f2fs/025: test to do sanity check section type correctly in f2fs GC
PATCH 1/1 f2fs/025: test to do sanity check section type correctly in f2fs GC
Without commit 520760b9f915 ("f2fs: optimize representative type determination in GC"), f2fs GC will report inconsistent segment type in large section issue, and then it will force to shutdown filesystem.
[ 768.190903] F2FS-fs (loop51): Inconsistent segment (3) type [1, 0] in SIT and SSA
The reason is f2fs kernel will assume all segment type inside large section is the same, during GC it loads type from one segment and compare it to other segments' type, however due to recovery flow, the chosen segment may has zero valid blocks w/ different segment type, since the segment is invalid(free) one, it will never be migrated, so that we should not treat such state as abnormal condition.
This testcase is created to simulate above condition to see whether f2fs kernel module can handle it correctly
---
v2:
- clear MKFS_OPTIONS and MOUNT_OPTIONS to guarantee block allocation is as expected.
tests/f2fs/025 | 92 ++++++++++++++++++++++++++++++++++++++++++++++ tests/f2fs/025.out | 2 + 2 files changed, 94 insertions(+) create mode 100644 tests/f2fs/025 create mode 100644 tests/f2fs/025.out
tests/f2fs/025 new file +92 -0
@@ -0,0 +1,92 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2026 Chao Yu <[email protected]> +# +# FS QA Test No. f2fs/025 +# +# Check whether f2fs will encounter cp_error (Inconsistent segment type) +# when doing sanity check on type of segments inside large section during +# garbage collection. +# +. ./common/preamble +_begin_fstest auto quick + +_fixed_by_kernel_commit 520760b9f915 \ + "f2fs: optimize representative type determination in GC" + +. ./common/filter + +_cleanup() +{ + cd / + rm -r -f $tmp.* +} + +_require_scratch +_require_xfs_io_command "pwrite" +_require_xfs_io_command "truncate" +_require_command "$F2FS_IO_PROG" f2fs_io +_require_check_dmesg + +# Clear options to avoid interference from external configurations +export MKFS_OPTIONS="" +export MOUNT_OPTIONS="" + +# Format with 96MB size and 2 segments per section +_scratch_mkfs_sized $((96 * 1024 * 1024)) "" "-s 2" >> $seqres.full 2>&1 + +# Mount with mode=lfs +_scratch_mount -o mode=lfs >> $seqres.full 2>&1 + +# Create files to fill whole filesystem, then segment type will be changed to node type +for ((i=0;i<5120;i++)) do + touch $SCRATCH_MNT/$i >> $seqres.full 2>&1 +done +sync + +# Remove all files to create free(empty) node segments +rm -f $SCRATCH_MNT/* +sync + +# Allocate free space so that we have chance to reuse free(empty) node segments +$XFS_IO_PROG -f -c "pwrite -b 4k 0 1928k" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -d -c "pwrite -b 4k 0 8M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 32K" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 2M" -c "fsync" $SCRATCH_MNT/file >> $seqres.full 2>&1 + +# Shutdown the filesystem without checkpoint +$F2FS_IO_PROG shutdown 2 $SCRATCH_MNT >> $seqres.full 2>&1 + +_scratch_unmount >> $seqres.full 2>&1 + +_scratch_mount -o mode=lfs >> $seqres.full 2>&1 + +# Run urgent_gc mode to trigger garbage collection +dev_name=$(_short_dev $SCRATCH_DEV) +if [ -f /sys/fs/f2fs/$dev_name/gc_urgent ]; then + echo 1 > /sys/fs/f2fs/$dev_name/gc_urgent +fi + +# Wait background GC thread to wake up to run and potentially encounter the inconsistency +sleep 5 + +_scratch_unmount >> $seqres.full 2>&1 + +# Check whether the dmesg has the warning indicating the bug +_check_dmesg_for "F2FS-fs \($dev_name\): Inconsistent segment" && \ + _fail "F2FS-fs ($dev_name): Inconsistent segment type detected in dmesg!" + +echo "Silence is golden" +status=0 +exit
tests/f2fs/025.out new file +2 -0
@@ -0,0 +1,2 @@ +QA output created by 025 +Silence is golden
Signature · 1 line
2.49.0
Quoted 13 lines · Chao Yu
Without commit
520760b9f915("f2fs: optimize representative type determination in GC"), f2fs GC will report inconsistent segment type in large section issue, and then it will force to shutdown filesystem.[ 768.190903] F2FS-fs (loop51): Inconsistent segment (3) type [1, 0] in SIT and SSAThe reason is f2fs kernel will assume all segment type inside large section is the same, during GC it loads type from one segment and compare it to other segments' type, however due to recovery flow, the chosen segment may has zero valid blocks w/ different segment type, since the segment is invalid(free) one, it will never be migrated, so that we should not treat such state as abnormal condition.
Quoted 7 lines
This testcase is created to simulate above condition to see whether f2fs kernel module can handle it correctly
Signed-off-by Chao Yu <[email protected]>---
v2:
- clearMKFS_OPTIONSandMOUNT_OPTIONSto guarantee block allocation is as expected.
Hi Chao,
Sorry, I just noticed your reply to my review on the previous patch version. Due to some unexpected shake-ups recently, I’ve been bogged down with setting up and modifying various new system environments, and I accidentally marked some unread emails as read.
The patch looks good to me, with just a few picky review points below:
Quoted 25 lines
tests/f2fs/025 | 92 ++++++++++++++++++++++++++++++++++++++++++++++ tests/f2fs/025.out | 2 + 2 files changed, 94 insertions(+) create mode 100644 tests/f2fs/025 create mode 100644 tests/f2fs/025.outtests/f2fs/025 new file +13
-0@@ -0,0 +1,92 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2026 Chao Yu <[email protected]> +# +# FS QA Test No. f2fs/025 +# +# Check whether f2fs will encounter cp_error (Inconsistent segment type) +# when doing sanity check on type of segments inside large section during +# garbage collection. +# +. ./common/preamble +_begin_fstest auto quick +
tests/f2fs/025 line 23
Quoted 10 lines
tests/f2fs/025 new file +10
-0+_fixed_by_kernel_commit 520760b9f915 \ + "f2fs: optimize representative type determination in GC" + +. ./common/filter + +_cleanup() +{ + cd / + rm -r -f $tmp.* +}
This _cleanup() function is same as default. It can be removed.
Quoted 6 lines
tests/f2fs/025 new file +6
-0+ +_require_scratch +_require_xfs_io_command "pwrite" +_require_xfs_io_command "truncate" +_require_command "$F2FS_IO_PROG" f2fs_io +_require_check_dmesg
tests/f2fs/025 line 39
Quoted 10 lines
tests/f2fs/025 new file +10
-0+ +# Clear options to avoid interference from external configurations +export MKFS_OPTIONS="" +export MOUNT_OPTIONS="" + +# Format with 96MB size and 2 segments per section +_scratch_mkfs_sized $((96 * 1024 * 1024)) "" "-s 2" >> $seqres.full 2>&1 + +# Mount with mode=lfs +_scratch_mount -o mode=lfs >> $seqres.full 2>&1
^^^^^^^^^^^^^^^^^^^^
It's helpless, due to if _scratch_mount fails, it exit() directly.
