Release v1.0.3

This release fixes a bug that results in significantly improved diff
creation times in certain cases.

Signed-off-by: Patrick McCarty <patrick.mccarty@intel.com>

Makefile.am

@@ -67,7 +67,8 @@ lib_LTLIBRARIES = \
 
 libbsdiff_la_SOURCES = \
 	src/diff.c \
-	src/patch.c
+	src/patch.c \
+	src/sufsort.c
 
 libbsdiff_la_LIBADD = \
 	$(zlib_LIBS)
@@ -146,6 +147,27 @@ dist_check_SCRIPTS = \
 	test/run.sh
 endif
 
+compliant:
+	@git diff --quiet --exit-code include src; ret=$$?; \
+	if [ $$ret -eq 1 ]; then \
+		echo "Error: can only check code style when include/ and src/ are clean."; \
+		echo "Stash or commit your changes and try again."; \
+		exit $$ret; \
+	elif [ $$ret -gt 1 ]; then \
+		exit $$ret; \
+	fi; \
+	clang-format -i -style=file include/*.h src/*.c; ret=$$?; \
+	if [ $$ret -ne 0 ]; then \
+		exit $$ret; \
+	fi; \
+	git diff --quiet --exit-code include src; ret=$$?; \
+	if [ $$ret -eq 1 ]; then \
+		echo "Code style issues found. Run 'git diff' to view issues."; \
+	elif [ $$ret -eq 0 ]; then \
+		echo "No code style issues found."; \
+	fi; \
+	exit $$ret
+
 release:
 	@git rev-parse v$(PACKAGE_VERSION) &> /dev/null; \
 	if [ "$$?" -eq 0 ]; then \

configure.ac

@@ -1,5 +1,5 @@
 AC_PREREQ([2.66])
-AC_INIT([bsdiff], [1.0.2], [patrick.mccarty@intel.com])
+AC_INIT([bsdiff], [1.0.3], [patrick.mccarty@intel.com])
 AC_CONFIG_MACRO_DIR([m4])
 AC_PROG_CC
 AC_PROG_CC_STDC

src/bsheader.h

@@ -2,6 +2,7 @@
 #define __INCLUDE_GUARD_BSHEADER_H
 
 #include <stdint.h>
+#include <sys/types.h> // for u_char
 
 #include "bsdiff.h"
 
@@ -58,7 +59,7 @@ struct header_v20 {
 	uint64_t extra_length;
 	uint64_t old_file_length;
 	uint64_t new_file_length;
-	uint64_t mtime;  /* unused */
+	uint64_t mtime; /* unused */
 	uint32_t file_mode;
 	uint32_t file_owner;
 	uint32_t file_group;
@@ -177,4 +178,6 @@ static inline int eblock_get_enc(enc_flags_t enc)
 	}
 }
 
+int qsufsort(int64_t *, int64_t *, u_char *, int64_t);
+
 #endif

src/diff.c

@@ -32,8 +32,6 @@ __FBSDID
 #define _GNU_SOURCE
 #include "config.h"
 
-#include <sys/types.h>
-
 #ifdef BSDIFF_WITH_BZIP2
 #include <bzlib.h>
 #endif
@@ -45,20 +43,20 @@ __FBSDID
 #include <lzma.h>
 #endif
 
+#include <assert.h>
+#include <endian.h>
+#include <grp.h>
+#include <pthread.h>
+#include <pwd.h>
+#include <stdint.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
+#include <sys/mman.h>
+#include <sys/stat.h>
+#include <sys/types.h>
 #include <unistd.h>
 #include <zlib.h>
-#include <endian.h>
-#include <stdint.h>
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <pwd.h>
-#include <grp.h>
-#include <pthread.h>
-#include <assert.h>
-#include <sys/mman.h>
 
 #include "bsheader.h"
 
@@ -76,172 +74,6 @@ static int bsdiff_fulldl;
 #undef MIN
 #define MIN(x, y) (((x) < (y)) ? (x) : (y))
 
