tree e600209d55a82929e025beac8d03e78261e647f5
parent 1f2208c7c6fe679ea2e4834b9784e9492c795f66
author Cristian Marussi <cristian.marussi@arm.com> 1655889559 +0100
committer Yang Xu <xuyang2018.jy@fujitsu.com> 1657100164 +0800

Fix mountns01/02/03/04 final umounts using more peculiar dir names

Running LTP20220527 release it appears that the recently re-written tests
mountns02/03/04 can now throw a warning on their final umount attempt in
some setup:

<<<test_output>>>
tst_test.c:1526: TINFO: Timeout per run is 0h 00m 30s
mountns04.c:38: TPASS: unbindable mount passed
tst_device.c:395: TWARN: umount('A') failed with EINVAL
mountns.h:36: TWARN: umount(A) failed: EINVAL (22)
tst_device.c:434: TINFO: No device is mounted at B

Moreover, the underlying safe_umount() then upgrades the TWARN emitted
from tst_umount to a TBROK, so causing the test to completely fail:

Summary:
passed   1
failed   0
broken   0
skipped  0
warnings 2
<<<execution_status>>>
initiation_status="ok"
duration=0 termination_type=exited termination_id=4 corefile=no

Even though the final SAFE_UMOUNTs in the test body properly unmount the
test created mountpoints, the final cleanup functions, that finally check
to see if those mountpoints are still mounted, can be fooled into falsely
thinking that test-chosen mountpoints "A" or "B" are still there: this is
due to the fact that the internal helper tst_is_mounted() uses a simple
strstr() on /proc/mounts to check if a directory is still mounted and
clearly the currently test-chosen names are far too much simple, being
one-letter, and they can be easily matched by other unrelated mountpoints
that happen to exist on a specific setup.

Use a more peculiar naming for the test chosen mountpoints and generalize
accordingly all the comments.

Cc: Andrea Cervesato <andrea.cervesato@suse.de>
Cc: Cyril Hrubis <chrubis@suse.cz>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Reviewed-by: Yang Xu <xuyang2018.jy@fujitsu.com>
Reviewed-by: Cyril Hrubis <chrubis@suse.cz>