SquashFS 3.1-r2 mounting 1.3-r3 filesystem

shint mosn — Tue, 16 Mar 2021 16:16:34 -0000

None of the squashfs-gear (Mksquashfs and Unsquashfs) are tested on Windows platforms like this. I actually have seen reviews that they paintings beneath Cygwin, but I can not verify this.

Looking for advice on the xz operation mode used in SquashFS

Yu Sun — Thu, 31 Oct 2019 15:14:11 -0000

When attempting to use SquashFS multi-threaded per-cpu decompressor, certain memory increase was observed as expected.

Since xz decompression is being used, changing the mode from XZ_PREALLOC to XZ_DYNALLOC would reduce the memory footprint, while maintaining our performance gain from using SquashFS multi-threaded decompressor gained in our system.

I would like to seek your opinion to see if back then XZ_PREALLOC was chosen for a specific reason that switching to XZ_DYNALLOC would have potential risk that I overlooked.

Any advice or input is highly appreciated!

Squashfs version 3.76 detected, only version 4.0 supported.

mikael monnier — Wed, 14 Nov 2018 00:48:20 -0000

Hi, everybody,

I have tried to mount a configuration file from a firmware of a cable/routeur.
Is the version 3.76 supported by any version of squashfs I can found around here ?
Thanks

Squashfs support for eMMC

balaji — Wed, 10 Jan 2018 10:08:13 -0000

Hi,
Is it preferarable to use squashfs for eMMC's read only file systemAre there any userspace utilities for squashfs debugging( like debugfs).Which compressor is best fit to reduce the impact on CPU,Memory and eMMC.

Alternatives to Squashfs (due to lack of uBoot support)

Colin Wernham — Thu, 15 Jun 2017 12:34:09 -0000

I have an eMMC device for which I require a compressed read-only filesystem that uBoot can access to load Kernel and DTB files from. Unfortunately, there is no u-boot support anywhere for SquashFS, so my question is:

What alternatives are possible? Remember, this is for eMMC, not NAND.

UBIFS is possible, but can this work on eMMC without UBI layer, and can uBoot load this without attaching UBI first?

squashfs4 byte order support

Alexandre Bouin — Wed, 12 Apr 2017 10:00:01 -0000

Last fix has to be done in get_block_length().
*length must have the correct endianess.

squashfs4 byte order support

Alexandre Bouin — Fri, 31 Mar 2017 08:14:38 -0000

Hi, I converted all leXX_to_cpu from fs//squashfs and mtdpart/mtdpart.c.
I get a crash in bigendian.
Any help woul appreciated ...

squashfs4 byte order support

Alexandre Bouin — Tue, 28 Mar 2017 10:05:49 -0000

Hi all,

In our company we are using squashfs4 with gzip and lzma compression to boot an OpenWRT file system, but our system is compiled in big endian, with a little endian squashfs filesystem. We would like to check it does not slow own the system by generating big endian squashfs filesystem.

On compilation, I observed mksquashfs4 is compiled against hist /usr/include headers. This results in having a mksqushfs generating archives with the same byte order than the host.
I hacked the BYTE_ORDER check to give my own byte order value, independantely of hosts current byte order. It enables having cross byte order archives (target byte order is no the same as host byte order).
On the other hand, on boot, kernel doesn't recognizes this archive. squashfs4 driver doesn't seems to know anything about endianness.

I may be wrong somewhere.
So I would appreiate any help by fixing my assumptions or giving some tips to get cross byte ordered filesystem.

Thanks by advances,
Alex.

BUG: I/O error reading the last fragment of a large file.

Jim Conyngham — Sun, 30 Oct 2016 22:36:48 -0000

OS: Ubuntu 14.04LTS
Kernel: 4.4.0-45-generic
mksquashfs version 4.3 (2014/05/12)
copyright (C) 2014 Phillip Lougher phillip@squashfs.org.uk

I am new to squashfs. I wanted to create a backup image of a large (8 TiB), but mostly empty hard drive on a much smaller drive, so I downloaded the latest squashfs from sourceforge and did make install without much ado.

Since I was unfamilar with squashfs, I wanted to do some experiments before trying such a large and long-running compression. I wrote a small C++ program to generate some compressable pseudorandom text for testing. (I'll paste the source below.)

After a few false starts, I got mksquashfs to run without any errors, compressing a pseudo-file input of 4000 MiB. There were no error messages generated by mksquashfs. But when I mount the resulting image file and try to verify the data, I get input/output errors reading the last 128K fragment of the file.

