[PATCH 1/3] erofs-utils: dump: add -S options for collecting filesystem statistics

[PATCH 1/3] erofs-utils: dump: add -S options for collecting filesystem statistics

[Guo Xuenan]

From: Wang Qi <mpiglet@outlook.com>

Signed-off-by: Guo Xuenan <guoxuenan@huawei.com>
Signed-off-by: Wang Qi <mpiglet@outlook.com>
---
 dump/Makefile.am         |   2 +-
 dump/main.c              | 358 ++++++++++++++++++++++++++++++++++++++-
 include/erofs/internal.h |   1 +
 lib/namei.c              |   2 +-
 4 files changed, 360 insertions(+), 3 deletions(-)

diff --git a/dump/Makefile.am b/dump/Makefile.am
index f0246d7..4759901 100644
--- a/dump/Makefile.am
+++ b/dump/Makefile.am
@@ -6,4 +6,4 @@ bin_PROGRAMS     = dump.erofs
 AM_CPPFLAGS = ${libuuid_CFLAGS}
 dump_erofs_SOURCES = main.c
 dump_erofs_CFLAGS = -Wall -Werror -I$(top_srcdir)/include
-dump_erofs_LDADD = $(top_builddir)/lib/liberofs.la ${libselinux_LIBS} ${libuuid_LIBS}
+dump_erofs_LDADD = $(top_builddir)/lib/liberofs.la ${libselinux_LIBS} ${libuuid_LIBS} ${liblz4_LIBS}
diff --git a/dump/main.c b/dump/main.c
index 5b7ac5c..98ee68b 100644
--- a/dump/main.c
+++ b/dump/main.c
@@ -18,9 +18,51 @@
 struct erofsdump_cfg {
 	unsigned int totalshow;
 	bool show_superblock;
+	bool show_statistics;
 };
 static struct erofsdump_cfg dumpcfg;
 
+static const char chart_format[] = "%-16s	%-11d %8.2f%% |%-50s|\n";
+static const char header_format[] = "%-16s %11s %16s |%-50s|\n";
+static char *file_types[] = {
+	".txt", ".so", ".xml", ".apk",
+	".odex", ".vdex", ".oat", ".rc",
+	".otf", ".txt", "others",
+};
+#define OTHERFILETYPE	ARRAY_SIZE(file_types)
+/* (1 << FILE_MAX_SIZE_BITS)KB */
+#define	FILE_MAX_SIZE_BITS	16
+
+static const char * const file_category_types[] = {
+	[EROFS_FT_UNKNOWN] = "unknown type",
+	[EROFS_FT_REG_FILE] = "regular file",
+	[EROFS_FT_DIR] = "directory",
+	[EROFS_FT_CHRDEV] = "char dev",
+	[EROFS_FT_BLKDEV] = "block dev",
+	[EROFS_FT_FIFO] = "FIFO file",
+	[EROFS_FT_SOCK] = "SOCK file",
+	[EROFS_FT_SYMLINK] = "symlink file",
+};
+
+struct erofs_statistics {
+	unsigned long files;
+	unsigned long compressed_files;
+	unsigned long uncompressed_files;
+	unsigned long files_total_size;
+	unsigned long files_total_origin_size;
+	double compress_rate;
+
+	/* [statistics] # of files based on inode_info->flags */
+	unsigned long file_category_stat[EROFS_FT_MAX];
+	/* [statistics] # of files based on file name extensions */
+	unsigned int file_type_stat[OTHERFILETYPE];
+	/* [statistics] # of files based on the original size of files */
+	unsigned int file_original_size[FILE_MAX_SIZE_BITS + 1];
+	/* [statistics] # of files based on the compressed size of files */
+	unsigned int file_comp_size[FILE_MAX_SIZE_BITS + 1];
+};
+static struct erofs_statistics stats;
+
 static struct option long_options[] = {
 	{"help", no_argument, 0, 1},
 	{0, 0, 0, 0},
@@ -39,10 +81,13 @@ static struct erofsdump_feature feature_lists[] = {
 	{ false, EROFS_FEATURE_INCOMPAT_CHUNKED_FILE, "chunked_file" },
 };
 
+static int erofs_read_dir(erofs_nid_t nid, erofs_nid_t parent_nid);
+
 static void usage(void)
 {
 	fputs("usage: [options] IMAGE\n\n"
 	      "Dump erofs layout from IMAGE, and [options] are:\n"
+	      " -S      show statistic information of the image\n"
 	      " -V      print the version number of dump.erofs and exit.\n"
 	      " -s      show information about superblock\n"
 	      " --help  display this help and exit.\n",
@@ -58,13 +103,17 @@ static int erofsdump_parse_options_cfg(int argc, char **argv)
 {
 	int opt;
 
-	while ((opt = getopt_long(argc, argv, "Vs",
+	while ((opt = getopt_long(argc, argv, "SVs",
 				  long_options, NULL)) != -1) {
 		switch (opt) {
 		case 's':
 			dumpcfg.show_superblock = true;
 			++dumpcfg.totalshow;
 			break;
+		case 'S':
+			dumpcfg.show_statistics = true;
+			++dumpcfg.totalshow;
+			break;
 		case 'V':
 			erofsdump_print_version();
 			exit(0);
@@ -90,6 +139,310 @@ static int erofsdump_parse_options_cfg(int argc, char **argv)
 	return 0;
 }
 