-/* NOTES:
- * I and V are chunks of memory (arrays) with length = (oldfile size +1) * sizeof(int64_t).
- * Additionally, we pass in arraylen now. The parent function qsufsort receives it, so it
- * should be available here as well for error checking.
- * start: is actually the point in the array sent in during the suffix sort, which sorts by
- * small blocks/chunks.
- * len: refers to the length of the current chunk being processed - NOT the array length(s).
- * h: will never be more than 8, and increases by *2 during suffix sort (h += h) */
-static void split(int64_t *I, int64_t *V, int64_t arraylen, int64_t start, int64_t len,
-		  int64_t h)
-{
-	int64_t i, j, k, x, tmp, jj, kk;
-
-	if (len < 16) {
-		for (k = start; k < start + len; k += j) {
-			j = 1;
-			x = V[I[k] + h];
-			for (i = 1; k + i < start + len; i++) {
-				if (V[I[k + i] + h] < x) {
-					x = V[I[k + i] + h];
-					j = 0;
-				}
-				if (V[I[k + i] + h] == x) {
-					tmp = I[k + j];
-					I[k + j] = I[k + i];
-					I[k + i] = tmp;
-					j++;
-				}
-			}
-			for (i = 0; i < j; i++) {
-				V[I[k + i]] = k + j - 1;
-			}
-			if (j == 1) {
-				I[k] = -1;
-			}
-		}
-		return;
-	}
-
-	x = V[I[start + len / 2] + h];
-	jj = 0;
-	kk = 0;
-	for (i = start; i < start + len; i++) {
-		if (V[I[i] + h] < x) {
-			jj++;
-		}
-		if (V[I[i] + h] == x) {
-			kk++;
-		}
-	}
-	jj += start;
-	kk += jj;
-
-	i = start;
-	j = 0;
-	k = 0;
-	while (i < jj) {
-		if (V[I[i] + h] < x) {
-			i++;
-		} else if (V[I[i] + h] == x) {
-			tmp = I[i];
-			I[i] = I[jj + j];
-			I[jj + j] = tmp;
-			j++;
-		} else {
-			tmp = I[i];
-			I[i] = I[kk + k];
-			I[kk + k] = tmp;
-			k++;
-		}
-	}
-
-	while (jj + j < kk) {
-		if (V[I[jj + j] + h] == x) {
-			j++;
-		} else {
-			tmp = I[jj + j];
-			I[jj + j] = I[kk + k];
-			I[kk + k] = tmp;
-			k++;
-		}
-	}
-
-	if (jj > start) {
-		split(I, V, arraylen, start, jj - start, h);
-	}
-
-	for (i = 0; i < kk - jj; i++) {
-		V[I[jj + i]] = kk - 1;
-	}
-	if (jj == kk - 1) {
-		I[jj] = -1;
-	}
-
-	if (start + len > kk) {
-		split(I, V, arraylen, kk, start + len - kk, h);
-	}
-}
-
-/* The old_data (previous file data) is passed into this suffix sort and sorted
- * accordingly using the I and V arrays, which are both of length oldsize +1. */
-static int qsufsort(int64_t *I, int64_t *V, u_char *old, int64_t oldsize)
-{
-	int64_t buckets[QSUF_BUCKET_SIZE];
-	int64_t i, h, len;
-
-	for (i = 0; i < QSUF_BUCKET_SIZE; i++) {
-		buckets[i] = 0;
-	}
-	for (i = 0; i < oldsize; i++) {
-		buckets[old[i]]++;
-	}
-	for (i = 1; i < QSUF_BUCKET_SIZE; i++) {
-		buckets[i] += buckets[i - 1];
-	}
-	for (i = QSUF_BUCKET_SIZE - 1; i > 0; i--) {
-		buckets[i] = buckets[i - 1];
-	}
-	buckets[0] = 0;
-
-	for (i = 0; i < oldsize; i++) {
-		if (buckets[old[i]] > oldsize + 1) {
-			return -1;
-		}
-		I[++buckets[old[i]]] = i;
-	}
-
-	for (i = 0; i < oldsize; i++) {
-		V[i] = buckets[old[i]];
-	}
-	V[oldsize] = 0;
-	for (i = 1; i < QSUF_BUCKET_SIZE; i++) {
-		if (buckets[i] == buckets[i - 1] + 1) {
-			I[buckets[i]] = -1;
-		}
-	}
-	I[0] = -1;
-
-	for (h = 1; I[0] != -(oldsize + 1); h += h) {
-		len = 0;
-		for (i = 0; i < oldsize + 1;) {
-			if (I[i] < 0) {
-				len -= I[i];
-				i -= I[i];
-			} else {
-				if (len) {
-					I[i - len] = -len;
-				}
-				len = V[I[i]] + 1 - i;
-				split(I, V, oldsize, i, len, h);
-				i += len;
-				len = 0;
-			}
-		}
-		if (len) {
-			I[i - len] = -len;
-		}
-	}
-
-	for (i = 0; i < oldsize + 1; i++) {
-		I[V[i]] = i;
-	}
-
-	return 0;
-}
-
 static int64_t matchlen(u_char *old, int64_t oldsize, u_char *new,
 			int64_t newsize)
 {
@@ -256,27 +88,58 @@ static int64_t matchlen(u_char *old, int64_t oldsize, u_char *new,
 	return i;
 }
 
