Re: [PATCH] Silly bitmap size accounting fix

[Nick Piggin <nickpiggin@yahoo.com.au>]

Steven Rostedt wrote:
> On Sat, 13 May 2006, Nick Piggin wrote:
> 
> 
>>Ingo Molnar wrote:
>>
>>>* Steven Rostedt <rostedt@goodmis.org> wrote:
>>>
>>>
>>>
>>>>-#define BITMAP_SIZE ((((MAX_PRIO+1+7)/8)+sizeof(long)-1)/sizeof(long))
>>>>+#define BITMAP_SIZE ((((MAX_PRIO+7)/8)+sizeof(long)-1)/sizeof(long))
>>>
>>>
>>>Acked-by: Ingo Molnar <mingo@elte.hu>
>>
>>Really?! What about the delimiter bit set at MAX_PRIO?
> 
> 
> 
> 		// delimiter for bitsearch
> 		__set_bit(MAX_PRIO, array->bitmap);
> 
> 
> Ah! I see what you mean.  New patch (add a comment).

That would have caused someone a world of pain 3 years ahead ;)

> 
> -- Steve
> 
> Signed-off-by: Steven Rostedt <rostedt@goodmis.org>

OK I guess. Does it help to also spell out exactly what's going on there?

> 
> Index: linux-2.6.17-rc3-mm1/kernel/sched.c
> ===================================================================
> --- linux-2.6.17-rc3-mm1.orig/kernel/sched.c	2006-05-12 04:02:32.000000000 -0400
> +++ linux-2.6.17-rc3-mm1/kernel/sched.c	2006-05-13 10:09:15.000000000 -0400
> @@ -192,6 +192,10 @@ static inline unsigned int task_timeslic
>   * These are the runqueue data structures:
>   */
> 
> +/*
> + * Calculate BITMAP_SIZE.
> + *  The bitmask holds MAX_PRIO bits + 1 for the delimiter.

+ * Calculation is to find the minimum number of longs that holds MAX_PRIO+1 bits:
+ *  size-in-chars = ceiling((MAX_PRIO+1) / CHAR_BITS)
+ *  size-in-longs = ceiling(size-in-chars / sizeof(long))

> + */
>  #define BITMAP_SIZE ((((MAX_PRIO+1+7)/8)+sizeof(long)-1)/sizeof(long))
> 

-- 
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com