+static int erofs_get_occupied_size(struct erofs_inode *inode,
+		erofs_off_t *size)
+{
+	*size = 0;
+	switch (inode->datalayout) {
+	case EROFS_INODE_FLAT_INLINE:
+	case EROFS_INODE_FLAT_PLAIN:
+		stats.uncompressed_files++;
+		*size = inode->i_size;
+		break;
+	case EROFS_INODE_FLAT_COMPRESSION_LEGACY:
+	case EROFS_INODE_FLAT_COMPRESSION:
+		stats.compressed_files++;
+		*size = inode->u.i_blocks * EROFS_BLKSIZ;
+		break;
+	default:
+		erofs_err("unknown datalayout");
+		return -1;
+	}
+	return 0;
+}
+
+static int erofs_getfile_extension(const char *filename)
+{
+	char *postfix = strrchr(filename, '.');
+	int type = 0;
+
+	if (postfix == NULL)
+		return OTHERFILETYPE - 1;
+	while (type < OTHERFILETYPE - 1) {
+		if (strcmp(postfix, file_types[type]) == 0)
+			break;
+		type++;
+	}
+	return type;
+}
+
+static void update_file_size_statatics(erofs_off_t occupied_size,
+		erofs_off_t original_size)
+{
+	int occupied_size_mark;
+	int original_size_mark;
+
+	original_size_mark = 0;
+	occupied_size_mark = 0;
+	occupied_size >>= 10;
+	original_size >>= 10;
+
+	while (occupied_size || original_size) {
+		if (occupied_size) {
+			occupied_size >>= 1;
+			occupied_size_mark++;
+		}
+		if (original_size) {
+			original_size >>= 1;
+			original_size_mark++;
+		}
+	}
+
+	if (original_size_mark >= FILE_MAX_SIZE_BITS)
+		stats.file_original_size[FILE_MAX_SIZE_BITS]++;
+	else
+		stats.file_original_size[original_size_mark]++;
+
+	if (occupied_size_mark >= FILE_MAX_SIZE_BITS)
+		stats.file_comp_size[FILE_MAX_SIZE_BITS]++;
+	else
+		stats.file_comp_size[occupied_size_mark]++;
+}
+
+static inline int erofs_checkdirent(struct erofs_dirent *de,
+		struct erofs_dirent *last_de,
+		u32 maxsize, const char *dname)
+{
+	int dname_len;
+	unsigned int nameoff = le16_to_cpu(de->nameoff);
+
+	if (nameoff < sizeof(struct erofs_dirent) ||
+			nameoff >= PAGE_SIZE) {
+		erofs_err("invalid de[0].nameoff %u @ nid %llu",
+				nameoff, de->nid | 0ULL);
+		return -EFSCORRUPTED;
+	}
+
+	dname_len = (de + 1 >= last_de) ? strnlen(dname, maxsize - nameoff) :
+				le16_to_cpu(de[1].nameoff) - nameoff;
+	/* a corrupted entry is found */
+	if (nameoff + dname_len > maxsize ||
+			dname_len > EROFS_NAME_LEN) {
+		erofs_err("bogus dirent @ nid %llu",
+				le64_to_cpu(de->nid) | 0ULL);
+		DBG_BUGON(1);
+		return -EFSCORRUPTED;
+	}
+	if (de->file_type >= EROFS_FT_MAX) {
+		erofs_err("invalid file type %llu", de->nid);
+		return -EFSCORRUPTED;
+	}
+	return dname_len;
+}
+
+static int erofs_read_dirent(struct erofs_dirent *de,
+		erofs_nid_t nid, erofs_nid_t parent_nid,
+		const char *dname)
+{
+	int err;
+	erofs_off_t occupied_size = 0;
+	struct erofs_inode inode = { .nid = de->nid };
+
+	stats.files++;
+	stats.file_category_stat[de->file_type]++;
+	err = erofs_read_inode_from_disk(&inode);
+	if (err) {
+		erofs_err("read file inode from disk failed!");
+		return err;
+	}
+
+	err = erofs_get_occupied_size(&inode, &occupied_size);
+	if (err) {
+		erofs_err("get file size failed\n");
+		return err;
+	}
+
+	if (de->file_type == EROFS_FT_REG_FILE) {
+		stats.files_total_origin_size += inode.i_size;
+		stats.file_type_stat[erofs_getfile_extension(dname)]++;
+		stats.files_total_size += occupied_size;
+		update_file_size_statatics(occupied_size, inode.i_size);
+	}
+
+	if ((de->file_type == EROFS_FT_DIR)
+				 && de->nid != nid && de->nid != parent_nid) {
+		err = erofs_read_dir(de->nid, nid);
+		if (err) {
+			erofs_err("parse dir nid %llu error occurred\n",
+					de->nid);
+			return err;
+		}
+	}
+	return 0;
+}
+
+
+static int erofs_read_dir(erofs_nid_t nid, erofs_nid_t parent_nid)
+{
+	struct erofs_inode vi = { .nid = nid};
+	int err;
+	char buf[EROFS_BLKSIZ];
+	erofs_off_t offset;
+
+	err = erofs_read_inode_from_disk(&vi);
+	if (err)
+		return err;
+
+	offset = 0;
+	while (offset < vi.i_size) {
+		erofs_off_t maxsize = min_t(erofs_off_t,
+			vi.i_size - offset, EROFS_BLKSIZ);
+		struct erofs_dirent *de = (void *)buf;
+		struct erofs_dirent *end;
+		unsigned int nameoff;
+
+		err = erofs_pread(&vi, buf, maxsize, offset);
+		if (err)
+			return err;
+
+		nameoff = le16_to_cpu(de->nameoff);
+		end = (void *)buf + nameoff;
+		while (de < end) {
+			const char *dname;
+			int ret;
+
+			/* skip "." and ".." dentry */
+			if (de->nid == nid || de->nid == parent_nid) {
+				de++;
+				continue;
+			}
+
+			dname = (char *)buf + nameoff;
+			ret = erofs_checkdirent(de, end, maxsize, dname);
+			if (ret < 0)
+				return ret;
+			ret = erofs_read_dirent(de, nid, parent_nid, dname);
+			if (ret < 0)
+				return ret;
+			++de;
+		}
+		offset += maxsize;
+	}
+	return 0;
+}
+
+static void erofsdump_print_chart_row(char *col1, unsigned int col2,
+		double col3, char *col4)
+{
+	char row[500] = {0};
+
+	sprintf(row, chart_format, col1, col2, col3, col4);
+	fprintf(stdout, row);
+}
+
+static void erofsdump_filesize_distribution(const char *title,
+		unsigned int *file_counts, unsigned int len)
+{
+	char col1[30];
+	unsigned int col2;
+	double col3;
+	char col4[400];
+	unsigned int lowerbound = 0;
+	unsigned int upperbound = 1;
+
+	fprintf(stdout, "\n%s file size distribution:\n", title);
+	fprintf(stdout, header_format, ">=(KB) .. <(KB) ", "count",
+			"ratio", "distribution");
+	for (int i = 0; i < len; i++) {
+		memset(col1, 0, sizeof(col1));
+		memset(col4, 0, sizeof(col4));
+		if (i == len - 1)
+			sprintf(col1, "%6d ..", lowerbound);
+		else if (i <= 6)
+			sprintf(col1, "%6d .. %-6d", lowerbound, upperbound);
+		else
+
+			sprintf(col1, "%6d .. %-6d", lowerbound, upperbound);
+		col2 = file_counts[i];
+		if (stats.file_category_stat[EROFS_FT_REG_FILE])
+			col3 = (double)(100 * col2) /
+				stats.file_category_stat[EROFS_FT_REG_FILE];
+		else
+			col3 = 0.0;
+		memset(col4, '#', col3 / 2);
+		erofsdump_print_chart_row(col1, col2, col3, col4);
+		lowerbound = upperbound;
+		upperbound <<= 1;
+	}
+}
+
+static void erofsdump_filetype_distribution(char **file_types, unsigned int len)
+{
+	char col1[30];
+	unsigned int col2;
+	double col3;
+	char col4[401];
+
+	fprintf(stdout, "\nFile type distribution:\n");
+	fprintf(stdout, header_format, "type", "count", "ratio",
+			"distribution");
+	for (int i = 0; i < len; i++) {
+		memset(col1, 0, sizeof(col1));
+		memset(col4, 0, sizeof(col4));
+		sprintf(col1, "%-17s", file_types[i]);
+		col2 = stats.file_type_stat[i];
+		if (stats.file_category_stat[EROFS_FT_REG_FILE])
+			col3 = (double)(100 * col2) /
+				stats.file_category_stat[EROFS_FT_REG_FILE];
+		else
+			col3 = 0.0;
+		memset(col4, '#', col3 / 2);
+		erofsdump_print_chart_row(col1, col2, col3, col4);
+	}
+}
+
+static void erofsdump_file_statistic(void)
+{
+	fprintf(stdout, "Filesystem total file count:		%lu\n",
+			stats.files);
+	for (int i = 0; i < EROFS_FT_MAX; i++)
+		fprintf(stdout, "Filesystem %s count:		%lu\n",
+			file_category_types[i], stats.file_category_stat[i]);
+
+	stats.compress_rate = (double)(100 * stats.files_total_size) /
+		(double)(stats.files_total_origin_size);
+	fprintf(stdout, "Filesystem compressed files:            %lu\n",
+			stats.compressed_files);
+	fprintf(stdout, "Filesystem uncompressed files:          %lu\n",
+			stats.uncompressed_files);
+	fprintf(stdout, "Filesystem total original file size:    %lu Bytes\n",
+			stats.files_total_origin_size);
+	fprintf(stdout, "Filesystem total file size:             %lu Bytes\n",
+			stats.files_total_size);
+	fprintf(stdout, "Filesystem compress rate:               %.2f%%\n",
+			stats.compress_rate);
+}
+
+static void erofsdump_print_statistic(void)
+{
+	int err;
+
+	err = erofs_read_dir(sbi.root_nid, sbi.root_nid);
+	if (err) {
+		erofs_err("read dir failed");
+		return;
+	}
+
+	erofsdump_file_statistic();
+	erofsdump_filesize_distribution("Original",
+			stats.file_original_size,
+			ARRAY_SIZE(stats.file_original_size));
+	erofsdump_filesize_distribution("On-disk",
+			stats.file_comp_size,
+			ARRAY_SIZE(stats.file_comp_size));
+	erofsdump_filetype_distribution(file_types, OTHERFILETYPE);
+}
+
 static void erofsdump_show_superblock(void)
 {
 	time_t time = sbi.build_time;
@@ -156,6 +509,9 @@ int main(int argc, char **argv)
 	if (dumpcfg.show_superblock)
 		erofsdump_show_superblock();
 
