kernel 4.0.2+ ext4 corruption

[poma]

On 20.05.2015 23:03, poma wrote:
> On 20.05.2015 15:28, Michael Cronenworth wrote:
>> On 05/20/2015 05:22 AM, Matthew Miller wrote:
>>> Michael, are you able to reproduce this reliably, by any chance, so we
>>> have a test case for any potential fix?
>>
>> I upgraded to 4.0.4 from a 4.0.0 build I was running. Initially everything seemed fine.
>>
>> The corruption occurred several hours later after I started an update run and during 
>> the run ldconfig was spitting out "bad ELF" errors. I corrected those that were 
>> reported, but upon a reboot it had corrupted even more files and the system was no 
>> longer bootable due to corruption to libXfont and a few other libraries.
>>
>> I was able to salvage the system after downgrading to 4.0.1 and re-installing rpm, 
>> yum, and a couple dozen other packages.
>>
>> The problem may only be related to >4.0.2 and mdadm RAID. I won't have time right 
>> this minute, but if no one else can easily reproduce it I will setup a VM with two 
>> drives and see if that's enough.
>>
> 
> 
> ext4: fix data corruption caused by unwritten and delayed extents
> https://git.kernel.org/cgit/linux/kernel/git/stable/linux-stable.git/commit/?id=d2dc317
> 
> RAID 0, RAID 1, RAID 4, RAID 5, RAID 6 or RAID 10?
> 
> Here is RAID1/EXT4, and everything is OK, although the kernel ain't the stock Fedora kernel,
> but 4.0.4 with "sched: always use blk_schedule_flush_plug in io_schedule_out"
> https://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/commit/kernel/sched/core.c?id=5596d0d
> https://bugzilla.redhat.com/show_bug.cgi?id=1220519
> 
> 

OK, 
https://bugzilla.redhat.com/show_bug.cgi?id=1223332#c3
RAID0 / RAID1 there.