83af29b4eb
When creating the nbd, we wait until it's ready, but we didn't do that when loading the configuration from JSON, which resulted in sporadic IO failures, as the device hasn't been initialized yet. This patch adds waitfornbd after each load_config call. Change-Id: Id350ae7b1afab11f5f3fbd131d938dbd65a8cb15 Signed-off-by: Konrad Sztyber <konrad.sztyber@intel.com> Reviewed-on: https://review.gerrithub.io/c/spdk/spdk/+/459616 Tested-by: SPDK CI Jenkins <sys_sgci@intel.com> Reviewed-by: Ben Walker <benjamin.walker@intel.com> Reviewed-by: Wojciech Malikowski <wojciech.malikowski@intel.com> Reviewed-by: Shuhei Matsumoto <shuhei.matsumoto.xt@hitachi.com> Reviewed-by: Mateusz Kozlowski <mateusz.kozlowski@intel.com> Reviewed-by: Paul Luse <paul.e.luse@intel.com> Reviewed-by: Darek Stojaczyk <dariusz.stojaczyk@intel.com>
104 lines
3.2 KiB
Bash
Executable File
104 lines
3.2 KiB
Bash
Executable File
#!/usr/bin/env bash
|
|
|
|
testdir=$(readlink -f $(dirname $0))
|
|
rootdir=$(readlink -f $testdir/../..)
|
|
source $rootdir/test/common/autotest_common.sh
|
|
|
|
rpc_py=$rootdir/scripts/rpc.py
|
|
pu_start=0
|
|
pu_end=3
|
|
additional_blocks=16
|
|
|
|
device=$1
|
|
uuid=$2
|
|
|
|
restore_kill() {
|
|
rm -f $testdir/config/ftl.json
|
|
rm -f $testdir/empty
|
|
rm -f $testdir/testblock
|
|
rm -f $testdir/testfile.md5
|
|
rm -f $testdir/testfile2.md5
|
|
|
|
$rpc_py delete_ftl_bdev -b nvme0 || true
|
|
killprocess $svcpid || true
|
|
rmmod nbd || true
|
|
}
|
|
|
|
trap "restore_kill; exit 1" SIGINT SIGTERM EXIT
|
|
|
|
# Extract chunk size
|
|
chunk_size=$($rootdir/examples/nvme/identify/identify -r "trtype:PCIe traddr:$device" |
|
|
grep 'Logical blks per chunk' | sed 's/[^0-9]//g')
|
|
|
|
band_size=$(($chunk_size*($pu_end-$pu_start+1)))
|
|
|
|
$rootdir/app/spdk_tgt/spdk_tgt & svcpid=$!
|
|
# Wait until spdk_tgt starts
|
|
waitforlisten $svcpid
|
|
|
|
if [ -n "$uuid" ]; then
|
|
$rpc_py construct_ftl_bdev -b nvme0 -a $device -l $pu_start-$pu_end -u $uuid -o
|
|
else
|
|
$rpc_py construct_ftl_bdev -b nvme0 -a $device -l $pu_start-$pu_end -o
|
|
fi
|
|
|
|
# Load the nbd driver
|
|
modprobe nbd
|
|
$rpc_py start_nbd_disk nvme0 /dev/nbd0
|
|
waitfornbd nbd0
|
|
|
|
$rpc_py save_config > $testdir/config/ftl.json
|
|
|
|
# Send band worth of data in 2 steps (some data should be written to 2nd band due to metadata overhead)
|
|
dd if=/dev/urandom of=/dev/nbd0 bs=4K count=$(($band_size - $chunk_size)) oflag=dsync
|
|
offset=$(($band_size - $chunk_size))
|
|
|
|
dd if=/dev/urandom of=/dev/nbd0 bs=4K count=$chunk_size seek=$offset oflag=dsync
|
|
offset=$(($offset + $chunk_size))
|
|
|
|
# Save md5 data of first batch (which should be fully on a closed band and recoverable)
|
|
dd if=/dev/nbd0 bs=4K count=$(($band_size - $chunk_size)) | md5sum > $testdir/testfile.md5
|
|
|
|
# Make sure the third batch of written data is fully on the second band
|
|
dd if=/dev/urandom of=/dev/nbd0 bs=4K count=$additional_blocks seek=$offset oflag=dsync
|
|
offset=$(($offset + $additional_blocks))
|
|
|
|
$rpc_py stop_nbd_disk /dev/nbd0
|
|
|
|
# Force kill bdev service (dirty shutdown) and start it again
|
|
kill -9 $svcpid
|
|
rm -f /dev/shm/spdk_tgt_trace.pid$svcpid
|
|
# TODO Adapt this after waitforlisten is expanded
|
|
sleep 5
|
|
|
|
$rootdir/app/spdk_tgt/spdk_tgt & svcpid=$!
|
|
# Wait until spdk_tgt starts
|
|
waitforlisten $svcpid
|
|
|
|
# Ftl should recover, though with a loss of data (-o config option)
|
|
$rpc_py load_config < $testdir/config/ftl.json
|
|
waitfornbd nbd0
|
|
|
|
# Write extra data after restore
|
|
dd if=/dev/urandom of=/dev/nbd0 bs=4K count=$chunk_size seek=$offset oflag=dsync
|
|
# Save md5 data
|
|
dd if=/dev/nbd0 bs=4K count=$chunk_size skip=$offset | md5sum > $testdir/testfile2.md5
|
|
|
|
# Make sure all data will be read from disk
|
|
echo 3 > /proc/sys/vm/drop_caches
|
|
|
|
# Without persistent cache, first batch of data should be recoverable
|
|
dd if=/dev/nbd0 bs=4K count=$(($band_size - $chunk_size)) | md5sum -c $testdir/testfile.md5
|
|
|
|
dd if=/dev/nbd0 of=$testdir/testblock bs=4k count=$additional_blocks skip=$band_size
|
|
# Last 4k blocks from before restore should be on second band, and return as 0s
|
|
dd if=/dev/zero of=$testdir/empty bs=4k count=$additional_blocks
|
|
cmp $testdir/empty $testdir/testblock
|
|
# Verify data written after restore is still there
|
|
dd if=/dev/nbd0 bs=4K count=$chunk_size skip=$(($band_size + $additional_blocks)) | md5sum -c $testdir/testfile2.md5
|
|
|
|
report_test_completion ftl_dirty_shutdown
|
|
|
|
trap - SIGINT SIGTERM EXIT
|
|
restore_kill
|