Archive for February, 2011

Ext2Fsd 0.50 was repacked !

Thursday, February 10th, 2011

Please re-download Ext2Fsd 0.50 package and re-install. Or update Ext2Fsd.sys manually from latest 0.50 package.

Download link:

http://sourceforge.net/projects/ext2fsd/files/Ext2fsd/0.50/

New packages checksums:

7c3942309ce986ac25b5ed648469399ab8a0fb6c *Ext2Fsd-0.50.7z
3dec201bd59aa516545e760e5c8a584bceab582d *Ext2Fsd-0.50.exe
2fb171ba41c31add4e4267a806d8596f2dfef4be *Ext2Fsd-0.50.src.7z
e0f465bf2237e0d846534736ec2305a74636f2bd *Ext2Fsd-0.50.src.zip
c13d95770818062887c2ddf2669433fc6a888bf2 *Ext2Fsd-0.50.zip

Changelog:

One dead-lock hang bug fixed: when copying big files (> 4G),
explorer would hang. This bug was introduced by fix #17 of
version 0.50: fastio doesn't update i_size.
Only file writing using fast i/o is affected.

Ext2Fsd 0.50 was released !

Saturday, February 5th, 2011
==================================
Release Notes for Ext2Fsd V0.50
==================================
Features implemented and bugfix from V0.48:
    1, Ext4 extent readonly support by Bo Branten. Writing is
       possible but with no size-extending
    2, Ext3 directory index (hash-tree) support
    3, Fast fsck (uninit_bg) and group block checksum support
    4, Ext4 64k block size support
    5, Symlink/special inodes open/read/deletion support
    6, Buffer head implemented over cache pages
    7, Memory allocation optimization for flexible-size inodes
    8, Improve file deletion: don't grab global lock when deleting
    9, FIXME: return zero-content for sparse file gaps
   10, FIXME: check available spaces before blocks allocation
   11, FIXME: refresh stale root dir content after journal replay
   12, FIXME: incompatible dentry management for 64k block size
   13, FIXME: don't do journal replay for devices set as readonly
   14, FIXME: Win7 cmd.exe always reports file sizes as zero
   15, FIXME: Win7 memory throttling issue calling CcPinRead
   16, FIXME: Ext3Fsd Build issues (SLIST/div64 for Win2k, browser files)
   17, FIXME: Compiling test failure (fastio doesn't update i_size)
   18, FIXME: Possible Mcb memory leak for symbolic links
Supported Ext3/4 features by 0.50:
    1, flexible inode size: > 128 bytes, up to block size
    2, dir_index:    htree directory index
    3, filetype:     extra file mode in dentry
    4, large_file:   > 4G files supported
    5, sparse_super: super block backup in group descriptor
    6, uninit_bg:    fast fsck and group checksum
    7, extent:       reading, writing with no extending.
    8, journal:      only support replay for internal journal
Unsupported Ext3/4 features:
    1, journal: log-based operations, external journal
    2, extent: size truncating & expanding, file deletion
    2, flex_bg: first metadata group
    3, EA (extended attributes), ACL support
Features to be implemented in future:
    1, EXT3/4 features support (extents, journal, flex_bg,)
    2, Documents improvement: clear and elaborate
    3, Performance improvement, code optimization
    4, Automatic check & update of new versions
    5, LVM support, e2fsprogs porting for Windows
Files descriptions:
    Ext2Fsd-0.50.exe:  setup wizard for Windows systems (Win2k, Winxp,
                       Vista, Win7, Server 2003, Server 2008/R2)
    Ext2Fsd-0.50.7z:   Batch setup package in 7-zip
    Ext2Fsd-0.50.zip:  Batch setup package in zip
    Ext2Fsd-0.50.src.7z:  Source codes of Ext3Fsd and Ext2Mgr in 7-zip
    Ext2Fsd-0.50.src.zip: Source codes of Ext3Fsd and Ext2Mgr in zip

Download link:
    http://sourceforge.net/projects/ext2fsd/files/Ext2fsd/0.50/

Files checksum:
    C:\Works\Ext2fsd\0.50>sha1sum Ext2Fsd-0.50*
    bec1b29452473a36dc99c163b7bedbc4bb53707d *Ext2Fsd-0.50.7z
    51cd8b7362ca3b401d455e2338f2c9eb9f97ae30 *Ext2Fsd-0.50.exe
    28f9f3456bfc104fb0773e26c26e8c2fc663663a *Ext2Fsd-0.50.src.7z
    844bce05d203d81d53f69493ae4a5b11dfb73667 *Ext2Fsd-0.50.src.zip
    747ed2f3eb22bf3f67fae7084f6c895f516e6058 *Ext2Fsd-0.50.zip
WARNINGS:
    The driver may crash your system and ruin your data unexpectedly,
    since there might be software conflicts and I could only test it
    on some of the popular platforms. You should use it with care and
    use it at your own risk!
Matt <matt at ext2fsd.com>
http://www.ext2fsd.com

Unformatted issue for Ext3Fsd 0.48

Saturday, February 5th, 2011

Several Ext2Fsd users reported this issue:  after upgrading to Ext2Fsd 0.48,  all the ext2/3 partitions that 0.46a could mount,  couldn’t be mounted any more. Though the drive letter was assigned successfully, Windows explorer would report the unformatted issue when trying to access the mount point.

This issue was caused by a new feature of Ext3Fsd 0.48:  in default, only selected volumes  (listed in registry: HKLM\SYSTEM\CurrentControlSet\Services\Ext2Fsd\Volumes) can be mounted.

You can make Ext3Fsd mount all Linux volumes in default,  do either of the following 2 guides  :

  1. Start Ext2Mgr, Click menu Tools -> Service Management, and then select the checkbox of “Assign Drive Letter Automatically” and save the changes. Rebooting is NOT necessary.
  2. Modify registry, change the DWORD value ( “HKLM\SYSTEM\CurrentControlSet\Services\Ext2Fsd\Parameters\AutoMount”)  to 1. Rebooting is required to make it take into effect.

Normally when assigning a drive letter with Ext2Mgr, Ext2Mgr will create a new entry of  in “HKLM\SYSTEM\CurrentControlSet\Services\Ext2Fsd\Volumes” to store the private settings of this volume/partition. Then Ext3Fsd will load this private settings first, only after that fails would Ext3Fsd use the global settings (readonly, codepage etc).

For truecrypt or filedisk volumes, Ext2Mgr won’t recognize them, so Ext2Mgr won’t create the setting entries in registry for these virtual volumes. In this case you can manually create  one or make Ext3Fsd mount all ext2/3 volumes in default. The format of the volume entry is like this:

Value name:  {C9-2E-AF-9C-FB-7D-45-7E-B5-E6-8D-44-0A-57-97-EA}
Value data:  Readonly;MountPoint=F:;CodePage=utf8;
Value name:  {CA-42-7A-17-D2-4E-4E-F8-9F-A6-72-8F-2A-07-48-2F}
Value data:  Ext3ForceWriting;MountPoint=H:;CodePage=utf8;

The value name is generated from the volume UUID in super_block.