[PATCH 0/3] btrfs-progs: tests: add new mkfs test for zoned device

[PATCH 0/3] btrfs-progs: tests: add new mkfs test for zoned device

[Naohiro Aota]

This series adds a new mkfs test for zoned device. The test is based on
mkfs-tests/001-basic-profiles, but uses nullb devices instead of loop
devices, to set a zoned profile.

Two preparation patches are necessaly for the test case. The first one
fixes the error handling of _find_free_index(). And, the second one adds
wait_for_nullbdevs and cond_wait_for_nullbdevs like the loop device's
counterpart.

Naohiro Aota (3):
  btrfs-progs: tests: check nullb index for the error case
  btrfs-progs: tests: add {,cond_}wait_for_nullbdevs
  btrfs-progs: tests: add new mkfs test for zoned device

 tests/common                                | 14 +++
 tests/mkfs-tests/039-zoned-profiles/test.sh | 98 +++++++++++++++++++++
 tests/nullb                                 |  3 +
 3 files changed, 115 insertions(+)
 create mode 100755 tests/mkfs-tests/039-zoned-profiles/test.sh

--
2.51.0

[PATCH 1/3] btrfs-progs: tests: check nullb index for the error case

[Naohiro Aota]

"_find_free_index" can return "ERROR: ...". Check the return value for the
case and fail the test.

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
---
 tests/nullb | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/tests/nullb b/tests/nullb
index 457ae0d8354a..bfc5640c4470 100755
--- a/tests/nullb
+++ b/tests/nullb
@@ -146,6 +146,9 @@ fi
 if [ "$CMD" = 'create' ]; then
 	_check_setup
 	index=$(_find_free_index)
+	if [[ "$index" = ERROR* ]]; then
+		_error "$index"
+	fi
 	name="nullb$index"
 	# size in MB
 	size=$(_parse_device_size "$@")
-- 
2.51.0

[PATCH 2/3] btrfs-progs: tests: add {,cond_}wait_for_nullbdevs

[Naohiro Aota]

It is a nullb version of {,cond_}wait_for_loopdevs. It waits for all the
nullb devices are ready to use.

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
---
 tests/common | 14 ++++++++++++++
 1 file changed, 14 insertions(+)

diff --git a/tests/common b/tests/common
index 2c90acb90cfc..1df37c390bf6 100644
--- a/tests/common
+++ b/tests/common
@@ -984,6 +984,20 @@ cleanup_nullbdevs()
 		name=$(basename "$dev")
 		run_check $SUDO_HELPER "$nullb" rm "$name"
 	done
+	unset nullb_devs
+}
+
+wait_for_nullbdevs()
+{
+	for dev in ${nullb_devs[@]}; do
+		run_mayfail $SUDO_HELPER "$TOP/btrfs" device ready "$dev"
+	done
+}
+
+cond_wait_for_nullbdevs() {
+	if [ -n "${nullb_devs[1]}" ]; then
+		wait_for_nullbdevs
+	fi
 }
 
 init_env()
-- 
2.51.0

[PATCH 3/3] btrfs-progs: tests: add new mkfs test for zoned device

[Naohiro Aota]

This new test is based on mkfs-tests/001-basic-profiles, and it goes
through the profiles to mkfs and do some basic checks.

Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
---
 tests/mkfs-tests/039-zoned-profiles/test.sh | 98 +++++++++++++++++++++
 1 file changed, 98 insertions(+)
 create mode 100755 tests/mkfs-tests/039-zoned-profiles/test.sh

