On Mon, Mar 04, 2019 at 02:54:54PM -0800, Omar Sandoval wrote:
> On Mon, Mar 04, 2019 at 04:15:01PM +0800, Ming Lei wrote:
> > Add one test to cover changes on block passthrough IO interface,
> > such as blk_rq_map_user(), blk_rq_map_user_iov(), blk_rq_unmap_user()
> > and blk_rq_map_kern().
> > 
> > Signed-off-by: Ming Lei <[email protected]>
> > ---
> >  tests/block/029     | 111 
> > ++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  tests/block/029.out |   5 +++
> >  2 files changed, 116 insertions(+)
> >  create mode 100755 tests/block/029
> >  create mode 100644 tests/block/029.out
> > 
> > diff --git a/tests/block/029 b/tests/block/029
> > new file mode 100755
> > index 000000000000..89d2de70833a
> > --- /dev/null
> > +++ b/tests/block/029
> > @@ -0,0 +1,111 @@
> > +#!/bin/bash
> > +# SPDX-License-Identifier: GPL-2.0+
> > +# Copyright (c) 2019 Ming Lei <[email protected]>
> > +#
> > +# Test userspace IO on NVMe loop device
> 
> Is this a regression test for one of your recent patches?

Yeah, it is for the following patch:

        https://marc.info/?l=linux-block&m=155161907811128&w=2

Also it will serve regression test for the following(not posted yet) patch
of 'enabling multi-bvec for passthrough IO'.

> 
> > +. tests/nvme/rc
> > +
> > +DESCRIPTION="test userspace IO via nvme-cli read/write interface"
> > +QUICK=1
> > +
> > +requires() {
> > +   _have_program nvme && _have_modules loop nvme-loop nvmet && \
> > +           _have_configfs
> > +}
> > +
> > +__test_user_io()
> > +{
> > +   DISK=$1
> > +   START=$2
> > +   CNT=$3
> > +
> > +   BS=`blockdev --getss $DISK`
> 
> $() instead of `` here and everywhere else, please. Also, $DISK needs to
> quoted, as do several other variables. Please run `make check`.
> 
> > +   SIZE=$(($CNT * $BS))
> > +
> > +   IMG=`mktemp /tmp/blk_img_XXXXXX`
> > +   IMG1=`mktemp /tmp/blk_img_XXXXXX`
> 
> Please make all of these variables local and lowercase.
> 
> > +   dd if=/dev/urandom of=$IMG bs=$BS count=$CNT status=none
> > +
> > +   let CNT--
> 
> $((CNT--))
> 
> > +   nvme write --start-block=$START --block-count=$CNT --data-size=$SIZE 
> > --data=$IMG $DISK
> > +   [ $? -ne 0 ] && return -1
> > +   nvme read --start-block=$START --block-count=$CNT --data-size=$SIZE 
> > --data=$IMG1 $DISK
> > +   [ $? -ne 0 ] && return -1
> > +
> > +   diff -q -u $IMG $IMG1
> > +   RES=$?
> > +
> > +   rm -f $IMG $IMG1
> > +   return $RES
> > +}
> > +
> > +test_user_io()
> > +{
> > +   DEV=$1
> > +
> > +   __test_user_io $DEV 1 512  > /dev/null 2>&1
> > +   [ -$? -ne 0 ] && echo "FAIL"
> > +
> > +   __test_user_io $DEV 1 511  > /dev/null 2>&1
> > +   [ -$? -ne 0 ] && echo "FAIL"
> > +
> > +   __test_user_io $DEV 1 513  > /dev/null 2>&1
> > +   [ -$? -ne 0 ] && echo "FAIL"
> > +
> > +   __test_user_io $DEV 511 1024  > /dev/null 2>&1
> > +   [ -$? -ne 0 ] && echo "FAIL"
> > +
> > +   __test_user_io $DEV 511 1023  > /dev/null 2>&1
> > +   [ -$? -ne 0 ] && echo "FAIL"
> > +
> > +   __test_user_io $DEV 511 1025  > /dev/null 2>&1
> > +   [ -$? -ne 0 ] && echo "FAIL"
> > +}
> > +
> > +test() {
> > +   echo "Running ${TEST_NAME}"
> > +
> > +   modprobe nvmet
> > +   modprobe nvme-loop
> > +
> > +   local port
> > +   local nvmedev
> > +   local loop_dev
> > +   local file_path="$TMPDIR/img"
> > +   local subsys_name="blktests-subsystem-1"
> > +
> > +   truncate -s 1G "${file_path}"
> > +
> > +   loop_dev="$(losetup -f --show "${file_path}")"
> > +
> > +   _create_nvmet_subsystem "${subsys_name}" "${loop_dev}" \
> > +            "91fdba0d-f87b-4c25-b80f-db7be1418b9e"
> > +   port="$(_create_nvmet_port "loop")"
> > +   _add_nvmet_subsys_to_port "${port}" "${subsys_name}"
> > +
> > +   nvme connect -t loop -n "${subsys_name}"
> > +
> > +   nvmedev="$(_find_nvme_loop_dev)"
> > +   cat "/sys/block/${nvmedev}n1/uuid"
> > +   cat "/sys/block/${nvmedev}n1/wwid"
> > +
> > +   test_user_io "/dev/${nvmedev}n1"
> > +
> > +   nvme disconnect -n "${subsys_name}"
> 
> We just changed the other nvme tests to redirect the output of
> disconnect like so:
> 
>       nvme disconnect -n "${subsys_name}" >> "$FULL" 2>&1
> 
> Let's do that for this one, too.
> 
> Thanks for the test!

All have been addressed in V2.

Thanks,
Ming

Reply via email to