2002-01-17 01:11:46

by Daniel Quinlan

[permalink] [raw]
Subject: [PATCH] cramfs updates for 2.4.17

Various cramfs updates for 2.4.17. It only touches cramfs stuff.
Your feedback is appreciated.

Note: This is not integrated with the big-endian patches yet, but
that's what I'm currently looking at.

------------------------------------------------------------------------

summary: cramfs updates

Documentation/filesystems/cramfs.txt
- remove comment about ROM size limit
- fix up magic

fs/cramfs/README
- add note about sorted directory entries
- mkcramfs compile-time DO_HOLES option replaced by run-time -z option
- update tools section
- add note about PAGE_CACHE_SIZE possibly changing on arm and ia64

cramfs_fs.h
- add CRAMFS_MAXPATHLEN (252)
- clarify CRAMFS_SUPPORTED_FLAGS definition

cramfsck.c
- some new tests (several based on tests in Peter Moulder's cramfsck)
- large-scale code cleanup, check more return codes, standardized
errors/warnings, etc.
- always try to mmap for crc test, fallback to read/write

mkcramfs.c
- bug fix: use CRAMFS_MAXPATHLEN not 255 as maximum pathname length
(mkcramfs generated invalid cramfs images if pathnames of length
253-255 were included, but cramfsck was able to detect the problem)
- medium-scale code cleanup, check more return codes, standardized
errors/warnings, etc.
- truncate UTF-8 filenames on character boundary (a longstanding TODO)
- use Z_BEST_COMPRESSION instead of default compression level

diff -ur linux-orig/Documentation/filesystems/cramfs.txt linux/Documentation/filesystems/cramfs.txt
--- linux-orig/Documentation/filesystems/cramfs.txt Thu Jul 19 16:14:53 2001
+++ linux/Documentation/filesystems/cramfs.txt Tue Jan 8 00:27:30 2002
@@ -19,9 +19,7 @@
File sizes are limited to less than 16MB.

Maximum filesystem size is a little over 256MB. (The last file on the
-filesystem is allowed to extend past 256MB.) (Comments in mkcramfs.c
-suggest that ROM sizes may be limited to 64MB, though that's not a
-limitation in cramfs code.)
+filesystem is allowed to extend past 256MB.)

Only the low 8 bits of gid are stored. The current version of
mkcramfs simply truncates to 8 bits, which is a potential security
@@ -48,18 +46,28 @@


For /usr/share/magic
-------------------
+--------------------

-0 long 0x28cd3d45 Linux cramfs
->4 long x size %d
->8 long x flags 0x%x
->12 long x future 0x%x
+0 ulelong 0x28cd3d45 Linux cramfs offset 0
+>4 ulelong x size %d
+>8 ulelong x flags 0x%x
+>12 ulelong x future 0x%x
>16 string >\0 signature "%.16s"
->32 long x fsid.crc 0x%x
->36 long x fsid.edition %d
->40 long x fsid.blocks %d
->44 long x fsid.files %d
+>32 ulelong x fsid.crc 0x%x
+>36 ulelong x fsid.edition %d
+>40 ulelong x fsid.blocks %d
+>44 ulelong x fsid.files %d
>48 string >\0 name "%.16s"
+512 ulelong 0x28cd3d45 Linux cramfs offset 512
+>516 ulelong x size %d
+>520 ulelong x flags 0x%x
+>524 ulelong x future 0x%x
+>528 string >\0 signature "%.16s"
+>544 ulelong x fsid.crc 0x%x
+>548 ulelong x fsid.edition %d
+>552 ulelong x fsid.blocks %d
+>556 ulelong x fsid.files %d
+>560 string >\0 name "%.16s"


Hacker Notes
diff -ur linux-orig/fs/cramfs/README linux/fs/cramfs/README
--- linux-orig/fs/cramfs/README Thu Jul 19 16:14:53 2001
+++ linux/fs/cramfs/README Sat Jan 12 22:15:55 2002
@@ -29,6 +29,10 @@
lines); put another way, the same order as `find -type d -exec
ls -AU1 {} \;'.

+Beginning in 2.4.7, directory entries are sorted. This optimization
+allows cramfs_lookup to return more quickly when a filename does not
+exist, speeds up user-space directory sorts, etc.
+
<data>:
One <file_data> for each file that's either a symlink or a
regular file of non-zero st_size.
@@ -63,17 +67,22 @@
This kernel supports cramfs holes (i.e. [efficient representation of]
blocks in uncompressed data consisting entirely of NUL bytes), but by
default mkcramfs doesn't test for & create holes, since cramfs in
-kernels up to at least 2.3.39 didn't support holes. Compile mkcramfs
-with -DDO_HOLES if you want it to create files that can have holes in
-them.
+kernels up to at least 2.3.39 didn't support holes. Run mkcramfs
+with -z if you want it to create files that can have holes in them.


Tools
-----

+The cramfsck program included in the scripts/cramfs directory can (and
+perhaps even should) be used to test cramfs filesystems, although it
+does not try to fix errors since cramfs is a read-only filesystem and
+you can just re-run mkcramfs after fixing any bugs. cramfsck can also
+extract cramfs filesystems without mounting them.
+
If you're hacking on cramfs, you might find useful some tools for
-testing cramfs at <http://cvs.bofh.asn.au/cramfs/>, including a
-rudimentary fsck for cramfs.
+testing cramfs at <http://cvs.bofh.asn.au/cramfs/>. Note: the
+cramfsck located there is superseded by the one in scripts/cramfs.


Future Development
@@ -103,8 +112,8 @@
PAGE_CACHE_SIZE (4096)' to `#include <asm/page.h>'. The disadvantage
is that the generated cramfs cannot always be shared between different
kernels, not even necessarily kernels of the same architecture if
-PAGE_CACHE_SIZE is subject to change between kernel versions.
-
+PAGE_CACHE_SIZE is subject to change between kernel versions
+(currently possible with arm and ia64).

The remaining options try to make cramfs more sharable.

diff -ur linux-orig/include/linux/cramfs_fs.h linux/include/linux/cramfs_fs.h
--- linux-orig/include/linux/cramfs_fs.h Thu Jul 19 16:14:53 2001
+++ linux/include/linux/cramfs_fs.h Tue Jan 8 00:29:38 2002
@@ -24,6 +24,12 @@
#define CRAMFS_OFFSET_WIDTH 26

/*
+ * Since inode.namelen is a unsigned 6-bit number, the maximum cramfs
+ * path length is 63 << 2 = 252.
+ */
+#define CRAMFS_MAXPATHLEN (((1 << CRAMFS_NAMELEN_WIDTH) - 1) << 2)
+
+/*
* Reasonably terse representation of the inode data.
*/
struct cramfs_inode {
@@ -52,14 +58,14 @@
* Superblock information at the beginning of the FS.
*/
struct cramfs_super {
- u32 magic; /* 0x28cd3d45 - random number */
- u32 size; /* length in bytes */
- u32 flags; /* 0 */
- u32 future; /* 0 */
- u8 signature[16]; /* "Compressed ROMFS" */
+ u32 magic; /* 0x28cd3d45 - random number */
+ u32 size; /* length in bytes */
+ u32 flags; /* feature flags */
+ u32 future; /* reserved for future use */
+ u8 signature[16]; /* "Compressed ROMFS" */
struct cramfs_info fsid; /* unique filesystem info */
- u8 name[16]; /* user-defined name */
- struct cramfs_inode root; /* Root inode data */
+ u8 name[16]; /* user-defined name */
+ struct cramfs_inode root; /* root inode data */
};

/*
@@ -79,7 +85,10 @@
* if (flags & ~CRAMFS_SUPPORTED_FLAGS). Maybe that should be
* changed to test super.future instead.
*/
-#define CRAMFS_SUPPORTED_FLAGS (0x7ff)
+#define CRAMFS_SUPPORTED_FLAGS ( 0x000000ff \
+ | CRAMFS_FLAG_HOLES \
+ | CRAMFS_FLAG_WRONG_SIGNATURE \
+ | CRAMFS_FLAG_SHIFTED_ROOT_OFFSET )

/* Uncompression interfaces to the underlying zlib */
int cramfs_uncompress_block(void *dst, int dstlen, void *src, int srclen);
diff -ur linux-orig/scripts/cramfs/cramfsck.c linux/scripts/cramfs/cramfsck.c
--- linux-orig/scripts/cramfs/cramfsck.c Thu Jul 19 16:14:53 2001
+++ linux/scripts/cramfs/cramfsck.c Sat Jan 12 22:19:44 2002
@@ -1,7 +1,7 @@
/*
* cramfsck - check a cramfs file system
*
- * Copyright (C) 2000-2001 Transmeta Corporation
+ * Copyright (C) 2000-2002 Transmeta Corporation
*
* 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
@@ -28,11 +28,14 @@
* 2000/07/11: Daniel Quinlan (file length tests, start at offset 0 or 512,
* fsck-compatible exit codes)
* 2000/07/15: Daniel Quinlan (initial support for block devices)
+ * 2002/01/10: Daniel Quinlan (additional checks, test more return codes,
+ * use read if mmap fails, standardize messages)
*/

/* compile-time options */
#define INCLUDE_FS_TESTS /* include cramfs checking and extraction */

