74cd76db68
This test reproduced issue #2428. It does the following: 1) submit some large writes (requiring R2T or RDMA READ) 2) submit fused compare (first fused) 3) variable number of qpair polls 4) submit fused write (second fused) The RDMA and TCP transports were not accounting for the fused bits, and would send a non-fused command down to the target layer between the two fused commands. By adding the variable amount of delay between the fused command submissions, it creates a window where a large write's payload is fetched from the host and submitted to the target between submission of the two fused commands. Signed-off-by: Jim Harris <james.r.harris@intel.com> Change-Id: I1abd5a6bca08386279c09f8135387826149d30ef Reviewed-on: https://review.spdk.io/gerrit/c/spdk/spdk/+/11963 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Community-CI: Broadcom CI <spdk-ci.pdl@broadcom.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Tomasz Zawadzki <tomasz.zawadzki@intel.com>
25 lines
866 B
Bash
Executable File
25 lines
866 B
Bash
Executable File
#!/usr/bin/env bash
|
|
|
|
testdir=$(readlink -f $(dirname $0))
|
|
rootdir=$(readlink -f $testdir/../../..)
|
|
source $rootdir/test/common/autotest_common.sh
|
|
source $rootdir/test/nvmf/common.sh
|
|
|
|
rpc_py="$rootdir/scripts/rpc.py"
|
|
|
|
nvmftestinit
|
|
nvmfappstart -m 0x2
|
|
|
|
$rpc_py nvmf_create_transport $NVMF_TRANSPORT_OPTS -u 8192
|
|
$rpc_py nvmf_create_subsystem nqn.2016-06.io.spdk:cnode1 -a -s SPDK00000000000001 -m 10
|
|
$rpc_py nvmf_subsystem_add_listener nqn.2016-06.io.spdk:cnode1 -t $TEST_TRANSPORT -a $NVMF_FIRST_TARGET_IP -s $NVMF_PORT
|
|
$rpc_py bdev_null_create NULL1 1000 512
|
|
$rpc_py bdev_wait_for_examine
|
|
$rpc_py nvmf_subsystem_add_ns nqn.2016-06.io.spdk:cnode1 NULL1
|
|
|
|
$rootdir/test/nvme/fused_ordering/fused_ordering -r "trtype:$TEST_TRANSPORT adrfam:IPv4 traddr:$NVMF_FIRST_TARGET_IP trsvcid:$NVMF_PORT subnqn:nqn.2016-06.io.spdk:cnode1"
|
|
|
|
trap - SIGINT SIGTERM EXIT
|
|
|
|
nvmftestfini
|