Quoted 20 lines
tests/f2fs/025 new file +20
-0+ +# Create files to fill whole filesystem, then segment type will be changed to node type +for ((i=0;i<5120;i++)) do + touch $SCRATCH_MNT/$i >> $seqres.full 2>&1 +done +sync + +# Remove all files to create free(empty) node segments +rm -f $SCRATCH_MNT/* +sync + +# Allocate free space so that we have chance to reuse free(empty) node segments +$XFS_IO_PROG -f -c "pwrite -b 4k 0 1928k" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1
tests/f2fs/025 line 69
Quoted 10 lines
tests/f2fs/025 new file +10
-0+sync + +$XFS_IO_PROG -d -c "pwrite -b 4k 0 8M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 32K" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 2M" -c "fsync" $SCRATCH_MNT/file >> $seqres.full 2>&1 + +# Shutdown the filesystem without checkpoint +$F2FS_IO_PROG shutdown 2 $SCRATCH_MNT >> $seqres.full 2>&1
I'm wondering if we can have f2fs supporting in common _scratch_shutdown helper :)
tests/f2fs/025 line 71
Quoted 2 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1
^^^^^^^^^^^^^^^^^^^^
If unmount fails, how about let it output the errors, to break the golden image?
tests/f2fs/025 line 73
Quoted 2 lines
tests/f2fs/025 new file +2
-0+ +_scratch_mount -o mode=lfs >> $seqres.full 2>&1
^^^^^^^^^^^^^^^^^^^^
helplesstests/f2fs/025 line 79
Quoted 6 lines
tests/f2fs/025 new file +6
-0+ +# Run urgent_gc mode to trigger garbage collection +dev_name=$(_short_dev $SCRATCH_DEV) +if [ -f /sys/fs/f2fs/$dev_name/gc_urgent ]; then + echo 1 > /sys/fs/f2fs/$dev_name/gc_urgent +fi
Hmm... what if there's not /sys/fs/f2fs/$dev_name/gc_urgent? Does it affect the test result?
If it does, this's a necessary requirement for this test, we shouldn't ignore it and keep running. Does $F2FS_IO_PROG provide a command to make a force GC? Or we need to check this file and _notrun if it's not existed.
tests/f2fs/025 line 82
Quoted 3 lines
tests/f2fs/025 new file +3
-0+ +# Wait background GC thread to wake up to run and potentially encounter the inconsistency +sleep 5
Does this sleep try to wait above "echo 1 > /sys/fs/f2fs/$dev_name/gc_urgent"? If so, it makes more sense to move it into the "if-then" logic.
tests/f2fs/025 line 84
Quoted 2 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1
^^^^^^^^^^^^^^^^^^^^
Same as above.
tests/f2fs/025 line 92
Quoted 8 lines
tests/f2fs/025 new file +8
-0+ +# Check whether the dmesg has the warning indicating the bug +_check_dmesg_for "F2FS-fs \($dev_name\): Inconsistent segment" && \ + _fail "F2FS-fs ($dev_name): Inconsistent segment type detected in dmesg!" + +echo "Silence is golden" +status=0 +exit
We've replaced "status=0;exit;" with "_exit 0".
Thanks,
Zorro
Quoted 11 lines
tests/f2fs/025.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 025 +Silence is goldenSignature · 1 line
2.49.0
Quoted 22 lines · Zorro Lang
Quoted 13 lines · Chao Yu
Without commit
520760b9f915("f2fs: optimize representative type determination in GC"), f2fs GC will report inconsistent segment type in large section issue, and then it will force to shutdown filesystem.[ 768.190903] F2FS-fs (loop51): Inconsistent segment (3) type [1, 0] in SIT and SSAThe reason is f2fs kernel will assume all segment type inside large section is the same, during GC it loads type from one segment and compare it to other segments' type, however due to recovery flow, the chosen segment may has zero valid blocks w/ different segment type, since the segment is invalid(free) one, it will never be migrated, so that we should not treat such state as abnormal condition.
Quoted 7 lines
This testcase is created to simulate above condition to see whether f2fs kernel module can handle it correctly
Signed-off-by Chao Yu <[email protected]>---
v2:
- clearMKFS_OPTIONSandMOUNT_OPTIONSto guarantee block allocation is as expected.
Quoted 6 lines
Hi Chao,
Sorry, I just noticed your reply to my review on the previous patch version. Due to some unexpected shake-ups recently, I’ve been bogged down with setting up and modifying various new system environments, and I accidentally marked some unread emails as read.
No worries. :)
Quoted 2 lines
The patch looks good to me, with just a few picky review points below:
Thanks Zorro for taking a look.
Quoted 30 lines
Quoted 30 lines
tests/f2fs/025 | 92 ++++++++++++++++++++++++++++++++++++++++++++++ tests/f2fs/025.out | 2 + 2 files changed, 94 insertions(+) create mode 100644 tests/f2fs/025 create mode 100644 tests/f2fs/025.outtests/f2fs/025 new file +17
-0@@ -0,0 +1,92 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2026 Chao Yu <[email protected]> +# +# FS QA Test No. f2fs/025 +# +# Check whether f2fs will encounter cp_error (Inconsistent segment type) +# when doing sanity check on type of segments inside large section during +# garbage collection. +# +. ./common/preamble +_begin_fstest auto quick + +_fixed_by_kernel_commit 520760b9f915 \ + "f2fs: optimize representative type determination in GC" + +. ./common/filter
Quoted 8 lines
tests/f2fs/025 line 23
Quoted 6 lines
tests/f2fs/025 new file +6
-0+ +_cleanup() +{ + cd / + rm -r -f $tmp.* +}This
_cleanup()function is same as default. It can be removed.
Will remove.
Quoted 18 lines
Quoted 7 lines
tests/f2fs/025 new file +6
-0+ +_require_scratch +_require_xfs_io_command "pwrite" +_require_xfs_io_command "truncate" +_require_command "$F2FS_IO_PROG" f2fs_io +_require_check_dmesgtests/f2fs/025 line 39
Quoted 10 lines
tests/f2fs/025 new file +10
-0+ +# Clear options to avoid interference from external configurations +export MKFS_OPTIONS="" +export MOUNT_OPTIONS="" + +# Format with 96MB size and 2 segments per section +_scratch_mkfs_sized $((96 * 1024 * 1024)) "" "-s 2" >> $seqres.full 2>&1 + +# Mount with mode=lfs +_scratch_mount -o mode=lfs >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^
Quoted 1 line
It's helpless, due to if
_scratch_mountfails, itexit()directly.
Right, will fix.