+#define _GNU_SOURCE
#include <sys/types.h>
#include <stdio.h>
#include <sys/stat.h>
@@ -43,7 +46,6 @@
#include <stdlib.h>
#include <errno.h>
#include <string.h>
-#include <assert.h>
#include <getopt.h>
#include <sys/sysmacros.h>
#include <utime.h>
@@ -53,25 +55,34 @@
#include <linux/cramfs_fs.h>
#include <zlib.h>

+/* Exit codes used by fsck-type programs */
+#define FSCK_OK 0 /* No errors */
+#define FSCK_NONDESTRUCT 1 /* File system errors corrected */
+#define FSCK_REBOOT 2 /* System should be rebooted */
+#define FSCK_UNCORRECTED 4 /* File system errors left uncorrected */
+#define FSCK_ERROR 8 /* Operational error */
+#define FSCK_USAGE 16 /* Usage or syntax error */
+#define FSCK_LIBRARY 128 /* Shared library error */
+
+#define PAD_SIZE 512
+#define PAGE_CACHE_SIZE (4096)
+
static const char *progname = "cramfsck";

static int fd; /* ROM image file descriptor */
static char *filename; /* ROM image filename */
-struct cramfs_super *super; /* just find the cramfs superblock once */
+struct cramfs_super super; /* just find the cramfs superblock once */
static int opt_verbose = 0; /* 1 = verbose (-v), 2+ = very verbose (-vv) */
#ifdef INCLUDE_FS_TESTS
-static int opt_extract = 0; /* extract cramfs (-x) */
-char *extract_dir = NULL; /* extraction directory (-x) */
-
-unsigned long start_inode = 1 << 28; /* start of first non-root inode */
-unsigned long end_inode = 0; /* end of the directory structure */
-unsigned long start_data = 1 << 28; /* start of the data (256 MB = max) */
-unsigned long end_data = 0; /* end of the data */
-/* true? cramfs_super < start_inode < end_inode <= start_data <= end_data */
+static int opt_extract = 0; /* extract cramfs (-x) */
+static char *extract_dir = "root"; /* extraction directory (-x) */
static uid_t euid; /* effective UID */

-#define PAD_SIZE 512
-#define PAGE_CACHE_SIZE (4096)
+/* (cramfs_super + start) <= start_dir < end_dir <= start_data <= end_data */
+static unsigned long start_dir = ~0UL; /* start of first non-root inode */
+static unsigned long end_dir = 0; /* end of the directory structure */
+static unsigned long start_data = ~0UL; /* start of the data (256 MB = max) */
+static unsigned long end_data = 0; /* end of the data */

/* Guarantee access to at least 8kB at a time */
#define ROMBUFFER_BITS 13
@@ -82,8 +93,10 @@

/* Uncompressing data structures... */
static char outbuffer[PAGE_CACHE_SIZE*2];
-z_stream stream;
+static z_stream stream;

+/* Prototypes */
+static void expand_fs(char *, struct cramfs_inode *);
#endif /* INCLUDE_FS_TESTS */

/* Input status of 0 to print help and exit without an error. */
@@ -100,8 +113,159 @@
exit(status);
}

+static void die(int status, int syserr, const char *fmt, ...)
+{
+ va_list arg_ptr;
+ int save = errno;
+
+ fflush(0);
+ va_start(arg_ptr, fmt);
+ fprintf(stderr, "%s: ", progname);
+ vfprintf(stderr, fmt, arg_ptr);
+ if (syserr) {
+ fprintf(stderr, ": %s", strerror(save));
+ }
+ fprintf(stderr, "\n");
+ va_end(arg_ptr);
+ exit(status);
+}
+
+static void test_super(int *start, size_t *length) {
+ struct stat st;
+
+ /* find the physical size of the file or block device */
+ if (stat(filename, &st) < 0) {
+ die(FSCK_ERROR, 1, "stat failed: %s", filename);
+ }
+ fd = open(filename, O_RDONLY);
+ if (fd < 0) {
+ die(FSCK_ERROR, 1, "open failed: %s", filename);
+ }
+ if (S_ISBLK(st.st_mode)) {
+ if (ioctl(fd, BLKGETSIZE, length) < 0) {
+ die(FSCK_ERROR, 1, "ioctl failed: unable to determine device size: %s", filename);
+ }
+ *length = *length * 512;
+ }
+ else if (S_ISREG(st.st_mode)) {
+ *length = st.st_size;
+ }
+ else {
+ die(FSCK_ERROR, 0, "not a block device or file: %s", filename);
+ }
+
+ if (*length < sizeof(struct cramfs_super)) {
+ die(FSCK_UNCORRECTED, 0, "file length too short");
+ }
+
+ /* find superblock */
+ if (read(fd, &super, sizeof(super)) != sizeof(super)) {
+ die(FSCK_ERROR, 1, "read failed: %s", filename);
+ }
+ if (super.magic == CRAMFS_MAGIC) {
+ *start = 0;
+ }
+ else if (*length >= (PAD_SIZE + sizeof(super))) {
+ lseek(fd, PAD_SIZE, SEEK_SET);
+ if (read(fd, &super, sizeof(super)) != sizeof(super)) {
+ die(FSCK_ERROR, 1, "read failed: %s", filename);
+ }
+ if (super.magic == CRAMFS_MAGIC) {
+ *start = PAD_SIZE;
+ }
+ }
+
+ /* superblock tests */
+ if (super.magic != CRAMFS_MAGIC) {
+ die(FSCK_UNCORRECTED, 0, "superblock magic not found");
+ }
+ if (super.flags & ~CRAMFS_SUPPORTED_FLAGS) {
+ die(FSCK_ERROR, 0, "unsupported filesystem features");
+ }
+ if (super.size < PAGE_CACHE_SIZE) {
+ die(FSCK_UNCORRECTED, 0, "superblock size (%d) too small", super.size);
+ }
+ if (super.flags & CRAMFS_FLAG_FSID_VERSION_2) {
+ if (super.fsid.files == 0) {
+ die(FSCK_UNCORRECTED, 0, "zero file count");
+ }
+ if (*length < super.size) {
+ die(FSCK_UNCORRECTED, 0, "file length too short");
+ }
+ else if (*length > super.size) {
+ fprintf(stderr, "warning: file extends past end of filesystem\n");
+ }
+ }
+ else {
+ fprintf(stderr, "warning: old cramfs format\n");
+ }
+}
+
+static void test_crc(int start)
+{
+ void *buf;
+ u32 crc;
+
+ if (!(super.flags & CRAMFS_FLAG_FSID_VERSION_2)) {
+#ifdef INCLUDE_FS_TESTS
+ return;
+#else /* not INCLUDE_FS_TESTS */
+ die(FSCK_USAGE, 0, "unable to test CRC: old cramfs format");
+#endif /* not INCLUDE_FS_TESTS */
+ }
+
+ crc = crc32(0L, Z_NULL, 0);
+
+ buf = mmap(NULL, super.size, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
+ if (buf == MAP_FAILED) {
+ buf = mmap(NULL, super.size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
+ if (buf != MAP_FAILED) {
+ lseek(fd, 0, SEEK_SET);
+ read(fd, buf, super.size);
+ }
+ }
+ if (buf != MAP_FAILED) {
+ ((struct cramfs_super *) (buf+start))->fsid.crc = crc32(0L, Z_NULL, 0);
+ crc = crc32(crc, buf+start, super.size-start);
+ munmap(buf, super.size);
+ }
+ else {
+ int retval;
+ size_t length = 0;
+
+ buf = malloc(4096);
+ if (!buf) {
+ die(FSCK_ERROR, 1, "malloc failed");
+ }
+ lseek(fd, start, SEEK_SET);
+ for (;;) {
+ retval = read(fd, buf, 4096);
+ if (retval < 0) {
+ die(FSCK_ERROR, 1, "read failed: %s", filename);
+ }
+ else if (retval == 0) {
+ break;
+ }
+ if (length == 0) {
+ ((struct cramfs_super *) buf)->fsid.crc = crc32(0L, Z_NULL, 0);
+ }
+ length += retval;
+ if (length > (super.size-start)) {
+ crc = crc32(crc, buf, retval - (length - (super.size-start)));
+ break;
+ }
+ crc = crc32(crc, buf, retval);
+ }
+ free(buf);
+ }
+
+ if (crc != super.fsid.crc) {
+ die(FSCK_UNCORRECTED, 0, "crc error");
+ }
+}
+
#ifdef INCLUDE_FS_TESTS
-void print_node(char type, struct cramfs_inode *i, char *name)
+static void print_node(char type, struct cramfs_inode *i, char *name)
{
char info[10];

@@ -135,6 +299,10 @@
static struct cramfs_inode *cramfs_iget(struct cramfs_inode * i)
{
struct cramfs_inode *inode = malloc(sizeof(struct cramfs_inode));
+
+ if (!inode) {
+ die(FSCK_ERROR, 1, "malloc failed");
+ }
*inode = *i;
return inode;
}
@@ -144,27 +312,27 @@
return cramfs_iget(romfs_read(ino));
}

-void iput(struct cramfs_inode *inode)
+static void iput(struct cramfs_inode *inode)
{
free(inode);
}

/*
- * Return the offset of the root directory,
- * or 0 if none.
+ * Return the offset of the root directory
*/
static struct cramfs_inode *read_super(void)
{
- unsigned long offset;
+ unsigned long offset = super.root.offset << 2;

