On Jun 21, 2007 11:54 +0200, Jan Kara wrote:
On Thu 21-06-07 03:33:43, Andreas Dilger wrote:
+ if (LINUX_S_ISCHR(inode.i_mode) || LINUX_S_ISBLK(inode.i_mode) ||
+ (LINUX_S_ISLNK(inode.i_mode)
+ ext2fs_inode_data_blocks(fs, inode) == 0))
+ return
Hi,
To start with, this was seen on a 2.6.18 kernel which was tainted. I
know how it doesn't make sense to work on a report from such an old code
base, never mind a tainted kernel, so feel free to ignore this if it's
not interesting..
I left a system running overnight, running stress --cpu 4
I have updated my patchset taking into account comments received from
Andreas and Jose, except for the last patch concerning the ACLs/EAs for
which I did not do anything.
As the changes are rather significant, I send here a new version for review.
Changes from the previous version:
- support of
This patch adds a new option '-D' to mkfs to allow creating ext4 filesystem.
lib/ext2fs/ext2_fs.h|7 +++
lib/ext2fs/initialize.c |1 +
misc/mke2fs.c | 18 --
3 files changed, 24 insertions(+), 2 deletions(-)
Index:
This patch introduces a new macro EXT2_DESC_PER_BLOCK to calculate the
number of group descriptors per block as two different structures are
defined.
Change from previous version:
use s_desc_size instead of sizeof(struct ext4_group_desc)
lib/ext2fs/closefs.c|2 +-
This patch introduces new bitmap interface for ext4 filesystems which
uses the new type blk64_t.
bitmaps.c | 156
--
bitops.c | 36 ++
bitops.h | 29 +++
ext2fs.h | 25 +
4 files changed, 231
This patch introduces new badblocks list functions for ext4 filesystems
which use the new type blk64_t.
badblocks.c | 471
+---
ext2fs.h| 28 ++-
ext2fsP.h | 13 +
freefs.c| 19 --
4 files changed, 423 insertions(+),
This patch contains the changes to support 48-bit block numbers in extents.
This patch is not complete, the new function block_iterate_extents() was
not updated.
e2fsck/pass1.c| 15 +--
lib/ext2fs/bmap.c |9 +
lib/ext2fs/ext3_extents.h | 28
This patch converts blk_t type to __u64 when compiling code with
_EXT4FS_ option.
Force the 64-bit feature when the filesystem is greater than 2**32 blocks.
Changes from the previous version:
- ask for confirmation before creating the ext4 filesystem
lib/ext2fs/ext2_fs.h |3 ++-
This patch converts the indirect block definitions from blkt_t type to
__u32.
Changes from the previous version:
- Fix errors found in the test suite make check
e2fsck/pass1.c |2 -
e2fsck/super.c |4 +-
lib/ext2fs/bb_inode.c | 10 +++---
lib/ext2fs/block.c |
This patch is intended to contain the changes to support 48-bit block
numbers for ACL. For now, it only contains modifications in the inode
structure. It has to be completed.
Andreas proposed to rename i_file_acl to i_file_ea
and i_file_acl_high to i_file_ea_hi.
ext2_fs.h | 10 ++
This patch converts some int or unsigned int block number definitions to
blk_t type. Most conversions are necessary for 64-bit porting.
Changes from the previous version:
- Fix errors found in the test suite make check
debugfs/debugfs.c | 12 +---
debugfs/logdump.c |
On Thu, Jun 21, 2007 at 04:40:55PM +0800, LIOU Payphone wrote:
my problem :
First of all, what version of e2fsprogs and kernel are you using?n
(1) when i want to format /dev/sda1 to be ext3-filesystem, i set
s_def_hash_version of struct ext3_super_block to be DX_HASH_LEGACY.
when the
Hi Kalpak,
I've broken up your patch and rewritten portions of them to
clean them up. As I've mentioned before, smaller patches are easier to
review and to cherry pick as necessary.
The changes from your patches are:
1) A smaller set of functionality (just returning the recommended
3 files changed, 26 insertions(+), 18 deletions(-)
lib/ext2fs/ext2fs.h|1 +
lib/ext2fs/mkjournal.c | 20
misc/util.c| 23 +--
# HG changeset patch
# User [EMAIL PROTECTED]
# Date 1182441546 14400
# Node ID
On Jun 20, 2007 17:26 +0530, Girish Shilamkar wrote:
crc = crc16(crc, (__u8 *)gdp + offset,
- sbi-s_es-s_desc_size - offset);
+ le16_to_cpu(sbi-s_es-s_desc_size)
+ -
Thanks, I've applied the blkid LUKS patch to the e2fsprogs SCM (modulo
a minor whitespace breakage which I fixed up).
