA-RISC and m68k:
> 
> Here's a patch to fix the compiler warnings. Is this OK for PA-RISC?
> 
> SUbject: [PATCH] HP input: kill warnings due to suseconds_t differences
> 
> Kill compiler warnings related to printf() formats in the input drivers for
> various HP9000 machines, which are shared between PA-RISC (suseconds_t is int)
> and m68k (suseconds_t is long). As both are 32-bit, it's safe to cast to int.
> 
> Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>

Thanks Geert.
Tested and OK for PA-RISC.

Acked-by: Helge Deller <deller@gmx.de>


> ---
>  drivers/input/misc/hp_sdc_rtc.c |   10 +++++-----
>  drivers/input/serio/hp_sdc.c    |    2 +-
>  2 files changed, 6 insertions(+), 6 deletions(-)
> 
> --- a/drivers/input/misc/hp_sdc_rtc.c
> +++ b/drivers/input/misc/hp_sdc_rtc.c
> @@ -458,35 +458,35 @@ static int hp_sdc_rtc_proc_output (char 
>  		p += sprintf(p, "i8042 rtc\t: READ FAILED!\n");
>  	} else {
>  		p += sprintf(p, "i8042 rtc\t: %ld.%02d seconds\n", 
> -			     tv.tv_sec, tv.tv_usec/1000);
> +			     tv.tv_sec, (int)tv.tv_usec/1000);
>  	}
>  
>  	if (hp_sdc_rtc_read_fhs(&tv)) {
>  		p += sprintf(p, "handshake\t: READ FAILED!\n");
>  	} else {
>          	p += sprintf(p, "handshake\t: %ld.%02d seconds\n", 
> -			     tv.tv_sec, tv.tv_usec/1000);
> +			     tv.tv_sec, (int)tv.tv_usec/1000);
>  	}
>  
>  	if (hp_sdc_rtc_read_mt(&tv)) {
>  		p += sprintf(p, "alarm\t\t: READ FAILED!\n");
>  	} else {
>  		p += sprintf(p, "alarm\t\t: %ld.%02d seconds\n", 
> -			     tv.tv_sec, tv.tv_usec/1000);
> +			     tv.tv_sec, (int)tv.tv_usec/1000);
>  	}
>  
>  	if (hp_sdc_rtc_read_dt(&tv)) {
>  		p += sprintf(p, "delay\t\t: READ FAILED!\n");
>  	} else {
>  		p += sprintf(p, "delay\t\t: %ld.%02d seconds\n", 
> -			     tv.tv_sec, tv.tv_usec/1000);
> +			     tv.tv_sec, (int)tv.tv_usec/1000);
>  	}
>  
>  	if (hp_sdc_rtc_read_ct(&tv)) {
>  		p += sprintf(p, "periodic\t: READ FAILED!\n");
>  	} else {
>  		p += sprintf(p, "periodic\t: %ld.%02d seconds\n", 
> -			     tv.tv_sec, tv.tv_usec/1000);
> +			     tv.tv_sec, (int)tv.tv_usec/1000);
>  	}
>  
>          p += sprintf(p,
> --- a/drivers/input/serio/hp_sdc.c
> +++ b/drivers/input/serio/hp_sdc.c
> @@ -323,7 +323,7 @@ static void hp_sdc_tasklet(unsigned long
>  			 * it back to the application. and be less verbose.
>  			 */
>  			printk(KERN_WARNING PREFIX "read timeout (%ius)!\n",
> -			       tv.tv_usec - hp_sdc.rtv.tv_usec);
> +			       (int)(tv.tv_usec - hp_sdc.rtv.tv_usec));
>  			curr->idx += hp_sdc.rqty;
>  			hp_sdc.rqty = 0;
>  			tmp = curr->seq[curr->actidx];
> 
> Gr{oetje,eeting}s,
> 
> 						Geert
> 
> --
> Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org
> 
> In personal conversations with technical people, I call myself a hacker. But
> when I'm talking to journalists I just say "programmer" or something like that.
> 							    -- Linus Torvalds
> 

Re: __kernel_suseconds_t is int on PA-RISC (and SPARC64)?

[David Miller]

From: Geert Uytterhoeven <geert@linux-m68k.org>
Date: Tue, 26 Aug 2008 22:11:25 +0200 (CEST)

> Apparently __kernel_suseconds_t is `int' on PA-RISC (and on SPARC64),
> while it's `long' on all other platforms.
> 
> Is there a specific reason for that?

An existing API probably decided it for us :-)

Anyways, no use in wondering why, it can't be changed at this point.