fixes for running on devices with protetion information

fixes for running on devices with protetion information

[Christoph Hellwig]

Hi all,

currently various tests either hang or fail when running on devices
with protection information.  This series makes tests using
unsupported device mapper targets _notrun instead of hanging or
failing with obscure errors.

Bote that to usefully run on devices with PI you'll also need my kernel
patches that fix a regression in PI information stacking and the
one-liner to add dm-error support for passing on integrity data.

Diffstat:
 common/dmthin     |   10 +++++-----
 tests/generic/081 |    3 ++-
 tests/generic/459 |    6 ++++--
 3 files changed, 11 insertions(+), 8 deletions(-)

[PATCH 1/3] dmthin: _notrun when dm-thin setup fails

[Christoph Hellwig]

dm-think does not work on device with protection information.  Instead
of failing the tests using it in obscure ways, _norun them.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 common/dmthin | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/common/dmthin b/common/dmthin
index 453e43a0499e..a1e1fb8763c0 100644
--- a/common/dmthin
+++ b/common/dmthin
@@ -105,13 +105,13 @@ _dmthin_init()
 	# Metadata device
 	DMTHIN_META_TABLE="0 $meta_dev_size linear $dm_backing_dev $meta_dev_offset"
 	_dmsetup_create $DMTHIN_META_NAME --table "$DMTHIN_META_TABLE" || \
-		_fatal "failed to create dm thin meta device"
+		_notrun "failed to create dm thin meta device"
 
 	# Data device
 	local data_dev_offset=$((meta_dev_offset + $meta_dev_size))
 	DMTHIN_DATA_TABLE="0 $data_dev_size linear $dm_backing_dev $data_dev_offset"
 	_dmsetup_create $DMTHIN_DATA_NAME --table "$DMTHIN_DATA_TABLE" || \
-		_fatal "failed to create dm thin data device"
+		_notrun "failed to create dm thin data device"
 
 	# Zap the pool metadata dev.  Explicitly fsync the zeroes to disk
 	# because a slow-running udev running concurrently with dd can maintain
@@ -124,16 +124,16 @@ _dmthin_init()
 	# "start length thin-pool metadata_dev data_dev data_block_size low_water_mark"
 	DMTHIN_POOL_TABLE="0 $data_dev_size thin-pool $DMTHIN_META_DEV $DMTHIN_DATA_DEV $cluster_size $low_water"
 	_dmsetup_create $DMTHIN_POOL_NAME --table "$DMTHIN_POOL_TABLE" || \
-		_fatal "failed to create dm thin pool device"
+		_notun "failed to create dm thin pool device"
 
 	# Thin volume
 	$DMSETUP_PROG message $DMTHIN_POOL_DEV 0 "create_thin $pool_id" || \
-		_fatal "failed to message pool device"
+		_notrun "failed to message pool device"
 
 	# start length thin pool_dev dev_id [external_origin_dev]
 	DMTHIN_VOL_TABLE="0 $virtual_size thin $DMTHIN_POOL_DEV $pool_id"
 	_dmsetup_create $DMTHIN_VOL_NAME --table "$DMTHIN_VOL_TABLE" || \
-		_fatal "failed to create dm thin volume device"
+		_notrun "failed to create dm thin volume device"
 
 }
 
-- 
2.47.2

[PATCH 2/3] generic/081: don't hang when running on devices with protection information

[Christoph Hellwig]