diff --git a/tests/mkfs-tests/039-zoned-profiles/test.sh b/tests/mkfs-tests/039-zoned-profiles/test.sh
new file mode 100755
index 000000000000..f40648cd06e1
--- /dev/null
+++ b/tests/mkfs-tests/039-zoned-profiles/test.sh
@@ -0,0 +1,98 @@
+#!/bin/bash
+# test various blockgroup profile combinations, use nullb devices as block
+# devices. This test is based on mkfs-tests/001-basic-profiles.
+
+source "$TEST_TOP/common" || exit
+
+check_prereq mkfs.btrfs
+check_prereq btrfs
+check_global_prereq blkzone
+
+setup_root_helper
+# Create one 128M device with 4M zones, 32 of them
+setup_nullbdevs 4 128 4
+prepare_nullbdevs
+dev1=${nullb_devs[1]}
+
+test_get_info()
+{
+	local tmp_out
+
+	tmp_out=$(_mktemp mkfs-get-info)
+	run_check $SUDO_HELPER "$TOP/btrfs" inspect-internal dump-super "$dev1"
+	run_check $SUDO_HELPER "$TOP/btrfs" check "$dev1"
+
+	# Work around for kernel bug that will treat SINGLE and single
+	# device RAID0 as the same.
+	# Thus kernel may create new SINGLE chunks, causing extra warning
+	# when testing single device RAID0.
+	cond_wait_for_nullbdevs
+	run_check $SUDO_HELPER mount -o ro "$dev1" "$TEST_MNT"
+	run_check_stdout "$TOP/btrfs" filesystem df "$TEST_MNT" > "$tmp_out"
+	if grep -q "Multiple block group profiles detected" "$tmp_out"; then
+		rm -- "$tmp_out"
+		_fail "temporary chunks are not properly cleaned up"
+	fi
+	rm -- "$tmp_out"
+	run_check $SUDO_HELPER "$TOP/btrfs" filesystem usage "$TEST_MNT"
+	run_check $SUDO_HELPER "$TOP/btrfs" device usage "$TEST_MNT"
+	run_check $SUDO_HELPER umount "$TEST_MNT"
+}
+
+test_do_mkfs()
+{
+	run_check $SUDO_HELPER "$TOP/mkfs.btrfs" -f "$@"
+	if run_check_stdout $SUDO_HELPER "$TOP/btrfs" check "$dev1" | grep -iq warning; then
+		_fail "warnings found in check output"
+	fi
+}
+
+test_mkfs_single()
+{
+	test_do_mkfs "$@" "$dev1"
+	test_get_info
+}
+test_mkfs_multi()
+{
+	test_do_mkfs "$@" ${nullb_devs[@]}
+	test_get_info
+}
+
+test_mkfs_single
+test_mkfs_single  -d  single  -m  single
+test_mkfs_single  -d  single  -m  dup
+
+test_mkfs_multi
+test_mkfs_multi   -d  single  -m  single
+
+if [ -f "/sys/fs/btrfs/features/raid_stripe_tree" ]; then
+	test_mkfs_single  -d  dup     -m  single
+	test_mkfs_single  -d  dup     -m  dup
+
+	test_mkfs_multi   -d  raid0   -m  raid0
+	test_mkfs_multi   -d  raid1   -m  raid1
+	test_mkfs_multi   -d  raid10  -m  raid10
+	# RAID5/6 are not yet supported.
+	# test_mkfs_multi   -d  raid5   -m  raid5
+	# test_mkfs_multi   -d  raid6   -m  raid6
+	test_mkfs_multi   -d  dup     -m  dup
+
+	if [ -f "/sys/fs/btrfs/features/raid1c34" ]; then
+		test_mkfs_multi   -d  raid1c3 -m  raid1c3
+		test_mkfs_multi   -d  raid1c4 -m  raid1c4
+	else
+		_log "skip mount test, missing support for raid1c34"
+		test_do_mkfs -d raid1c3 -m raid1c3 ${nullb_devs[@]}
+		test_do_mkfs -d raid1c4 -m raid1c4 ${nullb_devs[@]}
+	fi
+
+	# Non-standard profile/device combinations
+
+	# Single device raid0, two device raid10 (simple mount works on older kernels too)
+	test_do_mkfs -d raid0 -m raid0 "$dev1"
+	test_get_info
+	test_do_mkfs -d raid10 -m raid10 "${nullb_devs[1]}" "${nullb_devs[2]}"
+	test_get_info
+fi
+
+cleanup_nullbdevs
-- 
2.51.0

Re: [PATCH 1/3] btrfs-progs: tests: check nullb index for the error case

[Qu Wenruo]

在 2025/9/2 13:59, Naohiro Aota 写道:
> "_find_free_index" can return "ERROR: ...". Check the return value for the
> case and fail the test.
> 
> Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
> ---
>   tests/nullb | 3 +++
>   1 file changed, 3 insertions(+)
> 
> diff --git a/tests/nullb b/tests/nullb
> index 457ae0d8354a..bfc5640c4470 100755
> --- a/tests/nullb
> +++ b/tests/nullb
> @@ -146,6 +146,9 @@ fi
>   if [ "$CMD" = 'create' ]; then
>   	_check_setup
>   	index=$(_find_free_index)
> +	if [[ "$index" = ERROR* ]]; then
> +		_error "$index"
> +	fi

I think the bigger problem is that:

- _error() output into stdout instead of stderr

- the "exit 1" doesn't help in this case
   It will only kill the child bash, not the calling one.

