Re: [Fwd: [Bug 9732] New: oops in extent code via ext4_fallocate]

["Aneesh Kumar K.V" <aneesh.kumar@linux.vnet.ibm.com>]

On Fri, Jan 11, 2008 at 02:42:00PM -0600, Eric Sandeen wrote:

> 
> The BUG_ON is:
> 
>         BUG_ON(*logical < le32_to_cpu(ex->ee_block) + le16_to_cpu(ex->ee_len));
> 
> where these were the values:
> 
>         logical 8 ee_block 0 ee_len 32776
> 
> Haven't looked further into it yet.
> 

Not able to reproduce. Why is it calling ext4_ext_search_left for that
value of logical ? ext4_ext_get_blocks should have found that the
logical belong to an already existing extent and return from there. It
should have returned from

2365      if (create == EXT4_CREATE_UNINITIALIZED_EXT)
2366               goto out;


root@qemu-image:/ext4# /root/testfallocate  -f a 0 100
# FALLOCATE TEST REPORT #
        New blocks preallocated = 0.
        Number of bytes preallocated = 0
        Old file size = 100, New file size 100.
        Old num blocks = 4, New num blocks 4.


### TESTS PASSED ###
root@qemu-image:/ext4# ls -alh a
-rw-r--r-- 1 root root 100 Jan  9 16:23 a
root@qemu-image:/ext4# /root/truncate  ./a 50
root@qemu-image:/ext4# ls -alh a
-rw-r--r-- 1 root root 50 Jan  9 16:23 a
root@qemu-image:/ext4# /root/testfallocate  -f a 0 100
# FALLOCATE TEST REPORT #
        New blocks preallocated = 0.
        Number of bytes preallocated = 0
        Old file size = 50, New file size 100.
        Old num blocks = 4, New num blocks 4.


### TESTS PASSED ###
root@qemu-image:/ext4# ls -alh a
-rw-r--r-- 1 root root 100 Jan  9 16:23 a
root@qemu-image:/ext4# more /proc/mounts  | grep ext4dev
/dev/hdc /ext4 ext4dev rw,nodelalloc,data=writeback 0 0
root@qemu-image:/ext4#

The third fallocate should result in calling ext4_ext_get_blocks two
times. The first call with logical 0 and max 100 and it will return with
allocated 50. Then call again with logical 50 and max 50.

Something else is wrong.

-aneesh