qemu/tests/qemu-iotests
Fam Zheng e92f0e1910 block: Use bdrv_coroutine_enter to start I/O coroutines
BDRV_POLL_WHILE waits for the started I/O by releasing bs's ctx then polling
the main context, which relies on the yielded coroutine continuing on bs->ctx
before notifying qemu_aio_context with bdrv_wakeup().

Thus, using qemu_coroutine_enter to start I/O is wrong because if the coroutine
is entered from main loop, co->ctx will be qemu_aio_context, as a result of the
"release, poll, acquire" loop of BDRV_POLL_WHILE, race conditions happen when
both main thread and the iothread access the same BDS:

  main loop                                iothread
-----------------------------------------------------------------------
  blockdev_snapshot
    aio_context_acquire(bs->ctx)
                                           virtio_scsi_data_plane_handle_cmd
    bdrv_drained_begin(bs->ctx)
    bdrv_flush(bs)
      bdrv_co_flush(bs)                      aio_context_acquire(bs->ctx).enter
        ...
        qemu_coroutine_yield(co)
      BDRV_POLL_WHILE()
        aio_context_release(bs->ctx)
                                             aio_context_acquire(bs->ctx).return
                                               ...
                                                 aio_co_wake(co)
        aio_poll(qemu_aio_context)               ...
          co_schedule_bh_cb()                    ...
            qemu_coroutine_enter(co)             ...

              /* (A) bdrv_co_flush(bs)           /* (B) I/O on bs */
                      continues... */
                                             aio_context_release(bs->ctx)
        aio_context_acquire(bs->ctx)

Note that in above case, bdrv_drained_begin() doesn't do the "release,
poll, acquire" in BDRV_POLL_WHILE, because bs->in_flight == 0.

Fix this by using bdrv_coroutine_enter and enter coroutine in the right
context.

iotests 109 output is updated because the coroutine reenter flow during
mirror job complete is different (now through co_queue_wakeup, instead
of the unconditional qemu_coroutine_switch before), making the end job
len different.

Signed-off-by: Fam Zheng <famz@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
2017-04-11 20:07:15 +08:00
..
030
2017-02-24 16:09:23 +01:00
2016-04-15 17:22:11 +02:00
2017-02-28 20:40:36 +01:00
2015-01-23 12:41:32 -05:00
083
2016-05-12 15:33:23 +02:00
2016-01-07 21:30:17 +01:00
2015-12-18 14:36:17 +01:00
2016-05-19 16:45:31 +02:00
2016-03-30 12:16:03 +02:00
2016-02-22 09:49:46 +01:00
147
2017-04-03 17:11:39 +02:00
157
2016-07-13 13:41:39 +02:00
159
2016-09-20 22:10:57 +02:00
2016-09-20 22:10:57 +02:00
160
2016-09-20 22:10:57 +02:00
2016-09-20 22:10:57 +02:00
170
2016-09-20 22:10:57 +02:00
2016-09-20 22:10:57 +02:00
172
2016-10-27 16:29:14 -04:00
174
2017-02-12 00:47:42 +01:00
2017-02-12 00:47:42 +01:00
2017-04-03 17:11:40 +02:00

=== This is the QEMU I/O test suite ===

* Intro

This package contains a simple test suite for the I/O layer of qemu.
It does not require a guest, but only the qemu, qemu-img and qemu-io
binaries.  This does limit it to exercise the low-level I/O path only
but no actual block drivers like ide, scsi or virtio.

* Usage

Just run ./check to run all tests for the raw image format, or ./check
-qcow2 to test the qcow2 image format.  The output of ./check -h explains
additional options to test further image formats or I/O methods.

* Feedback and patches

Please send improvements to the test suite, general feedback or just
reports of failing tests cases to qemu-devel@nongnu.org with a CC:
to qemu-block@nongnu.org.