Re: [PATCH 3/4]ext4: Return exchanged blocks count to user space in failure

[Akira Fujita <a-fujita@rs.jp.nec.com>]

Hi Peng,

Peng Tao wrote:
>> I tried to reproduce your problem with the following steps,
>> but I couldn't. Please check my procedure.
>>
>> 1. Test environment
>> linux2.6.31-rc2 + two patches as follows:
>> http://marc.info/?l=linux-ext4&m=125186152727422&w=3
>> http://marc.info/?l=linux-ext4&m=125205817410548&w=3
>>
>> 2. Create ext4 filesystem and mount it
>> mke2fs -t ext4 -b 4096 /dev/XXX
>> mount -t ext4 -O rw,noatime,relatime,commit=360 /dev/XXX /mnt/XXX
>>
>> 3. Create orig and donor files
>> dd if=/dev/urandom of=/mnt/XXX/first.img bs=10M count=1
>> dd if=/dev/zero of=/mnt/XXX/first.img bs=10M count=0 seek=50
>> dd if=/dev/urandom of=/mnt/XXX/last.img bs=10M count=1 seek=49
>>
>> 4. Call EXT4_MOVE_MOVE_EXT ioctl to files with the following arguments
>> (orig_file = /mnt/XXX/first.img, donor_file = /mnt/XXX/last.img)
>> move_data.orig_start = 0;
>> move_data.donor_start = 0;
>> move_data.len = 12152;
>> ioctl(orig_fd, EXT4_IOC_MOVE_EXT, &move_data)
>>
>> However, EXT4_IOC_MOVE_EXT returned ENODATA (this is a fine result by my patch)
>> and it didn't occur the kernel panic which you got.
>> If I chose a wrong step, please correct me.
> Sorry, I didn't notice that I was at commit 7638d5322bd89d49e013a03fe2afaeb6d214fabd
> of linus tree that includes a few patches after 2.6.31-rc2.
>> I appreciate if you could test following environment.
>> - 2.6.31-rc8 + ext4 patch queue
>>  (commit:bdacccd94d08a3f89e7fca676b28aa262c6d75fa) + attached patch
> I can't reproduce the panic in the above environment. I believe the problem
> is fixed. Thanks for your work!

Ok, I will re-base the patch and then send it to the list.
Thanks for your work, too.

Regards,
Akira Fujita.