dedup

commit 8ed79b4311c4715717198c0e04dd06efb29be265
parent 499f62b1b9e3e6db6f9f77defdb0c2bb15c25778
Author: sin <sin@2f30.org>
Date:   Thu, 21 Feb 2019 13:48:22 +0000

Rework chunker interface

Eventually, this will be implemented using a ring buffer.

Diffstat:
M
chunker.c
 | 
56
+++++++++++++++++++++++++++++++-------------------------
M
dedup.c
 | 
2
+-
M
dedup.h
 | 
6
+++---

3 files changed, 35 insertions(+), 29 deletions(-)
diff --git a/chunker.c b/chunker.c
@@ -1,3 +1,4 @@
+#include <assert.h>
 #include <err.h>
 #include <stdint.h>
 #include <stdio.h>
@@ -9,8 +10,9 @@
 
 struct chunker {
 	uint8_t *buf;
-	size_t size;
-	size_t pos;
+	size_t cap;
+	size_t rpos;
+	size_t wpos;
 	int fd;
 };
 
@@ -22,8 +24,8 @@ get_chunk_size(struct chunker *chunker)
 	uint32_t fp;
 
 	/* buzhash should be at least WINSIZE */
-	if (chunker->pos < WINSIZE)
-		return chunker->pos;
+	if (chunker->wpos - chunker->rpos < WINSIZE)
+		return chunker->wpos - chunker->rpos;
 
 	bp = chunker->buf;
 
@@ -37,18 +39,18 @@ get_chunk_size(struct chunker *chunker)
 	 * WINSIZE the smallest possible block size.
 	 */
 	fp = buzh_init(bp, WINSIZE);
-	for (i = 0; i < chunker->pos - WINSIZE; i++) {
+	for (i = chunker->rpos; i < chunker->wpos - WINSIZE; i++) {
 		if (i > 0)
 			fp = buzh_update(fp, bp[i - 1], bp[WINSIZE + i - 1],
 			                 WINSIZE);
 		if ((fp & HASHMSK) == 0)
 			return i + WINSIZE;
 	}
-	return chunker->pos;
+	return chunker->wpos - chunker->rpos;
 }
 
 struct chunker *
-alloc_chunker(size_t size, int fd)
+alloc_chunker(size_t cap, int fd)
 {
 	struct chunker *chunker;
 
@@ -56,11 +58,12 @@ alloc_chunker(size_t size, int fd)
 	if (chunker == NULL)
 		err(1, "malloc");
 
-	chunker->buf = malloc(size);
+	chunker->buf = malloc(cap);
 	if (chunker->buf == NULL)
 		err(1, "malloc");
-	chunker->size = size;
-	chunker->pos = 0;
+	chunker->cap = cap;
+	chunker->rpos = 0;
+	chunker->wpos = 0;
 	chunker->fd = fd;
 
 	return chunker;
@@ -79,28 +82,31 @@ fill_chunker(struct chunker *chunker)
 	uint8_t *bp;
 	ssize_t n;
 
-	bp = &chunker->buf[chunker->pos];
-	n = read(chunker->fd, bp, chunker->size - chunker->pos);
+	bp = &chunker->buf[chunker->wpos];
+	n = read(chunker->fd, bp, chunker->cap - chunker->wpos);
 	if (n < 0)
 		err(1, "read");
-	chunker->pos += n;
-	return chunker->pos;
+	chunker->wpos += n;
+	return chunker->wpos;
+}
+
+uint8_t *
+get_chunk(struct chunker *chunker, size_t *chunk_size)
+{
+	assert(chunker->rpos <= chunker->wpos);
+	*chunk_size = get_chunk_size(chunker);
+	chunker->rpos += *chunk_size;
+	return chunker->buf;
 }
 
 void
-drain_chunker(struct chunker *chunker, size_t chunk_size)
+drain_chunker(struct chunker *chunker)
 {
 	uint8_t *src, *dst;
 
-	src = &chunker->buf[chunk_size];
+	src = &chunker->buf[chunker->rpos];
 	dst = chunker->buf;
-	memmove(dst, src, chunker->pos - chunk_size);
-	chunker->pos -= chunk_size;
-}
-
-uint8_t *
-get_chunk(struct chunker *chunker, size_t *size)
-{
-	*size = get_chunk_size(chunker);
-	return chunker->buf;
+	memmove(dst, src, chunker->wpos - chunker->rpos);
+	chunker->wpos -= chunker->rpos;
+	chunker->rpos = 0;
 }
diff --git a/dedup.c b/dedup.c
@@ -445,7 +445,7 @@ dedup(int fd, char *msg)
 		SHA256_Update(&ctx, chunkp, chunk_size);
 		snap = grow_snap(snap, snap->nr_blk_descs + 1);
 		dedup_chunk(snap, chunkp, chunk_size);
-		drain_chunker(chunker, chunk_size);
+		drain_chunker(chunker);
 	}
 	SHA256_Final(snap->md, &ctx);
 
diff --git a/dedup.h b/dedup.h
@@ -3,11 +3,11 @@
 struct chunker;
 
 /* chunker.c */
-struct chunker *alloc_chunker(size_t size, int fd);
+struct chunker *alloc_chunker(size_t cap, int fd);
 void free_chunker(struct chunker *chunker);
 ssize_t fill_chunker(struct chunker *chunker);
-void drain_chunker(struct chunker *chunker, size_t n);
-uint8_t *get_chunk(struct chunker *chunker, size_t *size);
+uint8_t *get_chunk(struct chunker *chunker, size_t *chunk_size);
+void drain_chunker(struct chunker *chunker);
 
 /* hash.c */
 uint32_t buzh_init(uint8_t *buf, size_t size);