diff options
author | Joel Becker <joel.becker@oracle.com> | 2008-12-16 03:24:33 +0100 |
---|---|---|
committer | Mark Fasheh <mfasheh@suse.com> | 2009-01-05 17:40:35 +0100 |
commit | 7bb458a58588f397068e4166c615e9fcc7480c16 (patch) | |
tree | a5d330c12f983837097f775e2d7c6c4d85fb9acd /fs/ocfs2 | |
parent | ocfs2: Don't hand-code xor in ocfs2_hamming_encode(). (diff) | |
download | linux-7bb458a58588f397068e4166c615e9fcc7480c16.tar.xz linux-7bb458a58588f397068e4166c615e9fcc7480c16.zip |
ocfs2: Another hamming code optimization.
In the calc_code_bit() function, we must find all powers of two beneath
the code bit number, *after* it's shifted by those powers of two. This
requires a loop to see where it ends up.
We can optimize it by starting at its most significant bit. This shaves
32% off the time, for a total of 67.6% shaved off of the original, naive
implementation.
Signed-off-by: Joel Becker <joel.becker@oracle.com>
Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Diffstat (limited to 'fs/ocfs2')
-rw-r--r-- | fs/ocfs2/blockcheck.c | 40 |
1 files changed, 39 insertions, 1 deletions
diff --git a/fs/ocfs2/blockcheck.c b/fs/ocfs2/blockcheck.c index 1d5083cef3a2..f102ec939c90 100644 --- a/fs/ocfs2/blockcheck.c +++ b/fs/ocfs2/blockcheck.c @@ -39,6 +39,35 @@ * c = # total code bits (d + p) */ + +/* + * Find the log base 2 of 32-bit v. + * + * Algorithm found on http://graphics.stanford.edu/~seander/bithacks.html, + * by Sean Eron Anderson. Code on the page is in the public domain unless + * otherwise noted. + * + * This particular algorithm is credited to Eric Cole. + */ +static int find_highest_bit_set(unsigned int v) +{ + + static const int MultiplyDeBruijnBitPosition[32] = + { + 0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8, + 31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9 + }; + + v |= v >> 1; /* first round down to power of 2 */ + v |= v >> 2; + v |= v >> 4; + v |= v >> 8; + v |= v >> 16; + v = (v >> 1) + 1; + + return MultiplyDeBruijnBitPosition[(u32)(v * 0x077CB531UL) >> 27]; +} + /* * Calculate the bit offset in the hamming code buffer based on the bit's * offset in the data buffer. Since the hamming code reserves all @@ -64,12 +93,21 @@ static unsigned int calc_code_bit(unsigned int i) b = i + 1; /* + * As a cheat, we know that all bits below b's highest bit must be + * parity bits, so we can start there. + */ + p = find_highest_bit_set(b); + b += p; + + /* * For every power of two below our bit number, bump our bit. * * We compare with (b + 1) becuase we have to compare with what b * would be _if_ it were bumped up by the parity bit. Capice? + * + * We start p at 2^p because of the cheat above. */ - for (p = 0; (1 << p) < (b + 1); p++) + for (p = (1 << p); p < (b + 1); p <<= 1) b++; return b; |