nvme/003: use _dmesg_since_test_start() to check dmesg output

To avoid a false-positive failure due to previous dmesg.

nvme/003 (test if we're sending keep-alives to a discovery controller) [failed]
    runtime  10.868s  ...  10.854s
    --- tests/nvme/003.out      2022-05-11 12:06:52.951599746 +0800
    +++ /home/lizhijian/blktests/results/nodev/nvme/003.out.bad 2022-05-30 10:36:07.775320133 +0800
    @@ -1,3 +1,4 @@
     Running nvme/003
    +Fail
     NQN:nqn.2014-08.org.nvmexpress.discovery disconnected 1 controller(s)
     Test complete

Signed-off-by: Li Zhijian <lizhijian@fujitsu.com>

diff --git a/tests/nvme/003 b/tests/nvme/003
index 101c1841c6df526bd7143b652c56c500c1ee58f1..c30042e3985f210e0bf623cf39599add26cd6d8c 100755 (executable)
--- a/tests/nvme/003
+++ b/tests/nvme/003
@@ -35,11 +35,11 @@ test() {
        # This is ugly but checking for the absence of error messages is ...
        sleep 10
 
-       if dmesg | grep -q "failed nvme_keep_alive_end_io error="; then
+       if _dmesg_since_test_start | grep -q "failed nvme_keep_alive_end_io error="; then
                echo "Fail"
        fi
 
-       if dmesg | grep -q "nvmet: unsupported cmd 24"; then
+       if _dmesg_since_test_start | grep -q "nvmet: unsupported cmd 24"; then
                echo "Fail"
        fi