Skip to contentSkip to conversation

generic: test PF_MEMALLOC interfering with accounting file write

PATCH In review Eric Biggers 2 participants 4 messages Apr 3, 2020

lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 generic: test PF_MEMALLOC interfering with accounting file write

From: Eric Biggers <[email protected]>

Add a regression test for the bug fixed by commit 10a98cb16d80 ("xfs:
clear PF_MEMALLOC before exiting xfsaild thread").

---

Changed since v1:
- Fix _require_bsd_process_accounting() to not leave process accounting enabled.
- Removed RFC tag since the kernel fix is now in mainline.

 common/config         |  1 +
 common/rc             | 12 +++++++++
 tests/generic/901     | 60 +++++++++++++++++++++++++++++++++++++++++++
 tests/generic/901.out |  2 ++
 tests/generic/group   |  1 +
 5 files changed, 76 insertions(+)
 create mode 100644 tests/generic/901
 create mode 100644 tests/generic/901.out
common/config +1 -0
index 1116cb99..8023273d 100644
@@ -220,6 +220,7 @@ export DUPEREMOVE_PROG="$(type -P duperemove)"
 export CC_PROG="$(type -P cc)"
 export FSVERITY_PROG="$(type -P fsverity)"
 export OPENSSL_PROG="$(type -P openssl)"
+export ACCTON_PROG="$(type -P accton)"
 
 # use 'udevadm settle' or 'udevsettle' to wait for lv to be settled.
 # newer systems have udevadm command but older systems like RHEL5 don't.
common/rc +12 -0
index 454f5ccf..7d6ea90c 100644
@@ -4168,6 +4168,18 @@ _check_xfs_scrub_does_unicode() {
 	return 0
 }
 
+# Require the 'accton' userspace tool and CONFIG_BSD_PROCESS_ACCT=y.
+_require_bsd_process_accounting()
+{
+	_require_command "$ACCTON_PROG" accton
+	$ACCTON_PROG on &> $tmp.test_accton
+	cat $tmp.test_accton >> $seqres.full
+	if grep 'Function not implemented' $tmp.test_accton; then
+		_notrun "BSD process accounting support unavailable"
+	fi
+	$ACCTON_PROG off >> $seqres.full
+}
+
 init_rc
 
 ################################################################################
tests/generic/901 new file +60 -0

60 lines not on this page · raw message

tests/generic/901.out new file +2 -0
new file mode 100644
index 00000000..b206bc11
@@ -0,0 +1,2 @@
+QA output created by 901
+Silence is golden
tests/generic/group +1 -0
index dc95b77b..61a67979 100644
@@ -595,3 +595,4 @@
 591 auto quick rw pipe splice
 592 auto quick encrypt
 593 auto quick encrypt
+901 auto quick
Signature · 1 line
2.26.0
Reviewed-by on v2 1/1
#1 lore raw
Quoted 108 lines · Eric Biggers

From: Eric Biggers <[email protected]>

Add a regression test for the bug fixed by commit 10a98cb16d80 ("xfs:
clear PF_MEMALLOC before exiting xfsaild thread").

---

Changed since v1:
- Fix _require_bsd_process_accounting() to not leave process accounting enabled.
- Removed RFC tag since the kernel fix is now in mainline.

 common/config         |  1 +
 common/rc             | 12 +++++++++
 tests/generic/901     | 60 +++++++++++++++++++++++++++++++++++++++++++
 tests/generic/901.out |  2 ++
 tests/generic/group   |  1 +
 5 files changed, 76 insertions(+)
 create mode 100644 tests/generic/901
 create mode 100644 tests/generic/901.out
common/config +1 -0
index 1116cb99..8023273d 100644
@@ -220,6 +220,7 @@ export DUPEREMOVE_PROG="$(type -P duperemove)"
 export CC_PROG="$(type -P cc)"
 export FSVERITY_PROG="$(type -P fsverity)"
 export OPENSSL_PROG="$(type -P openssl)"
+export ACCTON_PROG="$(type -P accton)"
 
 # use 'udevadm settle' or 'udevsettle' to wait for lv to be settled.
 # newer systems have udevadm command but older systems like RHEL5 don't.
common/rc +12 -0
index 454f5ccf..7d6ea90c 100644
@@ -4168,6 +4168,18 @@ _check_xfs_scrub_does_unicode() {
 	return 0
 }
 
