Re: [PATCH] staging/hv/osd: don't reimplement ALIGN macro

[Uwe Kleine-König]

Hello Jiri,

On Wed, Jan 19, 2011 at 01:43:57PM +0100, Jiri Slaby wrote:
> On 01/19/2011 09:54 AM, Uwe Kleine-König wrote:
> > On Tue, Jan 18, 2011 at 09:37:15PM -0800, Greg KH wrote:
> >> On Tue, Jan 18, 2011 at 04:39:11PM +0100, Uwe Kleine-König wrote:
> >>> Signed-off-by: Uwe Kleine-König <u.kleine-koenig@xxxxxxxxxxxxxx>
> >>> ---
> >>>  drivers/staging/hv/osd.h |    5 ++---
> >>>  1 files changed, 2 insertions(+), 3 deletions(-)
> >>>
> >>> diff --git a/drivers/staging/hv/osd.h b/drivers/staging/hv/osd.h
> >>> index ce064e8..61ae54c 100644
> >>> --- a/drivers/staging/hv/osd.h
> >>> +++ b/drivers/staging/hv/osd.h
> >>> @@ -28,10 +28,9 @@
> >>>  #include <linux/workqueue.h>
> >>>  
> >>>  /* Defines */
> >>> -#define ALIGN_UP(value, align)	(((value) & (align-1)) ?		\
> >>> -				 (((value) + (align-1)) & ~(align-1)) :	\
> >>> -				 (value))
> >>> +#define ALIGN_UP(value, align)		ALIGN((value), (align))
> >>
> >> How about dropping ALIGN_UP entirely and just using the built-in ALIGN()
> >> macro instead?
> > Can do.
> > 
> >>>  #define ALIGN_DOWN(value, align)	((value) & ~(align-1))
> >>
> >> Any chance to get rid of this as well with the ALIGN() macro, or is that
> >> really not possible?
> > it would be
> > 
> > 	#define ALIGN_DOWN(value, align) ALIGN((value) - (align) + 1, (align))
> > 
> > I think, but as it's only used once it might be easier to just use ALIGN
> > there, too.
> > 
> > BTW, it's used as follows:
> > 
> > 	#define NUM_PAGES_SPANNED(addr, len)    ((ALIGN(addr+len, PAGE_SIZE) - \
> > 						 ALIGN_DOWN(addr, PAGE_SIZE)) >>  \
> > 						 PAGE_SHIFT)
> 
> (DIV_ROUND_UP(addr+len, PAGE_SIZE) - ((addr) >> PAGE_SHIFT)))
> 
> or maybe better
> 
> (PAGE_ALIGN(addr+len) >> PAGE_SHIFT - ((addr) >> PAGE_SHIFT))
> 
> > I wonder if there is already a function yielding this value?
> > Wouldn't
> > 
> > 	((addr + len) >> PAGE_SHIFT) - (addr >> PAGE_SHIFT) + 1
> 
> No, this won't work (it's not equivalent).
Ah, this fails if addr + len is page aligned.

	((addr + len - 1) >> PAGE_SHIFT) - (addr >> PAGE_SHIFT) + 1

would work, but I like

	(PAGE_ALIGN(addr+len) >> PAGE_SHIFT - ((addr) >> PAGE_SHIFT))

better.  (In fact these two are equivalent:

	PAGE_ALIGN(addr+len) >> PAGE_SHIFT =
	__ALIGN_KERNEL(addr+len, PAGE_SIZE) >> PAGE_SHIFT =
	__ALIGN_KERNEL_MASK(addr+len, PAGE_SIZE - 1) >> PAGE_SHIFT =
	((addr + len + PAGE_SIZE - 1) & ~(PAGE_SIZE - 1)) >> PAGE_SHIFT =
	(addr + len + PAGE_SIZE - 1) >> PAGE_SHIFT =
	(addr + len - 1) >> PAGE_SHIFT + 1

Greg: should I send an updated patch or do you modify NUM_PAGES_SPANNED
accordingly?

Thanks
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
--
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/