On May 24, 2023 / 16:40, Ziyang Zhang wrote:
> On 2023/5/16 16:47, Shinichiro Kawasaki wrote:
> > On May 05, 2023 / 11:28, Ziyang Zhang wrote:
> >> It is very important to test ublk crash handling since the userspace
> >> part is not reliable. Especially we should test removing device, killing
> >> ublk daemons and user recovery feature.
> >>
> >> Add five new tests for ublk to cover these cases.
> >>
> >> Signed-off-by: Ziyang Zhang <ZiyangZhang@xxxxxxxxxxxxxxxxx>
> >> ---
>
> [...]
>
> >> diff --git a/tests/ublk/004 b/tests/ublk/004
> >> new file mode 100755
> >> index 0000000..84e01d1
> >> --- /dev/null
> >> +++ b/tests/ublk/004
> >> @@ -0,0 +1,50 @@
> >> +#!/bin/bash
> >> +# SPDX-License-Identifier: GPL-3.0+
> >> +# Copyright (C) 2023 Ziyang Zhang
> >> +#
> >> +# Test ublk crash with delete just after daemon kill
> >> +
> >> +. tests/ublk/rc
> >> +
> >> +DESCRIPTION="test ublk crash with delete just after daemon kill"
>
> [1]
>
> >> +
> >> +__run() {
> >> + local type=$1
> >> +
> >> + if [ "$type" == "null" ]; then
> >> + ${ublk_prog} add -t null -n 0 > "$FULL" 2>&1
> >> + else
> >> + truncate -s 1G "$TMPDIR/img"
> >> + ${ublk_prog} add -t loop -f "$TMPDIR/img" -n 0 > "$FULL" 2>&1
> >> + fi
> >> +
> >> + udevadm settle
> >> + if ! ${ublk_prog} list -n 0 >> "$FULL" 2>&1; then
> >> + echo "fail to list dev"
> >> + fi
> >> +
> >> + _run_fio_rand_io --filename=/dev/ublkb0 --time_based --runtime=30 >> "$FULL" 2>&1 &
> >
> > Nit: long line
> >
> >> + sleep 2
> >> +
> >> + kill -9 "$(__get_ublk_daemon_pid 0)"
> >
> > I think it would be the better to wait for the pid, to ensure that the ublk
> > daemon process completed.
>
> Hi Shinichiro,
>
> As the description[1] says, this test wants to delete ublk device just after killing
> the daemon. So we should not wait for the pid here.
Thanks for the clarification. I missed that point :)
