6 Description: FID_in_Dirent feature for OSD.
7 Details : Support for storing the FID in the directory entry has
8 been added to ext3 and ext4.
10 -------------------------------------------------------------------------------
12 2010-07-15 Oracle, Inc.
15 Severity : enhancement
17 Description: FID_in_Dirent feature for OSD.
18 Details : Support for storing the FID in the directory entry has
19 been added to ext3 and ext4.
21 -------------------------------------------------------------------------------
23 2009-11-05 Sun Microsystems, Inc.
27 Frequency : in recovery
29 Description: ext3-corrupted-orphans-2.6.patch added.
30 Details : Backport of ext3 patch to handle corrupted orphans.
32 -------------------------------------------------------------------------------
34 2009-07-31 Sun Microsystems, Inc.
38 Frequency : in recovery
40 Description: don't mix llog inodes with normal.
41 Details : allocate inodes for log in last inode group
43 -------------------------------------------------------------------------------
45 2009-04-20 Sun Microsystems, Inc.
50 Description: minor fixes and cleanups
51 Details : use EXT_UNSET_BLOCK to avoid confusion with EXT_MAX_BLOCK.
52 Initialize 'ix' variable in extents patch to stop compiler warning.
56 Description: update FIEMAP ioctl to match upstream kernel version
57 Details : the FIEMAP block-mapping ioctl had a prototype version in
58 ldiskfs 3.0.7 but this release updates it to match the
59 interface in the upstream kernel, with a new ioctl number.
62 Frequency : only if MMP is active and detects filesystem is in use
64 Description: if MMP startup fails, an oops is triggered
65 Details : if ldiskfs mounting doesn't succeed the error handling doesn't
66 clean up the MMP data correctly, causing an oops.
68 -------------------------------------------------------------------------------
70 2009-04-06 Sun Microsystems, Inc.
76 Description: fix racy locking of mballoc block bitmaps causing BUG
77 Details : The locking of the mballoc buddy bitmap and the in-memory
78 block bitmap was using two different spin locks in some
79 cases. This made it possible to incorrectly access the
80 mballoc bitmap while another process was modifying it,
81 causing a sanity assertion to fail. While no on-disk corruption
82 was reported, there was some risk of this happening.
84 -------------------------------------------------------------------------------
86 2009-02-07 Sun Microsystems, Inc.
89 Severity : enhancement
91 Description: Get RAID stripe size from superblock
92 Details : RAID striping parameters are now saved in the superblock itself,
93 so we should use these parameters instead of having to specify
94 a mount option each time.
97 Frequency : only if server is running on unsupported big-endian machine
99 Description: Disable big-endian ldiskfs server support.
100 Details : The ldiskfs code is not tested on big-endian machines, and
101 there are known compatibility problems in the extents code
102 when running for most of the kernels. Print an error message
103 and refuse to mount, in case anyone tests this. For existing
104 filesystems that might have been created in this way it is
105 possible to mount with the "bigendian_extents" option to
109 Frequency : only with software RAID-5
111 Description: MMP block reads may fail to detect another user of the filesystem
112 Details : with software RAID it is possible that the RAID driver will
113 cache the MMP block and not refetch it from disk. Force the
114 read to invalidate the RAID page cache and go directly to disk.
116 -------------------------------------------------------------------------------
118 2008-08-31 Sun Microsystems, Inc.
121 Severity : enhancement
123 Description: Interoperability at server side (Disk interoperability)
124 Details : Exported some ldiskfs functions which are required for iop
128 Description: OSS crashes frequently, e2fsck does not fix
129 Details : Add an extra check to ldiskfs extents code for the condition,
130 eh_entries = 0 & eh_depth != 0
134 Description: migrate ldiskfs - ldiskfs2 (extents on directories)
135 Details : disable preallocation for non-regular files.
138 Frequency : blocks per group < blocksize*8 and uninit_groups is enabled
140 Description: ldiskfs error: XXX blocks in bitmap, YYY in gd
141 Details : If blocks per group is less than blocksize*8, set rest of the
145 Frequency : only for filesystems larger than 8TB
147 Description: ldiskfs BUG ldiskfs_mb_use_best_found()
148 Details : The ldiskfs mballoc3 code was using a __u16 to store the group
149 number, but with 8TB+ filesystems there are more than 65536
150 groups, causing an oops.
