On Tue 21-10-14 16:43:55, Sasha Levin wrote:
> While the hash function used by the revoke hashtable is good somewhere else,
> it's not really good here.
> 
> The default hash shift (8) means that one third of the hashing function
> gets lost (and is undefined anyways (8 - 12 = negative shift)):
> 
>       "(block << (hash_shift - 12))) & (table->hash_size - 1)"
> 
> Instead, just use the kernel's generic hash function that gets used everywhere
> else.
  Thanks. Merged to my tree.

                                                                Honza
> 
> Signed-off-by: Sasha Levin <[email protected]>
> ---
>  fs/jbd/revoke.c |    7 ++-----
>  1 file changed, 2 insertions(+), 5 deletions(-)
> 
> diff --git a/fs/jbd/revoke.c b/fs/jbd/revoke.c
> index 8898bbd..dcead63 100644
> --- a/fs/jbd/revoke.c
> +++ b/fs/jbd/revoke.c
> @@ -93,6 +93,7 @@
>  #include <linux/bio.h>
>  #endif
>  #include <linux/log2.h>
> +#include <linux/hash.h>
>  
>  static struct kmem_cache *revoke_record_cache;
>  static struct kmem_cache *revoke_table_cache;
> @@ -129,15 +130,11 @@ static void flush_descriptor(journal_t *, struct 
> journal_head *, int, int);
>  
>  /* Utility functions to maintain the revoke table */
>  
> -/* Borrowed from buffer.c: this is a tried and tested block hash function */
>  static inline int hash(journal_t *journal, unsigned int block)
>  {
>       struct jbd_revoke_table_s *table = journal->j_revoke;
> -     int hash_shift = table->hash_shift;
>  
> -     return ((block << (hash_shift - 6)) ^
> -             (block >> 13) ^
> -             (block << (hash_shift - 12))) & (table->hash_size - 1);
> +     return hash_32(block, table->hash_shift);
>  }
>  
>  static int insert_revoke_hash(journal_t *journal, unsigned int blocknr,
> -- 
> 1.7.10.4
> 
-- 
Jan Kara <[email protected]>
SUSE Labs, CR
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Reply via email to