- offset = super->root.offset << 2;
- if (super->magic != CRAMFS_MAGIC)
- return NULL;
- if (memcmp(super->signature, CRAMFS_SIGNATURE, sizeof(super->signature)) != 0)
- return NULL;
- if (offset < sizeof(super))
- return NULL;
- return cramfs_iget(&super->root);
+ if (!S_ISDIR(super.root.mode))
+ die(FSCK_UNCORRECTED, 0, "root inode is not directory");
+ if (!(super.flags & CRAMFS_FLAG_SHIFTED_ROOT_OFFSET) &&
+ ((offset != sizeof(struct cramfs_super)) &&
+ (offset != PAD_SIZE + sizeof(struct cramfs_super))))
+ {
+ die(FSCK_UNCORRECTED, 0, "bad root offset (%lu)", offset);
+ }
+ return cramfs_iget(&super.root);
}

static int uncompress_block(void *src, int len)
@@ -179,38 +347,177 @@

inflateReset(&stream);

+ if (len > PAGE_CACHE_SIZE*2) {
+ die(FSCK_UNCORRECTED, 0, "data block too large");
+ }
err = inflate(&stream, Z_FINISH);
if (err != Z_STREAM_END) {
- fprintf(stderr, "%s: error %d while decompressing! %p(%d)\n",
- filename, err, src, len);
- exit(4);
+ die(FSCK_UNCORRECTED, 0, "decompression error %p(%d): %s",
+ zError(err), src, len);
}
return stream.total_out;
}

+static void do_uncompress(char *path, int fd, unsigned long offset, unsigned long size)
+{
+ unsigned long curr = offset + 4 * ((size + PAGE_CACHE_SIZE - 1) / PAGE_CACHE_SIZE);
+
+ do {
+ unsigned long out = PAGE_CACHE_SIZE;
+ unsigned long next = *(u32 *) romfs_read(offset);
+
+ if (next > end_data) {
+ end_data = next;
+ }
+
+ offset += 4;
+ if (curr == next) {
+ if (opt_verbose > 1) {
+ printf(" hole at %ld (%d)\n", curr, PAGE_CACHE_SIZE);
+ }
+ if (size < PAGE_CACHE_SIZE)
+ out = size;
+ memset(outbuffer, 0x00, out);
+ }
+ else {
+ if (opt_verbose > 1) {
+ printf(" uncompressing block at %ld to %ld (%ld)\n", curr, next, next - curr);
+ }
+ out = uncompress_block(romfs_read(curr), next - curr);
+ }
+ if (size >= PAGE_CACHE_SIZE) {
+ if (out != PAGE_CACHE_SIZE) {
+ die(FSCK_UNCORRECTED, 0, "non-block (%ld) bytes", out);
+ }
+ } else {
+ if (out != size) {
+ die(FSCK_UNCORRECTED, 0, "non-size (%ld vs %ld) bytes", out, size);
+ }
+ }
+ size -= out;
+ if (opt_extract) {
+ if (write(fd, outbuffer, out) < 0) {
+ die(FSCK_ERROR, 1, "write failed: %s", path);
+ }
+ }
+ curr = next;
+ } while (size);
+}
+
static void change_file_status(char *path, struct cramfs_inode *i)
{
struct utimbuf epoch = { 0, 0 };

if (euid == 0) {
if (lchown(path, i->uid, i->gid) < 0) {
- perror(path);
- exit(8);
+ die(FSCK_ERROR, 1, "lchown failed: %s", path);
}
if (S_ISLNK(i->mode))
return;
if ((S_ISUID | S_ISGID) & i->mode) {
if (chmod(path, i->mode) < 0) {
- perror(path);
- exit(8);
+ die(FSCK_ERROR, 1, "chown failed: %s", path);
}
}
}
if (S_ISLNK(i->mode))
return;
if (utime(path, &epoch) < 0) {
- perror(path);
- exit(8);
+ die(FSCK_ERROR, 1, "utime failed: %s", path);
+ }
+}
+
+static void do_directory(char *path, struct cramfs_inode *i)
+{
+ int pathlen = strlen(path);
+ int count = i->size;
+ unsigned long offset = i->offset << 2;
+ char *newpath = malloc(pathlen + 256);
+
+ if (!newpath) {
+ die(FSCK_ERROR, 1, "malloc failed");
+ }
+ if (offset == 0 && count != 0) {
+ die(FSCK_UNCORRECTED, 0, "directory inode has zero offset and non-zero size: %s", path);
+ }
+ if (offset != 0 && offset < start_dir) {
+ start_dir = offset;
+ }
+ /* TODO: Do we need to check end_dir for empty case? */
+ memcpy(newpath, path, pathlen);
+ newpath[pathlen] = '/';
+ pathlen++;
+ if (opt_verbose) {
+ print_node('d', i, path);
+ }
+ if (opt_extract) {
+ if (mkdir(path, i->mode) < 0) {
+ die(FSCK_ERROR, 1, "mkdir failed: %s", path);
+ }
+ change_file_status(path, i);
+ }
+ while (count > 0) {
+ struct cramfs_inode *child = iget(offset);
+ int size;
+ int newlen = child->namelen << 2;
+
+ size = sizeof(struct cramfs_inode) + newlen;
+ count -= size;
+
+ offset += sizeof(struct cramfs_inode);
+
+ memcpy(newpath + pathlen, romfs_read(offset), newlen);
+ newpath[pathlen + newlen] = 0;
+ if (newlen == 0) {
+ die(FSCK_UNCORRECTED, 0, "filename length is zero");
+ }
+ if ((pathlen + newlen) - strlen(newpath) > 3) {
+ die(FSCK_UNCORRECTED, 0, "bad filename length");
+ }
+ expand_fs(newpath, child);
+
+ offset += newlen;
+
+ if (offset <= start_dir) {
+ die(FSCK_UNCORRECTED, 0, "bad inode offset");
+ }
+ if (offset > end_dir) {
+ end_dir = offset;
+ }
+ iput(child); /* free(child) */
+ }
+ free(newpath);
+}
+
+static void do_file(char *path, struct cramfs_inode *i)
+{
+ unsigned long offset = i->offset << 2;
+ int fd = 0;
+
+ if (offset == 0 && i->size != 0) {
+ die(FSCK_UNCORRECTED, 0, "file inode has zero offset and non-zero size");
+ }
+ if (i->size == 0 && offset != 0) {
+ die(FSCK_UNCORRECTED, 0, "file inode has zero size and non-zero offset");
+ }
+ if (offset != 0 && offset < start_data) {
+ start_data = offset;
+ }
+ if (opt_verbose) {
+ print_node('f', i, path);
+ }
+ if (opt_extract) {
+ fd = open(path, O_WRONLY | O_CREAT | O_TRUNC, i->mode);
+ if (fd < 0) {
+ die(FSCK_ERROR, 1, "open failed: %s", path);
+ }
+ }
+ if (i->size) {
+ do_uncompress(path, fd, offset, i->size);
+ }
+ if (opt_extract) {
+ close(fd);
+ change_file_status(path, i);
}
}

@@ -221,32 +528,39 @@
unsigned long next = *(u32 *) romfs_read(offset);
unsigned long size;

+ if (offset == 0) {
+ die(FSCK_UNCORRECTED, 0, "symbolic link has zero offset");
+ }
+ if (i->size == 0) {
+ die(FSCK_UNCORRECTED, 0, "symbolic link has zero size");
+ }
+
+ if (offset < start_data) {
+ start_data = offset;
+ }
if (next > end_data) {
end_data = next;
}

size = uncompress_block(romfs_read(curr), next - curr);
if (size != i->size) {
- fprintf(stderr, "%s: size error in symlink `%s'\n",
- filename, path);
- exit(4);
+ die(FSCK_UNCORRECTED, 0, "size error in symlink: %s", path);
}
outbuffer[size] = 0;
if (opt_verbose) {
char *str;

- str = malloc(strlen(outbuffer) + strlen(path) + 5);
- strcpy(str, path);
- strncat(str, " -> ", 4);
- strncat(str, outbuffer, size);
-
+ asprintf(&str, "%s -> %s", path, outbuffer);
print_node('l', i, str);
if (opt_verbose > 1) {
printf(" uncompressing block at %ld to %ld (%ld)\n", curr, next, next - curr);
}
+ free(str);
}
if (opt_extract) {
- symlink(outbuffer, path);
+ if (symlink(outbuffer, path) < 0) {
+ die(FSCK_ERROR, 1, "symlink failed: %s", path);
+ }
change_file_status(path, i);
}
}
@@ -256,6 +570,9 @@
dev_t devtype = 0;
char type;

+ if (i->offset) { /* no need to shift offset */
+ die(FSCK_UNCORRECTED, 0, "special file has non-zero offset: %s", path);
+ }
if (S_ISCHR(i->mode)) {
devtype = i->size;
type = 'c';
@@ -264,13 +581,21 @@
devtype = i->size;
type = 'b';
}
- else if (S_ISFIFO(i->mode))
+ else if (S_ISFIFO(i->mode)) {
+ if (i->size != 0) {
+ die(FSCK_UNCORRECTED, 0, "fifo has non-zero size: %s", path);
+ }
type = 'p';
- else if (S_ISSOCK(i->mode))
+ }
+ else if (S_ISSOCK(i->mode)) {
+ if (i->size != 0) {
+ die(FSCK_UNCORRECTED, 0, "socket has non-zero size: %s", path);
+ }
type = 's';
+ }
else {
- fprintf(stderr, "%s: bogus mode on `%s' (%o)\n", filename, path, i->mode);
- exit(4);
+ die(FSCK_UNCORRECTED, 0, "bogus mode: %s (%o)", path, i->mode);
+ return; /* not reached */
}

if (opt_verbose) {
@@ -279,155 +604,62 @@

if (opt_extract) {
if (mknod(path, i->mode, devtype) < 0) {
- perror(path);
- exit(8);
+ die(FSCK_ERROR, 1, "mknod failed: %s", path);
}
change_file_status(path, i);
}
}