So I'd prefer to make _error() to output the warning to stderr, so that 
index will be empty on error.

Thanks,
Qu

>   	name="nullb$index"
>   	# size in MB
>   	size=$(_parse_device_size "$@")

Re: [PATCH 2/3] btrfs-progs: tests: add {,cond_}wait_for_nullbdevs

[Qu Wenruo]

在 2025/9/2 13:59, Naohiro Aota 写道:
> It is a nullb version of {,cond_}wait_for_loopdevs. It waits for all the
> nullb devices are ready to use.
> 
> Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
> ---
>   tests/common | 14 ++++++++++++++
>   1 file changed, 14 insertions(+)
> 
> diff --git a/tests/common b/tests/common
> index 2c90acb90cfc..1df37c390bf6 100644
> --- a/tests/common
> +++ b/tests/common
> @@ -984,6 +984,20 @@ cleanup_nullbdevs()
>   		name=$(basename "$dev")
>   		run_check $SUDO_HELPER "$nullb" rm "$name"
>   	done
> +	unset nullb_devs
> +}
> +
> +wait_for_nullbdevs()
> +{
> +	for dev in ${nullb_devs[@]}; do
> +		run_mayfail $SUDO_HELPER "$TOP/btrfs" device ready "$dev"
> +	done
> +}
> +
> +cond_wait_for_nullbdevs() {
> +	if [ -n "${nullb_devs[1]}" ]; then
> +		wait_for_nullbdevs
> +	fi

I guess we don't need cond_wait_for_nummbdevs()?

As if nullb_devs array is not defined/empty, the for loop inside 
wait_for_nullbdevs() will do nothing anyway.

Thanks,
Qu

>   }
>   
>   init_env()

Re: [PATCH 3/3] btrfs-progs: tests: add new mkfs test for zoned device

[Qu Wenruo]

在 2025/9/2 13:59, Naohiro Aota 写道:
> This new test is based on mkfs-tests/001-basic-profiles, and it goes
> through the profiles to mkfs and do some basic checks.
> 
> Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>

Despite the cond_wait_for_nullbdevs() usage, the test case itself looks 
good to me.

With the comment in patch 2 addressed, feel free to add:

