From: Randy Dunlap on
On Sun, 28 Mar 2010 14:57:27 -0700 Joe Perches wrote:

> Add a macro for the somewhat common use of
>
> (something + value) % value
>
> drivers/net/sundance.c includes a change from
> a for(;test;add) loop to while (test) {... add};
> because the test uses ADD_MOD
>
> Compile tested only
>
> Signed-off-by: Joe Perches <joe(a)perches.com>
> ---
> include/linux/kernel.h | 1 +
> drivers/ata/libata-eh.c | 2 +-
> drivers/media/video/ov511.c | 2 +-
> drivers/media/video/usbvideo/usbvideo.c | 2 +-
> drivers/net/dl2k.c | 8 ++++----
> drivers/net/sundance.c | 4 ++--
> drivers/pci/dmar.c | 6 +++---
> drivers/scsi/3w-9xxx.c | 2 +-
> drivers/staging/wavelan/wavelan_cs.c | 10 +++++-----
> drivers/usb/misc/sisusbvga/sisusb_con.c | 4 ++--
> drivers/video/console/vgacon.c | 4 ++--
> sound/oss/vwsnd.c | 6 +++---
> 12 files changed, 26 insertions(+), 25 deletions(-)
>
> diff --git a/include/linux/kernel.h b/include/linux/kernel.h
> index 7f07074..c96b1ac 100644
> --- a/include/linux/kernel.h
> +++ b/include/linux/kernel.h
> @@ -63,6 +63,7 @@ extern const char linux_proc_banner[];
> (((x) + ((__divisor) / 2)) / (__divisor)); \
> } \
> )
> +#define ADD_MOD(x, y) (((x) + (y)) % (y))
>
> #define _RET_IP_ (unsigned long)__builtin_return_address(0)
> #define _THIS_IP_ ({ __label__ __here; __here: (unsigned long)&&__here; })


It would be better not to evaluate y more than one time.

Also it's not safe for 64-bit 'y' on i386, right?
Looks like it would cause missing reference to __imoddi3 or whatever it is called.

so it can easily be misused IMO.

---
~Randy
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
From: Joe Perches on
On Mon, 2010-03-29 at 12:47 -0700, Randy Dunlap wrote:
> On Sun, 28 Mar 2010 14:57:27 -0700 Joe Perches wrote:
> > Add a macro for the somewhat common use of
> > (something + value) % value
> > because the test uses ADD_MOD
> > --- a/include/linux/kernel.h
> > +++ b/include/linux/kernel.h
> > +#define ADD_MOD(x, y) (((x) + (y)) % (y))
> It would be better not to evaluate y more than one time.

That could be said for nearly all the convenience
macros in kernel.h

> Also it's not safe for 64-bit 'y' on i386, right?
> Looks like it would cause missing reference to __imoddi3
> or whatever it is called.

Which is also true for nearly all the convenience
macros in kernel.h

> so it can easily be misused IMO.

etc.

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