[PATCH blktests] block/020: test IO req allocation

[PATCH blktests] block/020: test IO req allocation

[Ming Lei]

IO request allocation may hang for ever if the allocation process
migrages. This test covers the request allocation code path.

The following patch can fix this issue on linus kernel:

https://marc.info/?l=linux-block&m=152671586923185&w=2

Signed-off-by: Ming Lei <ming.lei@redhat.com>
---
 tests/block/020     | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/block/020.out |  2 ++
 2 files changed, 52 insertions(+)
 create mode 100755 tests/block/020
 create mode 100755 tests/block/020.out

diff --git a/tests/block/020 b/tests/block/020
new file mode 100755
index 000000000000..b102d7f12173
--- /dev/null
+++ b/tests/block/020
@@ -0,0 +1,50 @@
+#!/bin/bash
+#
+# Test blk-mq IO request allocation.
+#
+# Copyright (C) 2018 Ming Lei <ming.lei@redhat.com>
+#
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <http://www.gnu.org/licenses/>.
+
+DESCRIPTION="check if IO req allocation may hang"
+QUICK=1
+
+requires() {
+	_have_module null_blk && _have_fio
+}
+
+test() {
+	echo "Running ${TEST_NAME}"
+
+	modprobe -r null_blk
+	modprobe null_blk queue_mode=2 irqmode=2 completion_nsec=2000000 \
+			shared_tags=1 submit_queues=4 hw_queue_depth=1 || continue
+
+	local scheds
+	# shellcheck disable=SC2207
+	scheds=($(sed 's/[][]//g' /sys/block/nullb0/queue/scheduler))
+
+	for sched in "${scheds[@]}"; do
+		echo "Testing $sched" >> "$FULL"
+		echo "$sched" > /sys/block/nullb0/queue/scheduler
+		_fio_perf --bs=4k --ioengine=libaio --iodepth=1024 \
+				--numjobs="$(nproc)" --rw=randread --name=async \
+				--filename=/dev/nullb0 --size=1g --direct=1 \
+				--runtime=10
+	done
+
+	modprobe -r null_blk
+
+	echo "Test complete"
+}
diff --git a/tests/block/020.out b/tests/block/020.out
new file mode 100755
index 000000000000..dcb584d22e3e
--- /dev/null
+++ b/tests/block/020.out
@@ -0,0 +1,2 @@
+Running block/020
+Test complete
-- 
2.9.5

Re: [PATCH blktests] block/020: test IO req allocation

[Omar Sandoval]

On Tue, May 22, 2018 at 07:12:49PM +0800, Ming Lei wrote:
> IO request allocation may hang for ever if the allocation process
> migrages. This test covers the request allocation code path.
> 
> The following patch can fix this issue on linus kernel:
> 
> https://marc.info/?l=linux-block&m=152671586923185&w=2
> 
> Signed-off-by: Ming Lei <ming.lei@redhat.com>
> ---
>  tests/block/020     | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++
>  tests/block/020.out |  2 ++
>  2 files changed, 52 insertions(+)
>  create mode 100755 tests/block/020
>  create mode 100755 tests/block/020.out
> 
> diff --git a/tests/block/020 b/tests/block/020
> new file mode 100755
> index 000000000000..b102d7f12173
> --- /dev/null
> +++ b/tests/block/020
> @@ -0,0 +1,50 @@
> +#!/bin/bash
> +#
> +# Test blk-mq IO request allocation.

This isn't very descriptive. How about

Test blk-mq request allocation when hardware tags are limited.

Additionally, for regression tests for specific patches, please
reference what patch it is testing. In this case, it'd be

Regression test for patch "blk-mq: avoid to starve tag allocation after
allocation process migrates".

> +# Copyright (C) 2018 Ming Lei <ming.lei@redhat.com>
> +#
> +# This program is free software: you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation, either version 3 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License
> +# along with this program.  If not, see <http://www.gnu.org/licenses/>.
> +
> +DESCRIPTION="check if IO req allocation may hang"

A better description would say what this test actually does. In this
case, something like

DESCRIPTION="run null-blk on different schedulers with only one hardware tag"

I can fix this and apply when the discussion on the patch itself
settles. Thanks, Ming!