Reviewed-by: Qu Wenruo <wqu@suse.com>

Thanks,
Qu
> ---
>   tests/mkfs-tests/039-zoned-profiles/test.sh | 98 +++++++++++++++++++++
>   1 file changed, 98 insertions(+)
>   create mode 100755 tests/mkfs-tests/039-zoned-profiles/test.sh
> 
> diff --git a/tests/mkfs-tests/039-zoned-profiles/test.sh b/tests/mkfs-tests/039-zoned-profiles/test.sh
> new file mode 100755
> index 000000000000..f40648cd06e1
> --- /dev/null
> +++ b/tests/mkfs-tests/039-zoned-profiles/test.sh
> @@ -0,0 +1,98 @@
> +#!/bin/bash
> +# test various blockgroup profile combinations, use nullb devices as block
> +# devices. This test is based on mkfs-tests/001-basic-profiles.
> +
> +source "$TEST_TOP/common" || exit
> +
> +check_prereq mkfs.btrfs
> +check_prereq btrfs
> +check_global_prereq blkzone
> +
> +setup_root_helper
> +# Create one 128M device with 4M zones, 32 of them
> +setup_nullbdevs 4 128 4
> +prepare_nullbdevs
> +dev1=${nullb_devs[1]}
> +
> +test_get_info()
> +{
> +	local tmp_out
> +
> +	tmp_out=$(_mktemp mkfs-get-info)
> +	run_check $SUDO_HELPER "$TOP/btrfs" inspect-internal dump-super "$dev1"
> +	run_check $SUDO_HELPER "$TOP/btrfs" check "$dev1"
> +
> +	# Work around for kernel bug that will treat SINGLE and single
> +	# device RAID0 as the same.
> +	# Thus kernel may create new SINGLE chunks, causing extra warning
> +	# when testing single device RAID0.
> +	cond_wait_for_nullbdevs
> +	run_check $SUDO_HELPER mount -o ro "$dev1" "$TEST_MNT"
> +	run_check_stdout "$TOP/btrfs" filesystem df "$TEST_MNT" > "$tmp_out"
> +	if grep -q "Multiple block group profiles detected" "$tmp_out"; then
> +		rm -- "$tmp_out"
> +		_fail "temporary chunks are not properly cleaned up"
> +	fi
> +	rm -- "$tmp_out"
> +	run_check $SUDO_HELPER "$TOP/btrfs" filesystem usage "$TEST_MNT"
> +	run_check $SUDO_HELPER "$TOP/btrfs" device usage "$TEST_MNT"
> +	run_check $SUDO_HELPER umount "$TEST_MNT"
> +}
> +
> +test_do_mkfs()
> +{
> +	run_check $SUDO_HELPER "$TOP/mkfs.btrfs" -f "$@"
> +	if run_check_stdout $SUDO_HELPER "$TOP/btrfs" check "$dev1" | grep -iq warning; then
> +		_fail "warnings found in check output"
> +	fi
> +}
> +
> +test_mkfs_single()
> +{
> +	test_do_mkfs "$@" "$dev1"
> +	test_get_info
> +}
> +test_mkfs_multi()
> +{
> +	test_do_mkfs "$@" ${nullb_devs[@]}
> +	test_get_info
> +}
> +
> +test_mkfs_single
> +test_mkfs_single  -d  single  -m  single
> +test_mkfs_single  -d  single  -m  dup
> +
> +test_mkfs_multi
> +test_mkfs_multi   -d  single  -m  single
> +
> +if [ -f "/sys/fs/btrfs/features/raid_stripe_tree" ]; then
> +	test_mkfs_single  -d  dup     -m  single
> +	test_mkfs_single  -d  dup     -m  dup
> +
> +	test_mkfs_multi   -d  raid0   -m  raid0
> +	test_mkfs_multi   -d  raid1   -m  raid1
> +	test_mkfs_multi   -d  raid10  -m  raid10
> +	# RAID5/6 are not yet supported.
> +	# test_mkfs_multi   -d  raid5   -m  raid5
> +	# test_mkfs_multi   -d  raid6   -m  raid6
> +	test_mkfs_multi   -d  dup     -m  dup
> +
> +	if [ -f "/sys/fs/btrfs/features/raid1c34" ]; then
> +		test_mkfs_multi   -d  raid1c3 -m  raid1c3
> +		test_mkfs_multi   -d  raid1c4 -m  raid1c4
> +	else
> +		_log "skip mount test, missing support for raid1c34"
> +		test_do_mkfs -d raid1c3 -m raid1c3 ${nullb_devs[@]}
> +		test_do_mkfs -d raid1c4 -m raid1c4 ${nullb_devs[@]}
> +	fi
> +
> +	# Non-standard profile/device combinations
> +
> +	# Single device raid0, two device raid10 (simple mount works on older kernels too)
> +	test_do_mkfs -d raid0 -m raid0 "$dev1"
> +	test_get_info
> +	test_do_mkfs -d raid10 -m raid10 "${nullb_devs[1]}" "${nullb_devs[2]}"
> +	test_get_info
> +fi
> +
> +cleanup_nullbdevs

Re: [PATCH 3/3] btrfs-progs: tests: add new mkfs test for zoned device

[Johannes Thumshirn]

On 9/2/25 6:30 AM, Naohiro Aota wrote:
> +if [ -f "/sys/fs/btrfs/features/raid_stripe_tree" ]; then
> +	test_mkfs_single  -d  dup     -m  single
> +	test_mkfs_single  -d  dup     -m  dup
> +
> +	test_mkfs_multi   -d  raid0   -m  raid0
> +	test_mkfs_multi   -d  raid1   -m  raid1
> +	test_mkfs_multi   -d  raid10  -m  raid10
> +	# RAID5/6 are not yet supported.
> +	# test_mkfs_multi   -d  raid5   -m  raid5
> +	# test_mkfs_multi   -d  raid6   -m  raid6
> +	test_mkfs_multi   -d  dup     -m  dup
> +
> +	if [ -f "/sys/fs/btrfs/features/raid1c34" ]; then
> +		test_mkfs_multi   -d  raid1c3 -m  raid1c3
> +		test_mkfs_multi   -d  raid1c4 -m  raid1c4
> +	else
> +		_log "skip mount test, missing support for raid1c34"
> +		test_do_mkfs -d raid1c3 -m raid1c3 ${nullb_devs[@]}
> +		test_do_mkfs -d raid1c4 -m raid1c4 ${nullb_devs[@]}
> +	fi
> +
> +	# Non-standard profile/device combinations
> +
> +	# Single device raid0, two device raid10 (simple mount works on older kernels too)
> +	test_do_mkfs -d raid0 -m raid0 "$dev1"
> +	test_get_info
> +	test_do_mkfs -d raid10 -m raid10 "${nullb_devs[1]}" "${nullb_devs[2]}"
> +	test_get_info
> +fi
Wouldn't this need to check if mkfs.btrfs supports "-O raid-stripe-tree" 
as well?