Quoted 26 lines
Quoted 26 lines
tests/f2fs/025 new file +25
-0+ +# Create files to fill whole filesystem, then segment type will be changed to node type +for ((i=0;i<5120;i++)) do + touch $SCRATCH_MNT/$i >> $seqres.full 2>&1 +done +sync + +# Remove all files to create free(empty) node segments +rm -f $SCRATCH_MNT/* +sync + +# Allocate free space so that we have chance to reuse free(empty) node segments +$XFS_IO_PROG -f -c "pwrite -b 4k 0 1928k" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -d -c "pwrite -b 4k 0 8M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 32K" $SCRATCH_MNT/file >> $seqres.full 2>&1
Quoted 8 lines
tests/f2fs/025 line 69
Quoted 5 lines
tests/f2fs/025 new file +5
-0+$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 2M" -c "fsync" $SCRATCH_MNT/file >> $seqres.full 2>&1 + +# Shutdown the filesystem without checkpoint +$F2FS_IO_PROG shutdown 2 $SCRATCH_MNT >> $seqres.full 2>&1I'm wondering if we can have f2fs supporting in common
_scratch_shutdownhelper :)
I think we can change f2fs testcase to use _scratch_shutdown because the definition of nologflush shutdown interface is the same as xfs':
/*
* should be same as XFS_IOC_GOINGDOWN.
* Flags for going down operation used by FS_IOC_GOINGDOWN */
#define F2FS_IOC_SHUTDOWN _IOR('X', 125, __u32) /* Shutdown */
#define F2FS_GOING_DOWN_NOSYNC 0x2 /* going down */
#define XFS_IOC_GOINGDOWN _IOR ('X', 125, uint32_t)
#define XFS_FSOP_GOING_FLAGS_NOLOGFLUSH 0x2 /* don't flush log nor data */
Quoted 6 lines
tests/f2fs/025 line 71
Quoted 3 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^If unmount fails, how about let it output the errors, to break the golden image?
Yes, it's better.
Quoted 5 lines
tests/f2fs/025 line 73
Quoted 3 lines
tests/f2fs/025 new file +2
-0+ +_scratch_mount -o mode=lfs >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^ helpless
Will fix.
Quoted 11 lines
tests/f2fs/025 line 79
Quoted 7 lines
tests/f2fs/025 new file +6
-0+ +# Run urgent_gc mode to trigger garbage collection +dev_name=$(_short_dev $SCRATCH_DEV) +if [ -f /sys/fs/f2fs/$dev_name/gc_urgent ]; then + echo 1 > /sys/fs/f2fs/$dev_name/gc_urgent +fiHmm... what if there's not
/sys/fs/f2fs/$dev_name/gc_urgent? Does it affect the test result?
Quoted 4 lines
If it does, this's a necessary requirement for this test, we shouldn't ignore it and keep running. Does $F2FS_IO_PROG provide a command to make a force GC? Or we need to check this file and
_notrunif it's not existed.
Ah, right, that's good point!
We can use "$F2FS_IO_PROG gc_urgent <dev_name> run 5" instead, it will do below commands:
1. echo 1 > /sys/fs/f2fs/$dev_name/gc_urgent
2. sleep 5 seconds
3. echo 0 > /sys/fs/f2fs/$dev_name/gc_urgent
Quoted 8 lines
tests/f2fs/025 line 82
Quoted 4 lines
tests/f2fs/025 new file +3
-0+ +# Wait background GC thread to wake up to run and potentially encounter the inconsistency +sleep 5Does this sleep try to wait above "echo 1 >
/sys/fs/f2fs/$dev_name/gc_urgent"? If so, it makes more sense to move it into the "if-then" logic.
Quoted 4 lines
tests/f2fs/025 line 84
Quoted 2 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^Same as above.
Will fix.
Quoted 11 lines
tests/f2fs/025 line 92
Quoted 9 lines
tests/f2fs/025 new file +8
-0+ +# Check whether the dmesg has the warning indicating the bug +_check_dmesg_for "F2FS-fs \($dev_name\): Inconsistent segment" && \ + _fail "F2FS-fs ($dev_name): Inconsistent segment type detected in dmesg!" + +echo "Silence is golden" +status=0 +exitWe've replaced "status=0;exit;" with "
_exit0".
Will fix.
Thanks,
Quoted 15 lines
Thanks,
ZorroQuoted 11 lines
tests/f2fs/025.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 025 +Silence is goldenSignature · 1 line
2.49.0
Quoted 176 lines · Chao Yu
Quoted 22 lines · Zorro Lang
Quoted 13 lines · Chao Yu
Without commit
520760b9f915("f2fs: optimize representative type determination in GC"), f2fs GC will report inconsistent segment type in large section issue, and then it will force to shutdown filesystem.[ 768.190903] F2FS-fs (loop51): Inconsistent segment (3) type [1, 0] in SIT and SSAThe reason is f2fs kernel will assume all segment type inside large section is the same, during GC it loads type from one segment and compare it to other segments' type, however due to recovery flow, the chosen segment may has zero valid blocks w/ different segment type, since the segment is invalid(free) one, it will never be migrated, so that we should not treat such state as abnormal condition.
Quoted 7 lines
This testcase is created to simulate above condition to see whether f2fs kernel module can handle it correctly
Signed-off-by Chao Yu <[email protected]>---
v2:
- clearMKFS_OPTIONSandMOUNT_OPTIONSto guarantee block allocation is as expected.Quoted 6 lines
Hi Chao,
Sorry, I just noticed your reply to my review on the previous patch version. Due to some unexpected shake-ups recently, I’ve been bogged down with setting up and modifying various new system environments, and I accidentally marked some unread emails as read.
No worries. :)
Quoted 2 lines
The patch looks good to me, with just a few picky review points below:
Thanks Zorro for taking a look.
Quoted 30 lines
Quoted 30 lines
tests/f2fs/025 | 92 ++++++++++++++++++++++++++++++++++++++++++++++ tests/f2fs/025.out | 2 + 2 files changed, 94 insertions(+) create mode 100644 tests/f2fs/025 create mode 100644 tests/f2fs/025.outtests/f2fs/025 new file +17
-0@@ -0,0 +1,92 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2026 Chao Yu <[email protected]> +# +# FS QA Test No. f2fs/025 +# +# Check whether f2fs will encounter cp_error (Inconsistent segment type) +# when doing sanity check on type of segments inside large section during +# garbage collection. +# +. ./common/preamble +_begin_fstest auto quick + +_fixed_by_kernel_commit 520760b9f915 \ + "f2fs: optimize representative type determination in GC" + +. ./common/filterQuoted 8 lines
tests/f2fs/025 line 23
Quoted 6 lines
tests/f2fs/025 new file +6
-0+ +_cleanup() +{ + cd / + rm -r -f $tmp.* +}This
_cleanup()function is same as default. It can be removed.Will remove.