+int64_t max_len = 0;
+
+/**
+ * Finds the longest matching array of bytes between the OLD and NEW file. The
+ * old file is suffix-sorted; the suffix-sorted array is stored at I, and
+ * indices to search between are indicated by ST (start) and EN (end). Returns
+ * the length of the match, and POS is updated to the position of the match
+ * within OLD.
+ */
 static int64_t search(int64_t *I, u_char *old, int64_t oldsize,
 		      u_char *new, int64_t newsize, int64_t st, int64_t en,
 		      int64_t *pos)
 {
 	int64_t x, y;
 
+	/* Initialize max_len for the binary search */
+	if (st == 0 && en == oldsize) {
+		max_len = matchlen(old, oldsize, new, newsize);
+		*pos = I[st];
+	}
+
+	/* The binary search terminates here when "en" and "st" are adjacent
+	 * indices in the suffix-sorted array. */
 	if (en - st < 2) {
 		x = matchlen(old + I[st], oldsize - I[st], new, newsize);
-		y = matchlen(old + I[en], oldsize - I[en], new, newsize);
-
-		if (x > y) {
+		if (x > max_len) {
+			max_len = x;
 			*pos = I[st];
-			return x;
-		} else {
-			*pos = I[en];
-			return y;
 		}
+		y = matchlen(old + I[en], oldsize - I[en], new, newsize);
+		if (y > max_len) {
+			max_len = y;
+			*pos = I[en];
+		}
+
+		return max_len;
 	}
 
 	x = st + (en - st) / 2;
-	if (memcmp(old + I[x], new, MIN(oldsize - I[x], newsize)) < 0) {
+
+	int64_t length = MIN(oldsize - I[x], newsize);
+	u_char *oldoffset = old + I[x];
+
+	/* This match *could* be the longest one, so check for that here */
+	int64_t tmp = matchlen(oldoffset, length, new, length);
+	if (tmp > max_len) {
+		max_len = tmp;
+		*pos = I[x];
+	}
+
+	/* Determine how to continue the binary search */
+	if (memcmp(oldoffset, new, length) < 0) {
 		return search(I, old, oldsize, new, newsize, x, en, pos);
 	} else {
 		return search(I, old, oldsize, new, newsize, st, x, pos);

src/dump_main.c

@@ -30,12 +30,12 @@
  */
 
 #define _GNU_SOURCE
+#include <assert.h>
 #include <stdio.h>
 #include <stdlib.h>
-#include <unistd.h>
 #include <string.h>
-#include <assert.h>
 #include <time.h>
+#include <unistd.h>
 
