following up on jesper's yeoman efforts to tidy up a lot of the
*malloc calls, once upon a time, i whined enough to get the
<linux/log2.h> header file added so that a lot of painful bit tests
for power of 2 could be abbreviated.
the most useful test seems to have been:
static inline __attribute__((const))
bool is_power_of_2(unsigned long n)
{
return (n != 0 && ((n & (n - 1)) == 0));
}
i wrote a short script that looks for various forms of the old power
of 2 tests:
===== start =====
#!/bin/sh
DIR=${1-*}
echo -e " x & (x - 1):\n"
grep -Ern "([^\(\)]+) ?\& ?\(\1 ?- ?1\)" ${DIR}
echo -e " x & ((x) - 1):\n"
grep -Ern "([^\(\)]+) ?\& ?\(\(\1\) ?- ?1\)" ${DIR}
echo -e " (x) & (x - 1):\n"
grep -Ern "\(([^\(\)]+)\) ?\& ?\(\1 ?- ?1\)" ${DIR}
echo -e " (x) & ((x) - 1):\n"
grep -Ern "\(([^\(\)]+)\) ?\& ?\(\(\1\) ?- ?1\)" ${DIR}
===== end =====
and, unsurprisingly, while i once submitted a fair bit of that
cleanup, there's lots left -- just run the script at the top of the
source tree (give an optional subdir to restrict the search). there
are even a few places where code is still defining that test for
itself:
drivers/net/bna/bna.h:50:#define BNA_POWER_OF_2(x) (((x) & ((x) - 1)) == 0)
drivers/net/wireless/rt2x00/rt2x00reg.h:187:#define is_power_of_two(x) ( !((x) & ((x)-1)) )
if anyone wants to run the script on their favourite bit of the
tree, have at it.
rday
p.s. perhaps tests like this could be added to the checkpatch.pl
script. just a thought.
--
========================================================================
Robert P. J. Day Waterloo, Ontario, CANADA
http://crashcourse.ca
Twitter: http://twitter.com/rpjday
LinkedIn: http://ca.linkedin.com/in/rpjday
========================================================================