Adding upstream version 31+20240202.upstream/31+20240202

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

1 files changed, 3153 insertions, 0 deletions

diff --git a/tools/depmod.c b/tools/depmod.c
new file mode 100644
index 0000000..43fc354
--- /dev/null
+++ b/tools/depmod.c
@@ -0,0 +1,3153 @@
+/*
+ * kmod-depmod - calculate modules.dep  using libkmod.
+ *
+ * Copyright (C) 2011-2013  ProFUSION embedded systems
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <assert.h>
+#include <ctype.h>
+#include <dirent.h>
+#include <errno.h>
+#include <getopt.h>
+#include <limits.h>
+#include <regex.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <sys/stat.h>
+#include <sys/time.h>
+#include <sys/utsname.h>
+
+#include <shared/array.h>
+#include <shared/hash.h>
+#include <shared/macro.h>
+#include <shared/util.h>
+#include <shared/scratchbuf.h>
+
+#include <libkmod/libkmod-internal.h>
+
+#undef ERR
+#undef DBG
+
+#include "kmod.h"
+
+#define DEFAULT_VERBOSE LOG_WARNING
+static int verbose = DEFAULT_VERBOSE;
+
+static const char CFG_BUILTIN_KEY[] = "built-in";
+static const char CFG_EXTERNAL_KEY[] = "external";
+static const char *const default_cfg_paths[] = {
+	SYSCONFDIR "/depmod.d",
+	"/run/depmod.d",
+	"/usr/local/lib/depmod.d",
+	DISTCONFDIR "/depmod.d",
+	"/lib/depmod.d",
+	NULL
+};
+
+static const char cmdopts_s[] = "aAb:o:C:E:F:euqrvnP:wmVh";
+static const struct option cmdopts[] = {
+	{ "all", no_argument, 0, 'a' },
+	{ "quick", no_argument, 0, 'A' },
+	{ "basedir", required_argument, 0, 'b' },
+	{ "outdir", required_argument, 0, 'o' },
+	{ "config", required_argument, 0, 'C' },
+	{ "symvers", required_argument, 0, 'E' },
+	{ "filesyms", required_argument, 0, 'F' },
+	{ "errsyms", no_argument, 0, 'e' },
+	{ "unresolved-error", no_argument, 0, 'u' }, /* deprecated */
+	{ "quiet", no_argument, 0, 'q' }, /* deprecated */
+	{ "root", no_argument, 0, 'r' }, /* deprecated */
+	{ "verbose", no_argument, 0, 'v' },
+	{ "show", no_argument, 0, 'n' },
+	{ "dry-run", no_argument, 0, 'n' },
+	{ "symbol-prefix", required_argument, 0, 'P' },
+	{ "warn", no_argument, 0, 'w' },
+	{ "map", no_argument, 0, 'm' }, /* deprecated */
+	{ "version", no_argument, 0, 'V' },
+	{ "help", no_argument, 0, 'h' },
+	{ }
+};
+
+static void help(void)
+{
+	printf("Usage:\n"
+		"\t%s -[aA] [options] [forced_version]\n"
+		"\n"
+		"If no arguments (except options) are given, \"depmod -a\" is assumed\n"
+		"\n"
+		"depmod will output a dependency list suitable for the modprobe utility.\n"
+		"\n"
+		"Options:\n"
+		"\t-a, --all            Probe all modules\n"
+		"\t-A, --quick          Only does the work if there's a new module\n"
+		"\t-e, --errsyms        Report not supplied symbols\n"
+		"\t-n, --show           Write the dependency file on stdout only\n"
+		"\t-P, --symbol-prefix  Architecture symbol prefix\n"
+		"\t-C, --config=PATH    Read configuration from PATH\n"
+		"\t-v, --verbose        Enable verbose mode\n"
+		"\t-w, --warn           Warn on duplicates\n"
+		"\t-V, --version        show version\n"
+		"\t-h, --help           show this help\n"
+		"\n"
+		"The following options are useful for people managing distributions:\n"
+		"\t-b, --basedir=DIR    Use an image of a module tree.\n"
+		"\t-o, --outdir=DIR     Output directory for generated files.\n"
+		"\t-F, --filesyms=FILE  Use the file instead of the\n"
+		"\t                     current kernel symbols.\n"
+		"\t-E, --symvers=FILE   Use Module.symvers file to check\n"
+		"\t                     symbol versions.\n",
+		program_invocation_short_name);
+}
+
+_printf_format_(1, 2)
+static inline void _show(const char *fmt, ...)
+{
+	va_list args;
+
+	if (verbose <= DEFAULT_VERBOSE)
+		return;
+
+	va_start(args, fmt);
+	vfprintf(stdout, fmt, args);
+	fflush(stdout);
+	va_end(args);
+}
+#define SHOW(...) _show(__VA_ARGS__)
+
+
+/* binary index write *************************************************/
+#include <arpa/inet.h>
+/* BEGIN: code from module-init-tools/index.c just modified to compile here.
+ *
+ * Original copyright:
+ *   index.c: module index file shared functions for modprobe and depmod
+ *   Copyright (C) 2008  Alan Jenkins <alan-jenkins@tuffmail.co.uk>.
+ *
+ *   These programs are free software; you can redistribute it and/or modify
+ *   it under the terms of the GNU General Public License as published by
+ *   the Free Software Foundation; either version 2 of the License, or
+ *   (at your option) any later version.
+ *
+ *   This program is distributed in the hope that it will be useful,
+ *   but WITHOUT ANY WARRANTY; without even the implied warranty of
+ *   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ *   GNU General Public License for more details.
+ *
+ *   You should have received a copy of the GNU General Public License
+ *   along with these programs.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+/* see documentation in libkmod/libkmod-index.c */
+
+#define INDEX_MAGIC 0xB007F457
+#define INDEX_VERSION_MAJOR 0x0002
+#define INDEX_VERSION_MINOR 0x0001
+#define INDEX_VERSION ((INDEX_VERSION_MAJOR<<16)|INDEX_VERSION_MINOR)
+#define INDEX_CHILDMAX 128
+
+struct index_value {
+	struct index_value *next;
+	unsigned int priority;
+	char value[0];
+};
+
+/* In-memory index (depmod only) */
+struct index_node {
+	char *prefix;		/* path compression */
+	struct index_value *values;
+	unsigned char first;	/* range of child nodes */
+	unsigned char last;
+	struct index_node *children[INDEX_CHILDMAX]; /* indexed by character */
+};
+
+
+/* Format of node offsets within index file */
+enum node_offset {
+	INDEX_NODE_FLAGS    = 0xF0000000, /* Flags in high nibble */
+	INDEX_NODE_PREFIX   = 0x80000000,
+	INDEX_NODE_VALUES = 0x40000000,
+	INDEX_NODE_CHILDS   = 0x20000000,
+
+	INDEX_NODE_MASK     = 0x0FFFFFFF, /* Offset value */
+};
+
+static struct index_node *index_create(void)
+{
+	struct index_node *node;
+
+	node = NOFAIL(calloc(1, sizeof(struct index_node)));
+	node->prefix = NOFAIL(strdup(""));
+	node->first = INDEX_CHILDMAX;
+
+	return node;
+}
+
+static void index_values_free(struct index_value *values)
+{
+	while (values) {
+		struct index_value *value = values;
+
+		values = value->next;
+		free(value);
+	}
+}
+
+static void index_destroy(struct index_node *node)
+{
+	int c;
+
+	for (c = node->first; c <= node->last; c++) {
+		struct index_node *child = node->children[c];
+
+		if (child)
+			index_destroy(child);
+	}
+	index_values_free(node->values);
+	free(node->prefix);
+	free(node);
+}
+
+static void index__checkstring(const char *str)
+{
+	int i;
+
+	for (i = 0; str[i]; i++) {
+		int ch = str[i];
+
+		if (ch >= INDEX_CHILDMAX)
+			CRIT("Module index: bad character '%c'=0x%x - only 7-bit ASCII is supported:"
+			      "\n%s\n", (char) ch, (int) ch, str);
+	}
+}
+
+static int index_add_value(struct index_value **values,
+				const char *value, unsigned int priority)
+{
+	struct index_value *v;
+	int duplicate = 0;
+	int len;
+
+	/* report the presence of duplicate values */
+	for (v = *values; v; v = v->next) {
+		if (streq(v->value, value))
+			duplicate = 1;
+	}
+
+	/* find position to insert value */
+	while (*values && (*values)->priority < priority)
+		values = &(*values)->next;
+
+	len = strlen(value);
+	v = NOFAIL(calloc(1, sizeof(struct index_value) + len + 1));
+	v->next = *values;
+	v->priority = priority;
+	memcpy(v->value, value, len + 1);
+	*values = v;
+
+	return duplicate;
+}
+
+static int index_insert(struct index_node *node, const char *key,
+			const char *value, unsigned int priority)
+{
+	int i = 0; /* index within str */
+	int ch;
+
+	index__checkstring(key);
+	index__checkstring(value);
+
+	while(1) {
+		int j; /* index within node->prefix */
+
+		/* Ensure node->prefix is a prefix of &str[i].
+		   If it is not already, then we must split node. */
+		for (j = 0; node->prefix[j]; j++) {
+			ch = node->prefix[j];
+
+			if (ch != key[i+j]) {
+				char *prefix = node->prefix;
+				struct index_node *n;
+
+				/* New child is copy of node with prefix[j+1..N] */
+				n = NOFAIL(calloc(1, sizeof(struct index_node)));
+				memcpy(n, node, sizeof(struct index_node));
+				n->prefix = NOFAIL(strdup(&prefix[j+1]));
+
+				/* Parent has prefix[0..j], child at prefix[j] */
+				memset(node, 0, sizeof(struct index_node));
+				prefix[j] = '\0';
+				node->prefix = prefix;
+				node->first = ch;
+				node->last = ch;
+				node->children[ch] = n;
+
+				break;
+			}
+		}
+		/* j is now length of node->prefix */
+		i += j;
+
+		ch = key[i];
+		if(ch == '\0')
+			return index_add_value(&node->values, value, priority);
+
+		if (!node->children[ch]) {
+			struct index_node *child;
+
+			if (ch < node->first)
+				node->first = ch;
+			if (ch > node->last)
+				node->last = ch;
+			node->children[ch] = NOFAIL(calloc(1, sizeof(struct index_node)));
+
+			child = node->children[ch];
+			child->prefix = NOFAIL(strdup(&key[i+1]));
+			child->first = INDEX_CHILDMAX;
+			index_add_value(&child->values, value, priority);
+
+			return 0;
+		}
+
+		/* Descend into child node and continue */
+		node = node->children[ch];
+		i++;
+	}
+}
+
+static int index__haschildren(const struct index_node *node)
+{
+	return node->first < INDEX_CHILDMAX;
+}
+
+/* Recursive post-order traversal
+
+   Pre-order would make for better read-side buffering / readahead / caching.
+   (post-order means you go backwards in the file as you descend the tree).
+   However, index reading is already fast enough.
+   Pre-order is simpler for writing, and depmod is already slow.