Quoted 18 lines
Quoted 7 lines
tests/f2fs/025 new file +6
-0+ +_require_scratch +_require_xfs_io_command "pwrite" +_require_xfs_io_command "truncate" +_require_command "$F2FS_IO_PROG" f2fs_io +_require_check_dmesgtests/f2fs/025 line 39
Quoted 10 lines
tests/f2fs/025 new file +10
-0+ +# Clear options to avoid interference from external configurations +export MKFS_OPTIONS="" +export MOUNT_OPTIONS="" + +# Format with 96MB size and 2 segments per section +_scratch_mkfs_sized $((96 * 1024 * 1024)) "" "-s 2" >> $seqres.full 2>&1 + +# Mount with mode=lfs +_scratch_mount -o mode=lfs >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^Quoted 1 line
It's helpless, due to if
_scratch_mountfails, itexit()directly.Right, will fix.
Quoted 26 lines
Quoted 26 lines
tests/f2fs/025 new file +25
-0+ +# Create files to fill whole filesystem, then segment type will be changed to node type +for ((i=0;i<5120;i++)) do + touch $SCRATCH_MNT/$i >> $seqres.full 2>&1 +done +sync + +# Remove all files to create free(empty) node segments +rm -f $SCRATCH_MNT/* +sync + +# Allocate free space so that we have chance to reuse free(empty) node segments +$XFS_IO_PROG -f -c "pwrite -b 4k 0 1928k" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -d -c "pwrite -b 4k 0 8M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 32K" $SCRATCH_MNT/file >> $seqres.full 2>&1Quoted 8 lines
tests/f2fs/025 line 69
Quoted 5 lines
tests/f2fs/025 new file +5
-0+$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 2M" -c "fsync" $SCRATCH_MNT/file >> $seqres.full 2>&1 + +# Shutdown the filesystem without checkpoint +$F2FS_IO_PROG shutdown 2 $SCRATCH_MNT >> $seqres.full 2>&1I'm wondering if we can have f2fs supporting in common
_scratch_shutdownhelper :)I think we can change f2fs testcase to use
_scratch_shutdownbecause the definition of nologflush shutdown interface is the same as xfs':/*
* should be same asXFS_IOC_GOINGDOWN.
* Flags for going down operation used byFS_IOC_GOINGDOWN*/
#defineF2FS_IOC_SHUTDOWN_IOR('X', 125, __u32)/* Shutdown */
#defineF2FS_GOING_DOWN_NOSYNC0x2/* going down */#define
XFS_IOC_GOINGDOWN_IOR('X', 125,uint32_t)
#defineXFS_FSOP_GOING_FLAGS_NOLOGFLUSH0x2/* don't flush log nor data */Quoted 6 lines
tests/f2fs/025 line 71
Quoted 3 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^If unmount fails, how about let it output the errors, to break the golden image?
Yes, it's better.
Quoted 5 lines
tests/f2fs/025 line 73
Quoted 3 lines
tests/f2fs/025 new file +2
-0+ +_scratch_mount -o mode=lfs >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^ helplessWill fix.
Quoted 11 lines
tests/f2fs/025 line 79
Quoted 7 lines
tests/f2fs/025 new file +6
-0+ +# Run urgent_gc mode to trigger garbage collection +dev_name=$(_short_dev $SCRATCH_DEV) +if [ -f /sys/fs/f2fs/$dev_name/gc_urgent ]; then + echo 1 > /sys/fs/f2fs/$dev_name/gc_urgent +fiHmm... what if there's not
/sys/fs/f2fs/$dev_name/gc_urgent? Does it affect the test result?
Quoted 8 lines
Quoted 4 lines
If it does, this's a necessary requirement for this test, we shouldn't ignore it and keep running. Does $F2FS_IO_PROG provide a command to make a force GC? Or we need to check this file and
_notrunif it's not existed.Ah, right, that's good point!
We can use "$F2FS_IO_PROG
gc_urgent<dev_name> run 5" instead, it will do below commands:
Great, I just hope the *gc_urgent* isn't a new feature which needs something likes:_require_f2fs_io gc_urgent (there's not _require_f2fs_io:)
Quoted 3 lines
1. echo 1 >
/sys/fs/f2fs/$dev_name/gc_urgent
2. sleep 5 seconds
3. echo 0 >/sys/fs/f2fs/$dev_name/gc_urgent
It also depends on the /sys/fs/f2fs/$dev_name/gc_urgent too. So we have to face the same question:
If this file doesn't exist, should this test case _notrun?
Thanks,
Zorro
Quoted 49 lines
Quoted 9 lines
tests/f2fs/025 line 82
Quoted 5 lines
tests/f2fs/025 new file +3
-0+ +# Wait background GC thread to wake up to run and potentially encounter the inconsistency +sleep 5Does this sleep try to wait above "echo 1 >
/sys/fs/f2fs/$dev_name/gc_urgent"? If so, it makes more sense to move it into the "if-then" logic.Quoted 4 lines
tests/f2fs/025 line 84
Quoted 2 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^Same as above.
Will fix.
Quoted 11 lines
tests/f2fs/025 line 92
Quoted 9 lines
tests/f2fs/025 new file +8
-0+ +# Check whether the dmesg has the warning indicating the bug +_check_dmesg_for "F2FS-fs \($dev_name\): Inconsistent segment" && \ + _fail "F2FS-fs ($dev_name): Inconsistent segment type detected in dmesg!" + +echo "Silence is golden" +status=0 +exitWe've replaced "status=0;exit;" with "
_exit0".Will fix.
Thanks,
Quoted 15 lines
Thanks,
ZorroQuoted 11 lines
tests/f2fs/025.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 025 +Silence is goldenSignature · 1 line
2.49.0
Quoted 182 lines · Zorro Lang
Quoted 181 lines · Chao Yu
Quoted 22 lines · Zorro Lang
Quoted 13 lines · Chao Yu
Without commit
520760b9f915("f2fs: optimize representative type determination in GC"), f2fs GC will report inconsistent segment type in large section issue, and then it will force to shutdown filesystem.[ 768.190903] F2FS-fs (loop51): Inconsistent segment (3) type [1, 0] in SIT and SSAThe reason is f2fs kernel will assume all segment type inside large section is the same, during GC it loads type from one segment and compare it to other segments' type, however due to recovery flow, the chosen segment may has zero valid blocks w/ different segment type, since the segment is invalid(free) one, it will never be migrated, so that we should not treat such state as abnormal condition.
Quoted 7 lines
This testcase is created to simulate above condition to see whether f2fs kernel module can handle it correctly
Signed-off-by Chao Yu <[email protected]>---
v2:
- clearMKFS_OPTIONSandMOUNT_OPTIONSto guarantee block allocation is as expected.Quoted 6 lines
Hi Chao,
Sorry, I just noticed your reply to my review on the previous patch version. Due to some unexpected shake-ups recently, I’ve been bogged down with setting up and modifying various new system environments, and I accidentally marked some unread emails as read.
No worries. :)
Quoted 2 lines
The patch looks good to me, with just a few picky review points below:
Thanks Zorro for taking a look.
