On 08/19/2014 10:23 AM, David Sterba wrote:
> On Tue, Aug 19, 2014 at 07:58:20PM +0800, Fengguang Wu wrote:
>> We noticed an xfstests failure on commit
>>
>> 8d875f95da43c6a8f18f77869f2ef26e9594fecc ("btrfs: disable strict file flushes for renames and truncates")
>>
>> It's 100% reproducible in the 5 test runs.
> 
> Same here, different mkfs configurations.
> 
> generic/226 28s ...    [16:11:52] [16:12:55] - output mismatch (see /root/xfstests/results//generic/226.out.bad)
>     --- tests/generic/226.out   2013-05-29 17:16:03.000000000 +0200
>     +++ /root/xfstests/results//generic/226.out.bad     2014-08-19 16:12:55.000000000 +0200
>     @@ -1,6 +1,8 @@
>      QA output created by 226
>      --> mkfs 256m filesystem
>      --> 16 buffered 64m writes in a loop
>     -1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
>     +1 2 3 4 pwrite64: No space left on device
>     +5 6 7 8 9 10 11 12 pwrite64: No space left on device
>     +13 14 15 16
> 
> enospc on a small filesystem (256M)

I'm calling filemap flush more often, but otherwise everything else is
the same.  I'll take a look.

-chris