+ */
+static uint32_t index_write__node(const struct index_node *node, FILE *out)
+{
+	uint32_t *child_offs = NULL;
+	int child_count = 0;
+	long offset;
+
+	if (!node)
+		return 0;
+
+	/* Write children and save their offsets */
+	if (index__haschildren(node)) {
+		const struct index_node *child;
+		int i;
+
+		child_count = node->last - node->first + 1;
+		child_offs = NOFAIL(malloc(child_count * sizeof(uint32_t)));
+
+		for (i = 0; i < child_count; i++) {
+			child = node->children[node->first + i];
+			child_offs[i] = htonl(index_write__node(child, out));
+		}
+	}
+
+	/* Now write this node */
+	offset = ftell(out);
+
+	if (node->prefix[0]) {
+		fputs(node->prefix, out);
+		fputc('\0', out);
+		offset |= INDEX_NODE_PREFIX;
+	}
+
+	if (child_count) {
+		fputc(node->first, out);
+		fputc(node->last, out);
+		fwrite(child_offs, sizeof(uint32_t), child_count, out);
+		offset |= INDEX_NODE_CHILDS;
+	}
+
+	free(child_offs);
+
+	if (node->values) {
+		const struct index_value *v;
+		unsigned int value_count;
+		uint32_t u;
+
+		value_count = 0;
+		for (v = node->values; v != NULL; v = v->next)
+			value_count++;
+		u = htonl(value_count);
+		fwrite(&u, sizeof(u), 1, out);
+
+		for (v = node->values; v != NULL; v = v->next) {
+			u = htonl(v->priority);
+			fwrite(&u, sizeof(u), 1, out);
+			fputs(v->value, out);
+			fputc('\0', out);
+		}
+		offset |= INDEX_NODE_VALUES;
+	}
+
+	return offset;
+}
+
+static void index_write(const struct index_node *node, FILE *out)
+{
+	long initial_offset, final_offset;
+	uint32_t u;
+
+	u = htonl(INDEX_MAGIC);
+	fwrite(&u, sizeof(u), 1, out);
+	u = htonl(INDEX_VERSION);
+	fwrite(&u, sizeof(u), 1, out);
+
+	/* Second word is reserved for the offset of the root node */
+	initial_offset = ftell(out);
+	assert(initial_offset >= 0);
+	u = 0;
+	fwrite(&u, sizeof(uint32_t), 1, out);
+
+	/* Dump trie */
+	u = htonl(index_write__node(node, out));
+
+	/* Update first word */
+	final_offset = ftell(out);
+	assert(final_offset >= 0);
+	(void)fseek(out, initial_offset, SEEK_SET);
+	fwrite(&u, sizeof(uint32_t), 1, out);
+	(void)fseek(out, final_offset, SEEK_SET);
+}
+
+/* END: code from module-init-tools/index.c just modified to compile here.
+ */
+
+/* configuration parsing **********************************************/
+struct cfg_override {
+	struct cfg_override *next;
+	size_t len;
+	char path[];
+};
+
+enum search_type {
+	SEARCH_PATH,
+	SEARCH_BUILTIN,
+	SEARCH_EXTERNAL
+};
+
+struct cfg_search {
+	struct cfg_search *next;
+	enum search_type type;
+	size_t len;
+	char path[];
+};
+
+struct cfg_external {
+	struct cfg_external *next;
+	size_t len;
+	char path[];
+};
+
+struct cfg_exclude {
+	struct cfg_exclude *next;
+	char exclude_dir[];
+};
+
+struct cfg {
+	const char *kversion;
+	char dirname[PATH_MAX];
+	size_t dirnamelen;
+	char outdirname[PATH_MAX];
+	size_t outdirnamelen;
+	char sym_prefix;
+	uint8_t check_symvers;
+	uint8_t print_unknown;
+	uint8_t warn_dups;
+	struct cfg_override *overrides;
+	struct cfg_search *searches;
+	struct cfg_external *externals;
+	struct cfg_exclude *excludes;
+};
+
+static enum search_type cfg_define_search_type(const char *path)
+{
+	if (streq(path, CFG_BUILTIN_KEY))
+		return SEARCH_BUILTIN;
+	if (streq(path, CFG_EXTERNAL_KEY))
+		return SEARCH_EXTERNAL;
+	return SEARCH_PATH;
+}
+
+static int cfg_search_add(struct cfg *cfg, const char *path)
+{
+	struct cfg_search *s;
+	size_t len;
+	enum search_type type;
+
+	type = cfg_define_search_type(path);
+
+	if (type != SEARCH_PATH)
+		len = 0;
+	else
+		len = strlen(path) + 1;
+
+	s = malloc(sizeof(struct cfg_search) + len);
+	if (s == NULL) {
+		ERR("search add: out of memory\n");
+		return -ENOMEM;
+	}
+	s->type = type;
+	if (type != SEARCH_PATH)
+		s->len = 0;
+	else {
+		s->len = len - 1;
+		memcpy(s->path, path, len);
+	}
+
+	DBG("search add: %s, search type=%hhu\n", path, type);
+
+	s->next = cfg->searches;
+	cfg->searches = s;
+	return 0;
+}
+
+static void cfg_search_free(struct cfg_search *s)
+{
+	free(s);
+}
+
+static int cfg_override_add(struct cfg *cfg, const char *modname, const char *subdir)
+{
+	struct cfg_override *o;
+	size_t modnamelen = strlen(modname);
+	size_t subdirlen = strlen(subdir);
+	size_t i;
+
+	o = malloc(sizeof(struct cfg_override) + subdirlen + 1
+		   + modnamelen + 1);
+	if (o == NULL) {
+		ERR("override add: out of memory\n");
+		return -ENOMEM;
+	}
+	memcpy(o->path, subdir, subdirlen);
+	i = subdirlen;
+	o->path[i] = '/';
+	i++;
+
+	memcpy(o->path + i, modname, modnamelen);
+	i += modnamelen;
+	o->path[i] = '\0'; /* no extension, so we can match .ko/.ko.gz */
+
+	o->len = i;
+
+	DBG("override add: %s\n", o->path);
+
+	o->next = cfg->overrides;
+	cfg->overrides = o;
+	return 0;
+}
+
+static void cfg_override_free(struct cfg_override *o)
+{
+	free(o);
+}
+
+static int cfg_external_add(struct cfg *cfg, const char *path)
+{
+	struct cfg_external *ext;
+	size_t len = strlen(path);
+
+	ext = malloc(sizeof(struct cfg_external) + len + 1);
+	if (ext == NULL) {
+		ERR("external add: out of memory\n");
+		return -ENOMEM;
+	}
+
+	strcpy(ext->path, path);
+	ext->len = len;
+
+	DBG("external add: %s\n", ext->path);
+
+	ext->next = cfg->externals;
+	cfg->externals = ext;
+	return 0;
+}
+
+static void cfg_external_free(struct cfg_external *ext)
+{
+	free(ext);
+}
+
+static int cfg_exclude_add(struct cfg *cfg, const char *path)
+{
+	struct cfg_exclude *exc;
+	size_t len = strlen(path);
+
+	exc = malloc(sizeof(struct cfg_exclude) + len + 1);
+	if (exc == NULL) {
+		ERR("exclude add: out of memory\n");
+		return -ENOMEM;
+	}
+	memcpy(exc->exclude_dir, path, len + 1);
+
+	DBG("exclude add: %s\n", path);
+
+	exc->next = cfg->excludes;
+	cfg->excludes = exc;
+	return 0;
+}
+
+static void cfg_exclude_free(struct cfg_exclude *exc)
+{
+	free(exc);
+}
+
+static int cfg_kernel_matches(const struct cfg *cfg, const char *pattern)
+{
+	regex_t re;
+	int status;
+
+	/* old style */
+	if (streq(pattern, "*"))
+		return 1;
+
+	if (regcomp(&re, pattern, REG_EXTENDED|REG_NOSUB) != 0)
+		return 0;
+
+	status = regexec(&re, cfg->kversion, 0, NULL, 0);
+	regfree(&re);
+
+	return status == 0;
+}
+
+static int cfg_file_parse(struct cfg *cfg, const char *filename)
+{
+	char *line;
+	FILE *fp;
+	unsigned int linenum = 0;
+	int err;
+
+	fp = fopen(filename, "r");
+	if (fp == NULL) {
+		err = -errno;
+		ERR("file parse %s: %m\n", filename);
+		return err;
+	}
+
+	while ((line = freadline_wrapped(fp, &linenum)) != NULL) {
+		char *cmd, *saveptr;
+
+		if (line[0] == '\0' || line[0] == '#')
+			goto done_next;
+
+		cmd = strtok_r(line, "\t ", &saveptr);
+		if (cmd == NULL)
+			goto done_next;
+
+		if (streq(cmd, "search")) {
+			const char *sp;
+			while ((sp = strtok_r(NULL, "\t ", &saveptr)) != NULL) {
+				cfg_search_add(cfg, sp);
+			}
+		} else if (streq(cmd, "override")) {
+			const char *modname = strtok_r(NULL, "\t ", &saveptr);
+			const char *version = strtok_r(NULL, "\t ", &saveptr);
+			const char *subdir = strtok_r(NULL, "\t ", &saveptr);
+
+			if (modname == NULL || version == NULL ||
+					subdir == NULL)
+				goto syntax_error;
+
+			if (!cfg_kernel_matches(cfg, version)) {
+				INF("%s:%u: override kernel did not match %s\n",
+				    filename, linenum, version);
+				goto done_next;
+			}
+
+			cfg_override_add(cfg, modname, subdir);
+		} else if (streq(cmd, "external")) {
+			const char *version = strtok_r(NULL, "\t ", &saveptr);
+			const char *dir = strtok_r(NULL, "\t ", &saveptr);
+
+			if (version == NULL || dir == NULL)
+				goto syntax_error;
+
+			if (!cfg_kernel_matches(cfg, version)) {
+				INF("%s:%u: external directory did not match %s\n",
+				    filename, linenum, version);
+				goto done_next;
+			}
+
+			cfg_external_add(cfg, dir);
+		} else if (streq(cmd, "exclude")) {
+			const char *sp;
+			while ((sp = strtok_r(NULL, "\t ", &saveptr)) != NULL) {
+				cfg_exclude_add(cfg, sp);
+			}
+		} else if (streq(cmd, "include")
+				|| streq(cmd, "make_map_files")) {
+			INF("%s:%u: command %s not implemented yet\n",
+			    filename, linenum, cmd);
+		} else {
+syntax_error:
+			ERR("%s:%u: ignoring bad line starting with '%s'\n",
+			    filename, linenum, cmd);
+		}
+
+done_next:
+		free(line);
+	}
+
+	fclose(fp);
+
+	return 0;
+}
+
+static int cfg_files_filter_out(DIR *d, const char *dir, const char *name)
+{
+	size_t len = strlen(name);
+	struct stat st;
+
+	if (name[0] == '.')