+# Require the 'accton' userspace tool and CONFIG_BSD_PROCESS_ACCT=y.
+_require_bsd_process_accounting()
+{
+	_require_command "$ACCTON_PROG" accton
+	$ACCTON_PROG on &> $tmp.test_accton
+	cat $tmp.test_accton >> $seqres.full
+	if grep 'Function not implemented' $tmp.test_accton; then
+		_notrun "BSD process accounting support unavailable"
+	fi
+	$ACCTON_PROG off >> $seqres.full
+}
+
 init_rc
 
 ################################################################################
tests/generic/901 new file +44 -0
new file mode 100644
index 00000000..c59300f1
@@ -0,0 +1,60 @@
+#! /bin/bash
+# SPDX-License-Identifier: GPL-2.0
+# Copyright 2020 Google LLC
+#
+# FS QA Test No. 901
+#
+# Regression test for the bug fixed by commit 10a98cb16d80 ("xfs: clear
+# PF_MEMALLOC before exiting xfsaild thread").  If the bug exists, a kernel
+# WARNING should be triggered.  See the commit message for details.
+#
+seq=`basename $0`
+seqres=$RESULT_DIR/$seq
+echo "QA output created by $seq"
+
+here=`pwd`
+tmp=/tmp/$$
+status=1	# failure is the default!
+trap "_cleanup; exit \$status" 0 1 2 3 15
+
+_cleanup()
+{
+	$ACCTON_PROG off >> $seqres.full
+	rm -f $tmp.*
+}
+
+# get standard environment, filters and checks
+. ./common/rc
+. ./common/filter
+
+# remove previous $seqres.full before test
+rm -f $seqres.full
+
+# real QA test starts here
+_supported_fs generic
+_supported_os Linux
+_require_bsd_process_accounting
+_require_chattr S
+_require_test
+_require_scratch
+
+# To trigger the bug we must unmount a filesystem while BSD process accounting
+# is enabled.  The accounting file must also be located on a different
+# filesystem and have the sync flag set.
+

tests/generic/901 line 54

Quoted 10 lines
tests/generic/901 new file +10 -0
+accounting_file=$TEST_DIR/$seq
+
+rm -f $accounting_file
+touch $accounting_file
+$CHATTR_PROG +S $accounting_file
+
+_scratch_mkfs &>> $seqres.full
+$ACCTON_PROG $accounting_file >> $seqres.full
+_scratch_mount
+_scratch_unmount

_scratch_cycle_mount

With that fixed,

Thanks for submitting a regression test!

--D

Quoted 26 lines
tests/generic/901 new file +6 -0
+$ACCTON_PROG off >> $seqres.full
+
+echo "Silence is golden"
+
+status=0
+exit
tests/generic/901.out new file +2 -0
new file mode 100644
index 00000000..b206bc11
@@ -0,0 +1,2 @@
+QA output created by 901
+Silence is golden
tests/generic/group +1 -0
index dc95b77b..61a67979 100644
@@ -595,3 +595,4 @@
 591 auto quick rw pipe splice
 592 auto quick encrypt
 593 auto quick encrypt
+901 auto quick
Signature · 1 line
2.26.0
#2 lore raw
Quoted 8 lines · Darrick J. Wong

tests/generic/901 line 54

Quoted 2 lines
tests/generic/901 new file +2 -0
+_scratch_mount
+_scratch_unmount

_scratch_cycle_mount

With that fixed,

That doesn't work because _scratch_cycle_mount does unmount, then mount. This test wants mount, then unmount.

- Eric

#3 lore raw
Quoted 6 lines · Eric Biggers
Quoted 5 lines · Darrick J. Wong

tests/generic/901 line 54

Quoted 2 lines
tests/generic/901 new file +2 -0
+_scratch_mount
+_scratch_unmount

_scratch_cycle_mount

Quoted 6 lines
Quoted 3 lines

With that fixed,

That doesn't work because _scratch_cycle_mount does unmount, then mount. This test wants mount, then unmount.

LOL, and I even stumbled over the same thing last time. :(

Test looks fine as it is, and you can still add my RVB.

Ok, time to go do something else.

--D

--D

Quoted 1 line

- Eric