Remove parallel unpacking

Parallel unpacking didn't really improve the speed that much. Actually
sorting the files for optimized unpack order improved speed much more
than the parallel unpacker.

Furthermore, the fork based parallel unpacker was actually pretty messy
to begin with.

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

7 files changed, 14 insertions, 142 deletions

diff --git a/doc/rdsquashfs.1 b/doc/rdsquashfs.1
index 72c602a..338138a 100644
--- a/doc/rdsquashfs.1
+++ b/doc/rdsquashfs.1
@@ -57,13 +57,6 @@ Skip directories that would end up empty after applying the above rules.
 The following options are specific to unpacking files from a SquashFS image
 to disk:
 .TP
-\fB\-\-jobs\fR, \fB\-j\fR <count>
-Specify a number of parallel jobs to spawn for unpacking file data.
-The file hierarchy is created sequentially but the data unpacking is
-distributed over the given number of jobs so that each job has to unpack
-roughly the same amount of data. This can be used to speed up unpacking
-of large SquashFS archives.
-.TP
 \fB\-\-no\-sparse\fR, \fB\-Z\fR
 Do not create sparse files. Always unpack sparse files by
 writing blocks of zeros to disk.
diff --git a/include/fstree.h b/include/fstree.h
index 880a870..05e1ebe 100644
--- a/include/fstree.h
+++ b/include/fstree.h
@@ -295,12 +295,9 @@ tree_node_t *fstree_node_from_path(fstree_t *fs, const char *path);
 /*
   Optimize the order of the fstree file list for unpacking as to avoid
   unpacking fragment blocks more than once and to improve locality when
-  fetching data from disk. The resulting list is returned in 'out'.
-  If num_jobs is > 1, the list is split up for parallel processing.
+  fetching data from disk.
  */
-void optimize_unpack_order(fstree_t *fs, size_t num_jobs,
-			   file_info_t *out[num_jobs]);
-
+file_info_t *optimize_unpack_order(fstree_t *fs);
 
 /*
   Convert back to forward slashed, remove all preceeding and trailing slashes,
diff --git a/lib/fstree/optimize_unpack_order.c b/lib/fstree/optimize_unpack_order.c
index 51576bd..972d4b3 100644
--- a/lib/fstree/optimize_unpack_order.c
+++ b/lib/fstree/optimize_unpack_order.c
@@ -106,65 +106,14 @@ static file_info_t *list_sort(const fstree_t *fs, file_info_t *head)
 	return merge(fs, list_sort(fs, head), list_sort(fs, half));
 }
 
-static file_info_t *split_list(file_info_t *list, uint64_t threashold)
-{
-	file_info_t *it, *new = NULL;
-	uint64_t size = 0;
-
-	for (it = list; it != NULL; it = it->next) {
-		size += it->size - it->sparse;
-
-		if (size >= threashold) {
-			new = it->next;
-			it->next = NULL;
-			break;
-		}
-	}
-
-	return new;
-}
-
-static uint64_t total_size(file_info_t *list)
-{
-	uint64_t size = 0;
-	file_info_t *it;
-
-	for (it = list; it != NULL; it = it->next)
-		size += it->size - it->sparse;
-
-	return size;
-}
-
-void optimize_unpack_order(fstree_t *fs, size_t num_jobs,
-			   file_info_t *out[num_jobs])
+file_info_t *optimize_unpack_order(fstree_t *fs)
 {
 	file_info_t *file_list;
-	uint64_t threshold;
-	size_t i;
-
-	if (num_jobs < 1)
-		return;
-
-	for (i = 0; i < num_jobs; ++i)
-		out[i] = NULL;
 
 	file_list = list_sort(fs, fs->files);
 	while (file_list != NULL && file_list->input_file == NULL)
 		file_list = file_list->next;
 
 	fs->files = NULL;
-
-	if (num_jobs < 2) {
-		out[0] = file_list;
-		return;
-	}
-
-	threshold = total_size(file_list) / num_jobs;
-
-	for (i = 0; i < (num_jobs - 1); ++i) {
-		out[i] = file_list;
-		file_list = split_list(file_list, threshold);
-	}
-
-	out[i] = file_list;
+	return file_list;
 }
diff --git a/unpack/fill_files.c b/unpack/fill_files.c
index c0fc26c..df6ca7e 100644
--- a/unpack/fill_files.c
+++ b/unpack/fill_files.c
@@ -38,65 +38,17 @@ static int fill_files(data_reader_t *data, file_info_t *list, int flags)
 	return 0;
 }
 
-int fill_unpacked_files(fstree_t *fs, data_reader_t *data, int flags,
-			unsigned int num_jobs)
+int fill_unpacked_files(fstree_t *fs, data_reader_t *data, int flags)
 {
-	file_info_t **sublists, *it;
-	int exitstatus, status = 0;
-	unsigned int i;
-	pid_t pid;
+	file_info_t *list, *it;
+	int status = 0;
 
-	if (num_jobs < 1)
-		num_jobs = 1;
+	list = optimize_unpack_order(fs);
 
-	sublists = alloca(sizeof(sublists[0]) * num_jobs);
-	optimize_unpack_order(fs, num_jobs, sublists);
+	status = fill_files(data, list, flags);
 
