[PATCH blktests 0/3] improve block/011

[PATCH blktests 0/3] improve block/011

[Shin'ichiro Kawasaki]

As of today, three failure symptoms are observed with block/011 [1]. This series
addresses two of them. The first two patches address the disappearing system
disk (Symptom C in [1]). The third patch addresses the zero capacity NVME
devices. (Symptom B in [1]).

[1] https://lore.kernel.org/linux-block/rsmmxrchy6voi5qhl4irss5sprna3f5owkqtvybxglcv2pnylm@xmrnpfu3tfpe/

Shin'ichiro Kawasaki (3):
  common/rc: introduce _get_pci_from_dev_sysfs
  block/011: skip when mounted block devices are affected
  block/011: recover test target NVME device capacity

 common/rc       |  8 ++++++--
 tests/block/011 | 31 +++++++++++++++++++++++++++++++
 2 files changed, 37 insertions(+), 2 deletions(-)

-- 
2.40.1

[PATCH blktests 1/3] common/rc: introduce _get_pci_from_dev_sysfs

[Shin'ichiro Kawasaki]

To prepare for block/011 test case improvement, add the helper function
which gets PCI device from the given sysfs path of a block device.

Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
---
 common/rc | 8 ++++++--
 1 file changed, 6 insertions(+), 2 deletions(-)

diff --git a/common/rc b/common/rc
index 57e0f42..90122c0 100644
--- a/common/rc
+++ b/common/rc
@@ -313,12 +313,16 @@ _require_test_dev_is_pci() {
 	return 0
 }
 
-_get_pci_dev_from_blkdev() {
-	readlink -f "$TEST_DEV_SYSFS/device" | \
+_get_pci_from_dev_sysfs() {
+	readlink -f "$1/device" | \
 		grep -Eo '[0-9a-f]{4,5}:[0-9a-f]{2}:[0-9a-f]{2}\.[0-9a-f]' | \
 		tail -1
 }
 
+_get_pci_dev_from_blkdev() {
+	_get_pci_from_dev_sysfs "$TEST_DEV_SYSFS"
+}
+
 _get_pci_parent_from_blkdev() {
 	readlink -f "$TEST_DEV_SYSFS/device" | \
 		grep -Eo '[0-9a-f]{4,5}:[0-9a-f]{2}:[0-9a-f]{2}\.[0-9a-f]' | \
-- 
2.40.1

[PATCH blktests 2/3] block/011: skip when mounted block devices are affected

[Shin'ichiro Kawasaki]

The test case disables PCI device of the test target block device. When
the PCI device has other block devices mounted, those block devices are
disabled also. If the mounted device is the system disk, the test screws
up the system. To avoid such dangerous operation, check if the target
PCI device has mounted block devices. In that case, skip the test.

Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
---
 tests/block/011 | 17 +++++++++++++++++
 1 file changed, 17 insertions(+)

diff --git a/tests/block/011 b/tests/block/011
index 4f331b4..0699936 100755
--- a/tests/block/011
+++ b/tests/block/011
@@ -10,12 +10,29 @@ DESCRIPTION="disable PCI device while doing I/O"
 TIMED=1
 CAN_BE_ZONED=1
 
+pci_dev_mounted() {
+	local d dev p pdev
+
+	pdev="$(_get_pci_dev_from_blkdev)"
+	for d in /sys/block/*; do
+		dev=${d##*/}
+		p=$(_get_pci_from_dev_sysfs "$d")
+		[[ $p != "$pdev" ]] && continue
+		grep -qe "/dev/$dev" /proc/mounts && return 0
+	done
+	return 1
+}
+
 requires() {
 	_have_fio && _have_program setpci
 }
 
 device_requires() {
 	_require_test_dev_is_pci
+	if pci_dev_mounted; then
+		SKIP_REASONS+=("mounted block device exists on test target PCI device")
+		return 1
+	fi
 }
 
 test_device() {
-- 
2.40.1

[PATCH blktests 3/3] block/011: recover test target NVME device capacity

[Shin'ichiro Kawasaki]

The test case runs fio while disabling and enabling PCI device of the
test target block device. When the block device is a NVME PCI device,
the test triggers NVME controller reset. When an error happens during
the reset, NVME PCI driver marks zero capacity for the device. This
zero capacity device causes failures of the following test cases.

To avoid the failures by zero device capacity, check the capacity at the
test end. If it is zero, remove the device and rescan PCI bus to detect
the device again, and regain the correct capacity.

Signed-off-by: Shin'ichiro Kawasaki <shinichiro.kawasaki@wdc.com>
---
 tests/block/011 | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/tests/block/011 b/tests/block/011
index 0699936..a4230f4 100755
--- a/tests/block/011
+++ b/tests/block/011
@@ -59,4 +59,18 @@ test_device() {
 	done
 
 	echo "Test complete"
+
+	# This test triggers NVME controller resets. When any failure happens
+	# during the resets, the driver marks the NVME block devices with zero
+	# capacity. Then following tests fail with the zero capacity devices. To
+	# get back the correct capacity, remove and rescan the devices.
+	if ((!$(<"$TEST_DEV_SYSFS/size"))); then
+		echo "$TEST_DEV has zero capacity" >> "$FULL"
+		if [[ -w $TEST_DEV_SYSFS/device/device/remove ]] &&
+			   [[ -w /sys/bus/pci/rescan ]]; then
+			echo "Rescan to tegain the correct capacity" >> "$FULL"
+			echo 1 > "$TEST_DEV_SYSFS/device/device/remove"
+			echo 1 > /sys/bus/pci/rescan
+		fi
+	fi
 }
-- 
2.40.1

Re: [PATCH blktests 0/3] improve block/011

[Shinichiro Kawasaki]

On May 26, 2023 / 13:58, Shin'ichiro Kawasaki wrote:
> As of today, three failure symptoms are observed with block/011 [1]. This series
> addresses two of them. The first two patches address the disappearing system
> disk (Symptom C in [1]). The third patch addresses the zero capacity NVME
> devices. (Symptom B in [1]).
> 
> [1] https://lore.kernel.org/linux-block/rsmmxrchy6voi5qhl4irss5sprna3f5owkqtvybxglcv2pnylm@xmrnpfu3tfpe/

FYI, I've applied the patches.