152 Severity : enhancement
154 Description: Add a FIEMAP(FIle Extent MAP) ioctl for ldiskfs
155 Details : FIEMAP ioctl will allow an application to efficiently fetch the
156 extent information of a file. It can be used to map logical blocks
157 in a file to physical blocks in the block device.
161 Description: mkfs.lustre: Unable to mount /dev/cciss/c0d1:Cannot allocate memory
162 Details : correctly handle device paths using a subdirectory in /dev when
163 creating the per-device procfs directory under /proc/fs/ldiskfs.
165 -------------------------------------------------------------------------------
167 2008-04-26 Sun Microsystems, Inc.
172 Description: Kernel BUG at ... ldiskfs/ldiskfs/mballoc.c:3853
173 Details : Removing the underlying device (e.g. via a SCSI rescan) can cause a
174 kernel BUG in ldiskfs_mb_discard_inode_preallocations().
178 Description: deadlock in ext3_expand_extra_isize
179 Details : Inode expansion during ext3_mark_inode_dirty may lead to deadlock.
180 This is solved by making sure that ext3_expand_extra_isize isn't
181 called from ext3_xattr_set_handle.
185 Description: soft lockups on 1.6.2 MDS (is_subdir)
186 Details : don't add dentries with ".." to dcache and ignore such dentries in
190 Frequency : very rare, if additional xattrs are used on kernels >= 2.6.12
192 Description: files may lose file attributes in some cases
193 Details : If there are multiple extended attributes stored on the inode,
194 in particular ACLs, SELinux, or user attributes (if user_xattr
195 is specified for the client mount options) then there is a risk
196 of attribute loss. If an additional attribute is be stored
197 initially in the inode and then increase in size enough to be
198 moved to the external attribute block (e.g. ACL growing in size)
199 for the attribute to be lost.
203 Description: inode version not being initialized on new inodes
204 Details : The inode i_version field was not being initialized on disk.
205 This field is currently unused but will be needed for VBR.
207 --------------------------------------------------------------------------------
209 2008-01-11 Sun Microsystems, Inc.
214 Description: Add support for vanilla-2.6.22 kernel.
216 --------------------------------------------------------------------------------
218 2007-12-07 Cluster File Systems, Inc. <info@clusterfs.com>
223 Description: statfs speedup patches.
224 Details : The patches improve statfs performance for very large
228 Frequency : if the uninit_groups feature is enabled on ldiskfs
230 Description: e2fsck reports "invalid unused inodes count"
231 Details : If a new ldiskfs filesystem is created with the "uninit_groups"
232 feature and only a single inode is created in a group then the
233 "bg_unused_inodes" count is incorrectly updated. Creating a
234 second inode in that group would update it correctly.
236 --------------------------------------------------------------------------------
238 2007-09-27 Cluster File Systems, Inc. <info@clusterfs.com>
241 Severity : enhancement
243 Description: Add a FIEMAP(FIle Extent MAP) ioctl for ldiskfs
244 Details : FIEMAP ioctl will allow an application to efficiently fetch the
245 extent information of a file. It can be used to map logical blocks
246 in a file to physical blocks in the block device.
248 --------------------------------------------------------------------------------
250 2007-08-29 Cluster File Systems, Inc. <info@clusterfs.com>
256 Description: Multiple mount protection(MMP) support.
257 Details : This feature will protect the filesystem from being mounted more
258 than once simultaneously. It will also protect changes by e2fsprogs
259 to the filesystem if it is mounted. This assumes high importance
260 in a shared storage environment where multiple mounts can severely
261 corrupt the filesystem.
263 --------------------------------------------------------------------------------
265 2007-08-10 Cluster File Systems, Inc. <info@clusterfs.com>
267 * Initial release of ldiskfs as a separate package.
272 Description: Updated patchess for new RHEL4 kernel
273 Details : Add patch ext3-unlink-race.patch
274 Updated series file ldiskfs-2.6-rhel4.series