[PATCH -next v4 0/3] Optimize and bugfix for memory-hotplug

[PATCH -next v4 0/3] Optimize and bugfix for memory-hotplug

[Zhao Gongyi]

1. Add checking after online or offline
2. Restore memory before exit
3. Correct test's name

Changes in v4:
  - Remove redundant log information

Changes in v3:
  - Remove 2 obselute patches

Zhao Gongyi (3):
  selftests/memory-hotplug: Add checking after online or offline
  selftests/memory-hotplug: Restore memory before exit
  docs: notifier-error-inject: Correct test's name

 .../fault-injection/notifier-error-inject.rst |  4 +--
 .../memory-hotplug/mem-on-off-test.sh         | 27 +++++++++++++++----
 2 files changed, 24 insertions(+), 7 deletions(-)

--
2.17.1

[PATCH -next v4 1/3] selftests/memory-hotplug: Add checking after online or offline

[Zhao Gongyi]

Add checking for online_memory_expect_success()/
offline_memory_expect_success()/offline_memory_expect_fail(), or
the test would exit 0 although the functions return 1.

Signed-off-by: Zhao Gongyi <zhaogongyi@huawei.com>
---
 .../selftests/memory-hotplug/mem-on-off-test.sh      | 12 +++++++++---
 1 file changed, 9 insertions(+), 3 deletions(-)

diff --git a/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh b/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
index 46a97f318f58..1d87611a7d52 100755
--- a/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
+++ b/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
@@ -266,7 +266,9 @@ done
 #
 echo $error > $NOTIFIER_ERR_INJECT_DIR/actions/MEM_GOING_ONLINE/error
 for memory in `hotpluggable_offline_memory`; do
-	online_memory_expect_fail $memory
+	if ! online_memory_expect_fail $memory; then
+		retval=1
+	fi
 done

 #
@@ -274,7 +276,9 @@ done
 #
 echo 0 > $NOTIFIER_ERR_INJECT_DIR/actions/MEM_GOING_ONLINE/error
 for memory in `hotpluggable_offline_memory`; do
-	online_memory_expect_success $memory
+	if ! online_memory_expect_success $memory; then
+		retval=1
+	fi
 done

 #
@@ -283,7 +287,9 @@ done
 echo $error > $NOTIFIER_ERR_INJECT_DIR/actions/MEM_GOING_OFFLINE/error
 for memory in `hotpluggable_online_memory`; do
 	if [ $((RANDOM % 100)) -lt $ratio ]; then
-		offline_memory_expect_fail $memory
+		if ! offline_memory_expect_fail $memory; then
+			retval=1
+		fi
 	fi
 done

--
2.17.1

Re: [PATCH -next v4 1/3] selftests/memory-hotplug: Add checking after online or offline

[David Hildenbrand]

On 27.09.22 05:28, Zhao Gongyi wrote:
> Add checking for online_memory_expect_success()/
> offline_memory_expect_success()/offline_memory_expect_fail(), or
> the test would exit 0 although the functions return 1.
> 
> Signed-off-by: Zhao Gongyi <zhaogongyi@huawei.com>
> ---
>   .../selftests/memory-hotplug/mem-on-off-test.sh      | 12 +++++++++---
>   1 file changed, 9 insertions(+), 3 deletions(-)
> 
> diff --git a/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh b/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
> index 46a97f318f58..1d87611a7d52 100755
> --- a/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
> +++ b/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
> @@ -266,7 +266,9 @@ done
>   #
>   echo $error > $NOTIFIER_ERR_INJECT_DIR/actions/MEM_GOING_ONLINE/error
>   for memory in `hotpluggable_offline_memory`; do
> -	online_memory_expect_fail $memory
> +	if ! online_memory_expect_fail $memory; then
> +		retval=1
> +	fi
>   done
> 
>   #
> @@ -274,7 +276,9 @@ done
>   #
>   echo 0 > $NOTIFIER_ERR_INJECT_DIR/actions/MEM_GOING_ONLINE/error
>   for memory in `hotpluggable_offline_memory`; do
> -	online_memory_expect_success $memory
> +	if ! online_memory_expect_success $memory; then
> +		retval=1
> +	fi
>   done
> 
>   #
> @@ -283,7 +287,9 @@ done
>   echo $error > $NOTIFIER_ERR_INJECT_DIR/actions/MEM_GOING_OFFLINE/error
>   for memory in `hotpluggable_online_memory`; do
>   	if [ $((RANDOM % 100)) -lt $ratio ]; then
> -		offline_memory_expect_fail $memory
> +		if ! offline_memory_expect_fail $memory; then
> +			retval=1
> +		fi
>   	fi


LGTM

Reviewed-by: David Hildenbrand <david@redhat.com>


I am questioning the stability of the offlining test, though.
Offlining a random memory block can fail easily, because "->removable" is not
expressive:

# tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
Test scope: 2% hotplug memory
          online all hot-pluggable memory in offline state:
                  SKIPPED - no hot-pluggable memory in offline state
          offline 2% hot-pluggable memory in online state
          trying to offline 2 out of 96 memory block(s):
online->offline memory0
tools/testing/selftests/memory-hotplug/mem-on-off-test.sh: line 78: echo: write error: Invalid argument
offline_memory_expect_success 0: unexpected fail
online->offline memory10
online->offline memory11


I guess this test will almost always fail nowadays.

