David KastrupIn normal use cases, the performance wins are not overly impressive: we get something like 5-10% due to the slightly better locality of memory accesses using the packed structure.
However, since the data structure for index entries saves 33% of memory on 32-bit platforms and 40% on 64-bit platforms, the behavior when memory gets limited should be nicer.
This is a rather well-contained change. One obvious improvement would be sorting the elements in one bucket according to their hash, then using binary probing to find the elements with the right hash value.
As it stands, the output should be strictly the same as previously unless one uses the option for limiting the amount of used memory, in which case the created packs might be better.
Signed-off-by: David Kastrup <dak@gnu.org> --- diff-delta.c | 62 +++++++++++++++++++++++++++++++++++++++++++--------------- 1 files changed, 46 insertions(+), 16 deletions(-)
Show changes to diff-delta.c +46 −16
diff --git a/diff-delta.c b/diff-delta.c index 0dde2f2..cf0136f 100644 --- a/diff-delta.c +++ b/diff-delta.c @@ -115,9 +115,13 @@ static const unsigned int U[256] = { struct index_entry { const unsigned char *ptr; unsigned int val; - struct index_entry *next; }; +struct unpacked_index_entry { + struct index_entry entry; + struct unpacked_index_entry *next; +}; + struct delta_index { unsigned long memsize; const void *src_buf; @@ -131,7 +135,8 @@ struct delta_index * create_delta_index(const void *buf, unsigned long bufsize) unsigned int i, hsize, hmask, entries, prev_val, *hash_count; const unsigned char *data, *buffer = buf; struct delta_index *index; - struct index_entry *entry, **hash; + struct unpacked_index_entry *entry, **hash; + struct index_entry *aentry, **ahash; void *mem; unsigned long memsize; @@ -148,28 +153,21 @@ struct delta_index * create_delta_index(const void *buf, unsigned long bufsize) hmask = hsize - 1; /* allocate lookup index */ - memsize = sizeof(*index) + - sizeof(*hash) * hsize + + memsize = sizeof(*hash) * hsize + sizeof(*entry) * entries; mem = malloc(memsize); if (!mem) return NULL; - index = mem; - mem = index + 1; hash = mem; mem = hash + hsize; entry = mem; - index->memsize = memsize; - index->src_buf = buf; - index->src_size = bufsize; - index->hash_mask = hmask; memset(hash, 0, hsize * sizeof(*hash)); /* allocate an array to count hash entries */ hash_count = calloc(hsize, sizeof(*hash_count)); if (!hash_count) { - free(index); + free(hash); return NULL; } @@ -183,12 +181,13 @@ struct delta_index * create_delta_index(const void *buf, unsigned long bufsize) val = ((val << 8) | data[i]) ^ T[val >> RABIN_SHIFT]; if (val == prev_val) { /* keep the lowest of consecutive identical blocks */ - entry[-1].ptr = data + RABIN_WINDOW; + entry[-1].entry.ptr = data + RABIN_WINDOW; + --entries; } else { prev_val = val; i = val & hmask; - entry->ptr = data + RABIN_WINDOW; - entry->val = val; + entry->entry.ptr = data + RABIN_WINDOW; + entry->entry.val = val; entry->next = hash[i]; hash[i] = entry++; hash_count[i]++; @@ -212,15 +211,46 @@ struct delta_index * create_delta_index(const void *buf, unsigned long bufsize) continue; entry = hash[i]; do { - struct index_entry *keep = entry; + struct unpacked_index_entry *keep = entry; int skip = hash_count[i] / HASH_LIMIT; do { + --entries; entry = entry->next; } while(--skip && entry); + ++entries; keep->next = entry; } while(entry); } free(hash_count); + memsize = sizeof(*index) + +sizeof(*ahash) * (hsize+1) + + +sizeof(*aentry) * entries; + mem=malloc(memsize); + + if (!mem) { + free(hash); + return NULL; + } + + index = mem; + index->memsize = memsize; + index->src_buf = buf; + index->src_size = bufsize; + index->hash_mask = hmask; + + mem = index+1; + ahash = mem; + mem = ahash + (hsize+1); + aentry = mem; + + for (i=0; i<hsize; i++) { + ahash[i] = aentry; + for (entry=hash[i]; entry; entry=entry->next) + *aentry++ = entry->entry; + } + ahash[hsize] = aentry; + assert(aentry-(struct index_entry *)mem == entries); + free(hash); return index; } @@ -302,7 +332,7 @@ create_delta(const struct delta_index *index, val ^= U[data[-RABIN_WINDOW]]; val = ((val << 8) | *data) ^ T[val >> RABIN_SHIFT]; i = val & index->hash_mask; - for (entry = index->hash[i]; entry; entry = entry->next) { + for (entry = index->hash[i]; entry<index->hash[i+1]; entry++) { const unsigned char *ref = entry->ptr; const unsigned char *src = data; unsigned int ref_size = ref_top - ref;
-- 1.5.3.GIT