Quoted 30 lines
Quoted 30 lines
tests/f2fs/025 | 92 ++++++++++++++++++++++++++++++++++++++++++++++ tests/f2fs/025.out | 2 + 2 files changed, 94 insertions(+) create mode 100644 tests/f2fs/025 create mode 100644 tests/f2fs/025.outtests/f2fs/025 new file +17
-0@@ -0,0 +1,92 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2026 Chao Yu <[email protected]> +# +# FS QA Test No. f2fs/025 +# +# Check whether f2fs will encounter cp_error (Inconsistent segment type) +# when doing sanity check on type of segments inside large section during +# garbage collection. +# +. ./common/preamble +_begin_fstest auto quick + +_fixed_by_kernel_commit 520760b9f915 \ + "f2fs: optimize representative type determination in GC" + +. ./common/filterQuoted 8 lines
tests/f2fs/025 line 23
Quoted 6 lines
tests/f2fs/025 new file +6
-0+ +_cleanup() +{ + cd / + rm -r -f $tmp.* +}This
_cleanup()function is same as default. It can be removed.Will remove.
Quoted 18 lines
Quoted 7 lines
tests/f2fs/025 new file +6
-0+ +_require_scratch +_require_xfs_io_command "pwrite" +_require_xfs_io_command "truncate" +_require_command "$F2FS_IO_PROG" f2fs_io +_require_check_dmesgtests/f2fs/025 line 39
Quoted 10 lines
tests/f2fs/025 new file +10
-0+ +# Clear options to avoid interference from external configurations +export MKFS_OPTIONS="" +export MOUNT_OPTIONS="" + +# Format with 96MB size and 2 segments per section +_scratch_mkfs_sized $((96 * 1024 * 1024)) "" "-s 2" >> $seqres.full 2>&1 + +# Mount with mode=lfs +_scratch_mount -o mode=lfs >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^Quoted 1 line
It's helpless, due to if
_scratch_mountfails, itexit()directly.Right, will fix.
Quoted 26 lines
Quoted 26 lines
tests/f2fs/025 new file +25
-0+ +# Create files to fill whole filesystem, then segment type will be changed to node type +for ((i=0;i<5120;i++)) do + touch $SCRATCH_MNT/$i >> $seqres.full 2>&1 +done +sync + +# Remove all files to create free(empty) node segments +rm -f $SCRATCH_MNT/* +sync + +# Allocate free space so that we have chance to reuse free(empty) node segments +$XFS_IO_PROG -f -c "pwrite -b 4k 0 1928k" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -d -c "pwrite -b 4k 0 8M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 32K" $SCRATCH_MNT/file >> $seqres.full 2>&1Quoted 8 lines
tests/f2fs/025 line 69
Quoted 5 lines
tests/f2fs/025 new file +5
-0+$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 2M" -c "fsync" $SCRATCH_MNT/file >> $seqres.full 2>&1 + +# Shutdown the filesystem without checkpoint +$F2FS_IO_PROG shutdown 2 $SCRATCH_MNT >> $seqres.full 2>&1I'm wondering if we can have f2fs supporting in common
_scratch_shutdownhelper :)I think we can change f2fs testcase to use
_scratch_shutdownbecause the definition of nologflush shutdown interface is the same as xfs':/*
* should be same asXFS_IOC_GOINGDOWN.
* Flags for going down operation used byFS_IOC_GOINGDOWN*/
#defineF2FS_IOC_SHUTDOWN_IOR('X', 125, __u32)/* Shutdown */
#defineF2FS_GOING_DOWN_NOSYNC0x2/* going down */#define
XFS_IOC_GOINGDOWN_IOR('X', 125,uint32_t)
#defineXFS_FSOP_GOING_FLAGS_NOLOGFLUSH0x2/* don't flush log nor data */Quoted 6 lines
tests/f2fs/025 line 71
Quoted 3 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^If unmount fails, how about let it output the errors, to break the golden image?
Yes, it's better.
Quoted 5 lines
tests/f2fs/025 line 73
Quoted 3 lines
tests/f2fs/025 new file +2
-0+ +_scratch_mount -o mode=lfs >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^ helplessWill fix.
Quoted 15 lines
tests/f2fs/025 line 79
Quoted 7 lines
tests/f2fs/025 new file +6
-0+ +# Run urgent_gc mode to trigger garbage collection +dev_name=$(_short_dev $SCRATCH_DEV) +if [ -f /sys/fs/f2fs/$dev_name/gc_urgent ]; then + echo 1 > /sys/fs/f2fs/$dev_name/gc_urgent +fiHmm... what if there's not
/sys/fs/f2fs/$dev_name/gc_urgent? Does it affect the test result?If it does, this's a necessary requirement for this test, we shouldn't ignore it and keep running. Does $F2FS_IO_PROG provide a command to make a force GC? Or we need to check this file and
_notrunif it's not existed.
Quoted 5 lines
Quoted 3 lines
Ah, right, that's good point!
We can use "$F2FS_IO_PROG
gc_urgent<dev_name> run 5" instead, it will do below commands:Great, I just hope the *gc_urgent* isn't a new feature which needs something likes:
I guess it's not a new subcommand for f2fs_io,
commit 22d758e2e6af210dc9e6cdf99438f063383ba72f
Author: Jaegeuk Kim <[email protected]> Date: Tue Feb 19 19:07:21 2019 -0800
f2fs_io: add gc_urgent
e.g.,
f2fs_io gc_urgent dm-4 [start/end/run] [time in sec]
This controls sysfs/gc_urgent to run f2fs_gc urgently.
Quoted 1 line
_require_f2fs_io gc_urgent(there's not_require_f2fs_io:)
Agreed, we need to introduce _require_f2fs_io(), let me work on this.
Quoted 8 lines
Quoted 4 lines
1. echo 1 >
/sys/fs/f2fs/$dev_name/gc_urgent
2. sleep 5 seconds
3. echo 0 >/sys/fs/f2fs/$dev_name/gc_urgentIt also depends on the
/sys/fs/f2fs/$dev_name/gc_urgenttoo. So we have to face the same question:
If this file doesn't exist, should this test case_notrun?
Oh, right, maybe we can introduce _require_f2fs_sysfs() to check whether f2fs kernel module has supported target sysfs node?
Thanks,
Quoted 53 lines
Thanks,
ZorroQuoted 49 lines
Quoted 9 lines
tests/f2fs/025 line 82
Quoted 5 lines
tests/f2fs/025 new file +3
-0+ +# Wait background GC thread to wake up to run and potentially encounter the inconsistency +sleep 5Does this sleep try to wait above "echo 1 >
/sys/fs/f2fs/$dev_name/gc_urgent"? If so, it makes more sense to move it into the "if-then" logic.Quoted 4 lines
tests/f2fs/025 line 84
Quoted 2 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^Same as above.
Will fix.
Quoted 11 lines
tests/f2fs/025 line 92
Quoted 9 lines
tests/f2fs/025 new file +8
-0+ +# Check whether the dmesg has the warning indicating the bug +_check_dmesg_for "F2FS-fs \($dev_name\): Inconsistent segment" && \ + _fail "F2FS-fs ($dev_name): Inconsistent segment type detected in dmesg!" + +echo "Silence is golden" +status=0 +exitWe've replaced "status=0;exit;" with "
_exit0".Will fix.
