Ran into an interesting issue, and thought I'd report it. I created a
4TB file using posix_fallocate() on a freshly-created ext4 filesystem,
unmounted, and then ran e2fsck -f on it. Using e2fsprogs 1.41.9,
e2fsck ran through with no issues. Versions 1.41.10 and 1.41.11,
however, reported finding an error. Output was the same for both
1.41.10 and 1.41.11:
e2fsck 1.41.10 (10-Feb-2009)
Pass 1: Checking inodes, blocks, and sizes
Inode 12, i_blocks is 8589935432, should be 840. Fix? yes
Pass 2: Checking directory structure
Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information
c: ***** FILE SYSTEM WAS MODIFIED *****
c: 12/90523648 files (0.0% non-contiguous), 1079543383/1448361984 blocks
I'm in the process of trying it again using dd to create the large
file instead of posix_fallocate(), but I suspect the results will be
the same. Writing out such a huge file using dd takes a lot longer,
since as was discussed on this list a couple weeks ago, large
sequential writes on ext4 max out around 350MB/s. :)
-Justin
On 03/18/2010 04:25 PM, Justin Maggard wrote:
> Ran into an interesting issue, and thought I'd report it. I created a
> 4TB file using posix_fallocate() on a freshly-created ext4 filesystem,
> unmounted, and then ran e2fsck -f on it. Using e2fsprogs 1.41.9,
> e2fsck ran through with no issues. Versions 1.41.10 and 1.41.11,
> however, reported finding an error. Output was the same for both
> 1.41.10 and 1.41.11:
>
> e2fsck 1.41.10 (10-Feb-2009)
> Pass 1: Checking inodes, blocks, and sizes
> Inode 12, i_blocks is 8589935432, should be 840. Fix? yes
# bc
obase=16
8589935432
200000348
840
348
oops, so looks like another 32-bit overflow.
we go there if:
if ((pb.num_blocks != ext2fs_inode_i_blocks(fs, inode)) || ...
but:
struct process_block_struct {
ext2_ino_t ino;
unsigned is_dir:1, is_reg:1, clear:1, suppress:1,
fragmented:1, compressed:1, bbcheck:1;
blk_t num_blocks;
and:
typedef __u32 blk_t;
we can't fit 8589935432 into a u32; looks like this one needs a blk64_t
overhaul as well.
commmit 8a8f36540bbf5d4397cf476e216e9a720b5c1d8e added handling of
the high i_blocks number, but did not enlarge the container it
went into:
- if ((pb.num_blocks != inode->i_blocks) ||
+ if ((pb.num_blocks != ext2fs_inode_i_blocks(fs, inode)) ||
-Eric
> Pass 2: Checking directory structure
> Pass 3: Checking directory connectivity
> Pass 4: Checking reference counts
> Pass 5: Checking group summary information
>
> c: ***** FILE SYSTEM WAS MODIFIED *****
> c: 12/90523648 files (0.0% non-contiguous), 1079543383/1448361984 blocks
>
> I'm in the process of trying it again using dd to create the large
> file instead of posix_fallocate(), but I suspect the results will be
> the same. Writing out such a huge file using dd takes a lot longer,
> since as was discussed on this list a couple weeks ago, large
> sequential writes on ext4 max out around 350MB/s. :)
>
> -Justin
> --
> To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
On 03/18/2010 04:38 PM, Eric Sandeen wrote:
> commmit 8a8f36540bbf5d4397cf476e216e9a720b5c1d8e added handling of
> the high i_blocks number, but did not enlarge the container it
> went into:
>
> - if ((pb.num_blocks != inode->i_blocks) ||
> + if ((pb.num_blocks != ext2fs_inode_i_blocks(fs, inode)) ||
>
> -Eric
>
Actually, no, that was too quick a skim through, sorry. :)
does your filesystem end up with the EXT4_FEATURE_RO_COMPAT_HUGE_FILE
feature set after your posix_fallocate test?
-Eric