+		return 1;
+
+	if (len < 6 || !streq(name + len - 5, ".conf")) {
+		INF("All cfg files need .conf: %s/%s\n", dir, name);
+		return 1;
+	}
+
+	fstatat(dirfd(d), name, &st, 0);
+	if (S_ISDIR(st.st_mode)) {
+		ERR("Directories inside directories are not supported: %s/%s\n",
+		    dir, name);
+		return 1;
+	}
+
+	return 0;
+}
+
+struct cfg_file {
+	size_t dirlen;
+	size_t namelen;
+	const char *name;
+	char path[];
+};
+
+static void cfg_file_free(struct cfg_file *f)
+{
+	free(f);
+}
+
+static int cfg_files_insert_sorted(struct cfg_file ***p_files, size_t *p_n_files,
+					const char *dir, const char *name)
+{
+	struct cfg_file **files, *f;
+	size_t i, n_files, namelen, dirlen;
+	void *tmp;
+
+	dirlen = strlen(dir);
+	if (name != NULL)
+		namelen = strlen(name);
+	else {
+		name = basename(dir);
+		namelen = strlen(name);
+		dirlen -= namelen + 1;
+	}
+
+	n_files = *p_n_files;
+	files = *p_files;
+	for (i = 0; i < n_files; i++) {
+		int cmp = strcmp(name, files[i]->name);
+		if (cmp == 0) {
+			DBG("Ignoring duplicate config file: %.*s/%s\n",
+			    (int)dirlen, dir, name);
+			return -EEXIST;
+		} else if (cmp < 0)
+			break;
+	}
+
+	f = malloc(sizeof(struct cfg_file) + dirlen + namelen + 2);
+	if (f == NULL) {
+		ERR("files insert sorted: out of memory\n");
+		return -ENOMEM;
+	}
+
+	tmp = realloc(files, sizeof(struct cfg_file *) * (n_files + 1));
+	if (tmp == NULL) {
+		ERR("files insert sorted: out of memory\n");
+		free(f);
+		return -ENOMEM;
+	}
+	*p_files = files = tmp;
+
+	if (i < n_files) {
+		memmove(files + i + 1, files + i,
+			sizeof(struct cfg_file *) * (n_files - i));
+	}
+	files[i] = f;
+
+	f->dirlen = dirlen;
+	f->namelen = namelen;
+	f->name = f->path + dirlen + 1;
+	memcpy(f->path, dir, dirlen);
+	f->path[dirlen] = '/';
+	memcpy(f->path + dirlen + 1, name, namelen);
+	f->path[dirlen + 1 + namelen] = '\0';
+
+	*p_n_files = n_files + 1;
+	return 0;
+}
+
+/*
+ * Insert configuration files ignoring duplicates
+ */
+static int cfg_files_list(struct cfg_file ***p_files, size_t *p_n_files,
+				const char *path)
+{
+	struct dirent *dent;
+	DIR *d;
+	int err = 0;
+	struct stat st;
+
+	if (stat(path, &st) != 0) {
+		err = -errno;
+		DBG("could not stat '%s': %m\n", path);
+		return err;
+	}
+
+	if (!S_ISDIR(st.st_mode)) {
+		cfg_files_insert_sorted(p_files, p_n_files, path, NULL);
+		return 0;
+	}
+
+	d = opendir(path);
+	if (d == NULL) {
+		ERR("files list %s: %m\n", path);
+		return -EINVAL;
+	}
+
+	for (dent = readdir(d); dent != NULL; dent = readdir(d)) {
+		if (cfg_files_filter_out(d, path, dent->d_name))
+			continue;
+
+		cfg_files_insert_sorted(p_files, p_n_files, path, dent->d_name);
+	}
+
+	closedir(d);
+	DBG("parsed configuration files from %s\n", path);
+	return err;
+}
+
+static int cfg_load(struct cfg *cfg, const char * const *cfg_paths)
+{
+	size_t i, n_files = 0;
+	struct cfg_file **files = NULL;
+
+	if (cfg_paths == NULL)
+		cfg_paths = default_cfg_paths;
+
+	for (i = 0; cfg_paths[i] != NULL; i++)
+		cfg_files_list(&files, &n_files, cfg_paths[i]);
+
+	for (i = 0; i < n_files; i++) {
+		struct cfg_file *f = files[i];
+		cfg_file_parse(cfg, f->path);
+		cfg_file_free(f);
+	}
+	free(files);
+
+	/* For backward compatibility add "updates" to the head of the search
+	 * list here. But only if there was no "search" option specified.
+	 */
+	if (cfg->searches == NULL)
+		cfg_search_add(cfg, "updates");
+
+	return 0;
+}
+
+static void cfg_free(struct cfg *cfg)
+{
+	while (cfg->overrides) {
+		struct cfg_override *tmp = cfg->overrides;
+		cfg->overrides = cfg->overrides->next;
+		cfg_override_free(tmp);
+	}
+
+	while (cfg->searches) {
+		struct cfg_search *tmp = cfg->searches;
+		cfg->searches = cfg->searches->next;
+		cfg_search_free(tmp);
+	}
+
+	while (cfg->externals) {
+		struct cfg_external *tmp = cfg->externals;
+		cfg->externals = cfg->externals->next;
+		cfg_external_free(tmp);
+	}
+
+	while (cfg->excludes) {
+		struct cfg_exclude *tmp = cfg->excludes;
+		cfg->excludes = cfg->excludes->next;
+		cfg_exclude_free(tmp);
+	}
+}
+
+
+/* depmod calculations ***********************************************/
+struct vertex;
+struct mod {
+	struct kmod_module *kmod;
+	char *path;
+	const char *relpath; /* path relative to '$ROOT$MODULE_DIRECTORY/$VER/' */
+	char *uncrelpath; /* same as relpath but ending in .ko */
+	struct kmod_list *info_list;
+	struct kmod_list *dep_sym_list;
+	struct array deps; /* struct symbol */
+	size_t baselen; /* points to start of basename/filename */
+	size_t modnamesz;
+	int sort_idx; /* sort index using modules.order */
+	int dep_sort_idx; /* topological sort index */
+	uint16_t idx; /* index in depmod->modules.array */
+	uint16_t users; /* how many modules depend on this one */
+	bool visited; /* helper field to report cycles */
+	struct vertex *vertex; /* helper field to report cycles */
+	char modname[];
+};
+
+struct symbol {
+	struct mod *owner;
+	uint64_t crc;
+	char name[];
+};
+
+struct depmod {
+	const struct cfg *cfg;
+	struct kmod_ctx *ctx;
+	struct array modules;
+	struct hash *modules_by_uncrelpath;
+	struct hash *modules_by_name;
+	struct hash *symbols;
+};
+
+static void mod_free(struct mod *mod)
+{
+	DBG("free %p kmod=%p, path=%s\n", mod, mod->kmod, mod->path);
+	array_free_array(&mod->deps);
+	kmod_module_unref(mod->kmod);
+	kmod_module_info_free_list(mod->info_list);
+	kmod_module_dependency_symbols_free_list(mod->dep_sym_list);
+	free(mod->uncrelpath);
+	free(mod->path);
+	free(mod);
+}
+
+static int mod_add_dependency(struct mod *mod, struct symbol *sym)
+{
+	int err;
+
+	DBG("%s depends on %s %s\n", mod->path, sym->name,
+	    sym->owner != NULL ? sym->owner->path : "(unknown)");
+
+	if (sym->owner == NULL)
+		return 0;
+
+	err = array_append_unique(&mod->deps, sym->owner);
+	if (err == -EEXIST)
+		return 0;
+	if (err < 0)
+		return err;
+
+	sym->owner->users++;
+	SHOW("%s needs \"%s\": %s\n", mod->path, sym->name, sym->owner->path);
+	return 0;
+}
+
+static void symbol_free(struct symbol *sym)
+{
+	DBG("free %p sym=%s, owner=%p %s\n", sym, sym->name, sym->owner,
+	    sym->owner != NULL ? sym->owner->path : "");
+	free(sym);
+}
+
+static int depmod_init(struct depmod *depmod, struct cfg *cfg,
+							struct kmod_ctx *ctx)
+{
+	int err = 0;
+
+	depmod->cfg = cfg;
+	depmod->ctx = ctx;
+
+	array_init(&depmod->modules, 128);
+
+	depmod->modules_by_uncrelpath = hash_new(512, NULL);
+	if (depmod->modules_by_uncrelpath == NULL) {
+		err = -errno;
+		goto modules_by_uncrelpath_failed;
+	}
+
+	depmod->modules_by_name = hash_new(512, NULL);
+	if (depmod->modules_by_name == NULL) {
+		err = -errno;
+		goto modules_by_name_failed;
+	}
+
+	depmod->symbols = hash_new(2048, (void (*)(void *))symbol_free);
+	if (depmod->symbols == NULL) {
+		err = -errno;
+		goto symbols_failed;
+	}
+
+	return 0;
+
+symbols_failed:
+	hash_free(depmod->modules_by_name);
+modules_by_name_failed:
+	hash_free(depmod->modules_by_uncrelpath);
+modules_by_uncrelpath_failed:
+	return err;
+}
+
+static void depmod_shutdown(struct depmod *depmod)
+{
+	size_t i;
+
+	hash_free(depmod->symbols);
+
+	hash_free(depmod->modules_by_uncrelpath);
+
+	hash_free(depmod->modules_by_name);
+
+	for (i = 0; i < depmod->modules.count; i++)
+		mod_free(depmod->modules.array[i]);
+	array_free_array(&depmod->modules);
+
+	kmod_unref(depmod->ctx);
+}
+
+static int depmod_module_add(struct depmod *depmod, struct kmod_module *kmod)
+{
+	const struct cfg *cfg = depmod->cfg;
+	const char *modname, *lastslash;
+	size_t modnamesz;
+	struct mod *mod;
+	int err;
+
+	modname = kmod_module_get_name(kmod);
+	modnamesz = strlen(modname) + 1;
+
+	mod = calloc(1, sizeof(struct mod) + modnamesz);
+	if (mod == NULL)
+		return -ENOMEM;
+	mod->kmod = kmod;
+	mod->sort_idx = depmod->modules.count + 1;
+	mod->dep_sort_idx = INT32_MAX;
+	memcpy(mod->modname, modname, modnamesz);
+	mod->modnamesz = modnamesz;
+
+	array_init(&mod->deps, 4);
+
+	mod->path = strdup(kmod_module_get_path(kmod));
+	lastslash = strrchr(mod->path, '/');
+	mod->baselen = lastslash - mod->path;
+	if (strncmp(mod->path, cfg->dirname, cfg->dirnamelen) == 0 &&
+			mod->path[cfg->dirnamelen] == '/')
+		mod->relpath = mod->path + cfg->dirnamelen + 1;
+	else
+		mod->relpath = NULL;
+
+	err = hash_add_unique(depmod->modules_by_name, mod->modname, mod);
+	if (err < 0) {
+		ERR("hash_add_unique %s: %s\n", mod->modname, strerror(-err));
+		goto fail;
+	}
+
+	if (mod->relpath != NULL) {
+		size_t uncrelpathlen = lastslash - mod->relpath + modnamesz
+				       + strlen(KMOD_EXTENSION_UNCOMPRESSED);
+		mod->uncrelpath = memdup(mod->relpath, uncrelpathlen + 1);
+		mod->uncrelpath[uncrelpathlen] = '\0';
+		err = hash_add_unique(depmod->modules_by_uncrelpath,
+				      mod->uncrelpath, mod);
+		if (err < 0) {
+			ERR("hash_add_unique %s: %s\n",
+			    mod->uncrelpath, strerror(-err));
+			hash_del(depmod->modules_by_name, mod->modname);
+			goto fail;
+		}
+	}
+
+	DBG("add %p kmod=%p, path=%s\n", mod, kmod, mod->path);
+
+	return 0;
+
+fail:
+	free(mod->uncrelpath);
+	free(mod);
+	return err;
+}
+
+static int depmod_module_del(struct depmod *depmod, struct mod *mod)
+{
+	DBG("del %p kmod=%p, path=%s\n", mod, mod->kmod, mod->path);
+
+	if (mod->uncrelpath != NULL)
+		hash_del(depmod->modules_by_uncrelpath, mod->uncrelpath);
+
+	hash_del(depmod->modules_by_name, mod->modname);
+
+	mod_free(mod);
+	return 0;
+}
+
+static const char *search_to_string(const struct cfg_search *s)
+{
+	switch(s->type) {
+	case SEARCH_EXTERNAL:
+		return "external";
+	case SEARCH_BUILTIN:
+		return "built-in";
+	default:
+		return s->path;
+	}
+}
+
+static bool depmod_is_path_starts_with(const char *path,
+				       size_t pathlen,
+				       const char *prefix,
+				       size_t prefix_len)
+{
+	if (pathlen <= prefix_len)
+		return false;
+	if (path[prefix_len] != '/')
+		return false;
+	if (memcmp(path, prefix, prefix_len) != 0)
+		return false;
+
+	return true;
+}
+
+/* returns if existing module @mod is higher priority than newpath.