-	if (num_jobs < 2) {
-		status = fill_files(data, sublists[0], flags);
-		goto out;
-	}
-
-	for (i = 0; i < num_jobs; ++i) {
-		pid = fork();
-
-		if (pid == 0) {
-			/* Kill the child when the parent process dies */
-			prctl(PR_SET_PDEATHSIG, SIGKILL);
-
-			if (fill_files(data, sublists[i], flags))
-				exit(EXIT_FAILURE);
-			exit(EXIT_SUCCESS);
-		}
-
-		if (pid < 0) {
-			perror("fork");
-			status = -1;
-			break;
-		}
-	}
-
-	for (;;) {
-		errno = 0;
-		pid = waitpid(-1, &exitstatus, 0);
-
-		if (pid < 0) {
-			if (errno == EINTR)
-				continue;
-			if (errno == ECHILD)
-				break;
-		}
-
-		if (!WIFEXITED(exitstatus) ||
-		    WEXITSTATUS(exitstatus) != EXIT_SUCCESS) {
-			status = -1;
-		}
-	}
-out:
-	for (i = 0; i < num_jobs; ++i) {
-		for (it = sublists[i]; it != NULL; it = it->next)
-			free(it->input_file);
-	}
+	for (it = list; it != NULL; it = it->next)
+		free(it->input_file);
 
 	return status;
 }
diff --git a/unpack/options.c b/unpack/options.c
index e880656..c7689eb 100644
--- a/unpack/options.c
+++ b/unpack/options.c
@@ -22,7 +22,6 @@ static struct option long_opts[] = {
 	{ "set-xattr", no_argument, NULL, 'X' },
 #endif
 	{ "set-times", no_argument, NULL, 'T' },
-	{ "jobs", required_argument, NULL, 'j' },
 	{ "describe", no_argument, NULL, 'd' },
 	{ "chmod", no_argument, NULL, 'C' },
 	{ "chown", no_argument, NULL, 'O' },
@@ -74,7 +73,6 @@ static const char *help_string =
 #endif
 "  --set-times, -T           When unpacking files to disk, set the create\n"
 "                            and modify timestamps from the squashfs image.\n"
-"  --jobs, -j <count>        Number of parallel unpacking jobs to start.\n"
 "  --chmod, -C               Change permission flags of unpacked files to\n"
 "                            those store in the squashfs image.\n"
 "  --chown, -O               Change ownership of unpacked files to the\n"
@@ -110,7 +108,7 @@ static char *get_path(char *old, const char *arg)
 
 void process_command_line(options_t *opt, int argc, char **argv)
 {
-	int i, j;
+	int i;
 
 	opt->op = OP_NONE;
 	opt->rdtree_flags = 0;
@@ -118,7 +116,6 @@ void process_command_line(options_t *opt, int argc, char **argv)
 	opt->cmdpath = NULL;
 	opt->unpack_root = NULL;
 	opt->image_name = NULL;
-	opt->num_jobs = 1;
 
 	for (;;) {
 		i = getopt_long(argc, argv, short_opts, long_opts, NULL);
@@ -159,17 +156,6 @@ void process_command_line(options_t *opt, int argc, char **argv)
 		case 'T':
 			opt->flags |= UNPACK_SET_TIMES;
 			break;
-		case 'j':
-			for (j = 0; optarg[j] != '\0'; ++j) {
-				if (j > 6 || !isdigit(optarg[j]))
-					goto fail_num_jobs;
-			}
-
-			opt->num_jobs = atoi(optarg);
-
-			if (opt->num_jobs < 1)
-				goto fail_num_jobs;
-			break;
 		case 'c':
 			opt->op = OP_CAT;
 			opt->cmdpath = get_path(opt->cmdpath, optarg);
@@ -227,7 +213,4 @@ fail_arg:
 	fprintf(stderr, "Try `%s --help' for more information.\n", __progname);
 	free(opt->cmdpath);
 	exit(EXIT_FAILURE);
-fail_num_jobs:
-	fputs("Expected a positive integer for --jobs.\n", stderr);
-	goto fail_arg;
 }
diff --git a/unpack/rdsquashfs.c b/unpack/rdsquashfs.c
index e0cfaba..ea21318 100644
--- a/unpack/rdsquashfs.c
+++ b/unpack/rdsquashfs.c
@@ -115,7 +115,7 @@ int main(int argc, char **argv)
 		if (restore_fstree(n, opt.flags))
 			goto out;
 
-		if (fill_unpacked_files(&fs, data, opt.flags, opt.num_jobs))
+		if (fill_unpacked_files(&fs, data, opt.flags))
 			goto out;
 
 		if (update_tree_attribs(&fs, n, opt.flags))
diff --git a/unpack/rdsquashfs.h b/unpack/rdsquashfs.h
index 6805bed..b5570dd 100644
--- a/unpack/rdsquashfs.h
+++ b/unpack/rdsquashfs.h
@@ -57,7 +57,6 @@ typedef struct {
 	int op;
 	int rdtree_flags;
 	int flags;
-	unsigned int num_jobs;
 	char *cmdpath;
 	const char *unpack_root;
 	const char *image_name;
@@ -69,8 +68,7 @@ int restore_fstree(tree_node_t *root, int flags);
 
 int update_tree_attribs(fstree_t *fs, tree_node_t *root, int flags);
 
-int fill_unpacked_files(fstree_t *fs, data_reader_t *data, int flags,
-			unsigned int num_jobs);
+int fill_unpacked_files(fstree_t *fs, data_reader_t *data, int flags);
 
 int describe_tree(tree_node_t *root, const char *unpack_root);