dedup

commit 266d6321fe55b0b95675378a7621d0c259d88f06
parent d2dad2c14804f240b36322ceebe3a56b4b480a93
Author: sin <sin@2f30.org>
Date:   Wed,  1 May 2019 13:34:50 +0100

Add back blake2s support

Diffstat:
M
bstorage.c
 | 
50
+++++++++++++++++++++++++++++++++++++++++++++++---
M
dotest
 | 
18
++++++++++++++++++
M
dup-init.1
 | 
5
++---

3 files changed, 67 insertions(+), 6 deletions(-)
diff --git a/bstorage.c b/bstorage.c
@@ -48,6 +48,7 @@
 #define CNONETYPE	0
 #define CSNAPPYTYPE	1
 #define HBLAKE2BTYPE	0
+#define HBLAKE2STYPE	1
 
 extern ssize_t xread(int, void *, size_t);
 extern ssize_t xwrite(int, void *, size_t);
@@ -102,6 +103,7 @@ struct sctx {
 	struct bhdr bhdr;
 	int fd;
 	int rdonly;	/* when set to 1, the bssync() operation is a no-op */
+	int type;	/* hash algorithm for new blocks */
 };
 
 static int
@@ -120,7 +122,7 @@ static RB_PROTOTYPE(bdcache, bd, rbe, bd_cmp)
 static RB_GENERATE(bdcache, bd, rbe, bd_cmp)
 
 static int
-bhash(void *buf, size_t n, unsigned char *md)
+b2bhash(void *buf, size_t n, unsigned char *md)
 {
 	blake2b_state ctx;
 
@@ -131,6 +133,18 @@ bhash(void *buf, size_t n, unsigned char *md)
 	return blake2b_final(&ctx, md, MDSIZE);
 }
 
+static int
+b2shash(void *buf, size_t n, unsigned char *md)
+{
+	blake2s_state ctx;
+
+	if (blake2s_init(&ctx, MDSIZE) < 0)
+		return -1;
+	if (blake2s_update(&ctx, buf, n) < 0)
+		return -1;
+	return blake2s_final(&ctx, md, MDSIZE);
+}
+
 /* Read block header */
 static int
 unpackbhdr(int fd, struct bhdr *bhdr)
@@ -316,6 +330,8 @@ bscreat(struct bctx *bctx, char *path, int mode, struct bparam *bpar)
 	/* Set hash type */
 	if (strcmp(bpar->halgo, "blake2b") == 0) {
 		bhdr->flags |= HBLAKE2BTYPE << HALGOSHIFT;
+	} else if (strcmp(bpar->halgo, "blake2s") == 0) {
+		bhdr->flags |= HBLAKE2STYPE << HALGOSHIFT;
 	} else {
 		free(sctx);
 		close(fd);
@@ -323,6 +339,7 @@ bscreat(struct bctx *bctx, char *path, int mode, struct bparam *bpar)
 	}
 	bhdr->nbd = 0;
 	sctx->fd = fd;
+	sctx->type = (bhdr->flags >> HALGOSHIFT) & HALGOMASK;
 
 	if (packbhdr(fd, bhdr) < 0) {
 		free(sctx);
@@ -406,6 +423,9 @@ bsopen(struct bctx *bctx, char *path, int flags, int mode, struct bparam *bpar)
 	case HBLAKE2BTYPE:
 		bpar->halgo = "blake2b";
 		break;
+	case HBLAKE2STYPE:
+		bpar->halgo = "blake2s";
+		break;
 	default:
 		free(sctx);
 		close(fd);
@@ -414,6 +434,7 @@ bsopen(struct bctx *bctx, char *path, int flags, int mode, struct bparam *bpar)
 
 	sctx->fd = fd;
 	sctx->rdonly = flags == O_RDONLY;
+	sctx->type = halgo;
 
 	if (initbdcache(sctx) < 0) {
 		free(sctx);
@@ -434,8 +455,18 @@ bsput(struct bctx *bctx, void *buf, size_t n, unsigned char *md)
 
 	sctx = bctx->sctx;
 
-	if (bhash(buf, n, key.md) < 0)
+	switch (sctx->type) {
+	case HBLAKE2BTYPE:
+		if (b2bhash(buf, n, key.md) < 0)
+			return -1;
+		break;
+	case HBLAKE2STYPE:
+		if (b2shash(buf, n, key.md) < 0)
+			return -1;
+		break;
+	default:
 		return -1;
+	}
 
 	bd = RB_FIND(bdcache, &sctx->bdcache, &key);
 	if (bd != NULL) {
@@ -608,7 +639,20 @@ bscheck(struct bctx *bctx, unsigned char *md)
 		return -1;
 	}
 
-	if (bhash(buf, bd->size, key.md) < 0) {
+	switch (sctx->type) {
+	case HBLAKE2BTYPE:
+		if (b2bhash(buf, bd->size, key.md) < 0) {
+			free(buf);
+			return -1;
+		}
+		break;
+	case HBLAKE2STYPE:
+		if (b2shash(buf, bd->size, key.md) < 0) {
+			free(buf);
+			return -1;
+		}
+		break;
+	default:
 		free(buf);
 		return -1;
 	}
diff --git a/dotest b/dotest
@@ -89,6 +89,23 @@ test5()
 	rm -rf "$repo" "$data"
 }
 
+test6()
+{
+	repo=`mktemp -d`
+	data=`mktemp`
+	dd if=/dev/urandom of="$data" bs=1M count=64
+	./dup-init -H blake2s "$repo"
+	./dup-pack -r "$repo" snap0 < "$data"
+	./dup-pack -r "$repo" snap1 < "$data"
+	du -sh "$repo"
+	sum0=`sha1sum "$data" | awk '{print $1}'`
+	sum1=`./dup-unpack -r "$repo" snap0 | sha1sum | awk '{print $1}'`
+	sum2=`./dup-unpack -r "$repo" snap1 | sha1sum | awk '{print $1}'`
+	[ "$sum0" = "$sum1" ]
+	[ "$sum0" = "$sum2" ]
+	rm -rf "$repo" "$data"
+}
+
 make
 test0
 test1
@@ -96,3 +113,4 @@ test2
 test3
 test4
 test5
+test6
diff --git a/dup-init.1 b/dup-init.1
@@ -1,4 +1,4 @@
-.Dd April 25, 2019
+.Dd May 1, 2019
 .Dt DUP-INIT 1
 .Os
 .Sh NAME
@@ -21,8 +21,7 @@ is specified the current working directory is used.
 .It Fl H Ar hash
 The cryptographic hash function used to identify
 unique blocks in the store.
-The supported hash functions are blake2b.
-This flag only has an effect when initializing the repository.
+The supported hash functions are blake2b and blake2s.
 By default blake2b is used.
 .It Fl Z Ar compressor
 The compressor function used to compress the blocks