-static void do_uncompress(int fd, unsigned long offset, unsigned long size)
+static void expand_fs(char *path, struct cramfs_inode *inode)
{
- unsigned long curr = offset + 4 * ((size + PAGE_CACHE_SIZE - 1) / PAGE_CACHE_SIZE);
-
- do {
- unsigned long out = PAGE_CACHE_SIZE;
- unsigned long next = *(u32 *) romfs_read(offset);
-
- if (next > end_data) {
- end_data = next;
- }
-
- offset += 4;
- if (curr == next) {
- if (opt_verbose > 1) {
- printf(" hole at %ld (%d)\n", curr, PAGE_CACHE_SIZE);
- }
- if (size < PAGE_CACHE_SIZE)
- out = size;
- memset(outbuffer, 0x00, out);
- }
- else {
- if (opt_verbose > 1) {
- printf(" uncompressing block at %ld to %ld (%ld)\n", curr, next, next - curr);
- }
- out = uncompress_block(romfs_read(curr), next - curr);
- }
- if (size >= PAGE_CACHE_SIZE) {
- if (out != PAGE_CACHE_SIZE) {
- fprintf(stderr, "%s: Non-block (%ld) bytes\n", filename, out);
- exit(4);
- }
- } else {
- if (out != size) {
- fprintf(stderr, "%s: Non-size (%ld vs %ld) bytes\n", filename, out, size);
- exit(4);
- }
- }
- size -= out;
- if (opt_extract) {
- write(fd, outbuffer, out);
- }
- curr = next;
- } while (size);
+ if (S_ISDIR(inode->mode)) {
+ do_directory(path, inode);
+ }
+ else if (S_ISREG(inode->mode)) {
+ do_file(path, inode);
+ }
+ else if (S_ISLNK(inode->mode)) {
+ do_symlink(path, inode);
+ }
+ else {
+ do_special_inode(path, inode);
+ }
}

-static void expand_fs(int pathlen, char *path, struct cramfs_inode *inode)
+static void test_fs(int start)
{
- if (S_ISDIR(inode->mode)) {
- int count = inode->size;
- unsigned long offset = inode->offset << 2;
- char *newpath = malloc(pathlen + 256);
-
- if (count > 0 && offset < start_inode) {
- start_inode = offset;
- }
- /* XXX - need to check end_inode for empty case? */
- memcpy(newpath, path, pathlen);
- newpath[pathlen] = '/';
- pathlen++;
- if (opt_verbose) {
- print_node('d', inode, path);
- }
- if (opt_extract) {
- mkdir(path, inode->mode);
- change_file_status(path, inode);
- }
- while (count > 0) {
- struct cramfs_inode *child = iget(offset);
- int size;
- int newlen = child->namelen << 2;
-
- size = sizeof(struct cramfs_inode) + newlen;
- count -= size;
-
- offset += sizeof(struct cramfs_inode);
-
- memcpy(newpath + pathlen, romfs_read(offset), newlen);
- newpath[pathlen + newlen] = 0;
- if ((pathlen + newlen) - strlen(newpath) > 3) {
- fprintf(stderr, "%s: invalid cramfs--bad path length\n", filename);
- exit(4);
- }
- expand_fs(strlen(newpath), newpath, child);
-
- offset += newlen;
-
- if (offset > end_inode) {
- end_inode = offset;
- }
- }
- return;
- }
- if (S_ISREG(inode->mode)) {
- int fd = 0;
- unsigned long offset = inode->offset << 2;
+ struct cramfs_inode *root;

- if (offset > 0 && offset < start_data) {
- start_data = offset;
- }
- if (opt_verbose) {
- print_node('f', inode, path);
- }
- if (opt_extract) {
- fd = open(path, O_WRONLY | O_CREAT | O_TRUNC, inode->mode);
- }
- if (inode->size) {
- do_uncompress(fd, offset, inode->size);
+ root = read_super();
+ umask(0);
+ euid = geteuid();
+ stream.next_in = NULL;
+ stream.avail_in = 0;
+ inflateInit(&stream);
+ expand_fs(extract_dir, root);
+ inflateEnd(&stream);
+ if (start_data != ~0UL) {
+ if (start_data < (sizeof(struct cramfs_super) + start)) {
+ die(FSCK_UNCORRECTED, 0, "directory data start (%ld) < sizeof(struct cramfs_super) + start (%ld)", start_data, sizeof(struct cramfs_super) + start);
}
- if (opt_extract) {
- close(fd);
- change_file_status(path, inode);
+ if (end_dir != start_data) {
+ die(FSCK_UNCORRECTED, 0, "directory data end (%ld) != file data start (%ld)", end_dir, start_data);
}
- return;
}
- if (S_ISLNK(inode->mode)) {
- unsigned long offset = inode->offset << 2;
-
- if (offset < start_data) {
- start_data = offset;
+ if (super.flags & CRAMFS_FLAG_FSID_VERSION_2) {
+ if (end_data > super.size) {
+ die(FSCK_UNCORRECTED, 0, "invalid file data offset");
}
- do_symlink(path, inode);
- return;
- }
- else {
- do_special_inode(path, inode);
- return;
}
+ iput(root); /* free(root) */
}
#endif /* INCLUDE_FS_TESTS */

int main(int argc, char **argv)
{
- void *buf;
- size_t length;
- struct stat st;
- u32 crc_old, crc_new;
-#ifdef INCLUDE_FS_TESTS
- struct cramfs_inode *root;
-#endif /* INCLUDE_FS_TESTS */
int c; /* for getopt */
int start = 0;
+ size_t length;

if (argc)
progname = argv[0];
@@ -436,17 +668,14 @@
while ((c = getopt(argc, argv, "hx:v")) != EOF) {
switch (c) {
case 'h':
- usage(0);
+ usage(FSCK_OK);
case 'x':
#ifdef INCLUDE_FS_TESTS
opt_extract = 1;
- extract_dir = malloc(strlen(optarg) + 1);
- strcpy(extract_dir, optarg);
+ extract_dir = optarg;
break;
-#else /* not INCLUDE_FS_TESTS */
- fprintf(stderr, "%s: compiled without -x support\n",
- progname);
- exit(16);
+#else /* not INCLUDE_FS_TESTS */
+ die(FSCK_USAGE, 0, "compiled without -x support");
#endif /* not INCLUDE_FS_TESTS */
case 'v':
opt_verbose++;
@@ -455,134 +684,24 @@
}

if ((argc - optind) != 1)
- usage(16);
+ usage(FSCK_USAGE);
filename = argv[optind];

- /* find the physical size of the file or block device */
- if (lstat(filename, &st) < 0) {
- perror(filename);
- exit(8);
- }
- fd = open(filename, O_RDONLY);
- if (fd < 0) {
- perror(filename);
- exit(8);
- }
- if (S_ISBLK(st.st_mode)) {
- if (ioctl(fd, BLKGETSIZE, &length) < 0) {
- fprintf(stderr, "%s: warning--unable to determine filesystem size \n", filename);
- exit(4);
- }
- length = length * 512;
- }
- else if (S_ISREG(st.st_mode)) {
- length = st.st_size;
- }
- else {
- fprintf(stderr, "%s is not a block device or file\n", filename);
- exit(8);
- }
-
- if (length < sizeof(struct cramfs_super)) {
- fprintf(stderr, "%s: invalid cramfs--file length too short\n", filename);
- exit(4);
- }
-
- if (S_ISBLK(st.st_mode)) {
- /* nasty because mmap of block devices fails */
- buf = mmap(NULL, length, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);
- read(fd, buf, length);
- }
- else {
- /* nice and easy */
- buf = mmap(NULL, length, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
- }
-
- /* XXX - this could be cleaner... */
- if (((struct cramfs_super *) buf)->magic == CRAMFS_MAGIC) {
- start = 0;
- super = (struct cramfs_super *) buf;
- }
- else if (length >= (PAD_SIZE + sizeof(struct cramfs_super)) &&
- ((((struct cramfs_super *) (buf + PAD_SIZE))->magic == CRAMFS_MAGIC)))
- {
- start = PAD_SIZE;
- super = (struct cramfs_super *) (buf + PAD_SIZE);
- }
- else {
- fprintf(stderr, "%s: invalid cramfs--wrong magic\n", filename);
- exit(4);
- }
-
- if (super->flags & CRAMFS_FLAG_FSID_VERSION_2) {
- /* length test */
- if (length < super->size) {
- fprintf(stderr, "%s: invalid cramfs--file length too short\n", filename);
- exit(4);
- }
- else if (length > super->size) {
- fprintf(stderr, "%s: warning--file length too long, padded image?\n", filename);
- }
-
- /* CRC test */
- crc_old = super->fsid.crc;
- super->fsid.crc = crc32(0L, Z_NULL, 0);
- crc_new = crc32(0L, Z_NULL, 0);
- crc_new = crc32(crc_new, (unsigned char *) buf+start, super->size - start);
- if (crc_new != crc_old) {
- fprintf(stderr, "%s: invalid cramfs--crc error\n", filename);
- exit(4);
- }
- }
- else {
- fprintf(stderr, "%s: warning--old cramfs image, no CRC\n",
- filename);
- }
-
+ test_super(&start, &length);
+ test_crc(start);
#ifdef INCLUDE_FS_TESTS
- super = (struct cramfs_super *) malloc(sizeof(struct cramfs_super));
- if (((struct cramfs_super *) buf)->magic == CRAMFS_MAGIC) {
- memcpy(super, buf, sizeof(struct cramfs_super));
- }
- else if (length >= (PAD_SIZE + sizeof(struct cramfs_super)) &&
- ((((struct cramfs_super *) (buf + PAD_SIZE))->magic == CRAMFS_MAGIC)))
- {
- memcpy(super, (buf + PAD_SIZE), sizeof(struct cramfs_super));
- }
-
- munmap(buf, length);
-
- /* file format test, uses fake "blocked" accesses */
- root = read_super();
- umask(0);
- euid = geteuid();
- if (!root) {
- fprintf(stderr, "%s: invalid cramfs--bad superblock\n",
- filename);
- exit(4);
- }
- stream.next_in = NULL;
- stream.avail_in = 0;
- inflateInit(&stream);
-
- if (!extract_dir) {
- extract_dir = "root";
- }
-
- expand_fs(strlen(extract_dir), extract_dir, root);
- inflateEnd(&stream);
+ test_fs(start);
+#endif /* INCLUDE_FS_TESTS */