+	if (dumpcfg.show_statistics)
+		erofsdump_print_statistic();
+
 exit:
 	erofs_exit_configure();
 	return err;
diff --git a/include/erofs/internal.h b/include/erofs/internal.h
index 8b154ed..f19f839 100644
--- a/include/erofs/internal.h
+++ b/include/erofs/internal.h
@@ -263,6 +263,7 @@ struct erofs_map_blocks {
 int erofs_read_superblock(void);
 
 /* namei.c */
+int erofs_read_inode_from_disk(struct erofs_inode *vi);
 int erofs_ilookup(const char *path, struct erofs_inode *vi);
 
 /* data.c */
diff --git a/lib/namei.c b/lib/namei.c
index b4bdabf..56f199a 100644
--- a/lib/namei.c
+++ b/lib/namei.c
@@ -22,7 +22,7 @@ static dev_t erofs_new_decode_dev(u32 dev)
 	return makedev(major, minor);
 }
 
-static int erofs_read_inode_from_disk(struct erofs_inode *vi)
+int erofs_read_inode_from_disk(struct erofs_inode *vi)
 {
 	int ret, ifmt;
 	char buf[sizeof(struct erofs_inode_extended)];
-- 
2.31.1

[PATCH 2/3] erofs-utils: dump: add -i options to dump file information of specific inode number

[Guo Xuenan]

From: Wang Qi <mpiglet@outlook.com>

Signed-off-by: Guo Xuenan <guoxuenan@huawei.com>
Signed-off-by: Wang Qi <mpiglet@outlook.com>
---
 dump/main.c | 135 +++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 134 insertions(+), 1 deletion(-)

diff --git a/dump/main.c b/dump/main.c
index 98ee68b..8a31345 100644
--- a/dump/main.c
+++ b/dump/main.c
@@ -17,8 +17,10 @@
 
 struct erofsdump_cfg {
 	unsigned int totalshow;
+	bool show_inode;
 	bool show_superblock;
 	bool show_statistics;
+	u64 ino;
 };
 static struct erofsdump_cfg dumpcfg;
 
@@ -82,6 +84,9 @@ static struct erofsdump_feature feature_lists[] = {
 };
 
 static int erofs_read_dir(erofs_nid_t nid, erofs_nid_t parent_nid);
+static inline int erofs_checkdirent(struct erofs_dirent *de,
+		struct erofs_dirent *last_de,
+		u32 maxsize, const char *dname);
 
 static void usage(void)
 {
@@ -89,6 +94,7 @@ static void usage(void)
 	      "Dump erofs layout from IMAGE, and [options] are:\n"
 	      " -S      show statistic information of the image\n"
 	      " -V      print the version number of dump.erofs and exit.\n"
+	      " -i #    show the target inode info of nid #\n"
 	      " -s      show information about superblock\n"
 	      " --help  display this help and exit.\n",
 	      stderr);
@@ -102,10 +108,17 @@ static void erofsdump_print_version(void)
 static int erofsdump_parse_options_cfg(int argc, char **argv)
 {
 	int opt;
+	u64 i;
 
-	while ((opt = getopt_long(argc, argv, "SVs",
+	while ((opt = getopt_long(argc, argv, "SVi:s",
 				  long_options, NULL)) != -1) {
 		switch (opt) {
+		case 'i':
+			i = atoll(optarg);
+			dumpcfg.show_inode = true;
+			dumpcfg.ino = i;
+			++dumpcfg.totalshow;
+			break;
 		case 's':
 			dumpcfg.show_superblock = true;
 			++dumpcfg.totalshow;
@@ -331,6 +344,123 @@ static int erofs_read_dir(erofs_nid_t nid, erofs_nid_t parent_nid)
 	return 0;
 }
 
+static int erofs_get_pathname(erofs_nid_t nid, erofs_nid_t parent_nid,
+		erofs_nid_t target, char *path, unsigned int pos)
+{
+	int err;
+	struct erofs_inode inode = {.nid = nid};
+	erofs_off_t offset;
+	char buf[EROFS_BLKSIZ];
+
+	path[pos++] = '/';
+	if (target == sbi.root_nid)
+		return 0;
+
+	err = erofs_read_inode_from_disk(&inode);
+	if (err) {
+		erofs_err("read inode %lu failed", nid);
+		return err;
+	}
+
+	offset = 0;
+	while (offset < inode.i_size) {
+		erofs_off_t maxsize = min_t(erofs_off_t,
+					inode.i_size - offset, EROFS_BLKSIZ);
+		struct erofs_dirent *de = (void *)buf;
+		struct erofs_dirent *end;
+		unsigned int nameoff;
+
+		err = erofs_pread(&inode, buf, maxsize, offset);
+		if (err)
+			return err;
+
+		nameoff = le16_to_cpu(de->nameoff);
+		end = (void *)buf + nameoff;
+		while (de < end) {
+			const char *dname;
+			int len;
+
+			nameoff = le16_to_cpu(de->nameoff);
+			dname = (char *)buf + nameoff;
+			len = erofs_checkdirent(de, end, maxsize, dname);
+			if (len < 0)
+				return len;
+
+			if (de->nid == target) {
+				memcpy(path + pos, dname, len);
+				return 0;
+			}
+
+			if (de->file_type == EROFS_FT_DIR &&
+					de->nid != parent_nid &&
+					de->nid != nid) {
+				memcpy(path + pos, dname, len);
+				err = erofs_get_pathname(de->nid, nid,
+						target, path, pos + len);
+				if (!err)
+					return 0;
+				memset(path + pos, 0, len);
+			}
+			++de;
+		}
+		offset += maxsize;
+	}
+	return -1;
+}
+
+static void dumpfs_print_inode(void)
+{
+	int err;
+	erofs_off_t size;
+	u16 access_mode;
+	time_t t;
+	erofs_nid_t nid = dumpcfg.ino;
+	struct erofs_inode inode = {.nid = nid};
+	char path[PATH_MAX + 1] = {0};
+	char access_mode_str[] = "rwxrwxrwx";
+	char timebuf[128] = {0};
+
+	err = erofs_read_inode_from_disk(&inode);
+	if (err) {
+		erofs_err("read inode %lu from disk failed", nid);
+		return;
+	}
+
+	err = erofs_get_occupied_size(&inode, &size);
+	if (err) {
+		erofs_err("get file size failed\n");
+		return;
+	}
+
+	err = erofs_get_pathname(sbi.root_nid, sbi.root_nid, nid, path, 0);
+	if (err < 0) {
+		fprintf(stderr, "File: Path not found.\n");
+		return;
+	}
+
+	fprintf(stdout, "File : %s\n", path);
+	fprintf(stdout, "Inode: %lu  ", inode.nid);
+	fprintf(stdout, "Links: %u  ", inode.i_nlink);
+	fprintf(stdout, "Layout: %d\n", inode.datalayout);
+	fprintf(stdout, "File size: %lu   ", inode.i_size);
+	fprintf(stdout,	"On-disk size: %lu   ", size);
+	fprintf(stdout, "Compress rate: %.2f%%\n",
+			(double)(100 * size) / (double)(inode.i_size));
+	fprintf(stdout, "Inode size: %d   ", inode.inode_isize);
+	fprintf(stdout, "Extent size: %u   ", inode.extent_isize);
+	fprintf(stdout,	"Xattr size: %u\n", inode.xattr_isize);
+	access_mode = inode.i_mode & 0777;
+	t = inode.i_ctime;
+	strftime(timebuf, sizeof(timebuf),
+			"%Y-%m-%d %H:%M:%S", localtime(&t));
+	for (int i = 8; i >= 0; i--)
+		if (((access_mode >> i) & 1) == 0)
+			access_mode_str[8 - i] = '-';
+	fprintf(stdout, "Uid: %u   Gid: %u  ", inode.i_uid, inode.i_gid);
+	fprintf(stdout, "Access: %04o/%s\n", access_mode, access_mode_str);
+	fprintf(stdout, "Birth : %s\n", timebuf);
+}
+
 static void erofsdump_print_chart_row(char *col1, unsigned int col2,
 		double col3, char *col4)
 {
@@ -512,6 +642,9 @@ int main(int argc, char **argv)
 	if (dumpcfg.show_statistics)
 		erofsdump_print_statistic();
 
