On Tue, Jun 3, 2014 at 11:42 AM, Miklos Szeredi <miklos@szeredi.hu> wrote:
> On Tue, Jun 3, 2014 at 11:26 AM, Sedat Dilek <sedat.dilek@gmail.com> wrote:
>> On Tue, Jun 3, 2014 at 11:18 AM, Sedat Dilek <sedat.dilek@gmail.com> wrote:
>>> [...]
>>>>> [ NOTE-2: The call-trace I have seen once (TERMSLASH=0). ]
>>>>
>>>> Do you know for which operation?
>
> This still looks like the same annotation problem in
> generic_file_splice_write() I already sent a patch for.
>
> Fix now pushed to overlayfs.v22/overlayfs.current.
>

I have applied "vfs: fix wrong lockdep annotation in
generic_file_splice_write()" to my latest Linux-kernel and
truncate.test seems to be fine, now.
Feel free to add my Tested-by/Reported-by when you spinout a new
version of overlayfs in your Git tree.

For the sake of completeness my simple hack to identify the fs-op...

--- a/truncate.test
+++ b/truncate.test
@@ -33,6 +33,8 @@ for ((loop=0; loop<29; loop++)) {
        fi

        fs_op truncate $file $loop
+       echo "1st FS-OP: XXX"
+       dmesg | tail

        assert_is_upper $file
        post=`stat --printf %s $file`
@@ -48,5 +50,7 @@ for ((loop=0; loop<29; loop++)) {
        fi
     else
        fs_op truncate $file $loop -E ENOTDIR
+       echo "2nd FS-OP: XXX"
+       dmesg | tail
     fi
 }

All tests now run fine!

    # LC_ALL=C TEST_OVERLAYFS=1 TERMSLASH=1 ./run.sh

Thanks for all involved people.

- Sedat -