- if (start_data != 1 << 28 && end_inode != start_data) {
- fprintf(stderr, "%s: invalid cramfs--directory data end (%ld) != file data start (%ld)\n", filename, end_inode, start_data);
- exit(4);
- }
- if (super->flags & CRAMFS_FLAG_FSID_VERSION_2) {
- if (end_data > super->size) {
- fprintf(stderr, "%s: invalid cramfs--invalid file data offset\n", filename);
- exit(4);
- }
+ if (opt_verbose) {
+ printf("%s: OK\n", filename);
}
-#endif /* INCLUDE_FS_TESTS */

- exit(0);
+ exit(FSCK_OK);
}
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */
diff -ur linux-orig/scripts/cramfs/mkcramfs.c linux/scripts/cramfs/mkcramfs.c
--- linux-orig/scripts/cramfs/mkcramfs.c Thu Jul 19 16:14:53 2001
+++ linux/scripts/cramfs/mkcramfs.c Sun Jan 13 21:40:58 2002
@@ -1,7 +1,7 @@
/*
* mkcramfs - make a cramfs file system
*
- * Copyright (C) 1999-2001 Transmeta Corporation
+ * Copyright (C) 1999-2002 Transmeta Corporation
*
* 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
@@ -18,6 +18,10 @@
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
*/

+/*
+ * If you change the disk format of cramfs, please update fs/cramfs/README.
+ */
+
#include <sys/types.h>
#include <stdio.h>
#include <sys/stat.h>
@@ -28,39 +32,47 @@
#include <stdlib.h>
#include <errno.h>
#include <string.h>
-#include <assert.h>
#include <getopt.h>
+#include <stdarg.h>
#include <linux/cramfs_fs.h>
#include <zlib.h>

-#define PAD_SIZE 512 /* only 0 and 512 supported by kernel */
-
-static const char *progname = "mkcramfs";
+/* Exit codes used by mkfs-type programs */
+#define MKFS_OK 0 /* No errors */
+#define MKFS_ERROR 8 /* Operational error */
+#define MKFS_USAGE 16 /* Usage or syntax error */

-/* N.B. If you change the disk format of cramfs, please update fs/cramfs/README. */
+/* The kernel only supports PAD_SIZE of 0 and 512. */
+#define PAD_SIZE 512

-/* Input status of 0 to print help and exit without an error. */
-static void usage(int status)
-{
- FILE *stream = status ? stderr : stdout;
+/* The kernel assumes PAGE_CACHE_SIZE as block size. */
+#define PAGE_CACHE_SIZE (4096)

- fprintf(stream, "usage: %s [-h] [-e edition] [-i file] [-n name] dirname outfile\n"
- " -h print this help\n"
- " -E make all warnings errors (non-zero exit status)\n"
- " -e edition set edition number (part of fsid)\n"
- " -i file insert a file image into the filesystem (requires >= 2.4.0)\n"
- " -n name set name of cramfs filesystem\n"
- " -p pad by %d bytes for boot code\n"
- " -s sort directory entries (old option, ignored)\n"
- " -z make explicit holes (requires >= 2.3.39)\n"
- " dirname root of the filesystem to be compressed\n"
- " outfile output file\n", progname, PAD_SIZE);
+/*
+ * The longest filename component to allow for in the input directory tree.
+ * ext2fs (and many others) allow up to 255 bytes. A couple of filesystems
+ * allow longer (e.g. smbfs 1024), but there isn't much use in supporting
+ * >255-byte names in the input directory tree given that such names get
+ * truncated to CRAMFS_MAXPATHLEN (252 bytes) when written to cramfs.
+ *
+ * Old versions of mkcramfs generated corrupted filesystems if any input
+ * filenames exceeded CRAMFS_MAXPATHLEN (252 bytes), however old
+ * versions of cramfsck seem to have been able to detect the corruption.
+ */
+#define MAX_INPUT_NAMELEN 255

- exit(status);
-}
+/*
+ * Maximum size fs you can create is roughly 256MB. (The last file's
+ * data must begin within 256MB boundary but can extend beyond that.)
+ *
+ * Note that if you want it to fit in a ROM then you're limited to what the
+ * hardware and kernel can support.
+ */
+#define MAXFSLEN ((((1 << CRAMFS_OFFSET_WIDTH) - 1) << 2) /* offset */ \
+ + (1 << CRAMFS_SIZE_WIDTH) - 1 /* filesize */ \
+ + (1 << CRAMFS_SIZE_WIDTH) * 4 / PAGE_CACHE_SIZE /* block pointers */ )

-#define PAGE_CACHE_SIZE (4096)
-/* The kernel assumes PAGE_CACHE_SIZE as block size. */
+static const char *progname = "mkcramfs";
static unsigned int blksize = PAGE_CACHE_SIZE;
static long total_blocks = 0, total_nodes = 1; /* pre-count the root node */
static int image_length = 0;
@@ -72,31 +84,30 @@
*
* Note that kernels up to at least 2.3.39 don't support cramfs holes,
* which is why this is turned off by default.
+ *
+ * If opt_verbose is 1, be verbose. If it is higher, be even more verbose.
*/
-static int opt_edition = 0;
+static u32 opt_edition = 0;
static int opt_errors = 0;
static int opt_holes = 0;
static int opt_pad = 0;
+static int opt_verbose = 0;
static char *opt_image = NULL;
static char *opt_name = NULL;

static int warn_dev, warn_gid, warn_namelen, warn_skip, warn_size, warn_uid;

-#ifndef MIN
-# define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
-#endif
-
/* In-core version of inode / directory entry. */
struct entry {
/* stats */
- char *name;
+ unsigned char *name;
unsigned int mode, size, uid, gid;

/* FS data */
void *uncompressed;
- /* points to other identical file */
- struct entry *same;
- unsigned int offset; /* pointer to compressed data in archive */
+ /* points to other identical file */
+ struct entry *same;
+ unsigned int offset; /* pointer to compressed data in archive */
unsigned int dir_offset; /* Where in the archive is the directory entry? */

/* organization */
@@ -104,34 +115,67 @@
struct entry *next;
};

-/*
- * The longest file name component to allow for in the input directory tree.
- * Ext2fs (and many others) allow up to 255 bytes. A couple of filesystems
- * allow longer (e.g. smbfs 1024), but there isn't much use in supporting
- * >255-byte names in the input directory tree given that such names get
- * truncated to 255 bytes when written to cramfs.
- */
-#define MAX_INPUT_NAMELEN 255
+/* Input status of 0 to print help and exit without an error. */
+static void usage(int status)
+{
+ FILE *stream = status ? stderr : stdout;
+
+ fprintf(stream, "usage: %s [-h] [-e edition] [-i file] [-n name] dirname outfile\n"
+ " -h print this help\n"
+ " -E make all warnings errors (non-zero exit status)\n"
+ " -e edition set edition number (part of fsid)\n"
+ " -i file insert a file image into the filesystem (requires >= 2.4.0)\n"
+ " -n name set name of cramfs filesystem\n"
+ " -p pad by %d bytes for boot code\n"
+ " -s sort directory entries (old option, ignored)\n"
+ " -v be more verbose\n"
+ " -z make explicit holes (requires >= 2.3.39)\n"
+ " dirname root of the directory tree to be compressed\n"
+ " outfile output file\n", progname, PAD_SIZE);
+
+ exit(status);
+}
+
+static void die(int status, int syserr, const char *fmt, ...)
+{
+ va_list arg_ptr;
+ int save = errno;
+
+ fflush(0);
+ va_start(arg_ptr, fmt);
+ fprintf(stderr, "%s: ", progname);
+ vfprintf(stderr, fmt, arg_ptr);
+ if (syserr) {
+ fprintf(stderr, ": %s", strerror(save));
+ }
+ fprintf(stderr, "\n");
+ va_end(arg_ptr);
+ exit(status);
+}

static int find_identical_file(struct entry *orig,struct entry *newfile)
{
- if(orig==newfile) return 1;
- if(!orig) return 0;
- if(orig->size==newfile->size && orig->uncompressed && !memcmp(orig->uncompressed,newfile->uncompressed,orig->size)) {
- newfile->same=orig;
- return 1;
- }
- return find_identical_file(orig->child,newfile) ||
- find_identical_file(orig->next,newfile);
+ if (orig == newfile)
+ return 1;
+ if (!orig)
+ return 0;
+ if (orig->size == newfile->size && orig->uncompressed &&
+ !memcmp(orig->uncompressed, newfile->uncompressed, orig->size))
+ {
+ newfile->same = orig;
+ return 1;
+ }
+ return (find_identical_file(orig->child,newfile) ||
+ find_identical_file(orig->next,newfile));
}

static void eliminate_doubles(struct entry *root,struct entry *orig) {
- if(orig) {
- if(orig->size && orig->uncompressed)
+ if (orig) {
+ if (orig->size && orig->uncompressed)
find_identical_file(root,orig);
- eliminate_doubles(root,orig->child);
- eliminate_doubles(root,orig->next);
- }
+ eliminate_doubles(root,orig->child);
+ eliminate_doubles(root,orig->next);
+ }
}