 #include "bsdiff.h"
 #include "bsheader.h"
@@ -110,7 +110,7 @@ static void print_v20_header(struct header_v20 *h, FILE *f)
 	if (h->mtime == 0) {
 		printf("Mtime:\t(not set, as expected)\n");
 	} else {
-		printf("Mtime:\t%s (probably means there is a bug)\n", ctime((const time_t*)&h->mtime));
+		printf("Mtime:\t%s (probably means there is a bug)\n", ctime((const time_t *)&h->mtime));
 	}
 	printf("Mode:\t%4o\n", h->file_mode);
 	printf("Uid:\t%d\n", h->file_owner);

src/patch.c

@@ -44,23 +44,23 @@ __FBSDID
 #include <lzma.h>
 #endif
 
-#include <stdlib.h>
-#include <stdio.h>
-#include <string.h>
-#include <unistd.h>
-#include <zlib.h>
-#include <stdint.h>
-#include <sys/types.h>
-#include <sys/stat.h>
-#include <pwd.h>
-#include <grp.h>
-#include <fcntl.h>
-#include <limits.h>
-#include <linux/fs.h>
 #include <assert.h>
 #include <endian.h>
 #include <errno.h>
+#include <fcntl.h>
+#include <grp.h>
+#include <limits.h>
+#include <linux/fs.h>
+#include <pwd.h>
+#include <stdint.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
 #include <sys/mman.h>
+#include <sys/stat.h>
+#include <sys/types.h>
+#include <unistd.h>
+#include <zlib.h>
 
 #include "bsheader.h"
 
@@ -259,7 +259,7 @@ typedef struct {
 #ifdef BSDIFF_WITH_BZIP2
 		BZFILE *bz2; /* method = BZIP2 */
 #endif
-		gzFile gz;  /* method = GZIP */
+		gzFile gz; /* method = GZIP */
 #ifdef BSDIFF_WITH_LZMA
 		xzfile *xz; /* method = XZ */
 #endif