Re: [PATCH 3/3] btrfs-progs: tests: add new mkfs test for zoned device

[David Sterba]

On Tue, Sep 02, 2025 at 09:59:20AM +0000, Johannes Thumshirn wrote:
> On 9/2/25 6:30 AM, Naohiro Aota wrote:
> > +if [ -f "/sys/fs/btrfs/features/raid_stripe_tree" ]; then
> > +	test_mkfs_single  -d  dup     -m  single
> > +	test_mkfs_single  -d  dup     -m  dup
> > +
> > +	test_mkfs_multi   -d  raid0   -m  raid0
> > +	test_mkfs_multi   -d  raid1   -m  raid1
> > +	test_mkfs_multi   -d  raid10  -m  raid10
> > +	# RAID5/6 are not yet supported.
> > +	# test_mkfs_multi   -d  raid5   -m  raid5
> > +	# test_mkfs_multi   -d  raid6   -m  raid6
> > +	test_mkfs_multi   -d  dup     -m  dup
> > +
> > +	if [ -f "/sys/fs/btrfs/features/raid1c34" ]; then
> > +		test_mkfs_multi   -d  raid1c3 -m  raid1c3
> > +		test_mkfs_multi   -d  raid1c4 -m  raid1c4
> > +	else
> > +		_log "skip mount test, missing support for raid1c34"
> > +		test_do_mkfs -d raid1c3 -m raid1c3 ${nullb_devs[@]}
> > +		test_do_mkfs -d raid1c4 -m raid1c4 ${nullb_devs[@]}
> > +	fi
> > +
> > +	# Non-standard profile/device combinations
> > +
> > +	# Single device raid0, two device raid10 (simple mount works on older kernels too)
> > +	test_do_mkfs -d raid0 -m raid0 "$dev1"
> > +	test_get_info
> > +	test_do_mkfs -d raid10 -m raid10 "${nullb_devs[1]}" "${nullb_devs[2]}"
> > +	test_get_info
> > +fi
> Wouldn't this need to check if mkfs.btrfs supports "-O raid-stripe-tree" 
> as well?

The assumption of the environment is that kernel may not support all
features so there are the sysfs checks. For progs this assumes that it's
either a git version or the the testsuite version matches the installed
progs version.

We could possibly encode the assumptions simlar to what check_prereq()
does but for version-dependent features. I'm not sure if it's worth, so
far nobody else complained but it does make sense to make it more
robust.

Re: [PATCH 2/3] btrfs-progs: tests: add {,cond_}wait_for_nullbdevs

[Naohiro Aota]

On Tue Sep 2, 2025 at 6:15 PM JST, Qu Wenruo wrote:
>
>
> 在 2025/9/2 13:59, Naohiro Aota 写道:
>> It is a nullb version of {,cond_}wait_for_loopdevs. It waits for all the
>> nullb devices are ready to use.
>> 
>> Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
>> ---
>>   tests/common | 14 ++++++++++++++
>>   1 file changed, 14 insertions(+)
>> 
>> diff --git a/tests/common b/tests/common
>> index 2c90acb90cfc..1df37c390bf6 100644
>> --- a/tests/common
>> +++ b/tests/common
>> @@ -984,6 +984,20 @@ cleanup_nullbdevs()
>>   		name=$(basename "$dev")
>>   		run_check $SUDO_HELPER "$nullb" rm "$name"
>>   	done
>> +	unset nullb_devs
>> +}
>> +
>> +wait_for_nullbdevs()
>> +{
>> +	for dev in ${nullb_devs[@]}; do
>> +		run_mayfail $SUDO_HELPER "$TOP/btrfs" device ready "$dev"
>> +	done
>> +}
>> +
>> +cond_wait_for_nullbdevs() {
>> +	if [ -n "${nullb_devs[1]}" ]; then
>> +		wait_for_nullbdevs
>> +	fi
>
> I guess we don't need cond_wait_for_nummbdevs()?
>
> As if nullb_devs array is not defined/empty, the for loop inside 
> wait_for_nullbdevs() will do nothing anyway.