/*
@@ -155,20 +199,18 @@
/* TODO: Reuse the parent's buffer to save memcpy'ing and duplication. */
path = malloc(len + 1 + MAX_INPUT_NAMELEN + 1);
if (!path) {
- perror(NULL);
- exit(8);
+ die(MKFS_ERROR, 1, "malloc failed");
}
memcpy(path, name, len);
endpath = path + len;
*endpath = '/';
endpath++;

- /* read in the directory and sort */
- dircount = scandir(name, &dirlist, 0, cramsort);
+ /* read in the directory and sort */
+ dircount = scandir(name, &dirlist, 0, cramsort);

if (dircount < 0) {
- perror(name);
- exit(8);
+ die(MKFS_ERROR, 1, "scandir failed: %s", name);
}

/* process directory */
@@ -192,36 +234,38 @@
}
namelen = strlen(dirent->d_name);
if (namelen > MAX_INPUT_NAMELEN) {
- fprintf(stderr,
- "Very long (%u bytes) filename `%s' found.\n"
- " Please increase MAX_INPUT_NAMELEN in mkcramfs.c and recompile. Exiting.\n",
+ die(MKFS_ERROR, 0,
+ "very long (%u bytes) filename found: %s\n"
+ "please increase MAX_INPUT_NAMELEN in mkcramfs.c and recompile",
namelen, dirent->d_name);
- exit(8);
}
memcpy(endpath, dirent->d_name, namelen + 1);

if (lstat(path, &st) < 0) {
- perror(endpath);
warn_skip = 1;
continue;
}
entry = calloc(1, sizeof(struct entry));
if (!entry) {
- perror(NULL);
- exit(8);
+ die(MKFS_ERROR, 1, "calloc failed");
}
entry->name = strdup(dirent->d_name);
if (!entry->name) {
- perror(NULL);
- exit(8);
+ die(MKFS_ERROR, 1, "strdup failed");
}
- if (namelen > 255) {
- /* Can't happen when reading from ext2fs. */
-
- /* TODO: we ought to avoid chopping in half
- multi-byte UTF8 characters. */
- entry->name[namelen = 255] = '\0';
+ /* truncate multi-byte UTF-8 filenames on character boundary */
+ if (namelen > CRAMFS_MAXPATHLEN) {
+ namelen = CRAMFS_MAXPATHLEN;
warn_namelen = 1;
+ /* the first lost byte must not be a trail byte */
+ while ((entry->name[namelen] & 0xc0) == 0x80) {
+ namelen--;
+ /* are we reasonably certain it was UTF-8 ? */
+ if (entry->name[namelen] < 0x80 || !namelen) {
+ die(MKFS_ERROR, 0, "cannot truncate filenames not encoded in UTF-8");
+ }
+ }
+ entry->name[namelen] = '\0';
}
entry->mode = st.st_mode;
entry->size = st.st_size;
@@ -231,10 +275,10 @@
entry->gid = st.st_gid;
if (entry->gid >= 1 << CRAMFS_GID_WIDTH)
/* TODO: We ought to replace with a default
- gid instead of truncating; otherwise there
- are security problems. Maybe mode should
- be &= ~070. Same goes for uid once Linux
- supports >16-bit uids. */
+ gid instead of truncating; otherwise there
+ are security problems. Maybe mode should
+ be &= ~070. Same goes for uid once Linux
+ supports >16-bit uids. */
warn_gid = 1;
size = sizeof(struct cramfs_inode) + ((namelen + 3) & ~3);
*fslen_ub += size;
@@ -252,7 +296,6 @@
(which would just save a little memory). */
int fd = open(path, O_RDONLY);
if (fd < 0) {
- perror(path);
warn_skip = 1;
continue;
}
@@ -263,39 +306,38 @@
}

entry->uncompressed = mmap(NULL, entry->size, PROT_READ, MAP_PRIVATE, fd, 0);
- if (-1 == (int) (long) entry->uncompressed) {
- perror("mmap");
- exit(8);
+ if (entry->uncompressed == MAP_FAILED) {
+ die(MKFS_ERROR, 1, "mmap failed");
}
}
close(fd);
} else if (S_ISLNK(st.st_mode)) {
entry->uncompressed = malloc(entry->size);
if (!entry->uncompressed) {
- perror(NULL);
- exit(8);
+ die(MKFS_ERROR, 1, "malloc failed");
}
if (readlink(path, entry->uncompressed, entry->size) < 0) {
- perror(path);
warn_skip = 1;
continue;
}
} else if (S_ISFIFO(st.st_mode) || S_ISSOCK(st.st_mode)) {
/* maybe we should skip sockets */
entry->size = 0;
- } else {
+ } else if (S_ISCHR(st.st_mode) || S_ISBLK(st.st_mode)) {
entry->size = st.st_rdev;
if (entry->size & -(1<<CRAMFS_SIZE_WIDTH))
warn_dev = 1;
+ } else {
+ die(MKFS_ERROR, 0, "bogus file type: %s", entry->name);
}

if (S_ISREG(st.st_mode) || S_ISLNK(st.st_mode)) {
int blocks = ((entry->size - 1) / blksize + 1);

/* block pointers & data expansion allowance + data */
- if(entry->size)
+ if (entry->size)
*fslen_ub += (4+26)*blocks + entry->size + 3;
- }
+ }

/* Link it into the list */
*prev = entry;
@@ -313,9 +355,7 @@
struct cramfs_super *super = (struct cramfs_super *) base;
unsigned int offset = sizeof(struct cramfs_super) + image_length;

- if (opt_pad) {
- offset += opt_pad;
- }
+ offset += opt_pad; /* 0 if no padding */

super->magic = CRAMFS_MAGIC;
super->flags = CRAMFS_FLAG_FSID_VERSION_2 | CRAMFS_FLAG_SORTED_DIRS;
@@ -349,27 +389,66 @@
static void set_data_offset(struct entry *entry, char *base, unsigned long offset)
{
struct cramfs_inode *inode = (struct cramfs_inode *) (base + entry->dir_offset);
-#ifdef DEBUG
- assert ((offset & 3) == 0);
-#endif /* DEBUG */
+
+ if ((offset & 3) != 0) {
+ die(MKFS_ERROR, 0, "illegal offset of %lu bytes", offset);
+ }
if (offset >= (1 << (2 + CRAMFS_OFFSET_WIDTH))) {
- fprintf(stderr, "filesystem too big. Exiting.\n");
- exit(8);
+ die(MKFS_ERROR, 0, "filesystem too big");
}
inode->offset = (offset >> 2);
}

+/*
+ * TODO: Does this work for chars >= 0x80? Most filesystems use UTF-8
+ * encoding for filenames, whereas the console is a single-byte
+ * character set like iso-latin-1.
+ */
+static void print_node(struct entry *e)
+{
+ char info[10];
+ char type = '?';
+
+ if (S_ISREG(e->mode)) type = 'f';
+ else if (S_ISDIR(e->mode)) type = 'd';
+ else if (S_ISLNK(e->mode)) type = 'l';
+ else if (S_ISCHR(e->mode)) type = 'c';
+ else if (S_ISBLK(e->mode)) type = 'b';
+ else if (S_ISFIFO(e->mode)) type = 'p';
+ else if (S_ISSOCK(e->mode)) type = 's';
+
+ if (S_ISCHR(e->mode) || (S_ISBLK(e->mode))) {
+ /* major/minor numbers can be as high as 2^12 or 4096 */
+ snprintf(info, 10, "%4d,%4d", major(e->size), minor(e->size));
+ }
+ else {
+ /* size be as high as 2^24 or 16777216 */
+ snprintf(info, 10, "%9d", e->size);
+ }
+
+ printf("%c %04o %s %5d:%-3d %s\n",
+ type, e->mode & ~S_IFMT, info, e->uid, e->gid, e->name);
+}

/*
* We do a width-first printout of the directory
* entries, using a stack to remember the directories
* we've seen.
*/
-#define MAXENTRIES (100)
static unsigned int write_directory_structure(struct entry *entry, char *base, unsigned int offset)
{
int stack_entries = 0;
- struct entry *entry_stack[MAXENTRIES];
+ int stack_size = 64;
+ struct entry **entry_stack;
+
+ entry_stack = malloc(stack_size * sizeof(struct entry *));
+ if (!entry_stack) {
+ die(MKFS_ERROR, 1, "malloc failed");
+ }
+
+ if (opt_verbose) {
+ printf("root:\n");
+ }

for (;;) {
int dir_start = stack_entries;
@@ -398,15 +477,16 @@
inode->namelen = len >> 2;
offset += len;