+ * note this is the inverse of module-init-tools is_higher_priority()
+ */
+static int depmod_module_is_higher_priority(const struct depmod *depmod, const struct mod *mod, size_t baselen, size_t namelen, size_t modnamelen, const char *newpath)
+{
+	const struct cfg *cfg = depmod->cfg;
+	const struct cfg_override *ov;
+	const struct cfg_search *se;
+	const struct cfg_external *ext;
+
+	/* baselen includes the last '/' and mod->baselen doesn't. So it's
+	 * actually correct to use modnamelen in the first and modnamesz in
+	 * the latter */
+	size_t newlen = baselen + modnamelen;
+	size_t oldlen = mod->baselen + mod->modnamesz;
+	const char *oldpath = mod->path;
+	int i, bprio = -1, oldprio = -1, newprio = -1;
+	size_t relnewlen = 0;
+	size_t reloldlen = 0;
+	const char *relnewpath = NULL;
+	const char *reloldpath = NULL;
+
+	DBG("comparing priorities of %s and %s\n",
+	    oldpath, newpath);
+
+	if (strncmp(newpath, cfg->dirname, cfg->dirnamelen) == 0) {
+		relnewpath = newpath + cfg->dirnamelen + 1;
+		relnewlen = newlen - (cfg->dirnamelen + 1);
+	}
+	if (strncmp(oldpath, cfg->dirname, cfg->dirnamelen) == 0) {
+		reloldpath = oldpath + cfg->dirnamelen + 1;
+		reloldlen = oldlen - (cfg->dirnamelen + 1);
+	}
+
+	for (ov = cfg->overrides; ov != NULL; ov = ov->next) {
+		DBG("override %s\n", ov->path);
+		if (relnewlen == ov->len &&
+		    memcmp(ov->path, relnewpath, relnewlen) == 0)
+			return 0;
+		if (reloldlen == ov->len &&
+		    memcmp(ov->path, reloldpath, reloldlen) == 0)
+			return 1;
+	}
+
+	for (i = 0, se = cfg->searches; se != NULL; se = se->next, i++) {
+		DBG("search %s\n", search_to_string(se));
+		if (se->type == SEARCH_BUILTIN)
+			bprio = i;
+		else if (se->type == SEARCH_EXTERNAL) {
+			for (ext = cfg->externals; ext != NULL; ext = ext->next, i++) {
+				if (depmod_is_path_starts_with(newpath,
+							       newlen,
+							       ext->path,
+							       ext->len))
+					newprio = i;
+				if (depmod_is_path_starts_with(oldpath,
+							       oldlen,
+							       ext->path,
+							       ext->len))
+					oldprio = i;
+			}
+		} else if (relnewlen > se->len && relnewpath[se->len] == '/' &&
+			 memcmp(se->path, relnewpath, se->len) == 0)
+			newprio = i;
+		else if (reloldlen > se->len && reloldpath[se->len] == '/' &&
+			 memcmp(se->path, reloldpath, se->len) == 0)
+			oldprio = i;
+	}
+
+	if (newprio < 0)
+		newprio = bprio;
+	if (oldprio < 0)
+		oldprio = bprio;
+
+	DBG("priorities: built-in: %d, old: %d, new: %d\n",
+	    bprio, oldprio, newprio);
+
+	return newprio <= oldprio;
+}
+
+static int depmod_modules_search_file(struct depmod *depmod, size_t baselen, size_t namelen, const char *path)
+{
+	struct kmod_module *kmod;
+	struct mod *mod;
+	const char *relpath;
+	char modname[PATH_MAX];
+	size_t modnamelen;
+	int err;
+
+	if (!path_ends_with_kmod_ext(path + baselen, namelen))
+		return 0;
+
+	if (path_to_modname(path, modname, &modnamelen) == NULL) {
+		ERR("could not get modname from path %s\n", path);
+		return -EINVAL;
+	}
+
+	relpath = path + depmod->cfg->dirnamelen + 1;
+	DBG("try %s (%s)\n", relpath, modname);
+
+	mod = hash_find(depmod->modules_by_name, modname);
+	if (mod == NULL)
+		goto add;
+
+	if (depmod_module_is_higher_priority(depmod, mod, baselen,
+						namelen, modnamelen, path)) {
+		DBG("Ignored lower priority: %s, higher: %s\n",
+		    path, mod->path);
+		return 0;
+	}
+
+	DBG("Replace lower priority %s with new module %s\n",
+	    mod->relpath, relpath);
+	err = depmod_module_del(depmod, mod);
+	if (err < 0) {
+		ERR("could not del module %s: %s\n", mod->path, strerror(-err));
+		return err;
+	}
+
+add:
+	err = kmod_module_new_from_path(depmod->ctx, path, &kmod);
+	if (err < 0) {
+		ERR("could not create module %s: %s\n", path, strerror(-err));
+		return err;
+	}
+
+	err = depmod_module_add(depmod, kmod);
+	if (err < 0) {
+		ERR("could not add module %s: %s\n",
+		    path, strerror(-err));
+		kmod_module_unref(kmod);
+		return err;
+	}
+	return 0;
+}
+
+static bool should_exclude_dir(const struct cfg *cfg, const char *name)
+{
+	struct cfg_exclude *exc;
+
+	if (name[0] == '.' && (name[1] == '\0' ||
+			(name[1] == '.' && name[2] == '\0')))
+		return true;
+
+	if (streq(name, "build") || streq(name, "source"))
+		return true;
+
+	for (exc = cfg->excludes; exc != NULL; exc = exc->next) {
+		if (streq(name, exc->exclude_dir))
+			return true;
+	}
+
+	return false;
+}
+
+static int depmod_modules_search_dir(struct depmod *depmod, DIR *d, size_t baselen, struct scratchbuf *s_path)
+{
+	struct dirent *de;
+	int err = 0, dfd = dirfd(d);
+	char *path;
+
+	while ((de = readdir(d)) != NULL) {
+		const char *name = de->d_name;
+		size_t namelen;
+		uint8_t is_dir;
+
+		if (should_exclude_dir(depmod->cfg, name))
+			continue;
+
+		namelen = strlen(name);
+		if (scratchbuf_alloc(s_path, baselen + namelen + 2) < 0) {
+			err = -ENOMEM;
+			ERR("No memory\n");
+			continue;
+		}
+
+		path = scratchbuf_str(s_path);
+		memcpy(path + baselen, name, namelen + 1);
+
+		if (de->d_type == DT_REG)
+			is_dir = 0;
+		else if (de->d_type == DT_DIR)
+			is_dir = 1;
+		else {
+			struct stat st;
+			if (fstatat(dfd, name, &st, 0) < 0) {
+				ERR("fstatat(%d, %s): %m\n", dfd, name);
+				continue;
+			} else if (S_ISREG(st.st_mode))
+				is_dir = 0;
+			else if (S_ISDIR(st.st_mode))
+				is_dir = 1;
+			else {
+				ERR("unsupported file type %s: %o\n",
+				    path, st.st_mode & S_IFMT);
+				continue;
+			}
+		}
+
+		if (is_dir) {
+			int fd;
+			DIR *subdir;
+			fd = openat(dfd, name, O_RDONLY);
+			if (fd < 0) {
+				ERR("openat(%d, %s, O_RDONLY): %m\n",
+				    dfd, name);
+				continue;
+			}
+			subdir = fdopendir(fd);
+			if (subdir == NULL) {
+				ERR("fdopendir(%d): %m\n", fd);
+				close(fd);
+				continue;
+			}
+			path[baselen + namelen] = '/';
+			path[baselen + namelen + 1] = '\0';
+			err = depmod_modules_search_dir(depmod, subdir,
+							baselen + namelen + 1,
+							s_path);
+			closedir(subdir);
+		} else {
+			err = depmod_modules_search_file(depmod, baselen,
+							 namelen, path);
+		}
+
+		if (err < 0) {
+			path[baselen + namelen] = '\0';
+			ERR("failed %s: %s\n", path, strerror(-err));
+			err = 0; /* ignore errors */
+		}
+	}
+	return err;
+}
+
+static int depmod_modules_search_path(struct depmod *depmod,
+				      const char *path)
+{
+	char buf[256];
+	_cleanup_(scratchbuf_release) struct scratchbuf s_path_buf =
+		SCRATCHBUF_INITIALIZER(buf);
+	char *path_buf;
+	DIR *d;
+	size_t baselen;
+	int err;
+
+	d = opendir(path);
+	if (d == NULL) {
+		err = -errno;
+		ERR("could not open directory %s: %m\n", path);
+		return err;
+	}
+
+	baselen = strlen(path);
+
+	if (scratchbuf_alloc(&s_path_buf, baselen + 2) < 0) {
+		err = -ENOMEM;
+		goto out;
+	}
+	path_buf = scratchbuf_str(&s_path_buf);
+
+	memcpy(path_buf, path, baselen);
+	path_buf[baselen] = '/';
+	baselen++;
+	path_buf[baselen] = '\0';
+
+	err = depmod_modules_search_dir(depmod, d, baselen, &s_path_buf);
+out:
+	closedir(d);
+	return err;
+}
+
+static int depmod_modules_search(struct depmod *depmod)
+{
+	int err;
+	struct cfg_external *ext;
+
+	err = depmod_modules_search_path(depmod, depmod->cfg->dirname);
+	if (err < 0)
+		return err;
+
+	for (ext = depmod->cfg->externals; ext != NULL; ext = ext->next) {
+		err = depmod_modules_search_path(depmod, ext->path);
+		if (err < 0 && err == -ENOENT)
+			/* ignore external dir absense */
+			continue;
+	}
+
+	return 0;
+}
+
+static int mod_cmp(const void *pa, const void *pb) {
+	const struct mod *a = *(const struct mod **)pa;
+	const struct mod *b = *(const struct mod **)pb;
+	return a->sort_idx - b->sort_idx;
+}
+
+static int depmod_modules_build_array(struct depmod *depmod)
+{
+	struct hash_iter module_iter;
+	const void *v;
+	int err;
+
+	hash_iter_init(depmod->modules_by_name, &module_iter);
+	while (hash_iter_next(&module_iter, NULL, &v)) {
+		struct mod *mod = (struct mod *) v;
+		mod->idx = depmod->modules.count;
+		err = array_append(&depmod->modules, mod);
+		if (err < 0)
+			return err;
+	}
+
+	return 0;
+}
+
+static FILE *dfdopen(const char *dname, const char *filename, int flags,
+		     const char *mode)
+{
+	int fd, dfd;
+	FILE *ret;
+
+	dfd = open(dname, O_RDONLY);
+	if (dfd < 0) {
+		WRN("could not open directory %s: %m\n", dname);
+		return NULL;
+	}
+
+	fd = openat(dfd, filename, flags);
+	if (fd < 0) {
+		WRN("could not open %s at %s: %m\n", filename, dname);
+		ret = NULL;
+	} else {
+		ret = fdopen(fd, mode);
+		if (!ret) {
+			WRN("could not associate stream with %s: %m\n", filename);
+			close(fd);
+		}
+	}
+	close(dfd);
+	return ret;
+}
+
+
+
