Subject: [possible] fixes for ufs_dirhash.c
To: None <>
From: Blair Sadewitz <>
List: tech-kern
Date: 11/16/2007 01:28:32
Please see:  <>.

For the rationale for the changes to PR_NOWAIT, MALLOC/memset ->
malloc+M_ZERO, and MALLOC(dh ...) -> dh = malloc(dh ...), see the
commit messages
for revs. 1.1{7,6,4,3} at:


Also, daddr_t is int64_t; the hash table is two levels, so, if I
understand this correctly at all, we're allocating twice as much
memory as we need to be.