Skip to contentSkip to conversation

selftests: Fix runner.sh breakage

PATCH In review Mark Brown 2 participants 4 messages Apr 16 Kselftest

lore
Share & tools
lore

Mark Brown

author
#0 lore raw

cover selftests: Fix runner.sh breakage

Commit 2964f6b816c2 ("selftests: Use ktap helpers for runner.sh") introduced several regressions by introducing dependencies on bash and GNU coreutils breaking testing on Debian based systems. Roll these changes back.

---
Mark Brown (2):
selftests: Fix runner.sh busybox support
selftests: Fix runner.sh for non-bash shells

 tools/testing/selftests/kselftest/runner.sh | 26 ++++++++++++++------------
 1 file changed, 14 insertions(+), 12 deletions(-)

---
base-commit: 2964f6b816c25ee094df4a143eb5b8828910045f
change-id: 20260416-selftest-fix-readlink-e-da604f02686b

Best regards,

Signature · 1 line
Mark Brown <[email protected]>
author
#1 lore raw

PATCH 1/2 selftests: Fix runner.sh busybox support

Commit 2964f6b816c2 ("selftests: Use ktap helpers for runner.sh") added an import of ktap_helper.sh to runner.sh in order to standardise on these for output formatting. Rather than build on the existing requirement for the user to supply BASE_DIR to find the helpers it uses some magic which features a use of "readlink -e". Unfortunately the -e option is a GNU extension and is not available in at least busybox, meaning that runner.sh starts failing:

./run_kselftest.sh: 5: ./kselftest/runner.sh: Bad substitution ./run_kselftest.sh: 5: .: cannot open ./ktap_helpers.sh: No such file

Fix this by using the already required BASE_DIR to locate the helper library.

Fixes 2964f6b816c2 ("selftests: Use ktap helpers for runner.sh")
---
 tools/testing/selftests/kselftest/runner.sh | 18 ++++++++++--------
 1 file changed, 10 insertions(+), 8 deletions(-)
tools/testing/selftests/kselftest/runner.sh +10 -8
index 3eeec93c9da4..4dfc4271e4ec 100644
@@ -2,7 +2,16 @@
 # SPDX-License-Identifier: GPL-2.0
 #
 # Runs a set of tests in a given subdirectory.
-. $(dirname "$(readlink -e "${BASH_SOURCE[0]}")")/ktap_helpers.sh
+
+# There isn't a shell-agnostic way to find the path of a sourced file,
+# so we must rely on BASE_DIR being set to find other tools.
+if [ -z "$BASE_DIR" ]; then
+	echo "Error: BASE_DIR must be set before sourcing." >&2
+	exit 1
+fi
+
+. ${BASE_DIR}/kselftest/ktap_helpers.sh
+
 export timeout_rc=124
 export logfile=/dev/stdout
 export per_test_logging=
@@ -13,13 +22,6 @@ export RUN_IN_NETNS=
 # over our soft timeout limit.
 export kselftest_default_timeout=45
 
-# There isn't a shell-agnostic way to find the path of a sourced file,
-# so we must rely on BASE_DIR being set to find other tools.
-if [ -z "$BASE_DIR" ]; then
-	echo "Error: BASE_DIR must be set before sourcing." >&2
-	exit 1
-fi
-
 TR_CMD=$(command -v tr)
 
 # If Perl is unavailable, we must fall back to line-at-a-time prefixing
Signature · 1 line
2.47.3
author
#2 lore raw

PATCH 2/2 selftests: Fix runner.sh for non-bash shells

In reply to Mark Brown

Commit 2964f6b816c2 ("selftests: Use ktap helpers for runner.sh") added a number of bashisms and updated the interpreter specified for the script to be /bin/bash to reflect this. Unfortunately this does not actually achieve anything in production since the main way runner.sh is invoked is from the top level run_kselftest.sh which sources it rather than running it as a separate script and specifies the shell as /bin/sh. This means that on systems where /bin/sh is not bash (such as Debian where /bin/sh defaults to being dash) we see failures:

./run_kselftest.sh: 195: ./kselftest/runner.sh: Syntax error: "(" unexpected (expecting "}")

These bashisms come from this part of the change:

4. In runner.sh run_one(), get the return value and use ktap helpers for all pass/fail reporting. This allows counting pass/fail numbers in the main process.

which uses a bash array to track all the subtests being run. Convert this to use a simple flat variable instead.

Fixes 2964f6b816c2 ("selftests: Use ktap helpers for runner.sh")
---
 tools/testing/selftests/kselftest/runner.sh | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
tools/testing/selftests/kselftest/runner.sh +4 -4
index 4dfc4271e4ec..50919ce206f3 100644
@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/bin/sh
 # SPDX-License-Identifier: GPL-2.0
 #
 # Runs a set of tests in a given subdirectory.
@@ -192,7 +192,7 @@ run_many()
 	DIR="${PWD#${BASE_DIR}/}"
 	test_num=0
 	local rc
-	pids=()
+	pids=
 
 	for TEST in "$@"; do
 		BASENAME_TEST=$(basename $TEST)
@@ -203,7 +203,7 @@ run_many()
 		fi
 		if [ -n "$RUN_IN_NETNS" ]; then
 			run_in_netns &
-			pids+=($!)
+			pids="$pids $!"
 		else
 			run_one "$DIR" "$TEST" "$test_num"
 		fi
@@ -211,7 +211,7 @@ run_many()
 
 	# These variables are outputs of ktap_helpers.sh but since we've
 	# run the test in a subprocess we need to update them manually
-	for pid in "${pids[@]}"; do
+	for pid in $pids; do
 		wait "$pid"
 		rc=$?
 		case "$rc" in
Signature · 1 line
2.47.3
maintainer
#3 lore raw
In reply to Mark Brown
Quoted 13 lines · Mark Brown

Commit 2964f6b816c2 ("selftests: Use ktap helpers for runner.sh") introduced several regressions by introducing dependencies on bash and GNU coreutils breaking testing on Debian based systems. Roll these changes back.

---
Mark Brown (2):
selftests: Fix runner.sh busybox support
selftests: Fix runner.sh for non-bash shells

  tools/testing/selftests/kselftest/runner.sh | 26 ++++++++++++++------------
  1 file changed, 14 insertions(+), 12 deletions(-)
Quoted 8 lines

---
base-commit: 2964f6b816c25ee094df4a143eb5b8828910045f
change-id: 20260416-selftest-fix-readlink-e-da604f02686b

Best regards,
--
Mark Brown <[email protected]>

Thank you. I applied these two patches to linux-kselftest next and will send pr tomorrow.

thanks,
-- Shuah