Console log:

./somedata 4000 | tee >(md5sum > tmp.md5) >(sha1sum > tmp.sha1) | pv -ptres 4000m | mksquashfs emptydir/ squash.img -no-progress -p "image.dat f 444 jim jim cat"

This appears to work, but … mksquashfs creates the file with no error messages, but when you mount the squashfs, the very last block on the file cannot be read! [~/Downloads/squashfs/fubar/console_log_20161030A.txt]

$ ./somedata 4000 | tee >(md5sum > tmp.md5) >(sha1sum > tmp.sha1) | pv -ptres 4000m | mksquashfs emptydir/ squash.img -no-progress -p "image.dat f 444 jim jim cat"
Parallel mksquashfs: Using 4 processors
Creating 4.0 filesystem on squash.img, block size 131072.
0:02:01 [32.8MB/s] [==============================================================================================>] 100%

Exportable Squashfs 4.0 filesystem, gzip compressed, data block size 131072
compressed data, compressed metadata, compressed fragments, compressed xattrs
duplicates are removed
Filesystem size 589646.09 Kbytes (575.83 Mbytes)
14.40% of uncompressed filesystem size (4096125.27 Kbytes)
Inode table size 58231 bytes (56.87 Kbytes)
45.46% of uncompressed inode table size (128092 bytes)
Directory table size 29 bytes (0.03 Kbytes)
93.55% of uncompressed directory table size (31 bytes)
Number of duplicate files found 0
Number of inodes 2
Number of files 1
Number of fragments 1
Number of symbolic links 0
Number of device nodes 0
Number of fifo nodes 0
Number of socket nodes 0
Number of directories 1
Number of ids (unique uids + gids) 1
Number of uids 1
jim (1000)
Number of gids 1
jim (1000)
$ ls -l squash.img
-rw-r--r-- 1 jim jim 603799552 Oct 30 16:42 squash.img
$ sudo mount squash.img ./mnt -t squashfs -o loop
mount: warning: ./mnt seems to be mounted read-only.
$ ls -l mnt
total 4096000
-r--r--r-- 1 jim jim 4194304000 Oct 30 16:40 image.dat
$ md5sum mnt/image.dat
md5sum: mnt/image.dat: Input/output error
$ sha1sum mnt/image.dat
sha1sum: mnt/image.dat: Input/output error
$ dd if=mnt/image.dat of=/dev/null bs=1048576
dd: error reading ‘mnt/image.dat’: Input/output error
3999+1 records in
3999+1 records out
4194172928 bytes (4.2 GB) copied, 0.894814 s, 4.7 GB/s
$ dd if=mnt/image.dat of=/dev/null bs=131072
dd: error reading ‘mnt/image.dat’: Input/output error
31999+0 records in
31999+0 records out
4194172928 bytes (4.2 GB) copied, 0.815427 s, 5.1 GB/s
$ sudo umount ./mnt

.. where "./somedata 4000" write 4000 MiB of pseudorandom text to stdout.
md5sum: 2f809ee5927be88545390f877ac5b416 -
sha1sum: 1784be1a0e2c0888e16be82c7eea459e77c4df9f -

Source file somedata.cpp:

include <stdio.h>

include <stdlib.h>

include <string.h>

int usage ()
{
fprintf (stdout,
"\n"
"usage: somedata # | ...\n"
"\n"
"# - how many MiB of text to output\n"
"\n"
"This is a simple test data generator for testing a compression utility\n"
"like mksquashfs. The output is a pseudorandom sequence of ASCII characters\n"
"that should have a low information content and compress with high ratio.\n"
"\n");
return 1;
}

struct options_t
{
unsigned numMiB = 0;

bool parse (int argc, char ** argv)
{
    if (2 != argc)  return false;
    char * tailptr = nullptr;
    unsigned tmp = strtoul (argv[1], &tailptr, 10);
    if (((!tailptr) || (0 == strlen(tailptr))) 
    &&  (0 < tmp))  {
        numMiB = tmp;
        return true;
    }
    fprintf (stderr, "Invalid numeric value: %s\n", (const char *) argv[1]);
    return false;
}

} options;

int main (int argc, char ** argv)
{
if (!options.parse (argc, argv))
return usage();

enum { somebits = 0x55 };
static_assert (somebits == (RAND_MAX & somebits), "RAND_MAX problem");
char c = 'a';
for (unsigned n = (options.numMiB << 20) - 1; n-- > 0; )   {
    int r = rand();
    if (0 == (r & 0x3f)) {
        fputc ('\n', stdout);
        continue;
    }
    if (0 == (r & somebits))
        c = ' ' + int (float (rand()) * (float (0x5f) / float(RAND_MAX)));
    fputc (c, stdout);
}
fputc ('\n', stdout);
return 0;

}

Data integrity check for uncompressed data and metadata isn't implemented?

Denis Feklushkin — Tue, 12 Jul 2016 12:50:09 -0000

Hi!

I am tried to create SquashFS image without any encryption and then changed one byte in it. And nothing is happened when I am unsquashfs it - no errors or warnings, nothing.

Recent posts to Discussion

SquashFS 3.1-r2 mounting 1.3-r3 filesystem

Looking for advice on the xz operation mode used in SquashFS

Squashfs version 3.76 detected, only version 4.0 supported.

Hi, everybody,

Squashfs support for eMMC

Alternatives to Squashfs (due to lack of uBoot support)

squashfs4 byte order support

squashfs4 byte order support

squashfs4 byte order support

BUG: I/O error reading the last fragment of a large file.

include <stdio.h>

include <stdlib.h>

include <string.h>

Data integrity check for uncompressed data and metadata isn't implemented?