Thanks,
Quoted 15 lines
Thanks,
ZorroQuoted 11 lines
tests/f2fs/025.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 025 +Silence is goldenSignature · 1 line
2.49.0
Quoted 200 lines · Chao Yu
Quoted 182 lines · Zorro Lang
Quoted 181 lines · Chao Yu
Quoted 22 lines · Zorro Lang
Quoted 13 lines · Chao Yu
Without commit
520760b9f915("f2fs: optimize representative type determination in GC"), f2fs GC will report inconsistent segment type in large section issue, and then it will force to shutdown filesystem.[ 768.190903] F2FS-fs (loop51): Inconsistent segment (3) type [1, 0] in SIT and SSAThe reason is f2fs kernel will assume all segment type inside large section is the same, during GC it loads type from one segment and compare it to other segments' type, however due to recovery flow, the chosen segment may has zero valid blocks w/ different segment type, since the segment is invalid(free) one, it will never be migrated, so that we should not treat such state as abnormal condition.
Quoted 7 lines
This testcase is created to simulate above condition to see whether f2fs kernel module can handle it correctly
Signed-off-by Chao Yu <[email protected]>---
v2:
- clearMKFS_OPTIONSandMOUNT_OPTIONSto guarantee block allocation is as expected.Quoted 6 lines
Hi Chao,
Sorry, I just noticed your reply to my review on the previous patch version. Due to some unexpected shake-ups recently, I’ve been bogged down with setting up and modifying various new system environments, and I accidentally marked some unread emails as read.
No worries. :)
Quoted 2 lines
The patch looks good to me, with just a few picky review points below:
Thanks Zorro for taking a look.
Quoted 30 lines
Quoted 30 lines
tests/f2fs/025 | 92 ++++++++++++++++++++++++++++++++++++++++++++++ tests/f2fs/025.out | 2 + 2 files changed, 94 insertions(+) create mode 100644 tests/f2fs/025 create mode 100644 tests/f2fs/025.outtests/f2fs/025 new file +17
-0@@ -0,0 +1,92 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2026 Chao Yu <[email protected]> +# +# FS QA Test No. f2fs/025 +# +# Check whether f2fs will encounter cp_error (Inconsistent segment type) +# when doing sanity check on type of segments inside large section during +# garbage collection. +# +. ./common/preamble +_begin_fstest auto quick + +_fixed_by_kernel_commit 520760b9f915 \ + "f2fs: optimize representative type determination in GC" + +. ./common/filterQuoted 8 lines
tests/f2fs/025 line 23
Quoted 6 lines
tests/f2fs/025 new file +6
-0+ +_cleanup() +{ + cd / + rm -r -f $tmp.* +}This
_cleanup()function is same as default. It can be removed.Will remove.
Quoted 18 lines
Quoted 7 lines
tests/f2fs/025 new file +6
-0+ +_require_scratch +_require_xfs_io_command "pwrite" +_require_xfs_io_command "truncate" +_require_command "$F2FS_IO_PROG" f2fs_io +_require_check_dmesgtests/f2fs/025 line 39
Quoted 10 lines
tests/f2fs/025 new file +10
-0+ +# Clear options to avoid interference from external configurations +export MKFS_OPTIONS="" +export MOUNT_OPTIONS="" + +# Format with 96MB size and 2 segments per section +_scratch_mkfs_sized $((96 * 1024 * 1024)) "" "-s 2" >> $seqres.full 2>&1 + +# Mount with mode=lfs +_scratch_mount -o mode=lfs >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^Quoted 1 line
It's helpless, due to if
_scratch_mountfails, itexit()directly.Right, will fix.
Quoted 26 lines
Quoted 26 lines
tests/f2fs/025 new file +25
-0+ +# Create files to fill whole filesystem, then segment type will be changed to node type +for ((i=0;i<5120;i++)) do + touch $SCRATCH_MNT/$i >> $seqres.full 2>&1 +done +sync + +# Remove all files to create free(empty) node segments +rm -f $SCRATCH_MNT/* +sync + +# Allocate free space so that we have chance to reuse free(empty) node segments +$XFS_IO_PROG -f -c "pwrite -b 4k 0 1928k" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -d -c "pwrite -b 4k 0 8M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 32K" $SCRATCH_MNT/file >> $seqres.full 2>&1Quoted 8 lines
tests/f2fs/025 line 69
Quoted 5 lines
tests/f2fs/025 new file +5
-0+$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 2M" -c "fsync" $SCRATCH_MNT/file >> $seqres.full 2>&1 + +# Shutdown the filesystem without checkpoint +$F2FS_IO_PROG shutdown 2 $SCRATCH_MNT >> $seqres.full 2>&1I'm wondering if we can have f2fs supporting in common
_scratch_shutdownhelper :)I think we can change f2fs testcase to use
_scratch_shutdownbecause the definition of nologflush shutdown interface is the same as xfs':/*
* should be same asXFS_IOC_GOINGDOWN.
* Flags for going down operation used byFS_IOC_GOINGDOWN*/
#defineF2FS_IOC_SHUTDOWN_IOR('X', 125, __u32)/* Shutdown */
#defineF2FS_GOING_DOWN_NOSYNC0x2/* going down */#define
XFS_IOC_GOINGDOWN_IOR('X', 125,uint32_t)
#defineXFS_FSOP_GOING_FLAGS_NOLOGFLUSH0x2/* don't flush log nor data */Quoted 6 lines
tests/f2fs/025 line 71
Quoted 3 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^If unmount fails, how about let it output the errors, to break the golden image?
Yes, it's better.
Quoted 5 lines
tests/f2fs/025 line 73
Quoted 3 lines
tests/f2fs/025 new file +2
-0+ +_scratch_mount -o mode=lfs >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^ helplessWill fix.
Quoted 15 lines
tests/f2fs/025 line 79
Quoted 7 lines
tests/f2fs/025 new file +6
-0+ +# Run urgent_gc mode to trigger garbage collection +dev_name=$(_short_dev $SCRATCH_DEV) +if [ -f /sys/fs/f2fs/$dev_name/gc_urgent ]; then + echo 1 > /sys/fs/f2fs/$dev_name/gc_urgent +fiHmm... what if there's not
/sys/fs/f2fs/$dev_name/gc_urgent? Does it affect the test result?If it does, this's a necessary requirement for this test, we shouldn't ignore it and keep running. Does $F2FS_IO_PROG provide a command to make a force GC? Or we need to check this file and
_notrunif it's not existed.Quoted 5 lines
Quoted 3 lines
Ah, right, that's good point!
We can use "$F2FS_IO_PROG
gc_urgent<dev_name> run 5" instead, it will do below commands:Great, I just hope the *gc_urgent* isn't a new feature which needs something likes:
I guess it's not a new subcommand for
f2fs_io,commit
22d758e2e6af210dc9e6cdf99438f063383ba72fAuthor: Jaegeuk Kim <[email protected]> Date: Tue Feb 19 19:07:21 2019 -0800
f2fs_io: addgc_urgente.g.,
f2fs_iogc_urgentdm-4 [start/end/run] [time in sec]
Quoted 7 lines
This controls sysfs/gc_urgent to run
f2fs_gcurgently.Signed-off-by Jaegeuk Kim <[email protected]>Quoted 1 line
_require_f2fs_io gc_urgent(there's not_require_f2fs_io:)Agreed, we need to introduce
_require_f2fs_io(), let me work on this.