+static void depmod_modules_sort(struct depmod *depmod)
+{
+	char line[PATH_MAX];
+	const char *order_file = "modules.order";
+	FILE *fp;
+	unsigned idx = 0, total = 0;
+
+	fp = dfdopen(depmod->cfg->dirname, order_file, O_RDONLY, "r");
+	if (fp == NULL)
+		return;
+
+	while (fgets(line, sizeof(line), fp) != NULL) {
+		size_t len = strlen(line);
+		idx++;
+		if (len == 0)
+			continue;
+		if (line[len - 1] != '\n') {
+			ERR("%s/%s:%u corrupted line misses '\\n'\n",
+				depmod->cfg->dirname, order_file, idx);
+			goto corrupted;
+		}
+	}
+	total = idx + 1;
+	idx = 0;
+	fseek(fp, 0, SEEK_SET);
+	while (fgets(line, sizeof(line), fp) != NULL) {
+		size_t len = strlen(line);
+		struct mod *mod;
+
+		idx++;
+		if (len == 0)
+			continue;
+		line[len - 1] = '\0';
+
+		mod = hash_find(depmod->modules_by_uncrelpath, line);
+		if (mod == NULL)
+			continue;
+		mod->sort_idx = idx - total;
+	}
+
+	array_sort(&depmod->modules, mod_cmp);
+	for (idx = 0; idx < depmod->modules.count; idx++) {
+		struct mod *m = depmod->modules.array[idx];
+		m->idx = idx;
+	}
+
+corrupted:
+	fclose(fp);
+}
+
+static int depmod_symbol_add(struct depmod *depmod, const char *name,
+					bool prefix_skipped, uint64_t crc,
+					const struct mod *owner)
+{
+	size_t namelen;
+	int err;
+	struct symbol *sym;
+
+	if (!prefix_skipped && (name[0] == depmod->cfg->sym_prefix))
+		name++;
+
+	namelen = strlen(name) + 1;
+	sym = malloc(sizeof(struct symbol) + namelen);
+	if (sym == NULL)
+		return -ENOMEM;
+
+	sym->owner = (struct mod *)owner;
+	sym->crc = crc;
+	memcpy(sym->name, name, namelen);
+
+	err = hash_add(depmod->symbols, sym->name, sym);
+	if (err < 0) {
+		free(sym);
+		return err;
+	}
+
+	DBG("add %p sym=%s, owner=%p %s\n", sym, sym->name, owner,
+	    owner != NULL ? owner->path : "");
+
+	return 0;
+}
+
+static struct symbol *depmod_symbol_find(const struct depmod *depmod,
+							const char *name)
+{
+	if (name[0] == '.') /* PPC64 needs this: .foo == foo */
+		name++;
+	if (name[0] == depmod->cfg->sym_prefix)
+		name++;
+	return hash_find(depmod->symbols, name);
+}
+
+static int depmod_load_modules(struct depmod *depmod)
+{
+	struct mod **itr, **itr_end;
+
+	DBG("load symbols (%zd modules)\n", depmod->modules.count);
+
+	itr = (struct mod **)depmod->modules.array;
+	itr_end = itr + depmod->modules.count;
+	for (; itr < itr_end; itr++) {
+		struct mod *mod = *itr;
+		struct kmod_list *l, *list = NULL;
+		int err = kmod_module_get_symbols(mod->kmod, &list);
+		if (err < 0) {
+			if (err == -ENODATA)
+				DBG("ignoring %s: no symbols\n", mod->path);
+			else
+				ERR("failed to load symbols from %s: %s\n",
+						mod->path, strerror(-err));
+			goto load_info;
+		}
+		kmod_list_foreach(l, list) {
+			const char *name = kmod_module_symbol_get_symbol(l);
+			uint64_t crc = kmod_module_symbol_get_crc(l);
+			depmod_symbol_add(depmod, name, false, crc, mod);
+		}
+		kmod_module_symbols_free_list(list);
+
+load_info:
+		kmod_module_get_info(mod->kmod, &mod->info_list);
+		kmod_module_get_dependency_symbols(mod->kmod,
+						   &mod->dep_sym_list);
+		kmod_module_unref(mod->kmod);
+		mod->kmod = NULL;
+	}
+
+	DBG("loaded symbols (%zd modules, %u symbols)\n",
+	    depmod->modules.count, hash_get_count(depmod->symbols));
+
+	return 0;
+}
+
+static int depmod_load_module_dependencies(struct depmod *depmod, struct mod *mod)
+{
+	const struct cfg *cfg = depmod->cfg;
+	struct kmod_list *l;
+
+	DBG("do dependencies of %s\n", mod->path);
+	kmod_list_foreach(l, mod->dep_sym_list) {
+		const char *name = kmod_module_dependency_symbol_get_symbol(l);
+		uint64_t crc = kmod_module_dependency_symbol_get_crc(l);
+		int bindtype = kmod_module_dependency_symbol_get_bind(l);
+		struct symbol *sym = depmod_symbol_find(depmod, name);
+		uint8_t is_weak = bindtype == KMOD_SYMBOL_WEAK;
+
+		if (sym == NULL) {
+			DBG("%s needs (%c) unknown symbol %s\n",
+			    mod->path, bindtype, name);
+			if (cfg->print_unknown && !is_weak)
+				WRN("%s needs unknown symbol %s\n",
+				    mod->path, name);
+			continue;
+		}
+
+		if (cfg->check_symvers && sym->crc != crc && !is_weak) {
+			DBG("symbol %s (%#"PRIx64") module %s (%#"PRIx64")\n",
+			    sym->name, sym->crc, mod->path, crc);
+			if (cfg->print_unknown)
+				WRN("%s disagrees about version of symbol %s\n",
+				    mod->path, name);
+		}
+
+		mod_add_dependency(mod, sym);
+	}
+
+	return 0;
+}
+
+static int depmod_load_dependencies(struct depmod *depmod)
+{
+	struct mod **itr, **itr_end;
+
+	DBG("load dependencies (%zd modules, %u symbols)\n",
+	    depmod->modules.count, hash_get_count(depmod->symbols));
+
+	itr = (struct mod **)depmod->modules.array;
+	itr_end = itr + depmod->modules.count;
+	for (; itr < itr_end; itr++) {
+		struct mod *mod = *itr;
+
+		if (mod->dep_sym_list == NULL) {
+			DBG("ignoring %s: no dependency symbols\n", mod->path);
+			continue;
+		}
+
+		depmod_load_module_dependencies(depmod, mod);
+	}
+
+	DBG("loaded dependencies (%zd modules, %u symbols)\n",
+	    depmod->modules.count, hash_get_count(depmod->symbols));
+
+	return 0;
+}
+
+static int dep_cmp(const void *pa, const void *pb)
+{
+	const struct mod *a = *(const struct mod **)pa;
+	const struct mod *b = *(const struct mod **)pb;
+	return a->dep_sort_idx - b->dep_sort_idx;
+}
+
+static void depmod_sort_dependencies(struct depmod *depmod)
+{
+	struct mod **itr, **itr_end;
+	itr = (struct mod **)depmod->modules.array;
+	itr_end = itr + depmod->modules.count;
+	for (; itr < itr_end; itr++) {
+		struct mod *m = *itr;
+		if (m->deps.count > 1)
+			array_sort(&m->deps, dep_cmp);
+	}
+}
+
+struct vertex {
+	struct vertex *parent;
+	struct mod *mod;
+};
+
+static struct vertex *vertex_new(struct mod *mod, struct vertex *parent)
+{
+	struct vertex *v;
+
+	v = malloc(sizeof(*v));
+	if (v == NULL)
+		return NULL;
+
+	v->parent = parent;
+	v->mod = mod;
+	return v;
+}
+
+static void depmod_list_remove_data(struct kmod_list **list, void *data)
+{
+	struct kmod_list *l;
+
+	l = kmod_list_remove_data(*list, data);
+	*list = l;
+}
+
+static int depmod_report_one_cycle(struct depmod *depmod,
+				   struct vertex *vertex,
+				   struct kmod_list **roots,
+				   struct hash *loop_set)
+{
+	const char sep[] = " -> ";
+	size_t sz;
+	char *buf;
+	struct array reverse;
+	int i;
+	int n;
+	struct vertex *v;
+	int rc;
+
+	array_init(&reverse, 3);
+
+	sz = 0;
+	for (v = vertex->parent, n = 0;
+	     v != NULL;
+	     v = v->parent, n++) {
+
+		sz += v->mod->modnamesz - 1;
+		array_append(&reverse, v);
+		rc = hash_add(loop_set, v->mod->modname, NULL);
+		if (rc != 0)
+			return rc;
+		/* the hash will be freed where created */
+	}
+	sz += vertex->mod->modnamesz - 1;
+
+	buf = malloc(sz + n * strlen(sep) + 1);
+
+	sz = 0;
+	for (i = reverse.count - 1; i >= 0; i--) {
+		size_t len;
+
+		v = reverse.array[i];
+
+		len = v->mod->modnamesz - 1;
+		memcpy(buf + sz, v->mod->modname, len);
+		sz += len;
+		strcpy(buf + sz, sep);
+		sz += strlen(sep);
+
+		depmod_list_remove_data(roots, v->mod);
+	}
+	strcpy(buf + sz, vertex->mod->modname);
+	ERR("Cycle detected: %s\n", buf);
+
+	free(buf);
+	array_free_array(&reverse);
+
+	return 0;
+}
+
+static int depmod_report_cycles_from_root(struct depmod *depmod,
+					  struct mod *root_mod,
+					  struct kmod_list **roots,
+					  void **stack,
+					  size_t stack_size,
+					  struct hash *loop_set)
+{
+	struct kmod_list *free_list = NULL; /* struct vertex */
+	struct kmod_list *l;
+	struct vertex *root;
+	struct vertex *vertex;
+	struct vertex *v;
+	struct mod *m;
+	struct mod **itr, **itr_end;
+	size_t is;
+	int ret = -ENOMEM;
+
+	root = vertex_new(root_mod, NULL);
+	if (root == NULL) {
+		ERR("No memory to report cycles\n");
+		goto out;
+	}
+
+	l = kmod_list_append(free_list, root);
+	if (l == NULL) {
+		ERR("No memory to report cycles\n");
+		goto out;
+	}
+	free_list = l;
+
+	is = 0;
+	stack[is++] = (void *)root;
+
+	while (is > 0) {
+		vertex = stack[--is];
+		m = vertex->mod;
+		/*
+		 * because of the topological sort we can start only
+		 * from part of a loop or from a branch after a loop
+		 */
+		if (m->visited && m == root->mod) {
+			int rc;
+			rc = depmod_report_one_cycle(depmod, vertex,
+						     roots, loop_set);
+			if (rc != 0) {
+				ret = rc;
+				goto out;
+			}
+			continue;
+		}
+
+		m->visited = true;
+		if (m->deps.count == 0) {
+			/*
+			 * boundary condition: if there is more than one
+			 * single node branch (not a loop), it is
+			 * recognized as a loop by the code above:
+			 * m->visited because more then one,
+			 * m == root->mod since it is a single node.
+			 * So, prevent deeping into the branch second
+			 * time.