+	if (dumpcfg.show_inode)
+		dumpfs_print_inode();
+
 exit:
 	erofs_exit_configure();
 	return err;
-- 
2.31.1

[PATCH 3/3] erofs-utils: dump: add -e option to dump the inode extent info

[Guo Xuenan]

From: Wang Qi <mpiglet@outlook.com>

Signed-off-by: Guo Xuenan <guoxuenan@huawei.com>
Signed-off-by: Wang Qi <mpiglet@outlook.com>
---
 dump/main.c | 86 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 85 insertions(+), 1 deletion(-)

diff --git a/dump/main.c b/dump/main.c
index 8a31345..e4018a7 100644
--- a/dump/main.c
+++ b/dump/main.c
@@ -18,9 +18,11 @@
 struct erofsdump_cfg {
 	unsigned int totalshow;
 	bool show_inode;
+	bool show_inode_phy;
 	bool show_superblock;
 	bool show_statistics;
 	u64 ino;
+	u64 ino_phy;
 };
 static struct erofsdump_cfg dumpcfg;
 
@@ -94,6 +96,7 @@ static void usage(void)
 	      "Dump erofs layout from IMAGE, and [options] are:\n"
 	      " -S      show statistic information of the image\n"
 	      " -V      print the version number of dump.erofs and exit.\n"
+	      " -e #    show the inode extent info of nid #\n"
 	      " -i #    show the target inode info of nid #\n"
 	      " -s      show information about superblock\n"
 	      " --help  display this help and exit.\n",
@@ -110,9 +113,14 @@ static int erofsdump_parse_options_cfg(int argc, char **argv)
 	int opt;
 	u64 i;
 
-	while ((opt = getopt_long(argc, argv, "SVi:s",
+	while ((opt = getopt_long(argc, argv, "SVi:e:s",
 				  long_options, NULL)) != -1) {
 		switch (opt) {
+		case 'e':
+			dumpcfg.show_inode_phy = true;
+			dumpcfg.ino_phy = atoll(optarg);
+			++dumpcfg.totalshow;
+			break;
 		case 'i':
 			i = atoll(optarg);
 			dumpcfg.show_inode = true;
@@ -408,6 +416,79 @@ static int erofs_get_pathname(erofs_nid_t nid, erofs_nid_t parent_nid,
 	return -1;
 }
 
+static void erofsdump_show_inode_phy(void)
+{
+	int err;
+	erofs_nid_t nid = dumpcfg.ino_phy;
+	struct erofs_inode inode = {.nid = nid};
+	char path[PATH_MAX + 1] = {0};
+
+	err = erofs_read_inode_from_disk(&inode);
+	if (err < 0) {
+		erofs_err("read inode %lu from disk failed", nid);
+		return;
+	}
+
+	const erofs_off_t ibase = iloc(inode.nid);
+	const erofs_off_t pos = Z_EROFS_VLE_LEGACY_INDEX_ALIGN(
+			ibase + inode.inode_isize + inode.xattr_isize);
+	erofs_blk_t blocks = inode.u.i_blocks;
+	erofs_blk_t start = 0;
+	erofs_blk_t end = 0;
+	unsigned int extent_count;
+	struct erofs_map_blocks map = {
+		.index = UINT_MAX,
+		.m_la = 0,
+	};
+
+	fprintf(stdout, "Inode: %lu\n", nid);
+	err = erofs_get_pathname(sbi.root_nid, sbi.root_nid, nid, path, 0);
+	if (err) {
+		erofs_err("Path not found\n");
+		return;
+	}
+	fprintf(stdout, "File path: %s\n", path);
+	fprintf(stdout, "File size: %lu\n", inode.i_size);
+
+	switch (inode.datalayout) {
+	case EROFS_INODE_FLAT_INLINE:
+	case EROFS_INODE_FLAT_PLAIN:
+		if (inode.u.i_blkaddr == NULL_ADDR)
+			start = end = erofs_blknr(pos);
+		else {
+			start = inode.u.i_blkaddr;
+			end = start + BLK_ROUND_UP(inode.i_size) - 1;
+		}
+		fprintf(stdout, "Plain blknr: %u - %u\n", start, end);
+		break;
+
+	case EROFS_INODE_FLAT_COMPRESSION_LEGACY:
+	case EROFS_INODE_FLAT_COMPRESSION:
+		err = z_erofs_map_blocks_iter(&inode, &map, 0);
+		if (err)
+			erofs_err("get file blocks range failed");
+
+		start = erofs_blknr(map.m_pa);
+		end = start - 1 + blocks;
+		fprintf(stdout,
+				"Compressed blknr: %u - %u\n", start, end);
+		extent_count = 0;
+		map.m_la = 0;
+		fprintf(stdout, "Ext:  logical_offset/length:  physical_offset/length\n");
+		while (map.m_la < inode.i_size) {
+			err = z_erofs_map_blocks_iter(&inode, &map,
+					EROFS_GET_BLOCKS_FIEMAP);
+			fprintf(stdout, "%u: ", extent_count++);
+			fprintf(stdout, "%6lu..%-6lu / %-lu: ",
+					map.m_la, map.m_la + map.m_llen, map.m_llen);
+			fprintf(stdout, "%lu..%lu / %-lu\n",
+					map.m_pa, map.m_pa + map.m_plen, map.m_plen);
+			map.m_la += map.m_llen;
+		}
+		break;
+	}
+}
+
 static void dumpfs_print_inode(void)
 {
 	int err;
@@ -645,6 +726,9 @@ int main(int argc, char **argv)
 	if (dumpcfg.show_inode)
 		dumpfs_print_inode();
 
+	if (dumpcfg.show_inode_phy)
+		erofsdump_show_inode_phy();
+
 exit:
 	erofs_exit_configure();
 	return err;
-- 
2.31.1

Re: [PATCH 1/3] erofs-utils: dump: add -S options for collecting filesystem statistics

[Gao Xiang]

Hi Xuenan,

On Wed, Oct 27, 2021 at 08:03:49PM +0800, Guo Xuenan wrote:
> From: Wang Qi <mpiglet@outlook.com>
> 

This patch roughly looks good to me.

How about adding some info.. even it's obvious.
I think it's a good habit anyway.

Also, we need to add some manpage for dump.erofs as well finally.