BTW, there appears to be a problem here in udev regarding LUKS
identification:
https://bugs.launchpad.net/ubuntu/+source/e2fsprogs/+bug/93921
The problem is that udev sets its
On Jun 21, 2007 17:04 +0200, Valerie Clement wrote:
@@ -1094,6 +1095,14 @@ static void PRS(int argc, char *argv[])
exit(1);
}
break;
+ case 'D':
+ desc_size = strtoul(optarg, tmp, 0);
On Jun 21, 2007 17:15 +0200, Valerie Clement wrote:
-#define EXT2_DESC_PER_BLOCK(s) (EXT2_BLOCK_SIZE(s) / sizeof
(struct ext2_group_desc))
+#define EXT2_DESC_PER_BLOCK(s) (EXT2_BLOCK_SIZE(s) / \
+ ((EXT2_DESC_SIZE(s) = EXT2_MIN_DESC_SIZE_64BIT) ? \
+
Note - most of these comments relate to making these changes compatible
with existing e2fsprogs. I don't object to such patches for the short term
to make 64-bit usable e2fsprogs, but I don't think they should go into the
upstream e2fsprogs. I wonder if having _EXT4FS_ should force the .so
On Jun 21, 2007 17:25 +0200, Valerie Clement wrote:
@@ -26,8 +26,11 @@ void ext2fs_swap_super(struct ext2_super
sb-s_inodes_count = ext2fs_swab32(sb-s_inodes_count);
sb-s_blocks_count = ext2fs_swab32(sb-s_blocks_count);
+ sb-s_blocks_count_hi =
Andreas Dilger wrote:
Ted,
the attached patch adds an extra validity test in check_ext_attr(). If
an attribute's e_value_size is zero the current code does not allocate
a region for it and as a result the e_value_offs value is not verified.
However, if e_value_offs is very large then the
On Jun 21, 2007 17:29 +0200, Valerie Clement wrote:
@@ -91,6 +92,12 @@ void ext2fs_swap_group_desc(struct ext2_
gdp-bg_flags = ext2fs_swab16(gdp-bg_flags);
gdp-bg_itable_unused = ext2fs_swab16(gdp-bg_itable_unused);
gdp-bg_checksum = ext2fs_swab16(gdp-bg_checksum);
+ if
On Jun 21, 2007 17:31 +0200, Valerie Clement wrote:
+#define EXT4_EE_START(s, e) \
+ ((e)-ee_start +\
+ (((s)-s_feature_incompat EXT4_FEATURE_INCOMPAT_64BIT) \
+ (EXT2_BLOCKS_COUNT(s)
On Jun 21, 2007 17:42 +0200, Valerie Clement wrote:
@@ -1273,6 +1273,17 @@ static void PRS(int argc, char *argv[])
+ if (dev_size ((unsigned) 1 31)
+ fs_param.s_feature_incompat |=
+
On Sep 08, 2006 13:11 +0900, [EMAIL PROTECTED] wrote:
On July 7, 2006, sho wrote:
These patches support large blocksize up to PAGESIZE (max 64KB).
NOTE:
They limit the end of a directory block to 65532(64K - 4)
to avoid overflow only when using 64KB block.
Takashi, in light of the (very
On Jun 21, 2007 17:49 +0200, Valerie Clement wrote:
This patch is intended to contain the changes to support 48-bit block
numbers for ACL. For now, it only contains modifications in the inode
structure. It has to be completed.
Andreas proposed to rename i_file_acl to i_file_ea
and
Kernel build was failing with the uninitialized patches when
CONFIG_CRC16=m. The patch below resolves this issue.
Also, when testing the patches, filesystem mount fails:
EXT4-fs error (device sdc1): ext4_check_descriptors: Checksum for group 1
failed (0!=1)
EXT4-fs: group descriptors
Hello, Ted. I think I failed to represent it very well. But many thanks. :-)
What makes me puzzled is there are two entries both named 3220 with ino
3350 were shown under /mnt when I ls -li /mnt.
The version of kernel is 2.6.17-1.2142; and e2fsprogs-1.28.
You know that we can set
Hello!
I suggest developers to consider ext4 design from the point of view of these
papers:
IRON FILE SYSTEMS -
http://www.cs.wisc.edu/wind/Publications/vijayan-thesis06.pdf
IMHO - very impressive paper and developers of close future filesystems can't
ignore these problems and solutions.
and
30 matches
Mail list logo