-- 
Thanks,

David / dhildenb

[PATCH -next v4 2/3] selftests/memory-hotplug: Restore memory before exit

[Zhao Gongyi]

Some momory will be left in offline state when calling
offline_memory_expect_fail() failed. Restore it before exit.

Signed-off-by: Zhao Gongyi <zhaogongyi@huawei.com>
---
 .../memory-hotplug/mem-on-off-test.sh         | 21 ++++++++++++++-----
 1 file changed, 16 insertions(+), 5 deletions(-)

diff --git a/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh b/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
index 1d87611a7d52..91a7457616bb 100755
--- a/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
+++ b/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
@@ -134,6 +134,16 @@ offline_memory_expect_fail()
 	return 0
 }

+online_all_offline_memory()
+{
+	for memory in `hotpluggable_offline_memory`; do
+		if ! online_memory_expect_success $memory; then
+			echo "$FUNCNAME $memory: unexpected fail" >&2
+			retval=1
+		fi
+	done
+}
+
 error=-12
 priority=0
 # Run with default of ratio=2 for Kselftest run
@@ -275,11 +285,7 @@ done
 # Online all hot-pluggable memory
 #
 echo 0 > $NOTIFIER_ERR_INJECT_DIR/actions/MEM_GOING_ONLINE/error
-for memory in `hotpluggable_offline_memory`; do
-	if ! online_memory_expect_success $memory; then
-		retval=1
-	fi
-done
+online_all_offline_memory

 #
 # Test memory hot-remove error handling (online => offline)
@@ -296,4 +302,9 @@ done
 echo 0 > $NOTIFIER_ERR_INJECT_DIR/actions/MEM_GOING_OFFLINE/error
 /sbin/modprobe -q -r memory-notifier-error-inject

+#
+# Restore memory before exit
+#
+online_all_offline_memory
+
 exit $retval
--
2.17.1

Re: [PATCH -next v4 2/3] selftests/memory-hotplug: Restore memory before exit

[David Hildenbrand]

On 27.09.22 05:28, Zhao Gongyi wrote:
> Some momory will be left in offline state when calling
> offline_memory_expect_fail() failed. Restore it before exit.
> 
> Signed-off-by: Zhao Gongyi <zhaogongyi@huawei.com>
> ---
>   .../memory-hotplug/mem-on-off-test.sh         | 21 ++++++++++++++-----
>   1 file changed, 16 insertions(+), 5 deletions(-)
> 
> diff --git a/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh b/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
> index 1d87611a7d52..91a7457616bb 100755
> --- a/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
> +++ b/tools/testing/selftests/memory-hotplug/mem-on-off-test.sh
> @@ -134,6 +134,16 @@ offline_memory_expect_fail()
>   	return 0
>   }
> 
> +online_all_offline_memory()
> +{
> +	for memory in `hotpluggable_offline_memory`; do
> +		if ! online_memory_expect_success $memory; then
> +			echo "$FUNCNAME $memory: unexpected fail" >&2
> +			retval=1
> +		fi
> +	done
> +}
> +
>   error=-12
>   priority=0
>   # Run with default of ratio=2 for Kselftest run
> @@ -275,11 +285,7 @@ done
>   # Online all hot-pluggable memory
>   #
>   echo 0 > $NOTIFIER_ERR_INJECT_DIR/actions/MEM_GOING_ONLINE/error
> -for memory in `hotpluggable_offline_memory`; do
> -	if ! online_memory_expect_success $memory; then
> -		retval=1
> -	fi
> -done
> +online_all_offline_memory
> 
>   #
>   # Test memory hot-remove error handling (online => offline)
> @@ -296,4 +302,9 @@ done
>   echo 0 > $NOTIFIER_ERR_INJECT_DIR/actions/MEM_GOING_OFFLINE/error
>   /sbin/modprobe -q -r memory-notifier-error-inject
> 
> +#
> +# Restore memory before exit
> +#
> +online_all_offline_memory
> +
>   exit $retval
> --

Reviewed-by: David Hildenbrand <david@redhat.com>

-- 
Thanks,

David / dhildenb

[PATCH -next v4 3/3] docs: notifier-error-inject: Correct test's name

[Zhao Gongyi]

Correct test's name for mem-on-off-test.sh/cpu-on-off-test.sh.

Signed-off-by: Zhao Gongyi <zhaogongyi@huawei.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
---
 Documentation/fault-injection/notifier-error-inject.rst | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/Documentation/fault-injection/notifier-error-inject.rst b/Documentation/fault-injection/notifier-error-inject.rst
index 1668b6e48d3a..fdf2dc433ead 100644
--- a/Documentation/fault-injection/notifier-error-inject.rst
+++ b/Documentation/fault-injection/notifier-error-inject.rst
@@ -91,8 +91,8 @@ For more usage examples
 There are tools/testing/selftests using the notifier error injection features
 for CPU and memory notifiers.

- * tools/testing/selftests/cpu-hotplug/on-off-test.sh
- * tools/testing/selftests/memory-hotplug/on-off-test.sh
+ * tools/testing/selftests/cpu-hotplug/cpu-on-off-test.sh
+ * tools/testing/selftests/memory-hotplug/mem-on-off-test.sh

 These scripts first do simple online and offline tests and then do fault
 injection tests if notifier error injection module is available.
--
2.17.1