fix dedupe-range

3 files changed, 54 insertions, 35 deletions

diff --git a/lib/dedupe-range.c b/lib/dedupe-range.c
index cdb5044..c9e9bb0 100644
--- a/lib/dedupe-range.c
+++ b/lib/dedupe-range.c
@@ -13,17 +13,23 @@
 
 #define MIN(a, b) ((a) < (b) ? (a) : (b))
 
-void dedupe_range(struct range src, struct range *dsts) {
+void dedupe_range(uint64_t src_length, struct filepos src, struct filepos *dsts) {
+	// Count how many destination ranges we have
 	size_t dst_count;
 	for (dst_count = 0; dsts[dst_count].filename; dst_count++);
 
+	// Figure the maximum number of destination ranges we can fit
+	// in 1 ioctl
 	const size_t max_dst_count = (sysconf(_SC_PAGESIZE) - sizeof(struct file_dedupe_range))
 		/ sizeof(struct file_dedupe_range_info);
+	assert(max_dst_count > 0);
 
+	// Open the source file
 	int src_fd = open(src.filename, src.flags);
 	if (src_fd < 0)
 		error(EXIT_FAILURE, errno, "open src: %s", src.filename);
 
+	// Open the destination files
 	struct file_dedupe_range_info *range_info =
 		calloc(dst_count, sizeof(struct file_dedupe_range_info));
 	if (!range_info)
@@ -36,27 +42,36 @@ void dedupe_range(struct range src, struct range *dsts) {
 		range_info[i].dest_offset = dsts[i].offset;
 	}
 
+	// Do the work
+	struct file_dedupe_range *range = malloc(sysconf(_SC_PAGESIZE));
+	if (!range)
+		error(EXIT_FAILURE, errno, "malloc");
 	for (size_t files_deduped = 0; files_deduped < dst_count; ) {
-		uint16_t dest_count = MIN(dst_count - files_deduped, max_dst_count);
-		struct file_dedupe_range *range = malloc(sizeof(struct file_dedupe_range) + dest_count * sizeof(struct file_dedupe_range_info));
-		if (!range)
-			error(EXIT_FAILURE, errno, "malloc");
-		*range = (struct file_dedupe_range){
-			.src_offset = src.offset,
-			.src_length = src.length,
-			.dest_count = dest_count,
-			.reserved1 = 0,
-			.reserved2 = 0,
-		};
-		for (size_t i = 0; i < dest_count; i++)
+		// initialize the range structure
+		*range = (struct file_dedupe_range){0};
+		range->dest_count = MIN(dst_count - files_deduped, max_dst_count);
+		assert(range->dest_count > 0);
+		for (size_t i = 0;i < range->dest_count; i++)
 			range->info[i] = range_info[files_deduped+i];
 
-		bool erred = false;
-		while (range->src_length > 0) {
-			if (ioctl(src_fd, FIDEDUPERANGE, &range) < 0)
-				error(EXIT_FAILURE, errno, "ioctl (FIDEDUPERANGE)");
+		// call FIDEDUPERANGE repeatedly to dedupe all of src_length
+		uint64_t bytes_todo = src_length;
+		uint64_t bytes_done = 0;
+		while (bytes_todo > 0) {
+			range->src_offset = src.offset + bytes_done;
+			range->src_length = MIN(bytes_todo, 16*1024*1024/*16MiB*/); // XXX "silently ignore"?
+
+			if (ioctl(src_fd, FIDEDUPERANGE, range) < 0)
+				error(EXIT_FAILURE, errno, "FIDEDUPERANGE");
+
+			// how much actually got done?
 			uint64_t bytes_deduped = range->info[0].bytes_deduped;
 			assert(bytes_deduped <= range->src_length);
+			bytes_done += bytes_deduped;
+			bytes_todo -= bytes_deduped;
+
+			// verify that dedupe for each dst went ok
+			bool erred = false;
 			for (size_t i = 0; i < range->dest_count; i ++) {
 				if (range->info[i].bytes_deduped != bytes_deduped) {
 					error(0, errno, "dedupe: %"PRIu64" != %"PRIu64": %s",
@@ -79,11 +94,14 @@ void dedupe_range(struct range src, struct range *dsts) {
 			}
 			if (erred == true)
 				exit(EXIT_FAILURE);
-			range->src_offset += bytes_deduped;
-			range->src_length -= bytes_deduped;
 		}
 		files_deduped += range->dest_count;
-		free(range);
 	}
+	free(range);
+
+	for (size_t i = 0; i < dst_count; i++)
+		close(range_info[i].dest_fd);
 	free(range_info);
+
+	close(src_fd);
 }
diff --git a/lib/dedupe-range.h b/lib/dedupe-range.h
index 028fe08..142eafc 100644
--- a/lib/dedupe-range.h
+++ b/lib/dedupe-range.h
@@ -1,10 +1,9 @@
 #include <stdint.h>
 