src/sufsort.c

@@ -0,0 +1,193 @@
+/*-
+ * Copyright 2003-2005 Colin Percival
+ * All rights reserved
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted providing that the following conditions 
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
+ * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+ * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
+ * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
+ * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#include "bsheader.h"
+
+/* NOTES:
+ * I and V are chunks of memory (arrays) with length = (oldfile size +1) * sizeof(int64_t).
+ * Additionally, we pass in arraylen now. The parent function qsufsort receives it, so it
+ * should be available here as well for error checking.
+ * start: is actually the point in the array sent in during the suffix sort, which sorts by
+ * small blocks/chunks.
+ * len: refers to the length of the current chunk being processed - NOT the array length(s).
+ * h: will never be more than 8, and increases by *2 during suffix sort (h += h) */
+static void split(int64_t *I, int64_t *V, int64_t arraylen, int64_t start, int64_t len,
+		  int64_t h)
+{
+	int64_t i, j, k, x, tmp, jj, kk;
+
+	if (len < 16) {
+		for (k = start; k < start + len; k += j) {
+			j = 1;
+			x = V[I[k] + h];
+			for (i = 1; k + i < start + len; i++) {
+				if (V[I[k + i] + h] < x) {
+					x = V[I[k + i] + h];
+					j = 0;
+				}
+				if (V[I[k + i] + h] == x) {
+					tmp = I[k + j];
+					I[k + j] = I[k + i];
+					I[k + i] = tmp;
+					j++;
+				}
+			}
+			for (i = 0; i < j; i++) {
+				V[I[k + i]] = k + j - 1;
+			}
+			if (j == 1) {
+				I[k] = -1;
+			}
+		}
+		return;
+	}
+
+	x = V[I[start + len / 2] + h];
+	jj = 0;
+	kk = 0;
+	for (i = start; i < start + len; i++) {
+		if (V[I[i] + h] < x) {
+			jj++;
+		}
+		if (V[I[i] + h] == x) {
+			kk++;
+		}
+	}
+	jj += start;
+	kk += jj;
+
+	i = start;
+	j = 0;
+	k = 0;
+	while (i < jj) {
+		if (V[I[i] + h] < x) {
+			i++;
+		} else if (V[I[i] + h] == x) {
+			tmp = I[i];
+			I[i] = I[jj + j];
+			I[jj + j] = tmp;
+			j++;
+		} else {
+			tmp = I[i];
+			I[i] = I[kk + k];
+			I[kk + k] = tmp;
+			k++;
+		}
+	}
+
+	while (jj + j < kk) {
+		if (V[I[jj + j] + h] == x) {
+			j++;
+		} else {
+			tmp = I[jj + j];
+			I[jj + j] = I[kk + k];
+			I[kk + k] = tmp;
+			k++;
+		}
+	}
+
+	if (jj > start) {
+		split(I, V, arraylen, start, jj - start, h);
+	}
+
+	for (i = 0; i < kk - jj; i++) {
+		V[I[jj + i]] = kk - 1;
+	}
+	if (jj == kk - 1) {
+		I[jj] = -1;
+	}
+
+	if (start + len > kk) {
+		split(I, V, arraylen, kk, start + len - kk, h);
+	}
+}
+
+/* The old_data (previous file data) is passed into this suffix sort and sorted
+ * accordingly using the I and V arrays, which are both of length oldsize +1. */
+int qsufsort(int64_t *I, int64_t *V, u_char *old, int64_t oldsize)
+{
+	int64_t buckets[QSUF_BUCKET_SIZE];
+	int64_t i, h, len;
+
+	for (i = 0; i < QSUF_BUCKET_SIZE; i++) {
+		buckets[i] = 0;
+	}
+	for (i = 0; i < oldsize; i++) {
+		buckets[old[i]]++;
+	}
+	for (i = 1; i < QSUF_BUCKET_SIZE; i++) {
+		buckets[i] += buckets[i - 1];
+	}
+	for (i = QSUF_BUCKET_SIZE - 1; i > 0; i--) {
+		buckets[i] = buckets[i - 1];
+	}
+	buckets[0] = 0;
+
+	for (i = 0; i < oldsize; i++) {
+		if (buckets[old[i]] > oldsize + 1) {
+			return -1;
+		}
+		I[++buckets[old[i]]] = i;
+	}
+
+	for (i = 0; i < oldsize; i++) {
+		V[i] = buckets[old[i]];
+	}
+	V[oldsize] = 0;
+	for (i = 1; i < QSUF_BUCKET_SIZE; i++) {
+		if (buckets[i] == buckets[i - 1] + 1) {
+			I[buckets[i]] = -1;
+		}
+	}
+	I[0] = -1;
+
+	for (h = 1; I[0] != -(oldsize + 1); h += h) {
+		len = 0;
+		for (i = 0; i < oldsize + 1;) {
+			if (I[i] < 0) {
+				len -= I[i];
+				i -= I[i];
+			} else {
+				if (len) {
+					I[i - len] = -len;
+				}
+				len = V[I[i]] + 1 - i;
+				split(I, V, oldsize, i, len, h);
+				i += len;
+				len = 0;
+			}
+		}
+		if (len) {
+			I[i - len] = -len;
+		}
+	}
+
+	for (i = 0; i < oldsize + 1; i++) {
+		I[V[i]] = i;
+	}
+
+	return 0;
+}

test/run.sh

@@ -8,10 +8,15 @@ testnum=0
 
 sudo rm -f *.diff *.out
 
+VALGRIND="valgrind -q"
+if [ -n "$SKIP_VALGRIND" ]; then
+	VALGRIND=""
+fi
+
 libdir="$abs_builddir/.libs"
 ldpath="LD_LIBRARY_PATH=$libdir"
-BSDIFF="sudo $ldpath valgrind -q $libdir/bsdiff"
-BSPATCH="sudo $ldpath valgrind -q $libdir/bspatch"
+BSDIFF="sudo $ldpath $VALGRIND $libdir/bsdiff"
+BSPATCH="sudo $ldpath $VALGRIND $libdir/bspatch"
 
 # If exit status is 0, the test succeeded. Else it failed.
 check_success() {