- /* TODO: this may get it wrong for chars >= 0x80.
- Most filesystems use UTF8 encoding for filenames,
- whereas the console is a single-byte character
- set like iso-latin-1. */
- printf(" %s\n", entry->name);
+ if (opt_verbose)
+ print_node(entry);
+
if (entry->child) {
- if (stack_entries >= MAXENTRIES) {
- fprintf(stderr, "Exceeded MAXENTRIES. Raise this value in mkcramfs.c and recompile. Exiting.\n");
- exit(8);
+ if (stack_entries >= stack_size) {
+ stack_size *= 2;
+ entry_stack = realloc(entry_stack, stack_size * sizeof(struct entry *));
+ if (!entry_stack) {
+ die(MKFS_ERROR, 1, "realloc failed");
+ }
}
entry_stack[stack_entries] = entry;
stack_entries++;
@@ -416,9 +496,9 @@

/*
* Reverse the order the stack entries pushed during
- * this directory, for a small optimization of disk
- * access in the created fs. This change makes things
- * `ls -UR' order.
+ * this directory, for a small optimization of disk
+ * access in the created fs. This change makes things
+ * `ls -UR' order.
*/
{
struct entry **lo = entry_stack + dir_start;
@@ -439,29 +519,27 @@
entry = entry_stack[stack_entries];

set_data_offset(entry, base, offset);
- printf("'%s':\n", entry->name);
+ if (opt_verbose) {
+ printf("%s:\n", entry->name);
+ }
entry = entry->child;
}
+ free(entry_stack);
return offset;
}

static int is_zero(char const *begin, unsigned len)
{
- if (opt_holes)
- /* Returns non-zero iff the first LEN bytes from BEGIN are
- all NULs. */
- return (len-- == 0 ||
- (begin[0] == '\0' &&
- (len-- == 0 ||
- (begin[1] == '\0' &&
- (len-- == 0 ||
- (begin[2] == '\0' &&
- (len-- == 0 ||
- (begin[3] == '\0' &&
- memcmp(begin, begin + 4, len) == 0))))))));
- else
- /* Never create holes. */
- return 0;
+ /* Returns non-zero iff the first LEN bytes from BEGIN are all NULs. */
+ return (len-- == 0 ||
+ (begin[0] == '\0' &&
+ (len-- == 0 ||
+ (begin[1] == '\0' &&
+ (len-- == 0 ||
+ (begin[2] == '\0' &&
+ (len-- == 0 ||
+ (begin[3] == '\0' &&
+ memcmp(begin, begin + 4, len) == 0))))))));
}

/*
@@ -488,19 +566,23 @@
do {
unsigned long len = 2 * blksize;
unsigned int input = size;
+ int err;
+
if (input > blksize)
input = blksize;
size -= input;
- if (!is_zero (uncompressed, input)) {
- compress(base + curr, &len, uncompressed, input);
+ if (!(opt_holes && is_zero (uncompressed, input))) {
+ err = compress2(base + curr, &len, uncompressed, input, Z_BEST_COMPRESSION);
+ if (err != Z_OK) {
+ die(MKFS_ERROR, 0, "compression error: %s", zError(err));
+ }
curr += len;
}
uncompressed += input;

if (len > blksize*2) {
/* (I don't think this can happen with zlib.) */
- printf("AIEEE: block \"compressed\" to > 2*blocklength (%ld)\n", len);
- exit(8);
+ die(MKFS_ERROR, 0, "AIEEE: block \"compressed\" to > 2*blocklength (%ld)", len);
}

*(u32 *) (base + offset) = curr;
@@ -513,8 +595,10 @@
st_blocks * 512. But if you say that then perhaps
administrative data should also be included in both. */
change = new_size - original_size;
- printf("%6.2f%% (%+d bytes)\t%s\n",
- (change * 100) / (double) original_size, change, name);
+ if (opt_verbose > 1) {
+ printf("%6.2f%% (%+d bytes)\t%s\n",
+ (change * 100) / (double) original_size, change, name);
+ }

return curr;
}
@@ -529,18 +613,19 @@
{
do {
if (entry->uncompressed) {
- if(entry->same) {
- set_data_offset(entry, base, entry->same->offset);
- entry->offset=entry->same->offset;
- } else {
- set_data_offset(entry, base, offset);
- entry->offset=offset;
- offset = do_compress(base, offset, entry->name, entry->uncompressed, entry->size);
- }
+ if (entry->same) {
+ set_data_offset(entry, base, entry->same->offset);
+ entry->offset = entry->same->offset;
+ }
+ else {
+ set_data_offset(entry, base, offset);
+ entry->offset = offset;
+ offset = do_compress(base, offset, entry->name, entry->uncompressed, entry->size);
+ }
}
else if (entry->child)
offset = write_data(entry->child, base, offset);
- entry=entry->next;
+ entry=entry->next;
} while (entry);
return offset;
}
@@ -552,10 +637,12 @@

fd = open(file, O_RDONLY);
if (fd < 0) {
- perror(file);
- exit(8);
+ die(MKFS_ERROR, 1, "open failed: %s", file);
}
buf = mmap(NULL, image_length, PROT_READ, MAP_PRIVATE, fd, 0);
+ if (buf == MAP_FAILED) {
+ die(MKFS_ERROR, 1, "mmap failed");
+ }
memcpy(base + offset, buf, image_length);
munmap(buf, image_length);
close (fd);
@@ -567,27 +654,6 @@
return (offset + image_length);
}

-/*
- * Maximum size fs you can create is roughly 256MB. (The last file's
- * data must begin within 256MB boundary but can extend beyond that.)
- *
- * Note that if you want it to fit in a ROM then you're limited to what the
- * hardware and kernel can support (64MB?).
- */
-#define MAXFSLEN ((((1 << CRAMFS_OFFSET_WIDTH) - 1) << 2) /* offset */ \
- + (1 << CRAMFS_SIZE_WIDTH) - 1 /* filesize */ \
- + (1 << CRAMFS_SIZE_WIDTH) * 4 / PAGE_CACHE_SIZE /* block pointers */ )
-
-
-/*
- * Usage:
- *
- * mkcramfs directory-name outfile
- *
- * where "directory-name" is simply the root of the directory
- * tree that we want to generate a compressed filesystem out
- * of.
- */
int main(int argc, char **argv)
{
struct stat st; /* used twice... */
@@ -598,8 +664,9 @@
/* initial guess (upper-bound) of required filesystem size */
loff_t fslen_ub = sizeof(struct cramfs_super);
char const *dirname, *outfile;
- u32 crc = crc32(0L, Z_NULL, 0);
+ u32 crc;
int c; /* for getopt */
+ char *ep; /* for strtoul */

total_blocks = 0;

@@ -607,21 +674,23 @@
progname = argv[0];

/* command line options */
- while ((c = getopt(argc, argv, "hEe:i:n:psz")) != EOF) {
+ while ((c = getopt(argc, argv, "hEe:i:n:psvz")) != EOF) {
switch (c) {
case 'h':
- usage(0);
+ usage(MKFS_OK);
case 'E':
opt_errors = 1;
break;
case 'e':
- opt_edition = atoi(optarg);
+ errno = 0;
+ opt_edition = strtoul(optarg, &ep, 10);
+ if (errno || optarg[0] == '\0' || *ep != '\0')
+ usage(MKFS_USAGE);
break;
case 'i':
opt_image = optarg;
if (lstat(opt_image, &st) < 0) {
- perror(opt_image);
- exit(16);
+ die(MKFS_ERROR, 1, "lstat failed: %s", opt_image);
}
image_length = st.st_size; /* may be padded later */
fslen_ub += (image_length + 3); /* 3 is for padding */
@@ -636,6 +705,9 @@
case 's':
/* old option, ignored */
break;
+ case 'v':
+ opt_verbose++;
+ break;
case 'z':
opt_holes = 1;
break;
@@ -643,20 +715,21 @@
}

if ((argc - optind) != 2)
- usage(16);
+ usage(MKFS_USAGE);
dirname = argv[optind];
outfile = argv[optind + 1];

if (stat(dirname, &st) < 0) {
- perror(dirname);
- exit(16);
+ die(MKFS_USAGE, 1, "stat failed: %s", dirname);
}
fd = open(outfile, O_WRONLY | O_CREAT | O_TRUNC, 0666);
+ if (fd < 0) {
+ die(MKFS_USAGE, 1, "open failed: %s", outfile);
+ }

root_entry = calloc(1, sizeof(struct entry));
if (!root_entry) {
- perror(NULL);
- exit(8);
+ die(MKFS_ERROR, 1, "calloc failed");
}
root_entry->mode = st.st_mode;
root_entry->uid = st.st_uid;
@@ -665,35 +738,34 @@
root_entry->size = parse_directory(root_entry, dirname, &root_entry->child, &fslen_ub);

/* always allocate a multiple of blksize bytes because that's
- what we're going to write later on */
+ what we're going to write later on */
fslen_ub = ((fslen_ub - 1) | (blksize - 1)) + 1;

if (fslen_ub > MAXFSLEN) {
fprintf(stderr,
- "warning: guestimate of required size (upper bound) is %LdMB, but maximum image size is %uMB. We might die prematurely.\n",
+ "warning: estimate of required size (upper bound) is %LdMB, but maximum image size is %uMB, we might die prematurely\n",
fslen_ub >> 20,
MAXFSLEN >> 20);
fslen_ub = MAXFSLEN;
}