> Signed-off-by: Guo Xuenan <guoxuenan@huawei.com>
> Signed-off-by: Wang Qi <mpiglet@outlook.com>
> ---
>  dump/Makefile.am         |   2 +-
>  dump/main.c              | 358 ++++++++++++++++++++++++++++++++++++++-
>  include/erofs/internal.h |   1 +
>  lib/namei.c              |   2 +-
>  4 files changed, 360 insertions(+), 3 deletions(-)
> 
> diff --git a/dump/Makefile.am b/dump/Makefile.am
> index f0246d7..4759901 100644
> --- a/dump/Makefile.am
> +++ b/dump/Makefile.am
> @@ -6,4 +6,4 @@ bin_PROGRAMS     = dump.erofs
>  AM_CPPFLAGS = ${libuuid_CFLAGS}
>  dump_erofs_SOURCES = main.c
>  dump_erofs_CFLAGS = -Wall -Werror -I$(top_srcdir)/include
> -dump_erofs_LDADD = $(top_builddir)/lib/liberofs.la ${libselinux_LIBS} ${libuuid_LIBS}
> +dump_erofs_LDADD = $(top_builddir)/lib/liberofs.la ${libselinux_LIBS} ${libuuid_LIBS} ${liblz4_LIBS}
> diff --git a/dump/main.c b/dump/main.c
> index 5b7ac5c..98ee68b 100644
> --- a/dump/main.c
> +++ b/dump/main.c
> @@ -18,9 +18,51 @@
>  struct erofsdump_cfg {
>  	unsigned int totalshow;
>  	bool show_superblock;
> +	bool show_statistics;
>  };
>  static struct erofsdump_cfg dumpcfg;
>  
> +static const char chart_format[] = "%-16s	%-11d %8.2f%% |%-50s|\n";
> +static const char header_format[] = "%-16s %11s %16s |%-50s|\n";
> +static char *file_types[] = {
> +	".txt", ".so", ".xml", ".apk",
> +	".odex", ".vdex", ".oat", ".rc",
> +	".otf", ".txt", "others",
> +};
> +#define OTHERFILETYPE	ARRAY_SIZE(file_types)
> +/* (1 << FILE_MAX_SIZE_BITS)KB */
> +#define	FILE_MAX_SIZE_BITS	16
> +
> +static const char * const file_category_types[] = {
> +	[EROFS_FT_UNKNOWN] = "unknown type",
> +	[EROFS_FT_REG_FILE] = "regular file",
> +	[EROFS_FT_DIR] = "directory",
> +	[EROFS_FT_CHRDEV] = "char dev",
> +	[EROFS_FT_BLKDEV] = "block dev",
> +	[EROFS_FT_FIFO] = "FIFO file",
> +	[EROFS_FT_SOCK] = "SOCK file",
> +	[EROFS_FT_SYMLINK] = "symlink file",
> +};
> +
> +struct erofs_statistics {
> +	unsigned long files;
> +	unsigned long compressed_files;
> +	unsigned long uncompressed_files;
> +	unsigned long files_total_size;
> +	unsigned long files_total_origin_size;
> +	double compress_rate;
> +
> +	/* [statistics] # of files based on inode_info->flags */
> +	unsigned long file_category_stat[EROFS_FT_MAX];
> +	/* [statistics] # of files based on file name extensions */
> +	unsigned int file_type_stat[OTHERFILETYPE];
> +	/* [statistics] # of files based on the original size of files */
> +	unsigned int file_original_size[FILE_MAX_SIZE_BITS + 1];
> +	/* [statistics] # of files based on the compressed size of files */
> +	unsigned int file_comp_size[FILE_MAX_SIZE_BITS + 1];
> +};
> +static struct erofs_statistics stats;
> +
>  static struct option long_options[] = {
>  	{"help", no_argument, 0, 1},
>  	{0, 0, 0, 0},
> @@ -39,10 +81,13 @@ static struct erofsdump_feature feature_lists[] = {
>  	{ false, EROFS_FEATURE_INCOMPAT_CHUNKED_FILE, "chunked_file" },
>  };
>  
> +static int erofs_read_dir(erofs_nid_t nid, erofs_nid_t parent_nid);
> +
>  static void usage(void)
>  {
>  	fputs("usage: [options] IMAGE\n\n"
>  	      "Dump erofs layout from IMAGE, and [options] are:\n"
> +	      " -S      show statistic information of the image\n"
>  	      " -V      print the version number of dump.erofs and exit.\n"
>  	      " -s      show information about superblock\n"
>  	      " --help  display this help and exit.\n",
> @@ -58,13 +103,17 @@ static int erofsdump_parse_options_cfg(int argc, char **argv)
>  {
>  	int opt;
>  
> -	while ((opt = getopt_long(argc, argv, "Vs",
> +	while ((opt = getopt_long(argc, argv, "SVs",
>  				  long_options, NULL)) != -1) {
>  		switch (opt) {
>  		case 's':
>  			dumpcfg.show_superblock = true;
>  			++dumpcfg.totalshow;
>  			break;
> +		case 'S':
> +			dumpcfg.show_statistics = true;
> +			++dumpcfg.totalshow;
> +			break;
>  		case 'V':
>  			erofsdump_print_version();
>  			exit(0);
> @@ -90,6 +139,310 @@ static int erofsdump_parse_options_cfg(int argc, char **argv)
>  	return 0;
>  }
>  
> +static int erofs_get_occupied_size(struct erofs_inode *inode,
> +		erofs_off_t *size)
> +{
> +	*size = 0;
> +	switch (inode->datalayout) {
> +	case EROFS_INODE_FLAT_INLINE:
> +	case EROFS_INODE_FLAT_PLAIN:
> +		stats.uncompressed_files++;
> +		*size = inode->i_size;
> +		break;
> +	case EROFS_INODE_FLAT_COMPRESSION_LEGACY:
> +	case EROFS_INODE_FLAT_COMPRESSION:
> +		stats.compressed_files++;
> +		*size = inode->u.i_blocks * EROFS_BLKSIZ;
> +		break;

We need to support EROFS_INODE_CHUNK_BASED as well since it'll be used
for the chunk deduplication.


> +	default:
> +		erofs_err("unknown datalayout");
> +		return -1;
> +	}
> +	return 0;
> +}
> +
> +static int erofs_getfile_extension(const char *filename)
> +{
> +	char *postfix = strrchr(filename, '.');
> +	int type = 0;
> +
> +	if (postfix == NULL)
> +		return OTHERFILETYPE - 1;
> +	while (type < OTHERFILETYPE - 1) {

	for (type = 0; type < OTHERFILETYPE - 1; ++type)

> +		if (strcmp(postfix, file_types[type]) == 0)
> +			break;
> +		type++;
> +	}
> +	return type;
> +}
> +
> +static void update_file_size_statatics(erofs_off_t occupied_size,
> +		erofs_off_t original_size)
> +{
> +	int occupied_size_mark;
> +	int original_size_mark;

Better in the one line:
	int occupied_size_mark, original_size_mark;

> +
> +	original_size_mark = 0;
> +	occupied_size_mark = 0;
> +	occupied_size >>= 10;
> +	original_size >>= 10;
> +
> +	while (occupied_size || original_size) {
> +		if (occupied_size) {
> +			occupied_size >>= 1;
> +			occupied_size_mark++;
> +		}
> +		if (original_size) {
> +			original_size >>= 1;
> +			original_size_mark++;
> +		}
> +	}
> +
> +	if (original_size_mark >= FILE_MAX_SIZE_BITS)
> +		stats.file_original_size[FILE_MAX_SIZE_BITS]++;
> +	else
> +		stats.file_original_size[original_size_mark]++;
> +
> +	if (occupied_size_mark >= FILE_MAX_SIZE_BITS)
> +		stats.file_comp_size[FILE_MAX_SIZE_BITS]++;
> +	else
> +		stats.file_comp_size[occupied_size_mark]++;
> +}
> +
> +static inline int erofs_checkdirent(struct erofs_dirent *de,
> +		struct erofs_dirent *last_de,
> +		u32 maxsize, const char *dname)
> +{
> +	int dname_len;
> +	unsigned int nameoff = le16_to_cpu(de->nameoff);
> +
> +	if (nameoff < sizeof(struct erofs_dirent) ||
> +			nameoff >= PAGE_SIZE) {
> +		erofs_err("invalid de[0].nameoff %u @ nid %llu",
> +				nameoff, de->nid | 0ULL);
> +		return -EFSCORRUPTED;
> +	}
> +
> +	dname_len = (de + 1 >= last_de) ? strnlen(dname, maxsize - nameoff) :
> +				le16_to_cpu(de[1].nameoff) - nameoff;
> +	/* a corrupted entry is found */
> +	if (nameoff + dname_len > maxsize ||
> +			dname_len > EROFS_NAME_LEN) {
> +		erofs_err("bogus dirent @ nid %llu",
> +				le64_to_cpu(de->nid) | 0ULL);
> +		DBG_BUGON(1);
> +		return -EFSCORRUPTED;
> +	}
> +	if (de->file_type >= EROFS_FT_MAX) {
> +		erofs_err("invalid file type %llu", de->nid);
> +		return -EFSCORRUPTED;
> +	}
> +	return dname_len;
> +}
> +
> +static int erofs_read_dirent(struct erofs_dirent *de,
> +		erofs_nid_t nid, erofs_nid_t parent_nid,
> +		const char *dname)
> +{
> +	int err;
> +	erofs_off_t occupied_size = 0;
> +	struct erofs_inode inode = { .nid = de->nid };
> +
> +	stats.files++;
> +	stats.file_category_stat[de->file_type]++;
> +	err = erofs_read_inode_from_disk(&inode);
> +	if (err) {
> +		erofs_err("read file inode from disk failed!");
> +		return err;
> +	}
> +
> +	err = erofs_get_occupied_size(&inode, &occupied_size);
> +	if (err) {
> +		erofs_err("get file size failed\n");
> +		return err;
> +	}
> +
> +	if (de->file_type == EROFS_FT_REG_FILE) {
> +		stats.files_total_origin_size += inode.i_size;
> +		stats.file_type_stat[erofs_getfile_extension(dname)]++;
> +		stats.files_total_size += occupied_size;
> +		update_file_size_statatics(occupied_size, inode.i_size);
> +	}
> +
> +	if ((de->file_type == EROFS_FT_DIR)
> +				 && de->nid != nid && de->nid != parent_nid) {
> +		err = erofs_read_dir(de->nid, nid);
> +		if (err) {
> +			erofs_err("parse dir nid %llu error occurred\n",
> +					de->nid);
> +			return err;
> +		}
> +	}
> +	return 0;
> +}
> +
> +
> +static int erofs_read_dir(erofs_nid_t nid, erofs_nid_t parent_nid)
> +{
> +	struct erofs_inode vi = { .nid = nid};

