Re: 2.6.0-test9-mm1 -- DIO and AIO Tests results
From: Daniel McNeil
Date: Fri Oct 31 2003 - 18:36:41 EST
I have re-run the AIO tests on test9, test9-mm1, and
test9-mm1+retry-based-aio-dio-patch;
I ran the tests on an ext3 file system running on a 2-proc XEON
to an ide disk drive.
2.6.0-test9:
===========
As reported earlier, test9 is seeing uninitialized data and the
kiocbs are being modified after free:
$ dio_sparse
non zero buffer at buf[0] => 0xaa,aa,aa,aa
non-zero read at offset 69533696
$ aiodio_sparse
non zero buffer at buf[0] => 0xaa,aa,aa,aa
non-zero read at offset 81854464
With slab debug on, test9 still gets
slab error in check_poison_obj(): cache `kiocb': object was modified after freeing
2.6.0-test9-mm1:
===============
dio_spares - run without seeing uninitialized data
aiodio_sparse - get AIO errors and these kernel messages:
aio_run_iocb: iocb->ki_retry = NULL
Maximal retry count. Bytes done 0
Slab corruption: start=e8676d68, expend=e8676e37, problemat=e8676d68
Last user: [<c018f883>](__aio_put_req+0xa8/0x1b3)
...
slab error in check_poison_obj(): cache `kiocb': object was modified after freeing
2.6.0-test9-mm1+retry-based-aio-dio-patch (from suparna)
=======================================================
dio_sparse - runs successfully and does not see uninitialized data
aiodio_sparse - runs successfully, does not see uninitialized data,
and no slab corruption.
As before, my updated tests are here:
http://developer.osdl.org/daniel/AIO/TESTS/
Daniel
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/