Ah, yes, that's correct. I followed cond_wait_for_loopdevs here. Maybe,
we should drop wait_for_loopdevs and convert the callers as wel.

>
> Thanks,
> Qu
>
>>   }
>>   
>>   init_env()

Re: [PATCH 3/3] btrfs-progs: tests: add new mkfs test for zoned device

[Naohiro Aota]

On Wed Sep 3, 2025 at 3:57 AM JST, David Sterba wrote:
> On Tue, Sep 02, 2025 at 09:59:20AM +0000, Johannes Thumshirn wrote:
>> On 9/2/25 6:30 AM, Naohiro Aota wrote:
>> > +if [ -f "/sys/fs/btrfs/features/raid_stripe_tree" ]; then
>> > +	test_mkfs_single  -d  dup     -m  single
>> > +	test_mkfs_single  -d  dup     -m  dup
>> > +
>> > +	test_mkfs_multi   -d  raid0   -m  raid0
>> > +	test_mkfs_multi   -d  raid1   -m  raid1
>> > +	test_mkfs_multi   -d  raid10  -m  raid10
>> > +	# RAID5/6 are not yet supported.
>> > +	# test_mkfs_multi   -d  raid5   -m  raid5
>> > +	# test_mkfs_multi   -d  raid6   -m  raid6
>> > +	test_mkfs_multi   -d  dup     -m  dup
>> > +
>> > +	if [ -f "/sys/fs/btrfs/features/raid1c34" ]; then
>> > +		test_mkfs_multi   -d  raid1c3 -m  raid1c3
>> > +		test_mkfs_multi   -d  raid1c4 -m  raid1c4
>> > +	else
>> > +		_log "skip mount test, missing support for raid1c34"
>> > +		test_do_mkfs -d raid1c3 -m raid1c3 ${nullb_devs[@]}
>> > +		test_do_mkfs -d raid1c4 -m raid1c4 ${nullb_devs[@]}
>> > +	fi
>> > +
>> > +	# Non-standard profile/device combinations
>> > +
>> > +	# Single device raid0, two device raid10 (simple mount works on older kernels too)
>> > +	test_do_mkfs -d raid0 -m raid0 "$dev1"
>> > +	test_get_info
>> > +	test_do_mkfs -d raid10 -m raid10 "${nullb_devs[1]}" "${nullb_devs[2]}"
>> > +	test_get_info
>> > +fi
>> Wouldn't this need to check if mkfs.btrfs supports "-O raid-stripe-tree" 
>> as well?
>
> The assumption of the environment is that kernel may not support all
> features so there are the sysfs checks. For progs this assumes that it's
> either a git version or the the testsuite version matches the installed
> progs version.
>
> We could possibly encode the assumptions simlar to what check_prereq()
> does but for version-dependent features. I'm not sure if it's worth, so
> far nobody else complained but it does make sense to make it more
> robust.

I remembered that raid-stripe-tree is still only available with the
EXPERIMENTAL config. I'll add "_test_config EXPERIMENTAL" for this part.

Re: [PATCH 1/3] btrfs-progs: tests: check nullb index for the error case

[Naohiro Aota]

On Tue Sep 2, 2025 at 6:11 PM JST, Qu Wenruo wrote:
>
>
> 在 2025/9/2 13:59, Naohiro Aota 写道:
>> "_find_free_index" can return "ERROR: ...". Check the return value for the
>> case and fail the test.
>> 
>> Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
>> ---
>>   tests/nullb | 3 +++
>>   1 file changed, 3 insertions(+)
>> 
>> diff --git a/tests/nullb b/tests/nullb
>> index 457ae0d8354a..bfc5640c4470 100755
>> --- a/tests/nullb
>> +++ b/tests/nullb
>> @@ -146,6 +146,9 @@ fi
>>   if [ "$CMD" = 'create' ]; then
>>   	_check_setup
>>   	index=$(_find_free_index)
>> +	if [[ "$index" = ERROR* ]]; then
>> +		_error "$index"
>> +	fi
>
> I think the bigger problem is that:
>
> - _error() output into stdout instead of stderr
>
> - the "exit 1" doesn't help in this case
>    It will only kill the child bash, not the calling one.
>
> So I'd prefer to make _error() to output the warning to stderr, so that 
> index will be empty on error.

Ah, yes. It's safer to do so. I'll revise the patch.

>
> Thanks,
> Qu
>
>>   	name="nullb$index"
>>   	# size in MB
>>   	size=$(_parse_device_size "$@")