+			 */
+			depmod_list_remove_data(roots, m);
+
+			continue;
+		}
+
+		itr = (struct mod **) m->deps.array;
+		itr_end = itr + m->deps.count;
+		for (; itr < itr_end; itr++) {
+			struct mod *dep = *itr;
+			v = vertex_new(dep, vertex);
+			if (v == NULL) {
+				ERR("No memory to report cycles\n");
+				goto out;
+			}
+			assert(is < stack_size);
+			stack[is++] = v;
+
+			l = kmod_list_append(free_list, v);
+			if (l == NULL) {
+				ERR("No memory to report cycles\n");
+				goto out;
+			}
+			free_list = l;
+
+		}
+	}
+	ret = 0;
+
+out:
+	while (free_list) {
+		v = free_list->data;
+		l = kmod_list_remove(free_list);
+		free_list = l;
+		free(v);
+	}
+
+	return ret;
+}
+
+static void depmod_report_cycles(struct depmod *depmod, uint16_t n_mods,
+				 uint16_t *users)
+{
+	int num_cyclic = 0;
+	struct kmod_list *roots = NULL; /* struct mod */
+	struct kmod_list *l;
+	size_t n_r; /* local n_roots */
+	int i;
+	int err;
+	_cleanup_free_ void **stack = NULL;
+	struct mod *m;
+	struct mod *root;
+	struct hash *loop_set;
+
+	for (i = 0, n_r = 0; i < n_mods; i++) {
+		if (users[i] <= 0)
+			continue;
+		m = depmod->modules.array[i];
+		l = kmod_list_append(roots, m);
+		if (l == NULL) {
+			ERR("No memory to report cycles\n");
+			goto out_list;
+		}
+		roots = l;
+		n_r++;
+	}
+
+	stack = malloc(n_r * sizeof(void *));
+	if (stack == NULL) {
+		ERR("No memory to report cycles\n");
+		goto out_list;
+	}
+
+	loop_set = hash_new(16, NULL);
+	if (loop_set == NULL) {
+		ERR("No memory to report cycles\n");
+		goto out_list;
+	}
+
+	while (roots != NULL) {
+		root = roots->data;
+		l = kmod_list_remove(roots);
+		roots = l;
+		err = depmod_report_cycles_from_root(depmod,
+						     root,
+						     &roots,
+						     stack, n_r, loop_set);
+		if (err < 0)
+			goto out_hash;
+	}
+
+	num_cyclic = hash_get_count(loop_set);
+	ERR("Found %d modules in dependency cycles!\n", num_cyclic);
+
+out_hash:
+	hash_free(loop_set);
+out_list:
+	while (roots != NULL) {
+		/* no need to free data, come from outside */
+		roots = kmod_list_remove(roots);
+	}
+}
+
+static int depmod_calculate_dependencies(struct depmod *depmod)
+{
+	const struct mod **itrm;
+	uint16_t *users, *roots, *sorted;
+	uint16_t i, n_roots = 0, n_sorted = 0, n_mods = depmod->modules.count;
+	int ret = 0;
+
+	users = malloc(sizeof(uint16_t) * n_mods * 3);
+	if (users == NULL)
+		return -ENOMEM;
+	roots = users + n_mods;
+	sorted = roots + n_mods;
+
+	DBG("calculate dependencies and ordering (%hu modules)\n", n_mods);
+
+	assert(depmod->modules.count < UINT16_MAX);
+
+	/* populate modules users (how many modules uses it) */
+	itrm = (const struct mod **)depmod->modules.array;
+	for (i = 0; i < n_mods; i++, itrm++) {
+		const struct mod *m = *itrm;
+		users[i] = m->users;
+		if (users[i] == 0) {
+			roots[n_roots] = i;
+			n_roots++;
+		}
+	}
+
+	/* topological sort (outputs modules without users first) */
+	while (n_roots > 0) {
+		const struct mod **itr_dst, **itr_dst_end;
+		struct mod *src;
+		uint16_t src_idx = roots[--n_roots];
+
+		src = depmod->modules.array[src_idx];
+		src->dep_sort_idx = n_sorted;
+		sorted[n_sorted] = src_idx;
+		n_sorted++;
+
+		itr_dst = (const struct mod **)src->deps.array;
+		itr_dst_end = itr_dst + src->deps.count;
+		for (; itr_dst < itr_dst_end; itr_dst++) {
+			const struct mod *dst = *itr_dst;
+			uint16_t dst_idx = dst->idx;
+			assert(users[dst_idx] > 0);
+			users[dst_idx]--;
+			if (users[dst_idx] == 0) {
+				roots[n_roots] = dst_idx;
+				n_roots++;
+			}
+		}
+	}
+
+	if (n_sorted < n_mods) {
+		depmod_report_cycles(depmod, n_mods, users);
+		ret = -EINVAL;
+		goto exit;
+	}
+
+	depmod_sort_dependencies(depmod);
+
+	DBG("calculated dependencies and ordering (%hu modules)\n", n_mods);
+
+exit:
+	free(users);
+	return ret;
+}
+
+static int depmod_load(struct depmod *depmod)
+{
+	int err;
+
+	err = depmod_load_modules(depmod);
+	if (err < 0)
+		return err;
+
+	err = depmod_load_dependencies(depmod);
+	if (err < 0)
+		return err;
+
+	err = depmod_calculate_dependencies(depmod);
+	if (err < 0)
+		return err;
+
+	return 0;
+}
+
+static size_t mod_count_all_dependencies(const struct mod *mod)
+{
+	size_t i, count = 0;
+	for (i = 0; i < mod->deps.count; i++) {
+		const struct mod *d = mod->deps.array[i];
+		count += 1 + mod_count_all_dependencies(d);
+	}
+	return count;
+}
+
+static int mod_fill_all_unique_dependencies(const struct mod *mod, const struct mod **deps, size_t n_deps, size_t *last)
+{
+	size_t i;
+	int err = 0;
+	for (i = 0; i < mod->deps.count; i++) {
+		const struct mod *d = mod->deps.array[i];
+		size_t j;
+		uint8_t exists = 0;
+
+		for (j = 0; j < *last; j++) {
+			if (deps[j] == d) {
+				exists = 1;
+				break;
+			}
+		}
+
+		if (exists)
+			continue;
+
+		if (*last >= n_deps)
+			return -ENOSPC;
+		deps[*last] = d;
+		(*last)++;
+		err = mod_fill_all_unique_dependencies(d, deps, n_deps, last);
+		if (err < 0)
+			break;
+	}
+	return err;
+}
+
+static const struct mod **mod_get_all_sorted_dependencies(const struct mod *mod, size_t *n_deps)
+{
+	const struct mod **deps;
+	size_t last = 0;
+
+	*n_deps = mod_count_all_dependencies(mod);
+	if (*n_deps == 0)
+		return NULL;
+
+	deps = malloc(sizeof(struct mod *) * (*n_deps));
+	if (deps == NULL)
+		return NULL;
+
+	if (mod_fill_all_unique_dependencies(mod, deps, *n_deps, &last) < 0) {
+		free(deps);
+		return NULL;
+	}
+
+	qsort(deps, last, sizeof(struct mod *), dep_cmp);
+	*n_deps = last;
+	return deps;
+}
+
+static inline const char *mod_get_compressed_path(const struct mod *mod)
+{
+	if (mod->relpath != NULL)
+		return mod->relpath;
+	return mod->path;
+}
+
+static int output_deps(struct depmod *depmod, FILE *out)
+{
+	size_t i;
+
+	for (i = 0; i < depmod->modules.count; i++) {
+		const struct mod **deps, *mod = depmod->modules.array[i];
+		const char *p = mod_get_compressed_path(mod);
+		size_t j, n_deps;
+
+		fprintf(out, "%s:", p);
+
+		if (mod->deps.count == 0)
+			goto end;
+
+		deps = mod_get_all_sorted_dependencies(mod, &n_deps);
+		if (deps == NULL) {
+			ERR("could not get all sorted dependencies of %s\n", p);
+			goto end;
+		}
+
+		for (j = 0; j < n_deps; j++) {
+			const struct mod *d = deps[j];
+			fprintf(out, " %s", mod_get_compressed_path(d));
+		}
+		free(deps);
+	end:
+		putc('\n', out);
+	}
+
+	return 0;
+}
+
+static int output_deps_bin(struct depmod *depmod, FILE *out)
+{
+	struct index_node *idx;
+	size_t i;
+
+	if (out == stdout)
+		return 0;
+
+	idx = index_create();
+	if (idx == NULL)
+		return -ENOMEM;
+
+	for (i = 0; i < depmod->modules.count; i++) {
+		const struct mod **deps, *mod = depmod->modules.array[i];
+		const char *p = mod_get_compressed_path(mod);
+		char *line;
+		size_t j, n_deps, linepos, linelen, slen;
+		int duplicate;
+
+		deps = mod_get_all_sorted_dependencies(mod, &n_deps);
+		if (deps == NULL && n_deps > 0) {
+			ERR("could not get all sorted dependencies of %s\n", p);
+			continue;
+		}
+
+		linelen = strlen(p) + 1;
+		for (j = 0; j < n_deps; j++) {
+			const struct mod *d = deps[j];
+			linelen += 1 + strlen(mod_get_compressed_path(d));
+		}
+
+		line = malloc(linelen + 1);
+		if (line == NULL) {
+			free(deps);
+			ERR("modules.deps.bin: out of memory\n");
+			continue;
+		}
+
+		linepos = 0;
+		slen = strlen(p);
+		memcpy(line + linepos, p, slen);
+		linepos += slen;
+		line[linepos] = ':';
+		linepos++;
+
+		for (j = 0; j < n_deps; j++) {
+			const struct mod *d = deps[j];
+			const char *dp;
+
+			line[linepos] = ' ';
+			linepos++;
+
+			dp = mod_get_compressed_path(d);
+			slen = strlen(dp);
+			memcpy(line + linepos, dp, slen);
+			linepos += slen;
+		}
+		line[linepos] = '\0';
+
+		duplicate = index_insert(idx, mod->modname, line, mod->idx);
+		if (duplicate && depmod->cfg->warn_dups)
+			WRN("duplicate module deps:\n%s\n", line);
+		free(line);
+		free(deps);
+	}
+
+	index_write(idx, out);
+	index_destroy(idx);
+
+	return 0;
+}
+
+static int output_aliases(struct depmod *depmod, FILE *out)
+{
+	size_t i;
+
+	fputs("# Aliases extracted from modules themselves.\n", out);
+
+	for (i = 0; i < depmod->modules.count; i++) {
+		const struct mod *mod = depmod->modules.array[i];
+		struct kmod_list *l;
+
+		kmod_list_foreach(l, mod->info_list) {
+			const char *key = kmod_module_info_get_key(l);
+			const char *value = kmod_module_info_get_value(l);
+
+			if (!streq(key, "alias"))
+				continue;
+
+			fprintf(out, "alias %s %s\n", value, mod->modname);
+		}
+	}
+
+	return 0;
+}
+
+static int output_aliases_bin(struct depmod *depmod, FILE *out)
+{
+	struct index_node *idx;
+	size_t i;
+
+	if (out == stdout)
+		return 0;
+
+	idx = index_create();
+	if (idx == NULL)
+		return -ENOMEM;
+
+	for (i = 0; i < depmod->modules.count; i++) {
+		const struct mod *mod = depmod->modules.array[i];
+		struct kmod_list *l;
+
+		kmod_list_foreach(l, mod->info_list) {
+			const char *key = kmod_module_info_get_key(l);
+			const char *value = kmod_module_info_get_value(l);
+			char buf[PATH_MAX];
+			const char *alias;
+			int duplicate;
+
+			if (!streq(key, "alias"))
+				continue;
+
+			if (alias_normalize(value, buf, NULL) < 0) {
+				WRN("Unmatched bracket in %s\n", value);
+				continue;
+			}
+			alias = buf;
+
+			duplicate = index_insert(idx, alias, mod->modname,