> +QUICK=1
> +
> +requires() {
> +	_have_module null_blk && _have_fio
> +}
> +
> +test() {
> +	echo "Running ${TEST_NAME}"
> +
> +	modprobe -r null_blk
> +	modprobe null_blk queue_mode=2 irqmode=2 completion_nsec=2000000 \
> +			shared_tags=1 submit_queues=4 hw_queue_depth=1 || continue
> +
> +	local scheds
> +	# shellcheck disable=SC2207
> +	scheds=($(sed 's/[][]//g' /sys/block/nullb0/queue/scheduler))
> +
> +	for sched in "${scheds[@]}"; do
> +		echo "Testing $sched" >> "$FULL"
> +		echo "$sched" > /sys/block/nullb0/queue/scheduler
> +		_fio_perf --bs=4k --ioengine=libaio --iodepth=1024 \
> +				--numjobs="$(nproc)" --rw=randread --name=async \
> +				--filename=/dev/nullb0 --size=1g --direct=1 \
> +				--runtime=10
> +	done
> +
> +	modprobe -r null_blk
> +
> +	echo "Test complete"
> +}
> diff --git a/tests/block/020.out b/tests/block/020.out
> new file mode 100755
> index 000000000000..dcb584d22e3e
> --- /dev/null
> +++ b/tests/block/020.out
> @@ -0,0 +1,2 @@
> +Running block/020
> +Test complete
> -- 
> 2.9.5
> 

Re: [PATCH blktests] block/020: test IO req allocation

[Ming Lei]

On Tue, May 22, 2018 at 05:01:54PM -0700, Omar Sandoval wrote:
> On Tue, May 22, 2018 at 07:12:49PM +0800, Ming Lei wrote:
> > IO request allocation may hang for ever if the allocation process
> > migrages. This test covers the request allocation code path.
> > 
> > The following patch can fix this issue on linus kernel:
> > 
> > https://marc.info/?l=linux-block&m=152671586923185&w=2
> > 
> > Signed-off-by: Ming Lei <ming.lei@redhat.com>
> > ---
> >  tests/block/020     | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++
> >  tests/block/020.out |  2 ++
> >  2 files changed, 52 insertions(+)
> >  create mode 100755 tests/block/020
> >  create mode 100755 tests/block/020.out
> > 
> > diff --git a/tests/block/020 b/tests/block/020
> > new file mode 100755
> > index 000000000000..b102d7f12173
> > --- /dev/null
> > +++ b/tests/block/020
> > @@ -0,0 +1,50 @@
> > +#!/bin/bash
> > +#
> > +# Test blk-mq IO request allocation.
> 
> This isn't very descriptive. How about
> 
> Test blk-mq request allocation when hardware tags are limited.

It is fine, except the issue isn't limited to small hw queue tags,
and it might be triggered on NVMe if there are huge IO processes.

> 
> Additionally, for regression tests for specific patches, please
> reference what patch it is testing. In this case, it'd be
> 
> Regression test for patch "blk-mq: avoid to starve tag allocation after
> allocation process migrates".

OK.

> 
> > +# Copyright (C) 2018 Ming Lei <ming.lei@redhat.com>
> > +#
> > +# This program is free software: you can redistribute it and/or modify
> > +# it under the terms of the GNU General Public License as published by
> > +# the Free Software Foundation, either version 3 of the License, or
> > +# (at your option) any later version.
> > +#
> > +# This program is distributed in the hope that it will be useful,
> > +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> > +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> > +# GNU General Public License for more details.
> > +#
> > +# You should have received a copy of the GNU General Public License
> > +# along with this program.  If not, see <http://www.gnu.org/licenses/>.
> > +
> > +DESCRIPTION="check if IO req allocation may hang"
> 
> A better description would say what this test actually does. In this
> case, something like
> 
> DESCRIPTION="run null-blk on different schedulers with only one hardware tag"
> 
> I can fix this and apply when the discussion on the patch itself
> settles. Thanks, Ming!

OK, thanks!

Thanks,
Ming