For some odd geometries*, mkfs will try to allocate inode tables off
the end of the block group and fail, rather than warning that too
many inodes have been requested.
This is because when ext2fs_initialize calculates metadata overhead,
it is only adding in group descriptor blocks and the superblock
if the *last* bg contains them - but the first bg also has all of
the various metadata bits taking up space.
Unconditionally adding those counts into the overhead seems to fix
it properly.
*for example "mke2fs -b 1024 -m 0 -g 256 -N 3745 fsfile 1024"
(Note, the test here is a little funky; the expected output is
actually a mkfs failure - but a proper failure instead of the
allocator catching the problem at the last minute)
Signed-off-by: Eric Sandeen <[email protected]>
Index: e2fsprogs-1.40.2/lib/ext2fs/initialize.c
===================================================================
--- e2fsprogs-1.40.2.orig/lib/ext2fs/initialize.c
+++ e2fsprogs-1.40.2/lib/ext2fs/initialize.c
@@ -299,15 +299,13 @@ ipg_retry:
/*
* Overhead is the number of bookkeeping blocks per group. It
- * includes the superblock backup, the group descriptor
- * backups, the inode bitmap, the block bitmap, and the inode
- * table.
+ * includes the superblock (or backup), the block bitmap, the
+ * inode bitmap, the inode table, the group descriptor (or
+ * backups), and the reserved gdt blocks.
*/
- overhead = (int) (2 + fs->inode_blocks_per_group);
-
- if (ext2fs_bg_has_super(fs, fs->group_desc_count - 1))
- overhead += 1 + fs->desc_blocks + super->s_reserved_gdt_blocks;
+ overhead = (int) (3 + fs->inode_blocks_per_group +
+ fs->desc_blocks + super->s_reserved_gdt_blocks);
/* This can only happen if the user requested too many inodes */
if (overhead > super->s_blocks_per_group)Index: e2fsprogs-1.40.2/tests/m_mkfs_overhead/expect.1
===================================================================
--- /dev/null
+++ e2fsprogs-1.40.2/tests/m_mkfs_overhead/expect.1
@@ -0,0 +1,10 @@
+./test.img: Cannot create filesystem with requested number of inodes while setting up superblock
+./test.img: Attempt to read block from filesystem resulted in short read while opening filesystem
+features: Filesystem not open
+
+../e2fsck/e2fsck: Attempt to read block from filesystem resulted in short read while trying to open ./test.img
+Could this be a zero-length partition?
+Exit status is 8
+
+../misc/dumpe2fs: Attempt to read block from filesystem resulted in short read while trying to open ./test.img
+Couldn't find valid filesystem superblock.
Index: e2fsprogs-1.40.2/tests/m_mkfs_overhead/script
===================================================================
--- /dev/null
+++ e2fsprogs-1.40.2/tests/m_mkfs_overhead/script
@@ -0,0 +1,5 @@
+DESCRIPTION="test bg overhead calculation"
+FS_SIZE=1024
+MKE2FS_OPTS="-b 1024 -m 0 -g 256 -N 3745"
+. $cmd_dir/run_mke2fs
+
On Fri, Jul 20, 2007 at 04:47:14PM -0500, Eric Sandeen wrote:
> For some odd geometries*, mkfs will try to allocate inode tables off
> the end of the block group and fail, rather than warning that too
> many inodes have been requested.
>
> This is because when ext2fs_initialize calculates metadata overhead,
> it is only adding in group descriptor blocks and the superblock
> if the *last* bg contains them - but the first bg also has all of
> the various metadata bits taking up space.
>
> Unconditionally adding those counts into the overhead seems to fix
> it properly.
Yes, but it screws up the test right after that. We need to calculate
the overheads for both the first and last block group. When I respun
your patch and applied it to git, I fixed this.
I also added an explicit "Addresses-Red-Hat-Bugzilla: " entry to the
commit log. (This is mainly for your convenience when you're
cherry-picking patches.)
Finally, your patch didn't apply because you apparently nuked a
newline in the patch here:
> + overhead = (int) (3 + fs->inode_blocks_per_group +
> + fs->desc_blocks + super->s_reserved_gdt_blocks);
>
> /* This can only happen if the user requested too many inodes */
> if (overhead > super->s_blocks_per_group)Index: e2fsprogs-1.40.2/tests/m_mkfs_overhead/expect.1
^^^^^^
The patch I applied to git looks like this. (Note that due to a bug
in the test scripts, there are significant ^M characters, so you can't
just apply this patch that you get via e-mail. You'll need to pull it
from git, or reuse the m_mkfs_overhead/expect.1 file from your patch.)
- Ted
commit 5a92a627f1e067ecfaa2478cfafeca2817cdc69f
Author: Theodore Ts'o <[email protected]>
Date: Sun Jul 22 15:07:13 2007 -0400
Properly calculate overhead in ext2fs_initialize()
For some odd geometries*, mkfs will try to allocate inode tables off
the end of the block group and fail, rather than warning that too
many inodes have been requested.