Oh, 2019 was 7 years ago. No one should be complaining about this unless they're on a super old downstream f2fs-tools. But anyway, having _require_f2fs_io is definitely good for future f2fs testing :)
Therefore, _require_f2fs_io is not strictly urgent for this patch. It's up to you whether to include it now or handle it in a later update.
Quoted 6 lines
Quoted 6 lines
Quoted 5 lines
1. echo 1 >
/sys/fs/f2fs/$dev_name/gc_urgent
2. sleep 5 seconds
3. echo 0 >/sys/fs/f2fs/$dev_name/gc_urgent
Quoted 6 lines
Quoted 3 lines
It also depends on the
/sys/fs/f2fs/$dev_name/gc_urgenttoo. So we have to face the same question:
If this file doesn't exist, should this test case_notrun?Oh, right, maybe we can introduce
_require_f2fs_sysfs()to check whether f2fs kernel module has supported target sysfs node?
There's a _require_fs_sysfs_attr, so...
if you don't care about $SCRATCH_DEV is mounted or not, you can:
_require_fs_sysfs_attr $TEST_DEV gc_urgent
or after _scratch_mount:
_require_fs_sysfs_attr $SCRATCH_DEV gc_urgent
Then I think you can either use $F2FS_IO_PROG gc_urgent, or if you're worried about its compatibility, you can:
_set_fs_sysfs_attr $SCRATCH_DEV gc_urgent 1
sleep 5
Thanks,
Zorro
Quoted 58 lines
Thanks,
Quoted 53 lines
Thanks,
ZorroQuoted 49 lines
Quoted 9 lines
tests/f2fs/025 line 82
Quoted 5 lines
tests/f2fs/025 new file +3
-0+ +# Wait background GC thread to wake up to run and potentially encounter the inconsistency +sleep 5Does this sleep try to wait above "echo 1 >
/sys/fs/f2fs/$dev_name/gc_urgent"? If so, it makes more sense to move it into the "if-then" logic.Quoted 4 lines
tests/f2fs/025 line 84
Quoted 2 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^Same as above.
Will fix.
Quoted 11 lines
tests/f2fs/025 line 92
Quoted 9 lines
tests/f2fs/025 new file +8
-0+ +# Check whether the dmesg has the warning indicating the bug +_check_dmesg_for "F2FS-fs \($dev_name\): Inconsistent segment" && \ + _fail "F2FS-fs ($dev_name): Inconsistent segment type detected in dmesg!" + +echo "Silence is golden" +status=0 +exitWe've replaced "status=0;exit;" with "
_exit0".Will fix.
Thanks,
Quoted 15 lines
Thanks,
ZorroQuoted 11 lines
tests/f2fs/025.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 025 +Silence is goldenSignature · 1 line
2.49.0
Quoted 205 lines · Zorro Lang
Quoted 204 lines · Chao Yu
Quoted 182 lines · Zorro Lang
Quoted 181 lines · Chao Yu
Quoted 22 lines · Zorro Lang
Quoted 13 lines · Chao Yu
Without commit
520760b9f915("f2fs: optimize representative type determination in GC"), f2fs GC will report inconsistent segment type in large section issue, and then it will force to shutdown filesystem.[ 768.190903] F2FS-fs (loop51): Inconsistent segment (3) type [1, 0] in SIT and SSAThe reason is f2fs kernel will assume all segment type inside large section is the same, during GC it loads type from one segment and compare it to other segments' type, however due to recovery flow, the chosen segment may has zero valid blocks w/ different segment type, since the segment is invalid(free) one, it will never be migrated, so that we should not treat such state as abnormal condition.
Quoted 7 lines
This testcase is created to simulate above condition to see whether f2fs kernel module can handle it correctly
Signed-off-by Chao Yu <[email protected]>---
v2:
- clearMKFS_OPTIONSandMOUNT_OPTIONSto guarantee block allocation is as expected.Quoted 6 lines
Hi Chao,
Sorry, I just noticed your reply to my review on the previous patch version. Due to some unexpected shake-ups recently, I’ve been bogged down with setting up and modifying various new system environments, and I accidentally marked some unread emails as read.
No worries. :)
Quoted 2 lines
The patch looks good to me, with just a few picky review points below:
Thanks Zorro for taking a look.
Quoted 30 lines
Quoted 30 lines
tests/f2fs/025 | 92 ++++++++++++++++++++++++++++++++++++++++++++++ tests/f2fs/025.out | 2 + 2 files changed, 94 insertions(+) create mode 100644 tests/f2fs/025 create mode 100644 tests/f2fs/025.outtests/f2fs/025 new file +17
-0@@ -0,0 +1,92 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2026 Chao Yu <[email protected]> +# +# FS QA Test No. f2fs/025 +# +# Check whether f2fs will encounter cp_error (Inconsistent segment type) +# when doing sanity check on type of segments inside large section during +# garbage collection. +# +. ./common/preamble +_begin_fstest auto quick + +_fixed_by_kernel_commit 520760b9f915 \ + "f2fs: optimize representative type determination in GC" + +. ./common/filterQuoted 8 lines
tests/f2fs/025 line 23
Quoted 6 lines
tests/f2fs/025 new file +6
-0+ +_cleanup() +{ + cd / + rm -r -f $tmp.* +}This
_cleanup()function is same as default. It can be removed.Will remove.
Quoted 18 lines
Quoted 7 lines
tests/f2fs/025 new file +6
-0+ +_require_scratch +_require_xfs_io_command "pwrite" +_require_xfs_io_command "truncate" +_require_command "$F2FS_IO_PROG" f2fs_io +_require_check_dmesgtests/f2fs/025 line 39
Quoted 10 lines
tests/f2fs/025 new file +10
-0+ +# Clear options to avoid interference from external configurations +export MKFS_OPTIONS="" +export MOUNT_OPTIONS="" + +# Format with 96MB size and 2 segments per section +_scratch_mkfs_sized $((96 * 1024 * 1024)) "" "-s 2" >> $seqres.full 2>&1 + +# Mount with mode=lfs +_scratch_mount -o mode=lfs >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^Quoted 1 line
It's helpless, due to if
_scratch_mountfails, itexit()directly.Right, will fix.