Some style problem.

> +	int err;
> +	char buf[EROFS_BLKSIZ];
> +	erofs_off_t offset;
> +
> +	err = erofs_read_inode_from_disk(&vi);
> +	if (err)
> +		return err;
> +
> +	offset = 0;
> +	while (offset < vi.i_size) {
> +		erofs_off_t maxsize = min_t(erofs_off_t,
> +			vi.i_size - offset, EROFS_BLKSIZ);

Some style problem, two extra tabs here are preferable.

Thanks,
Gao Xiang

> +		struct erofs_dirent *de = (void *)buf;
> +		struct erofs_dirent *end;
> +		unsigned int nameoff;
> +
> +		err = erofs_pread(&vi, buf, maxsize, offset);
> +		if (err)
> +			return err;
> +
> +		nameoff = le16_to_cpu(de->nameoff);
> +		end = (void *)buf + nameoff;
> +		while (de < end) {
> +			const char *dname;
> +			int ret;
> +
> +			/* skip "." and ".." dentry */
> +			if (de->nid == nid || de->nid == parent_nid) {
> +				de++;
> +				continue;
> +			}
> +
> +			dname = (char *)buf + nameoff;
> +			ret = erofs_checkdirent(de, end, maxsize, dname);
> +			if (ret < 0)
> +				return ret;
> +			ret = erofs_read_dirent(de, nid, parent_nid, dname);
> +			if (ret < 0)
> +				return ret;
> +			++de;
> +		}
> +		offset += maxsize;
> +	}
> +	return 0;
> +}
> +
> +static void erofsdump_print_chart_row(char *col1, unsigned int col2,
> +		double col3, char *col4)
> +{
> +	char row[500] = {0};
> +
> +	sprintf(row, chart_format, col1, col2, col3, col4);
> +	fprintf(stdout, row);
> +}
> +
> +static void erofsdump_filesize_distribution(const char *title,
> +		unsigned int *file_counts, unsigned int len)
> +{
> +	char col1[30];
> +	unsigned int col2;
> +	double col3;
> +	char col4[400];
> +	unsigned int lowerbound = 0;
> +	unsigned int upperbound = 1;
> +
> +	fprintf(stdout, "\n%s file size distribution:\n", title);
> +	fprintf(stdout, header_format, ">=(KB) .. <(KB) ", "count",
> +			"ratio", "distribution");
> +	for (int i = 0; i < len; i++) {
> +		memset(col1, 0, sizeof(col1));
> +		memset(col4, 0, sizeof(col4));
> +		if (i == len - 1)
> +			sprintf(col1, "%6d ..", lowerbound);
> +		else if (i <= 6)
> +			sprintf(col1, "%6d .. %-6d", lowerbound, upperbound);
> +		else
> +
> +			sprintf(col1, "%6d .. %-6d", lowerbound, upperbound);
> +		col2 = file_counts[i];
> +		if (stats.file_category_stat[EROFS_FT_REG_FILE])
> +			col3 = (double)(100 * col2) /
> +				stats.file_category_stat[EROFS_FT_REG_FILE];
> +		else
> +			col3 = 0.0;
> +		memset(col4, '#', col3 / 2);
> +		erofsdump_print_chart_row(col1, col2, col3, col4);
> +		lowerbound = upperbound;
> +		upperbound <<= 1;
> +	}
> +}
> +
> +static void erofsdump_filetype_distribution(char **file_types, unsigned int len)
> +{
> +	char col1[30];
> +	unsigned int col2;
> +	double col3;
> +	char col4[401];
> +
> +	fprintf(stdout, "\nFile type distribution:\n");
> +	fprintf(stdout, header_format, "type", "count", "ratio",
> +			"distribution");
> +	for (int i = 0; i < len; i++) {
> +		memset(col1, 0, sizeof(col1));
> +		memset(col4, 0, sizeof(col4));
> +		sprintf(col1, "%-17s", file_types[i]);
> +		col2 = stats.file_type_stat[i];
> +		if (stats.file_category_stat[EROFS_FT_REG_FILE])
> +			col3 = (double)(100 * col2) /
> +				stats.file_category_stat[EROFS_FT_REG_FILE];
> +		else
> +			col3 = 0.0;
> +		memset(col4, '#', col3 / 2);
> +		erofsdump_print_chart_row(col1, col2, col3, col4);
> +	}
> +}
> +
> +static void erofsdump_file_statistic(void)
> +{
> +	fprintf(stdout, "Filesystem total file count:		%lu\n",
> +			stats.files);
> +	for (int i = 0; i < EROFS_FT_MAX; i++)
> +		fprintf(stdout, "Filesystem %s count:		%lu\n",
> +			file_category_types[i], stats.file_category_stat[i]);
> +
> +	stats.compress_rate = (double)(100 * stats.files_total_size) /
> +		(double)(stats.files_total_origin_size);
> +	fprintf(stdout, "Filesystem compressed files:            %lu\n",
> +			stats.compressed_files);
> +	fprintf(stdout, "Filesystem uncompressed files:          %lu\n",
> +			stats.uncompressed_files);
> +	fprintf(stdout, "Filesystem total original file size:    %lu Bytes\n",
> +			stats.files_total_origin_size);
> +	fprintf(stdout, "Filesystem total file size:             %lu Bytes\n",
> +			stats.files_total_size);
> +	fprintf(stdout, "Filesystem compress rate:               %.2f%%\n",
> +			stats.compress_rate);
> +}
> +
> +static void erofsdump_print_statistic(void)
> +{
> +	int err;
> +
> +	err = erofs_read_dir(sbi.root_nid, sbi.root_nid);
> +	if (err) {
> +		erofs_err("read dir failed");
> +		return;
> +	}
> +
> +	erofsdump_file_statistic();
> +	erofsdump_filesize_distribution("Original",
> +			stats.file_original_size,
> +			ARRAY_SIZE(stats.file_original_size));
> +	erofsdump_filesize_distribution("On-disk",
> +			stats.file_comp_size,
> +			ARRAY_SIZE(stats.file_comp_size));
> +	erofsdump_filetype_distribution(file_types, OTHERFILETYPE);
> +}
> +
>  static void erofsdump_show_superblock(void)
>  {
>  	time_t time = sbi.build_time;
> @@ -156,6 +509,9 @@ int main(int argc, char **argv)
>  	if (dumpcfg.show_superblock)
>  		erofsdump_show_superblock();
>  
> +	if (dumpcfg.show_statistics)
> +		erofsdump_print_statistic();
> +
>  exit:
>  	erofs_exit_configure();
>  	return err;
> diff --git a/include/erofs/internal.h b/include/erofs/internal.h
> index 8b154ed..f19f839 100644
> --- a/include/erofs/internal.h
> +++ b/include/erofs/internal.h
> @@ -263,6 +263,7 @@ struct erofs_map_blocks {
>  int erofs_read_superblock(void);
>  
>  /* namei.c */
> +int erofs_read_inode_from_disk(struct erofs_inode *vi);
>  int erofs_ilookup(const char *path, struct erofs_inode *vi);
>  
>  /* data.c */
> diff --git a/lib/namei.c b/lib/namei.c
> index b4bdabf..56f199a 100644
> --- a/lib/namei.c
> +++ b/lib/namei.c
> @@ -22,7 +22,7 @@ static dev_t erofs_new_decode_dev(u32 dev)
>  	return makedev(major, minor);
>  }
>  
> -static int erofs_read_inode_from_disk(struct erofs_inode *vi)
> +int erofs_read_inode_from_disk(struct erofs_inode *vi)
>  {
>  	int ret, ifmt;
>  	char buf[sizeof(struct erofs_inode_extended)];
> -- 
> 2.31.1