This is because when ext2fs_initialize calculates metadata overhead,
it is only adding in group descriptor blocks and the superblock
if the *last* bg contains them - but the first bg also has all of
the various metadata bits taking up space.
We need to calculate the overhead both for the first block group and
the last block groups separately, since the two different tests need
to know what the overheads are for those two cases, which may be
different.
*for example "mke2fs -b 1024 -m 0 -g 256 -N 3745 fsfile 1024"
(Note, the test here is a little funky; the expected output is
actually a mkfs failure - but a proper failure instead of the
allocator catching the problem at the last minute)
Addresses-Red-Hat-Bugzilla: #241767
Signed-off-by: Eric Sandeen <[email protected]>
Signed-off-by: "Theodore Ts'o" <[email protected]>
diff --git a/lib/ext2fs/initialize.c b/lib/ext2fs/initialize.c
index 9cc3d12..16e9eaa 100644
--- a/lib/ext2fs/initialize.c
+++ b/lib/ext2fs/initialize.c
@@ -298,16 +298,13 @@ ipg_retry:
}
/*
- * Overhead is the number of bookkeeping blocks per group. It
- * includes the superblock backup, the group descriptor
- * backups, the inode bitmap, the block bitmap, and the inode
- * table.
+ * Calculate the maximum number of bookkeeping blocks per
+ * group. It includes the superblock, the block group
+ * descriptors, the block bitmap, the inode bitmap, the inode
+ * table, and the reserved gdt blocks.
*/
-
- overhead = (int) (2 + fs->inode_blocks_per_group);
-
- if (ext2fs_bg_has_super(fs, fs->group_desc_count - 1))
- overhead += 1 + fs->desc_blocks + super->s_reserved_gdt_blocks;
+ overhead = (int) (3 + fs->inode_blocks_per_group +
+ fs->desc_blocks + super->s_reserved_gdt_blocks);
/* This can only happen if the user requested too many inodes */
if (overhead > super->s_blocks_per_group)
@@ -316,8 +313,13 @@ ipg_retry:
/*
* See if the last group is big enough to support the
* necessary data structures. If not, we need to get rid of
- * it.
+ * it. We need to recalculate the overhead for the last block
+ * group, since it might or might not have a superblock
+ * backup.
*/
+ overhead = (int) (2 + fs->inode_blocks_per_group);
+ if (ext2fs_bg_has_super(fs, fs->group_desc_count - 1))
+ overhead += 1 + fs->desc_blocks + super->s_reserved_gdt_blocks;
rem = ((super->s_blocks_count - super->s_first_data_block) %
super->s_blocks_per_group);
if ((fs->group_desc_count == 1) && rem && (rem < overhead))
diff --git a/tests/m_mkfs_overhead/expect.1 b/tests/m_mkfs_overhead/expect.1
new file mode 100644
index 0000000..fff3684
--- /dev/null
+++ b/tests/m_mkfs_overhead/expect.1
@@ -0,0 +1,10 @@
+./test.img: Cannot create filesystem with requested number of inodes while setting up superblock
+./test.img: Attempt to read block from filesystem resulted in short read while opening filesystem
+features: Filesystem not open
+
+../e2fsck/e2fsck: Attempt to read block from filesystem resulted in short read while trying to open ./test.img
+Could this be a zero-length partition?
+Exit status is 8
+
+../misc/dumpe2fs: Attempt to read block from filesystem resulted in short read while trying to open ./test.img
+Couldn't find valid filesystem superblock.
diff --git a/tests/m_mkfs_overhead/script b/tests/m_mkfs_overhead/script
new file mode 100644
index 0000000..8cd9134
--- /dev/null
+++ b/tests/m_mkfs_overhead/script
@@ -0,0 +1,5 @@
+DESCRIPTION="test bg overhead calculation"
+FS_SIZE=1024
+MKE2FS_OPTS="-b 1024 -m 0 -g 256 -N 3745"
+. $cmd_dir/run_mke2fs
+
Theodore Tso wrote:
> On Fri, Jul 20, 2007 at 04:47:14PM -0500, Eric Sandeen wrote:
>> For some odd geometries*, mkfs will try to allocate inode tables off
>> the end of the block group and fail, rather than warning that too
>> many inodes have been requested.
>>
>> This is because when ext2fs_initialize calculates metadata overhead,
>> it is only adding in group descriptor blocks and the superblock
>> if the *last* bg contains them - but the first bg also has all of
>> the various metadata bits taking up space.
>>
>> Unconditionally adding those counts into the overhead seems to fix
>> it properly.
>
> Yes, but it screws up the test right after that. We need to calculate
> the overheads for both the first and last block group. When I respun
> your patch and applied it to git, I fixed this.
Yep, I missed that. Thanks for catching....
> I also added an explicit "Addresses-Red-Hat-Bugzilla: " entry to the
> commit log. (This is mainly for your convenience when you're
> cherry-picking patches.)
Thanks, I appreciate that.
> Finally, your patch didn't apply because you apparently nuked a
> newline in the patch here:
Urhg, sorry. Not sure how that happened.
Thanks,
-Eric