-struct range {
+struct filepos {
 	char *filename;
 	int flags; /* to pass to open(2) */
 	uint64_t offset;
-	uint64_t length;
 };
 
-void dedupe_range(struct range src, struct range *dsts);
+void dedupe_range(uint64_t length, struct filepos src, struct filepos *dsts);
diff --git a/src/cow-dedupe-range.c b/src/cow-dedupe-range.c
index bffa8f1..b96ead1 100644
--- a/src/cow-dedupe-range.c
+++ b/src/cow-dedupe-range.c
@@ -34,9 +34,9 @@ bool atou64(const char *str, uint64_t *ret) {
 }
 
 void usage() {
-	printf("Usage: %2$*1$s [OPTIONS] SRC_FILENAME SRC_OFFSET SRC_LENGTH \\\n"
-	       "       %3$*1$s           DST_FILENAME DST_OFFSET \\\n"
-	       "       %3$*1$s          [DST_FILENAME DST_OFFSET]...\n"
+	printf("Usage: %2$*1$s [OPTIONS] LENGTH SRC_FILENAME SRC_OFFSET \\\n"
+	       "       %3$*1$s                  DST_FILENAME DST_OFFSET \\\n"
+	       "       %3$*1$s                 [DST_FILENAME DST_OFFSET]...\n"
 	       "Submit a file deduplication request to the kernel.\n"
 	       "If the file ranges are not duplicates, the kernel will ignore this request.\n"
 	       "\n"
@@ -76,19 +76,21 @@ int main(int argc, char *argv[]) {
 	}
 	if (argc - optind < 5)
 		errusage("too few arguments");
-	if ((argc - optind - 3) % 2 != 0)
+	if ((argc - optind - 1) % 2 != 0)
 		errusage("wrong number of arguments");
 
-	struct range src;
-	src.filename = argv[optind];
+	uint64_t src_length;
+	if (!atou64(argv[optind], &src_length))
+		error(2, errno, "invalid length '%s'", argv[optind]);
+
+	struct filepos src;
+	src.filename = argv[optind+1];
 	src.flags = O_RDONLY;
-	if (!atou64(argv[optind+1], &src.offset))
-		error(2, errno, "invalid offset '%s'", argv[optind+1]);
-	if (!atou64(argv[optind+2], &src.length))
-		error(2, errno, "invalid length '%s'", argv[optind+2]);
+	if (!atou64(argv[optind+2], &src.offset))
+		error(2, errno, "invalid offset '%s'", argv[optind+2]);
 	
 	const size_t dst_count = (argc - optind - 3) / 2;
-	struct range *dsts = calloc(dst_count + 1, sizeof(struct range));
+	struct filepos *dsts = calloc(dst_count + 1, sizeof(struct filepos));
 	for (size_t i = 0; i < dst_count; i++) {
 		dsts[i].filename = argv[optind+3+(i*2)];
 		dsts[i].flags = ro ? O_RDONLY : O_RDWR;
@@ -96,7 +98,7 @@ int main(int argc, char *argv[]) {
 			error(2, errno, "invalid offset '%s'", argv[optind+3+(i*2)+1]);
 	}
 
-	dedupe_range(src, dsts);
+	dedupe_range(src_length, src, dsts);
 	free(dsts);
 	return EXIT_SUCCESS;
 }