Re: [PATCH 2/3] erofs-utils: dump: add -i options to dump file information of specific inode number

[Gao Xiang]

On Wed, Oct 27, 2021 at 08:03:50PM +0800, Guo Xuenan wrote:
> From: Wang Qi <mpiglet@outlook.com>
> 
> Signed-off-by: Guo Xuenan <guoxuenan@huawei.com>
> Signed-off-by: Wang Qi <mpiglet@outlook.com>
> ---
>  dump/main.c | 135 +++++++++++++++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 134 insertions(+), 1 deletion(-)
> 
> diff --git a/dump/main.c b/dump/main.c
> index 98ee68b..8a31345 100644
> --- a/dump/main.c
> +++ b/dump/main.c
> @@ -17,8 +17,10 @@
>  
>  struct erofsdump_cfg {
>  	unsigned int totalshow;
> +	bool show_inode;
>  	bool show_superblock;
>  	bool show_statistics;
> +	u64 ino;

erofs_nid_t nid;

>  };
>  static struct erofsdump_cfg dumpcfg;
>  
> @@ -82,6 +84,9 @@ static struct erofsdump_feature feature_lists[] = {
>  };
>  
>  static int erofs_read_dir(erofs_nid_t nid, erofs_nid_t parent_nid);
> +static inline int erofs_checkdirent(struct erofs_dirent *de,
> +		struct erofs_dirent *last_de,
> +		u32 maxsize, const char *dname);
>  
>  static void usage(void)
>  {
> @@ -89,6 +94,7 @@ static void usage(void)
>  	      "Dump erofs layout from IMAGE, and [options] are:\n"
>  	      " -S      show statistic information of the image\n"
>  	      " -V      print the version number of dump.erofs and exit.\n"
> +	      " -i #    show the target inode info of nid #\n"
>  	      " -s      show information about superblock\n"
>  	      " --help  display this help and exit.\n",
>  	      stderr);
> @@ -102,10 +108,17 @@ static void erofsdump_print_version(void)
>  static int erofsdump_parse_options_cfg(int argc, char **argv)
>  {
>  	int opt;
> +	u64 i;

erofs_nid_t i;

>  
> -	while ((opt = getopt_long(argc, argv, "SVs",
> +	while ((opt = getopt_long(argc, argv, "SVi:s",
>  				  long_options, NULL)) != -1) {
>  		switch (opt) {
> +		case 'i':

I perfer "--nid" for this, since the inode number is actually the field
in the on-disk inode (Although we report nid for now.)

> +			i = atoll(optarg);
> +			dumpcfg.show_inode = true;
> +			dumpcfg.ino = i;
> +			++dumpcfg.totalshow;
> +			break;
>  		case 's':
>  			dumpcfg.show_superblock = true;
>  			++dumpcfg.totalshow;
> @@ -331,6 +344,123 @@ static int erofs_read_dir(erofs_nid_t nid, erofs_nid_t parent_nid)
>  	return 0;
>  }
>  
> +static int erofs_get_pathname(erofs_nid_t nid, erofs_nid_t parent_nid,
> +		erofs_nid_t target, char *path, unsigned int pos)
> +{
> +	int err;
> +	struct erofs_inode inode = {.nid = nid};
> +	erofs_off_t offset;
> +	char buf[EROFS_BLKSIZ];
> +
> +	path[pos++] = '/';
> +	if (target == sbi.root_nid)
> +		return 0;
> +
> +	err = erofs_read_inode_from_disk(&inode);
> +	if (err) {
> +		erofs_err("read inode %lu failed", nid);
> +		return err;
> +	}
> +
> +	offset = 0;
> +	while (offset < inode.i_size) {
> +		erofs_off_t maxsize = min_t(erofs_off_t,
> +					inode.i_size - offset, EROFS_BLKSIZ);
> +		struct erofs_dirent *de = (void *)buf;
> +		struct erofs_dirent *end;
> +		unsigned int nameoff;
> +
> +		err = erofs_pread(&inode, buf, maxsize, offset);
> +		if (err)
> +			return err;
> +
> +		nameoff = le16_to_cpu(de->nameoff);
> +		end = (void *)buf + nameoff;
> +		while (de < end) {
> +			const char *dname;
> +			int len;
> +
> +			nameoff = le16_to_cpu(de->nameoff);
> +			dname = (char *)buf + nameoff;
> +			len = erofs_checkdirent(de, end, maxsize, dname);
> +			if (len < 0)
> +				return len;
> +
> +			if (de->nid == target) {
> +				memcpy(path + pos, dname, len);
> +				return 0;
> +			}
> +
> +			if (de->file_type == EROFS_FT_DIR &&
> +					de->nid != parent_nid &&
> +					de->nid != nid) {
> +				memcpy(path + pos, dname, len);
> +				err = erofs_get_pathname(de->nid, nid,
> +						target, path, pos + len);
> +				if (!err)
> +					return 0;
> +				memset(path + pos, 0, len);
> +			}
> +			++de;
> +		}
> +		offset += maxsize;
> +	}
> +	return -1;
> +}
> +
> +static void dumpfs_print_inode(void)
> +{
> +	int err;
> +	erofs_off_t size;
> +	u16 access_mode;
> +	time_t t;
> +	erofs_nid_t nid = dumpcfg.ino;
> +	struct erofs_inode inode = {.nid = nid};
> +	char path[PATH_MAX + 1] = {0};
> +	char access_mode_str[] = "rwxrwxrwx";
> +	char timebuf[128] = {0};
> +
> +	err = erofs_read_inode_from_disk(&inode);
> +	if (err) {
> +		erofs_err("read inode %lu from disk failed", nid);
> +		return;
> +	}
> +
> +	err = erofs_get_occupied_size(&inode, &size);
> +	if (err) {
> +		erofs_err("get file size failed\n");
> +		return;
> +	}
> +
> +	err = erofs_get_pathname(sbi.root_nid, sbi.root_nid, nid, path, 0);
> +	if (err < 0) {
> +		fprintf(stderr, "File: Path not found.\n");
> +		return;
> +	}
> +
> +	fprintf(stdout, "File : %s\n", path);
> +	fprintf(stdout, "Inode: %lu  ", inode.nid);
> +	fprintf(stdout, "Links: %u  ", inode.i_nlink);
> +	fprintf(stdout, "Layout: %d\n", inode.datalayout);
> +	fprintf(stdout, "File size: %lu   ", inode.i_size);
> +	fprintf(stdout,	"On-disk size: %lu   ", size);
> +	fprintf(stdout, "Compress rate: %.2f%%\n",
> +			(double)(100 * size) / (double)(inode.i_size));
> +	fprintf(stdout, "Inode size: %d   ", inode.inode_isize);
> +	fprintf(stdout, "Extent size: %u   ", inode.extent_isize);
> +	fprintf(stdout,	"Xattr size: %u\n", inode.xattr_isize);
> +	access_mode = inode.i_mode & 0777;
> +	t = inode.i_ctime;

Only extented inodes have their own timestamps.