Quoted 26 lines
Quoted 26 lines
tests/f2fs/025 new file +25
-0+ +# Create files to fill whole filesystem, then segment type will be changed to node type +for ((i=0;i<5120;i++)) do + touch $SCRATCH_MNT/$i >> $seqres.full 2>&1 +done +sync + +# Remove all files to create free(empty) node segments +rm -f $SCRATCH_MNT/* +sync + +# Allocate free space so that we have chance to reuse free(empty) node segments +$XFS_IO_PROG -f -c "pwrite -b 4k 0 1928k" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 16M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +sync + +$XFS_IO_PROG -d -c "pwrite -b 4k 0 8M" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 32K" $SCRATCH_MNT/file >> $seqres.full 2>&1Quoted 8 lines
tests/f2fs/025 line 69
Quoted 5 lines
tests/f2fs/025 new file +5
-0+$XFS_IO_PROG -c "truncate 0" $SCRATCH_MNT/file >> $seqres.full 2>&1 +$XFS_IO_PROG -d -c "pwrite -b 4k 0 2M" -c "fsync" $SCRATCH_MNT/file >> $seqres.full 2>&1 + +# Shutdown the filesystem without checkpoint +$F2FS_IO_PROG shutdown 2 $SCRATCH_MNT >> $seqres.full 2>&1I'm wondering if we can have f2fs supporting in common
_scratch_shutdownhelper :)I think we can change f2fs testcase to use
_scratch_shutdownbecause the definition of nologflush shutdown interface is the same as xfs':/*
* should be same asXFS_IOC_GOINGDOWN.
* Flags for going down operation used byFS_IOC_GOINGDOWN*/
#defineF2FS_IOC_SHUTDOWN_IOR('X', 125, __u32)/* Shutdown */
#defineF2FS_GOING_DOWN_NOSYNC0x2/* going down */#define
XFS_IOC_GOINGDOWN_IOR('X', 125,uint32_t)
#defineXFS_FSOP_GOING_FLAGS_NOLOGFLUSH0x2/* don't flush log nor data */Quoted 6 lines
tests/f2fs/025 line 71
Quoted 3 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^If unmount fails, how about let it output the errors, to break the golden image?
Yes, it's better.
Quoted 5 lines
tests/f2fs/025 line 73
Quoted 3 lines
tests/f2fs/025 new file +2
-0+ +_scratch_mount -o mode=lfs >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^ helplessWill fix.
Quoted 15 lines
tests/f2fs/025 line 79
Quoted 7 lines
tests/f2fs/025 new file +6
-0+ +# Run urgent_gc mode to trigger garbage collection +dev_name=$(_short_dev $SCRATCH_DEV) +if [ -f /sys/fs/f2fs/$dev_name/gc_urgent ]; then + echo 1 > /sys/fs/f2fs/$dev_name/gc_urgent +fiHmm... what if there's not
/sys/fs/f2fs/$dev_name/gc_urgent? Does it affect the test result?If it does, this's a necessary requirement for this test, we shouldn't ignore it and keep running. Does $F2FS_IO_PROG provide a command to make a force GC? Or we need to check this file and
_notrunif it's not existed.Quoted 5 lines
Quoted 3 lines
Ah, right, that's good point!
We can use "$F2FS_IO_PROG
gc_urgent<dev_name> run 5" instead, it will do below commands:Great, I just hope the *gc_urgent* isn't a new feature which needs something likes:
I guess it's not a new subcommand for
f2fs_io,commit
22d758e2e6af210dc9e6cdf99438f063383ba72fAuthor: Jaegeuk Kim <[email protected]> Date: Tue Feb 19 19:07:21 2019 -0800
f2fs_io: addgc_urgente.g.,
f2fs_iogc_urgentdm-4 [start/end/run] [time in sec]This controls sysfs/gc_urgent to run
f2fs_gcurgently.Signed-off-by Jaegeuk Kim <[email protected]>
Quoted 7 lines
Quoted 3 lines
Quoted 1 line
_require_f2fs_io gc_urgent(there's not_require_f2fs_io:)Agreed, we need to introduce
_require_f2fs_io(), let me work on this.Oh, 2019 was 7 years ago. No one should be complaining about this unless they're on a super old downstream f2fs-tools. But anyway, having
_require_f2fs_iois definitely good for future f2fs testing :)
Yeah, agreed.
Quoted 3 lines
Therefore,
_require_f2fs_iois not strictly urgent for this patch. It's up to you whether to include it now or handle it in a later update.
Let me update a bit later.
Quoted 11 lines
Quoted 11 lines
Quoted 10 lines
Quoted 6 lines
1. echo 1 >
/sys/fs/f2fs/$dev_name/gc_urgent
2. sleep 5 seconds
3. echo 0 >/sys/fs/f2fs/$dev_name/gc_urgentIt also depends on the
/sys/fs/f2fs/$dev_name/gc_urgenttoo. So we have to face the same question:
If this file doesn't exist, should this test case_notrun?
Quoted 7 lines
Quoted 2 lines
Oh, right, maybe we can introduce
_require_f2fs_sysfs()to check whether f2fs kernel module has supported target sysfs node?There's a
_require_fs_sysfs_attr, so...if you don't care about $SCRATCH_DEV is mounted or not, you can:
_require_fs_sysfs_attr$TEST_DEVgc_urgent
gc_urgent is not a per-image feature, so above version looks fine.
Quoted 6 lines
or after
_scratch_mount:
_require_fs_sysfs_attr$SCRATCH_DEVgc_urgentThen I think you can either use
$F2FS_IO_PROG gc_urgent, or if you're worried about its compatibility, you can:
I'm not worried about the compatibility.
Quoted 2 lines
_set_fs_sysfs_attr$SCRATCH_DEVgc_urgent1
sleep 5
Thanks for the suggestion. :)
Let me update w/:
_require_fs_sysfs_attr $TEST_DEV gc_urgent
...
$F2FS_IO_PROG gc_urgent
Thanks,
Quoted 62 lines
Thanks,
ZorroQuoted 58 lines
Thanks,
Quoted 53 lines
Thanks,
ZorroQuoted 49 lines
Quoted 9 lines
tests/f2fs/025 line 82
Quoted 5 lines
tests/f2fs/025 new file +3
-0+ +# Wait background GC thread to wake up to run and potentially encounter the inconsistency +sleep 5Does this sleep try to wait above "echo 1 >
/sys/fs/f2fs/$dev_name/gc_urgent"? If so, it makes more sense to move it into the "if-then" logic.Quoted 4 lines
tests/f2fs/025 line 84
Quoted 2 lines
tests/f2fs/025 new file +2
-0+ +_scratch_unmount >> $seqres.full 2>&1^^^^^^^^^^^^^^^^^^^^Same as above.
Will fix.
Quoted 11 lines
tests/f2fs/025 line 92
Quoted 9 lines
tests/f2fs/025 new file +8
-0+ +# Check whether the dmesg has the warning indicating the bug +_check_dmesg_for "F2FS-fs \($dev_name\): Inconsistent segment" && \ + _fail "F2FS-fs ($dev_name): Inconsistent segment type detected in dmesg!" + +echo "Silence is golden" +status=0 +exitWe've replaced "status=0;exit;" with "
_exit0".Will fix.
Thanks,
Quoted 15 lines
Thanks,
ZorroQuoted 11 lines
tests/f2fs/025.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 025 +Silence is goldenSignature · 1 line
2.49.0