[2.4.19] fix for fuzzy hash <linux/ghash.h> [Attempt 2]
PATCH 1/1 [2.4.19] fix for fuzzy hash <linux/ghash.h> [Attempt 2]
[ Originally posted on 2002/09/11; I just looked in 2.4.20-pre9 and it was not there, hence the repost. ]
The DEF_HASH_FUZZY macro allows the user to template their hash; it takes on a paramter for the hashing-function, namely HASHFN. When used with a hashing-function named anything other than 'hashfn()', a module using the kernel's fuzzy hash implementation will not compile.
None of the in-kernel 2.4.x drivers use this primitive (yet) so it's no wonder no one has spotted it. The patch is very trivial and makes me think that I am the very first user of the include/linux/ghash.h
hash-table primitive. ;)
Bart.
---
include/linux/ghash.h +2 -2
@@ -106,7 +106,7 @@ \ LINKAGE TYPE * find_##NAME##_hash(struct NAME##_table * tbl, KEYTYPE pos)\ {\ - int ix = hashfn(pos);\ + int ix = HASHFN(pos);\ TYPE * ptr = tbl->hashtable[ix];\ while(ptr && KEYCMP(ptr->KEY, pos))\ ptr = ptr->PTRS.next_hash;\
@@ -206,7 +206,7 @@ \ LINKAGE TYPE * find_##NAME##_hash(struct NAME##_table * tbl, KEYTYPE pos)\ {\ - int ix = hashfn(pos);\ + int ix = HASHFN(pos);\ TYPE * ptr = tbl->hashtable[ix];\ while(ptr && KEYCMP(ptr->KEY, pos))\ ptr = ptr->PTRS.next_hash;\
Em Sun, Oct 06, 2002 at 05:01:24PM -0400, Bart Trojanowski escreveu:
Quoted 3 lines
wonder no one has spotted it. The patch is very trivial and makes me think that I am the very first user of the
include/linux/ghash.h
hash-table primitive. ;)
Somebody told me that this was used in when dentry was introduced to the kernel, but then after rewrites it stopped being used, I was even thinking about submitting a patch removing it from the tree, but now there is one user, you :-)
- Arnaldo
Quoted 9 lines · Arnaldo Carvalho de Melo
Em Sun, Oct 06, 2002 at 05:01:24PM -0400, Bart Trojanowski escreveu:
Quoted 3 lines
wonder no one has spotted it. The patch is very trivial and makes me think that I am the very first user of the
include/linux/ghash.h
hash-table primitive. ;)Somebody told me that this was used in when dentry was introduced to the kernel, but then after rewrites it stopped being used, I was even thinking about submitting a patch removing it from the tree, but now there is one user, you :-)
Too bad, should have acted faster ;-)
This attempt is much like the single linked lists: it looks like something you ought to be able to generalize, but somehow it never quite works. Writing the code out in full gives you a more efficient, more compact result every time.
Signature · 1 line
Daniel