+						 mod->idx);
+			if (duplicate && depmod->cfg->warn_dups)
+				WRN("duplicate module alias:\n%s %s\n",
+				    alias, mod->modname);
+		}
+	}
+
+	index_write(idx, out);
+	index_destroy(idx);
+
+	return 0;
+}
+
+static int output_softdeps(struct depmod *depmod, FILE *out)
+{
+	size_t i;
+
+	fputs("# Soft dependencies extracted from modules themselves.\n", out);
+
+	for (i = 0; i < depmod->modules.count; i++) {
+		const struct mod *mod = depmod->modules.array[i];
+		struct kmod_list *l;
+
+		kmod_list_foreach(l, mod->info_list) {
+			const char *key = kmod_module_info_get_key(l);
+			const char *value = kmod_module_info_get_value(l);
+
+			if (!streq(key, "softdep"))
+				continue;
+
+			fprintf(out, "softdep %s %s\n", mod->modname, value);
+		}
+	}
+
+	return 0;
+}
+
+static int output_symbols(struct depmod *depmod, FILE *out)
+{
+	struct hash_iter iter;
+	const void *v;
+
+	fputs("# Aliases for symbols, used by symbol_request().\n", out);
+
+	hash_iter_init(depmod->symbols, &iter);
+
+	while (hash_iter_next(&iter, NULL, &v)) {
+		const struct symbol *sym = v;
+		if (sym->owner == NULL)
+			continue;
+
+		fprintf(out, "alias symbol:%s %s\n",
+					sym->name, sym->owner->modname);
+	}
+
+	return 0;
+}
+
+static int output_symbols_bin(struct depmod *depmod, FILE *out)
+{
+	struct index_node *idx;
+	char alias[1024];
+	_cleanup_(scratchbuf_release) struct scratchbuf salias =
+		SCRATCHBUF_INITIALIZER(alias);
+	size_t baselen = sizeof("symbol:") - 1;
+	struct hash_iter iter;
+	const void *v;
+	int ret = 0;
+
+	if (out == stdout)
+		return 0;
+
+	idx = index_create();
+	if (idx == NULL)
+		return -ENOMEM;
+
+	memcpy(alias, "symbol:", baselen);
+
+	hash_iter_init(depmod->symbols, &iter);
+
+	while (hash_iter_next(&iter, NULL, &v)) {
+		int duplicate;
+		const struct symbol *sym = v;
+		size_t len;
+
+		if (sym->owner == NULL)
+			continue;
+
+		len = strlen(sym->name);
+
+		if (scratchbuf_alloc(&salias, baselen + len + 1) < 0) {
+			ret = -ENOMEM;
+			goto err_scratchbuf;
+		}
+		memcpy(scratchbuf_str(&salias) + baselen, sym->name, len + 1);
+		duplicate = index_insert(idx, alias, sym->owner->modname,
+							sym->owner->idx);
+
+		if (duplicate && depmod->cfg->warn_dups)
+			WRN("duplicate module syms:\n%s %s\n",
+						alias, sym->owner->modname);
+	}
+
+	index_write(idx, out);
+
+err_scratchbuf:
+	index_destroy(idx);
+
+	if (ret < 0)
+		ERR("output symbols: %s\n", strerror(-ret));
+
+	return ret;
+}
+
+static int output_builtin_bin(struct depmod *depmod, FILE *out)
+{
+	FILE *in;
+	struct index_node *idx;
+	char line[PATH_MAX], modname[PATH_MAX];
+
+	if (out == stdout)
+		return 0;
+
+	in = dfdopen(depmod->cfg->dirname, "modules.builtin", O_RDONLY, "r");
+	if (in == NULL)
+		return 0;
+
+	idx = index_create();
+	if (idx == NULL) {
+		fclose(in);
+		return -ENOMEM;
+	}
+
+	while (fgets(line, sizeof(line), in) != NULL) {
+		if (!isalpha(line[0])) {
+			ERR("Invalid modules.builtin line: %s\n", line);
+			continue;
+		}
+
+		path_to_modname(line, modname, NULL);
+		index_insert(idx, modname, "", 0);
+	}
+
+	index_write(idx, out);
+	index_destroy(idx);
+	fclose(in);
+
+	return 0;
+}
+
+static int flush_stream(FILE *in, int endchar)
+{
+	size_t i = 0;
+	int c;
+
+	for (c = fgetc(in);
+	     c != EOF && c != endchar && c != '\0';
+	     c = fgetc(in))
+		;
+
+	return c == endchar ? i : 0;
+}
+
+static int flush_stream_to(FILE *in, int endchar, char *dst, size_t dst_sz)
+{
+	size_t i = 0;
+	int c;
+
+	for (c = fgetc(in);
+	     c != EOF && c != endchar && c != '\0' && i < dst_sz;
+	     c = fgetc(in))
+		dst[i++] = c;
+
+	if (i == dst_sz) {
+		WRN("Could not flush stream: %d. Partial content: %.*s\n",
+		    ENOSPC, (int) dst_sz, dst);
+		i--;
+	}
+
+	return c == endchar ? i : 0;
+}
+
+static int output_builtin_alias_bin(struct depmod *depmod, FILE *out)
+{
+	FILE *in;
+	struct index_node *idx;
+	int ret;
+
+	if (out == stdout)
+		return 0;
+
+	in = dfdopen(depmod->cfg->dirname, "modules.builtin.modinfo", O_RDONLY, "r");
+	if (in == NULL)
+		return 0;
+
+	idx = index_create();
+	if (idx == NULL) {
+		fclose(in);
+		return -ENOMEM;
+	}
+
+	/* format: modname.key=value\0 */
+	while (!feof(in) && !ferror(in)) {
+		char alias[PATH_MAX];
+		char modname[PATH_MAX];
+		char value[PATH_MAX];
+		size_t len;
+
+		len = flush_stream_to(in, '.', modname, sizeof(modname));
+		modname[len] = '\0';
+		if (!len)
+			continue;
+
+		len = flush_stream_to(in, '=', value, sizeof(value));
+		value[len] = '\0';
+		if (!streq(value, "alias")) {
+			flush_stream(in, '\0');
+			continue;
+		}
+
+		len = flush_stream_to(in, '\0', value, sizeof(value));
+		value[len] = '\0';
+		if (!len)
+			continue;
+
+		alias[0] = '\0';
+		if (alias_normalize(value, alias, NULL) < 0) {
+			WRN("Unmatched bracket in %s\n", value);
+			continue;
+		}
+
+		index_insert(idx, alias, modname, 0);
+	}
+
+	if (ferror(in)) {
+		ret = -EINVAL;
+	} else {
+		index_write(idx, out);
+		ret = 0;
+	}
+
+	index_destroy(idx);
+	fclose(in);
+
+	return ret;
+}
+
+static int output_devname(struct depmod *depmod, FILE *out)
+{
+	size_t i;
+	bool empty = true;
+
+	for (i = 0; i < depmod->modules.count; i++) {
+		const struct mod *mod = depmod->modules.array[i];
+		struct kmod_list *l;
+		const char *devname = NULL;
+		char type = '\0';
+		unsigned int major = 0, minor = 0;
+
+		kmod_list_foreach(l, mod->info_list) {
+			const char *key = kmod_module_info_get_key(l);
+			const char *value = kmod_module_info_get_value(l);
+			unsigned int maj, min;
+
+			if (!streq(key, "alias"))
+				continue;
+
+			if (strstartswith(value, "devname:"))
+				devname = value + sizeof("devname:") - 1;
+			else if (sscanf(value, "char-major-%u-%u",
+						&maj, &min) == 2) {
+				type = 'c';
+				major = maj;
+				minor = min;
+			} else if (sscanf(value, "block-major-%u-%u",
+						&maj, &min) == 2) {
+				type = 'b';
+				major = maj;
+				minor = min;
+			}
+
+			if (type != '\0' && devname != NULL)
+				break;
+		}
+
+		if (devname != NULL) {
+			if (type != '\0') {
+				if (empty) {
+					fputs("# Device nodes to trigger on-demand module loading.\n",
+					      out);
+					empty = false;
+				}
+				fprintf(out, "%s %s %c%u:%u\n", mod->modname,
+					devname, type, major, minor);
+                        } else
+				ERR("Module '%s' has devname (%s) but "
+				    "lacks major and minor information. "
+				    "Ignoring.\n", mod->modname, devname);
+		}
+	}
+
+	return 0;
+}
+
+static int depmod_output(struct depmod *depmod, FILE *out)
+{
+	static const struct depfile {
+		const char *name;
+		int (*cb)(struct depmod *depmod, FILE *out);
+	} *itr, depfiles[] = {
+		{ "modules.dep", output_deps },
+		{ "modules.dep.bin", output_deps_bin },
+		{ "modules.alias", output_aliases },
+		{ "modules.alias.bin", output_aliases_bin },
+		{ "modules.softdep", output_softdeps },
+		{ "modules.symbols", output_symbols },
+		{ "modules.symbols.bin", output_symbols_bin },
+		{ "modules.builtin.bin", output_builtin_bin },
+		{ "modules.builtin.alias.bin", output_builtin_alias_bin },
+		{ "modules.devname", output_devname },
+		{ }
+	};
+	const char *dname = depmod->cfg->outdirname;
+	int dfd, err = 0;
+	struct timeval tv;
+
+	gettimeofday(&tv, NULL);
+
+	if (out != NULL)
+		dfd = -1;
+	else {
+		err = mkdir_p(dname, strlen(dname), 0755);
+		if (err < 0) {
+			CRIT("could not create directory %s: %m\n", dname);
+			return err;
+		}
+		dfd = open(dname, O_RDONLY);
+		if (dfd < 0) {
+			err = -errno;
+			CRIT("could not open directory %s: %m\n", dname);
+			return err;
+		}
+	}
+
+	for (itr = depfiles; itr->name != NULL; itr++) {
+		FILE *fp = out;
+		char tmp[NAME_MAX] = "";
+		int r, ferr;
+
+		if (fp == NULL) {
+			int flags = O_CREAT | O_EXCL | O_WRONLY;
+			int mode = 0644;
+			int fd;
+
+			snprintf(tmp, sizeof(tmp), "%s.%i.%li.%li", itr->name, getpid(),
+					tv.tv_usec, tv.tv_sec);
+			fd = openat(dfd, tmp, flags, mode);
+			if (fd < 0) {
+				ERR("openat(%s, %s, %o, %o): %m\n",
+				    dname, tmp, flags, mode);
+				continue;
+			}
+			fp = fdopen(fd, "wb");
+			if (fp == NULL) {
+				ERR("fdopen(%d=%s/%s): %m\n", fd, dname, tmp);
+				close(fd);
+				continue;
+			}
+		}
+
+		r = itr->cb(depmod, fp);
+		if (fp == out)
+			continue;
+
+		ferr = ferror(fp) | fclose(fp);
+
+		if (r < 0) {
+			if (unlinkat(dfd, tmp, 0) != 0)
+				ERR("unlinkat(%s, %s): %m\n", dname, tmp);
+
+			ERR("Could not write index '%s': %s\n", itr->name,
+								strerror(-r));
+			err = -errno;
+			break;
+		}
+
+		if (renameat(dfd, tmp, dfd, itr->name) != 0) {
+			err = -errno;
+			CRIT("renameat(%s, %s, %s, %s): %m\n",
+					dname, tmp, dname, itr->name);
+			break;
+		}
+
+		if (ferr) {
+			err = -ENOSPC;
+			ERR("Could not create index '%s'. Output is truncated: %s\n",
+						itr->name, strerror(-err));
+			break;
+		}
+	}
+
+	if (dfd >= 0)
+		close(dfd);
+
+	return err;
+}
+
+static void depmod_add_fake_syms(struct depmod *depmod)
+{
+	/* __this_module is magic inserted by kernel loader. */
+	depmod_symbol_add(depmod, "__this_module", true, 0, NULL);
+	/* On S390, this is faked up too */
+	depmod_symbol_add(depmod, "_GLOBAL_OFFSET_TABLE_", true, 0, NULL);
+	/* On PowerPC64 ABIv2, .TOC. is more or less _GLOBAL_OFFSET_TABLE_ */
+	if (!depmod_symbol_find(depmod, "TOC."))