> +	strftime(timebuf, sizeof(timebuf),
> +			"%Y-%m-%d %H:%M:%S", localtime(&t));

place
	access_mode = inode.i_mode & 0777;

here.

> +	for (int i = 8; i >= 0; i--)
> +		if (((access_mode >> i) & 1) == 0)
> +			access_mode_str[8 - i] = '-';
> +	fprintf(stdout, "Uid: %u   Gid: %u  ", inode.i_uid, inode.i_gid);
> +	fprintf(stdout, "Access: %04o/%s\n", access_mode, access_mode_str);
> +	fprintf(stdout, "Birth : %s\n", timebuf);

"Timestamp :" here is preferred.

Thanks,
Gao Xiang

> +}
> +
>  static void erofsdump_print_chart_row(char *col1, unsigned int col2,
>  		double col3, char *col4)
>  {
> @@ -512,6 +642,9 @@ int main(int argc, char **argv)
>  	if (dumpcfg.show_statistics)
>  		erofsdump_print_statistic();
>  
> +	if (dumpcfg.show_inode)
> +		dumpfs_print_inode();
> +
>  exit:
>  	erofs_exit_configure();
>  	return err;
> -- 
> 2.31.1

Re: [PATCH 3/3] erofs-utils: dump: add -e option to dump the inode extent info

[Gao Xiang]

Hi Xuenan,

On Wed, Oct 27, 2021 at 08:03:51PM +0800, Guo Xuenan wrote:
> From: Wang Qi <mpiglet@outlook.com>
> 
> Signed-off-by: Guo Xuenan <guoxuenan@huawei.com>
> Signed-off-by: Wang Qi <mpiglet@outlook.com>

Please help resend the FIEMAP report patch as well. So I could use b4
tool to merge the whole patchset....

> ---
>  dump/main.c | 86 ++++++++++++++++++++++++++++++++++++++++++++++++++++-
>  1 file changed, 85 insertions(+), 1 deletion(-)
> 
> diff --git a/dump/main.c b/dump/main.c
> index 8a31345..e4018a7 100644
> --- a/dump/main.c
> +++ b/dump/main.c
> @@ -18,9 +18,11 @@
>  struct erofsdump_cfg {
>  	unsigned int totalshow;
>  	bool show_inode;
> +	bool show_inode_phy;
>  	bool show_superblock;
>  	bool show_statistics;
>  	u64 ino;
> +	u64 ino_phy;
>  };
>  static struct erofsdump_cfg dumpcfg;
>  
> @@ -94,6 +96,7 @@ static void usage(void)
>  	      "Dump erofs layout from IMAGE, and [options] are:\n"
>  	      " -S      show statistic information of the image\n"
>  	      " -V      print the version number of dump.erofs and exit.\n"
> +	      " -e #    show the inode extent info of nid #\n"
>  	      " -i #    show the target inode info of nid #\n"
>  	      " -s      show information about superblock\n"
>  	      " --help  display this help and exit.\n",
> @@ -110,9 +113,14 @@ static int erofsdump_parse_options_cfg(int argc, char **argv)
>  	int opt;
>  	u64 i;
>  
> -	while ((opt = getopt_long(argc, argv, "SVi:s",
> +	while ((opt = getopt_long(argc, argv, "SVi:e:s",
>  				  long_options, NULL)) != -1) {
>  		switch (opt) {
> +		case 'e':
> +			dumpcfg.show_inode_phy = true;
> +			dumpcfg.ino_phy = atoll(optarg);
> +			++dumpcfg.totalshow;
> +			break;
>  		case 'i':
>  			i = atoll(optarg);
>  			dumpcfg.show_inode = true;
> @@ -408,6 +416,79 @@ static int erofs_get_pathname(erofs_nid_t nid, erofs_nid_t parent_nid,
>  	return -1;
>  }
>  
> +static void erofsdump_show_inode_phy(void)
> +{
> +	int err;
> +	erofs_nid_t nid = dumpcfg.ino_phy;
> +	struct erofs_inode inode = {.nid = nid};
> +	char path[PATH_MAX + 1] = {0};
> +
> +	err = erofs_read_inode_from_disk(&inode);
> +	if (err < 0) {
> +		erofs_err("read inode %lu from disk failed", nid);
> +		return;
> +	}
> +
> +	const erofs_off_t ibase = iloc(inode.nid);
> +	const erofs_off_t pos = Z_EROFS_VLE_LEGACY_INDEX_ALIGN(
> +			ibase + inode.inode_isize + inode.xattr_isize);
> +	erofs_blk_t blocks = inode.u.i_blocks;
> +	erofs_blk_t start = 0;
> +	erofs_blk_t end = 0;
> +	unsigned int extent_count;
> +	struct erofs_map_blocks map = {
> +		.index = UINT_MAX,
> +		.m_la = 0,
> +	};

let's avoid variable definitions here (please help define them at the
beginning of this function instead.)

> +
> +	fprintf(stdout, "Inode: %lu\n", nid);
> +	err = erofs_get_pathname(sbi.root_nid, sbi.root_nid, nid, path, 0);
> +	if (err) {
> +		erofs_err("Path not found\n");
> +		return;
> +	}
> +	fprintf(stdout, "File path: %s\n", path);
> +	fprintf(stdout, "File size: %lu\n", inode.i_size);
> +
> +	switch (inode.datalayout) {
> +	case EROFS_INODE_FLAT_INLINE:
> +	case EROFS_INODE_FLAT_PLAIN:

we need to implement EROFS_INODE_CHUNK_BASED:

> +		if (inode.u.i_blkaddr == NULL_ADDR)
> +			start = end = erofs_blknr(pos);
> +		else {
> +			start = inode.u.i_blkaddr;
> +			end = start + BLK_ROUND_UP(inode.i_size) - 1;
> +		}
> +		fprintf(stdout, "Plain blknr: %u - %u\n", start, end);
> +		break;
> +
> +	case EROFS_INODE_FLAT_COMPRESSION_LEGACY:
> +	case EROFS_INODE_FLAT_COMPRESSION:

Actually I guess these can be merged into one helper erofs_map_block()
instead.

Thanks,
Gao Xiang

> +		err = z_erofs_map_blocks_iter(&inode, &map, 0);
> +		if (err)
> +			erofs_err("get file blocks range failed");
> +
> +		start = erofs_blknr(map.m_pa);
> +		end = start - 1 + blocks;
> +		fprintf(stdout,
> +				"Compressed blknr: %u - %u\n", start, end);
> +		extent_count = 0;
> +		map.m_la = 0;
> +		fprintf(stdout, "Ext:  logical_offset/length:  physical_offset/length\n");
> +		while (map.m_la < inode.i_size) {
> +			err = z_erofs_map_blocks_iter(&inode, &map,
> +					EROFS_GET_BLOCKS_FIEMAP);
> +			fprintf(stdout, "%u: ", extent_count++);
> +			fprintf(stdout, "%6lu..%-6lu / %-lu: ",
> +					map.m_la, map.m_la + map.m_llen, map.m_llen);
> +			fprintf(stdout, "%lu..%lu / %-lu\n",
> +					map.m_pa, map.m_pa + map.m_plen, map.m_plen);
> +			map.m_la += map.m_llen;
> +		}
> +		break;
> +	}
> +}
> +
>  static void dumpfs_print_inode(void)
>  {
>  	int err;
> @@ -645,6 +726,9 @@ int main(int argc, char **argv)
>  	if (dumpcfg.show_inode)
>  		dumpfs_print_inode();
>  
> +	if (dumpcfg.show_inode_phy)
> +		erofsdump_show_inode_phy();
> +
>  exit:
>  	erofs_exit_configure();
>  	return err;
> -- 
> 2.31.1