- /* find duplicate files. TODO: uses the most inefficient algorithm
- possible. */
- eliminate_doubles(root_entry,root_entry);
+ /* find duplicate files. TODO: uses the most inefficient algorithm
+ possible. */
+ eliminate_doubles(root_entry,root_entry);

/* TODO: Why do we use a private/anonymous mapping here
- followed by a write below, instead of just a shared mapping
- and a couple of ftruncate calls? Is it just to save us
- having to deal with removing the file afterwards? If we
- really need this huge anonymous mapping, we ought to mmap
- in smaller chunks, so that the user doesn't need nn MB of
- RAM free. If the reason is to be able to write to
- un-mmappable block devices, then we could try shared mmap
- and revert to anonymous mmap if the shared mmap fails. */
+ followed by a write below, instead of just a shared mapping
+ and a couple of ftruncate calls? Is it just to save us
+ having to deal with removing the file afterwards? If we
+ really need this huge anonymous mapping, we ought to mmap
+ in smaller chunks, so that the user doesn't need nn MB of
+ RAM free. If the reason is to be able to write to
+ un-mmappable block devices, then we could try shared mmap
+ and revert to anonymous mmap if the shared mmap fails. */
rom_image = mmap(NULL, fslen_ub?fslen_ub:1, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANONYMOUS, -1, 0);

- if (-1 == (int) (long) rom_image) {
- perror("ROM image map");
- exit(8);
+ if (rom_image == MAP_FAILED) {
+ die(MKFS_ERROR, 1, "mmap failed");
}

/* Skip the first opt_pad bytes for boot loader code */
@@ -715,7 +787,7 @@
offset = write_data(root_entry, rom_image, offset);

/* We always write a multiple of blksize bytes, so that
- losetup works. */
+ losetup works. */
offset = ((offset - 1) | (blksize - 1)) + 1;
printf("Everything: %d kilobytes\n", offset >> 10);

@@ -724,53 +796,58 @@
printf("Super block: %d bytes\n", sizeof(struct cramfs_super));

/* Put the checksum in. */
+ crc = crc32(0L, Z_NULL, 0);
crc = crc32(crc, (rom_image+opt_pad), (offset-opt_pad));
((struct cramfs_super *) (rom_image+opt_pad))->fsid.crc = crc;
printf("CRC: %x\n", crc);

/* Check to make sure we allocated enough space. */
if (fslen_ub < offset) {
- fprintf(stderr, "not enough space allocated for ROM image (%Ld allocated, %d used)\n",
- fslen_ub, offset);
- exit(8);
+ die(MKFS_ERROR, 0, "not enough space allocated for ROM image (%Ld allocated, %d used)", fslen_ub, offset);
}

written = write(fd, rom_image, offset);
if (written < 0) {
- perror("ROM image");
- exit(8);
+ die(MKFS_ERROR, 1, "write failed");
}
if (offset != written) {
- fprintf(stderr, "ROM image write failed (%d %d)\n", written, offset);
- exit(8);
+ die(MKFS_ERROR, 0, "ROM image write failed (wrote %d of %d bytes)", written, offset);
}

/* (These warnings used to come at the start, but they scroll off the
- screen too quickly.) */
- if (warn_namelen) /* (can't happen when reading from ext2fs) */
- fprintf(stderr, /* bytes, not chars: think UTF8. */
- "warning: filenames truncated to 255 bytes.\n");
+ screen too quickly.) */
+ if (warn_namelen)
+ fprintf(stderr, /* bytes, not chars: think UTF-8. */
+ "warning: filenames truncated to %d bytes (possibly less if multi-byte UTF-8)\n",
+ CRAMFS_MAXPATHLEN);
if (warn_skip)
- fprintf(stderr, "warning: files were skipped due to errors.\n");
+ fprintf(stderr, "warning: files were skipped due to errors\n");
if (warn_size)
fprintf(stderr,
- "warning: file sizes truncated to %luMB (minus 1 byte).\n",
+ "warning: file sizes truncated to %luMB (minus 1 byte)\n",
1L << (CRAMFS_SIZE_WIDTH - 20));
if (warn_uid) /* (not possible with current Linux versions) */
fprintf(stderr,
- "warning: uids truncated to %u bits. (This may be a security concern.)\n",
+ "warning: uids truncated to %u bits (this may be a security concern)\n",
CRAMFS_UID_WIDTH);
if (warn_gid)
fprintf(stderr,
- "warning: gids truncated to %u bits. (This may be a security concern.)\n",
+ "warning: gids truncated to %u bits (this may be a security concern)\n",
CRAMFS_GID_WIDTH);
if (warn_dev)
fprintf(stderr,
- "WARNING: device numbers truncated to %u bits. This almost certainly means\n"
- "that some device files will be wrong.\n",
+ "WARNING: device numbers truncated to %u bits (this almost certainly means\n"
+ "that some device files will be wrong)\n",
CRAMFS_OFFSET_WIDTH);
if (opt_errors &&
(warn_namelen||warn_skip||warn_size||warn_uid||warn_gid||warn_dev))
- exit(8);
- return 0;
+ exit(MKFS_ERROR);
+
+ exit(MKFS_OK);
}
+
+/*
+ * Local variables:
+ * c-file-style: "linux"
+ * End:
+ */

--
Daniel Quinlan (at work)
[email protected]


2002-01-17 01:21:46

by Tomasz Kłoczko

[permalink] [raw]
Subject: Re: [PATCH] cramfs updates for 2.4.17

On Wed, 16 Jan 2002, Daniel Quinlan wrote:
[..]
> mkcramfs.c
[..]

Why not move this tool to util-linux ?

kloczek
--
-----------------------------------------------------------
*Ludzie nie maj? problem?w, tylko sobie sami je stwarzaj?*
-----------------------------------------------------------
Tomasz K?oczko, sys adm @zie.pg.gda.pl|*e-mail: [email protected]*


2002-01-17 02:36:32

by Daniel Quinlan

[permalink] [raw]
Subject: Re: [PATCH] cramfs updates for 2.4.17

=?ISO-8859-2?Q?Tomasz=5FK=B3oczko?= writes:

> On Wed, 16 Jan 2002, Daniel Quinlan wrote:
> [..]
> > mkcramfs.c
> [..]
>
> Why not move this tool to util-linux ?

I'm actually ready to move the tools to sourceforge (the cramfs tools
CVS tree is there).

http://sourceforge.net/projects/cramfs/

Assuming it's okay with Linus and Marcelo, I'll remove scripts/cramfs
in the next version of the patch (which should be fine for 2.5 too).

Dan

2002-01-17 10:19:10

by Johan Adolfsson

[permalink] [raw]
Subject: Re: [PATCH] cramfs updates for 2.4.17


> =?ISO-8859-2?Q?Tomasz=5FK=B3oczko?= writes:
>
> > On Wed, 16 Jan 2002, Daniel Quinlan wrote:
> > [..]
> > > mkcramfs.c
> > [..]
> >
> > Why not move this tool to util-linux ?
>
> I'm actually ready to move the tools to sourceforge (the cramfs tools
> CVS tree is there).
>
> http://sourceforge.net/projects/cramfs/
>
> Assuming it's okay with Linus and Marcelo, I'll remove scripts/cramfs
> in the next version of the patch (which should be fine for 2.5 too).
>
> Dan

Why move it?
Having it in the tree makes it nice and easy to use it,
and easy keeping it in sync.
BTW: I have an improvement wish for cramfs:
Add the date of filesystemgeneration to the image header somewhere
and use that to update the time info in the inode, that way you could
get a cheap way of having some valid time stamp for the files without
it taking a lot of space (helps if you have a webserver and the browser
caches stuff)
/Johan


2002-01-17 12:32:30

by Erik Mouw

[permalink] [raw]
Subject: Re: [PATCH] cramfs updates for 2.4.17

On Thu, Jan 17, 2002 at 11:20:06AM +0100, Johan Adolfsson wrote:
> Daniel Quinlan wrote:
> > Assuming it's okay with Linus and Marcelo, I'll remove scripts/cramfs
> > in the next version of the patch (which should be fine for 2.5 too).
>
> Why move it?

Because leaving it in the kernel tree would be an invitation to move
mke2fs, mkreiserfs, mkxfsfs, etc. into the kernel source tree as well.


Erik

--
J.A.K. (Erik) Mouw, Information and Communication Theory Group, Faculty
of Information Technology and Systems, Delft University of Technology,
PO BOX 5031, 2600 GA Delft, The Netherlands Phone: +31-15-2783635
Fax: +31-15-2781843 Email: [email protected]
WWW: http://www-ict.its.tudelft.nl/~erik/

2002-01-18 08:10:26

by Horst von Brand

[permalink] [raw]
Subject: Re: [PATCH] cramfs updates for 2.4.17

Daniel Quinlan <[email protected]> said:

[...]

> fs/cramfs/README
> - add note about sorted directory entries
> - mkcramfs compile-time DO_HOLES option replaced by run-time -z option

I'd do it the other way around: No-holes is for backward compatibility, and
should someday go away. Let the legacy uses pay the price...
--
Horst von Brand http://counter.li.org # 22616