lkml.org 
[lkml]   [2002]   [Jan]   [4]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: hashed waitqueues
Date
On January 4, 2002 06:40 pm, William Lee Irwin III wrote:

> + unsigned long hash = (unsigned long)page;

Just to be anal here, 'long' doesn't add anything useful to this
declaration. In fact, u32 is what you want since you've chosen your
multiplier with a 32 bit register in mind - on 64bit arch I think you'll get
distinctly non-random results as it stands.

> + hash *= GOLDEN_RATIO_PRIME;
> + hash >>= BITS_PER_LONG - zone->wait_table_bits;
> + hash &= zone->wait_table_size - 1;

Nice hash! For arches with expensive multiplies you might want to look
for a near-golden ratio multiplier that has a simple contruction in terms of
1's & 0's so it can be computed with 2 or 3 shift-adds, dumping the
efficiency problem on the compiler. You don't have to restrict your search
to the neighbourhood of 32 bits, you can go a few bits less than that and
substract from a smaller value than BITS_PER_LONG (actually, you should just
write '32' there, since that's what you really have in mind).

--
Daniel


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

\
 
 \ /
  Last update: 2005-03-22 13:18    [W:0.232 / U:9.748 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site