+		depmod_symbol_add(depmod, "TOC.", true, 0, NULL);
+}
+
+static int depmod_load_symvers(struct depmod *depmod, const char *filename)
+{
+	char line[10240];
+	FILE *fp;
+	unsigned int linenum = 0;
+
+	fp = fopen(filename, "r");
+	if (fp == NULL) {
+		int err = -errno;
+		DBG("load symvers: %s: %m\n", filename);
+		return err;
+	}
+	DBG("load symvers: %s\n", filename);
+
+	/* eg. "0xb352177e\tfind_first_bit\tvmlinux\tEXPORT_SYMBOL" */
+	while (fgets(line, sizeof(line), fp) != NULL) {
+		const char *ver, *sym, *where;
+		char *verend;
+		uint64_t crc;
+
+		linenum++;
+
+		ver = strtok(line, " \t");
+		sym = strtok(NULL, " \t");
+		where = strtok(NULL, " \t");
+		if (!ver || !sym || !where)
+			continue;
+
+		if (!streq(where, "vmlinux"))
+			continue;
+
+		crc = strtoull(ver, &verend, 16);
+		if (verend[0] != '\0') {
+			ERR("%s:%u Invalid symbol version %s: %m\n",
+			    filename, linenum, ver);
+			continue;
+		}
+
+		depmod_symbol_add(depmod, sym, false, crc, NULL);
+	}
+	depmod_add_fake_syms(depmod);
+
+	DBG("loaded symvers: %s\n", filename);
+
+	fclose(fp);
+	return 0;
+}
+
+static int depmod_load_system_map(struct depmod *depmod, const char *filename)
+{
+	const char ksymstr[] = "__ksymtab_";
+	const size_t ksymstr_len = sizeof(ksymstr) - 1;
+	char line[10240];
+	FILE *fp;
+	unsigned int linenum = 0;
+
+	fp = fopen(filename, "r");
+	if (fp == NULL) {
+		int err = -errno;
+		DBG("load System.map: %s: %m\n", filename);
+		return err;
+	}
+	DBG("load System.map: %s\n", filename);
+
+	/* eg. c0294200 R __ksymtab_devfs_alloc_devnum */
+	while (fgets(line, sizeof(line), fp) != NULL) {
+		char *p, *end;
+
+		linenum++;
+
+		p = strchr(line, ' ');
+		if (p == NULL)
+			goto invalid_syntax;
+		p++;
+		p = strchr(p, ' ');
+		if (p == NULL)
+			goto invalid_syntax;
+		p++;
+
+		/* skip prefix */
+		if (p[0] == depmod->cfg->sym_prefix)
+			p++;
+
+		/* Covers gpl-only and normal symbols. */
+		if (strncmp(p, ksymstr, ksymstr_len) != 0)
+			continue;
+
+		end = strchr(p, '\n');
+		if (end != NULL)
+			*end = '\0';
+
+		depmod_symbol_add(depmod, p + ksymstr_len, true, 0, NULL);
+		continue;
+
+	invalid_syntax:
+		ERR("%s:%u: invalid line: %s\n", filename, linenum, line);
+	}
+	depmod_add_fake_syms(depmod);
+
+	DBG("loaded System.map: %s\n", filename);
+
+	fclose(fp);
+	return 0;
+}
+
+
+static int depfile_up_to_date_dir(DIR *d, time_t mtime, size_t baselen, char *path)
+{
+	struct dirent *de;
+	int err = 1, dfd = dirfd(d);
+
+	while ((de = readdir(d)) != NULL) {
+		const char *name = de->d_name;
+		size_t namelen;
+		struct stat st;
+
+		if (name[0] == '.' && (name[1] == '\0' ||
+				       (name[1] == '.' && name[2] == '\0')))
+			continue;
+		if (streq(name, "build") || streq(name, "source"))
+			continue;
+		namelen = strlen(name);
+		if (baselen + namelen + 2 >= PATH_MAX) {
+			path[baselen] = '\0';
+			ERR("path is too long %s%s\n", path, name);
+			continue;
+		}
+
+		if (fstatat(dfd, name, &st, 0) < 0) {
+			ERR("fstatat(%d, %s): %m\n", dfd, name);
+			continue;
+		}
+
+		if (S_ISDIR(st.st_mode)) {
+			int fd;
+			DIR *subdir;
+			memcpy(path + baselen, name, namelen + 1);
+			if (baselen + namelen + 2 + NAME_MAX >= PATH_MAX) {
+				ERR("directory path is too long %s\n", path);
+				continue;
+			}
+			fd = openat(dfd, name, O_RDONLY);
+			if (fd < 0) {
+				ERR("openat(%d, %s, O_RDONLY): %m\n",
+				    dfd, name);
+				continue;
+			}
+			subdir = fdopendir(fd);
+			if (subdir == NULL) {
+				ERR("fdopendir(%d): %m\n", fd);
+				close(fd);
+				continue;
+			}
+			path[baselen + namelen] = '/';
+			path[baselen + namelen + 1] = '\0';
+			err = depfile_up_to_date_dir(subdir, mtime,
+						     baselen + namelen + 1,
+						     path);
+			closedir(subdir);
+		} else if (S_ISREG(st.st_mode)) {
+			if (!path_ends_with_kmod_ext(name, namelen))
+				continue;
+
+			memcpy(path + baselen, name, namelen + 1);
+			err = st.st_mtime <= mtime;
+			if (err == 0) {
+				DBG("%s %"PRIu64" is newer than %"PRIu64"\n",
+				    path, (uint64_t)st.st_mtime,
+				    (uint64_t)mtime);
+			}
+		} else {
+			ERR("unsupported file type %s: %o\n",
+			    path, st.st_mode & S_IFMT);
+			continue;
+		}
+
+		if (err == 0)
+			break; /* outdated! */
+		else if (err < 0) {
+			path[baselen + namelen] = '\0';
+			ERR("failed %s: %s\n", path, strerror(-err));
+			err = 1; /* ignore errors */
+		}
+	}
+
+	return err;
+}
+
+/* uptodate: 1, outdated: 0, errors < 0 */
+static int depfile_up_to_date(const char *dirname)
+{
+	char path[PATH_MAX];
+	DIR *d = opendir(dirname);
+	struct stat st;
+	size_t baselen;
+	int err;
+	if (d == NULL) {
+		err = -errno;
+		ERR("could not open directory %s: %m\n", dirname);
+		return err;
+	}
+
+	if (fstatat(dirfd(d), "modules.dep", &st, 0) != 0) {
+		err = -errno;
+		ERR("could not fstatat(%s, modules.dep): %m\n", dirname);
+		closedir(d);
+		return err;
+	}
+
+	baselen = strlen(dirname);
+	memcpy(path, dirname, baselen);
+	path[baselen] = '/';
+	baselen++;
+	path[baselen] = '\0';
+
+	err = depfile_up_to_date_dir(d, st.st_mtime, baselen, path);
+	closedir(d);
+	return err;
+}
+
+static int is_version_number(const char *version)
+{
+	unsigned int d1, d2;
+	return (sscanf(version, "%u.%u", &d1, &d2) == 2);
+}
+
+static int do_depmod(int argc, char *argv[])
+{
+	FILE *out = NULL;
+	int err = 0, all = 0, maybe_all = 0, n_config_paths = 0;
+	_cleanup_free_ char *root = NULL;
+	_cleanup_free_ char *out_root = NULL;
+	_cleanup_free_ const char **config_paths = NULL;
+	const char *system_map = NULL;
+	const char *module_symvers = NULL;
+	const char *null_kmod_config = NULL;
+	struct utsname un;
+	struct kmod_ctx *ctx = NULL;
+	struct cfg cfg;
+	struct depmod depmod;
+
+	memset(&cfg, 0, sizeof(cfg));
+	memset(&depmod, 0, sizeof(depmod));
+
+	for (;;) {
+		int c, idx = 0;
+		c = getopt_long(argc, argv, cmdopts_s, cmdopts, &idx);
+		if (c == -1)
+			break;
+		switch (c) {
+		case 'a':
+			all = 1;
+			break;
+		case 'A':
+			maybe_all = 1;
+			break;
+		case 'b':
+			if (root)
+				free(root);
+			root = path_make_absolute_cwd(optarg);
+			break;
+		case 'o':
+			if (out_root)
+				free(out_root);
+			out_root = path_make_absolute_cwd(optarg);
+			break;
+		case 'C': {
+			size_t bytes = sizeof(char *) * (n_config_paths + 2);
+			void *tmp = realloc(config_paths, bytes);
+			if (!tmp) {
+				fputs("Error: out-of-memory\n", stderr);
+				goto cmdline_failed;
+			}
+			config_paths = tmp;
+			config_paths[n_config_paths] = optarg;
+			n_config_paths++;
+			config_paths[n_config_paths] = NULL;
+			break;
+		}
+		case 'E':
+			module_symvers = optarg;
+			cfg.check_symvers = 1;
+			break;
+		case 'F':
+			system_map = optarg;
+			break;
+		case 'e':
+			cfg.print_unknown = 1;
+			break;
+		case 'v':
+			verbose++;
+			break;
+		case 'n':
+			out = stdout;
+			break;
+		case 'P':
+			if (optarg[1] != '\0') {
+				CRIT("-P only takes a single char\n");
+				goto cmdline_failed;
+			}
+			cfg.sym_prefix = optarg[0];
+			break;
+		case 'w':
+			cfg.warn_dups = 1;
+			break;
+		case 'u':
+		case 'q':
+		case 'r':
+		case 'm':
+			if (idx > 0)
+				WRN("Ignored deprecated option --%s\n",
+				    cmdopts[idx].name);
+			else
+				WRN("Ignored deprecated option -%c\n", c);
+
+			break;
+		case 'h':
+			help();
+			return EXIT_SUCCESS;
+		case 'V':
+			puts(PACKAGE " version " VERSION);
+			puts(KMOD_FEATURES);
+			return EXIT_SUCCESS;
+		case '?':
+			goto cmdline_failed;
+		default:
+			ERR("unexpected getopt_long() value '%c'.\n", c);
+			goto cmdline_failed;
+		}
+	}
+
+	if (optind < argc) {
+		if (!is_version_number(argv[optind])) {
+			ERR("Bad version passed %s\n", argv[optind]);
+			goto cmdline_failed;
+		}
+		cfg.kversion = argv[optind];
+		optind++;
+	} else {
+		if (uname(&un) < 0) {
+			CRIT("uname() failed: %s\n", strerror(errno));
+			goto cmdline_failed;
+		}
+		cfg.kversion = un.release;
+	}
+
+	cfg.dirnamelen = snprintf(cfg.dirname, PATH_MAX,
+				  "%s" MODULE_DIRECTORY "/%s",
+				  root ?: "", cfg.kversion);
+
+	cfg.outdirnamelen = snprintf(cfg.outdirname, PATH_MAX,
+				     "%s" MODULE_DIRECTORY "/%s",
+				     out_root ?: (root ?: ""), cfg.kversion);
+
+	if (optind == argc)
+		all = 1;
+
+	if (maybe_all) {
+		if (out == stdout)
+			goto done;
+		/* ignore up-to-date errors (< 0) */
+		if (depfile_up_to_date(cfg.dirname) == 1)
+			goto done;
+		all = 1;
+	}
+
+	ctx = kmod_new(cfg.dirname, &null_kmod_config);
+	if (ctx == NULL) {
+		CRIT("kmod_new(\"%s\", {NULL}) failed: %m\n", cfg.dirname);
+		goto cmdline_failed;
+	}
+
+	log_setup_kmod_log(ctx, verbose);
+
+	err = depmod_init(&depmod, &cfg, ctx);
+	if (err < 0) {
+		CRIT("depmod_init: %s\n", strerror(-err));
+		goto depmod_init_failed;
+	}
+	ctx = NULL; /* owned by depmod */
+
+	if (module_symvers != NULL) {
+		err = depmod_load_symvers(&depmod, module_symvers);
+		if (err < 0) {
+			CRIT("could not load %s: %s\n", module_symvers,
+			     strerror(-err));
+			goto cmdline_failed;
+		}
+	} else if (system_map != NULL) {
+		err = depmod_load_system_map(&depmod, system_map);
+		if (err < 0) {
+			CRIT("could not load %s: %s\n", system_map,
+			     strerror(-err));
+			goto cmdline_failed;
+		}
+	} else if (cfg.print_unknown) {
+		WRN("-e needs -E or -F\n");
+		cfg.print_unknown = 0;
+	}
+
+	if (all) {
+		err = cfg_load(&cfg, config_paths);
+		if (err < 0) {
+			CRIT("could not load configuration files\n");
+			goto cmdline_modules_failed;
+		}
+		err = depmod_modules_search(&depmod);
+		if (err < 0) {
+			CRIT("could not search modules: %s\n", strerror(-err));
+			goto cmdline_modules_failed;
+		}
+	} else {
+		int i;
+
+		for (i = optind; i < argc; i++) {
+			const char *path = argv[i];
+			struct kmod_module *mod;
+
+			if (path[0] != '/') {
+				CRIT("%s: not absolute path.\n", path);
+				goto cmdline_modules_failed;
+			}
+
+			err = kmod_module_new_from_path(depmod.ctx, path, &mod);
+			if (err < 0) {
+				CRIT("could not create module %s: %s\n",
+				     path, strerror(-err));
+				goto cmdline_modules_failed;
+			}
+
+			err = depmod_module_add(&depmod, mod);
+			if (err < 0) {
+				CRIT("could not add module %s: %s\n",
+				     path, strerror(-err));
+				kmod_module_unref(mod);
+				goto cmdline_modules_failed;
+			}
+		}
+	}
+
+	err = depmod_modules_build_array(&depmod);
+	if (err < 0) {
+		CRIT("could not build module array: %s\n",
+		     strerror(-err));
+		goto cmdline_modules_failed;
+	}
+
+	depmod_modules_sort(&depmod);
+	err = depmod_load(&depmod);
+	if (err < 0)
+		goto cmdline_modules_failed;
+
+	err = depmod_output(&depmod, out);
+
+done:
+	depmod_shutdown(&depmod);
+	cfg_free(&cfg);
+	return err >= 0 ? EXIT_SUCCESS : EXIT_FAILURE;
+
+cmdline_modules_failed:
+	depmod_shutdown(&depmod);
+depmod_init_failed:
+	if (ctx != NULL)
+		kmod_unref(ctx);
+cmdline_failed:
+	cfg_free(&cfg);
+	return EXIT_FAILURE;
+}
+
+const struct kmod_cmd kmod_cmd_compat_depmod = {
+	.name = "depmod",
+	.cmd = do_depmod,
+	.help = "compat depmod command",
+};