The dm snapshot target does not work on devices with protection
information, and the udevadm settle command simply hangs when lvcreate
failed because of that.  Avoid that by doing a strategic _notrun when
lvcreate fails.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 tests/generic/081 | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tests/generic/081 b/tests/generic/081
index 97928bf4a435..00280e9cff3b 100755
--- a/tests/generic/081
+++ b/tests/generic/081
@@ -73,7 +73,8 @@ $LVM_PROG vgcreate -f $vgname $SCRATCH_DEV >>$seqres.full 2>&1 || \
 	_notrun "LVM is too stupid for this device"
 # We use yes pipe instead of 'lvcreate --yes' because old version of lvm
 # (like 2.02.95 in RHEL6) don't support --yes option
-yes | $LVM_PROG lvcreate -L ${lvsize}M -n $lvname $vgname >>$seqres.full 2>&1
+yes | $LVM_PROG lvcreate -L ${lvsize}M -n $lvname $vgname >>$seqres.full 2>&1 \
+	|| _notrun "Can't set up LVM"
 _udev_wait /dev/mapper/$vgname-$lvname
 
 
-- 
2.47.2

[PATCH 3/3] generic/450: don't hang when running on devices with protection information

[Christoph Hellwig]

The dm thinp target does not work on devices with protection information,
and the udevadm settle command simply hangs when lvcreate failed because
of that.  Avoid that by doing a strategic _notrun when lvcreate fails.

Signed-off-by: Christoph Hellwig <hch@lst.de>
---
 tests/generic/459 | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/tests/generic/459 b/tests/generic/459
index 3f5f0f2b7dc1..48520f9f4af0 100755
--- a/tests/generic/459
+++ b/tests/generic/459
@@ -93,12 +93,14 @@ $LVM_PROG vgcreate -f $vgname $SCRATCH_DEV >>$seqres.full 2>&1
 
 $LVM_PROG lvcreate  --thinpool $poolname  --errorwhenfull y \
 		    --zero n -L $origpsize \
-		    --poolmetadatasize 4M $vgname >>$seqres.full 2>&1
+		    --poolmetadatasize 4M $vgname >>$seqres.full 2>&1 || \
+	_notrun "Can't create thin pool"
 
 # Create a overprovisioned 300MB dm-thin virt. device
 $LVM_PROG lvcreate  --virtualsize $virtsize \
 		    -T $vgname/$poolname \
-		    -n $lvname >>$seqres.full 2>&1
+		    -n $lvname >>$seqres.full 2>&1 || \
+	_notrun "Can't create thin volume"
 _udev_wait /dev/mapper/$vgname-$lvname
 _mkfs_dev /dev/mapper/$vgname-$lvname >>$seqres.full 2>&1
 
-- 
2.47.2

Re: [PATCH 1/3] dmthin: _notrun when dm-thin setup fails

[Darrick J. Wong]

On Mon, Aug 18, 2025 at 09:27:16AM +0200, Christoph Hellwig wrote:
> dm-think does not work on device with protection information.  Instead
> of failing the tests using it in obscure ways, _norun them.

Heh, I wish we had a thinking device mapper target. :)

> Signed-off-by: Christoph Hellwig <hch@lst.de>

Anyhow, failure to set up a dm-thinp volume sounds like failing to meet
a test precondition, sooo

Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>

--D

> ---
>  common/dmthin | 10 +++++-----
>  1 file changed, 5 insertions(+), 5 deletions(-)
> 
> diff --git a/common/dmthin b/common/dmthin
> index 453e43a0499e..a1e1fb8763c0 100644
> --- a/common/dmthin
> +++ b/common/dmthin
> @@ -105,13 +105,13 @@ _dmthin_init()
>  	# Metadata device
>  	DMTHIN_META_TABLE="0 $meta_dev_size linear $dm_backing_dev $meta_dev_offset"
>  	_dmsetup_create $DMTHIN_META_NAME --table "$DMTHIN_META_TABLE" || \
> -		_fatal "failed to create dm thin meta device"
> +		_notrun "failed to create dm thin meta device"
>  
>  	# Data device
>  	local data_dev_offset=$((meta_dev_offset + $meta_dev_size))
>  	DMTHIN_DATA_TABLE="0 $data_dev_size linear $dm_backing_dev $data_dev_offset"
>  	_dmsetup_create $DMTHIN_DATA_NAME --table "$DMTHIN_DATA_TABLE" || \
> -		_fatal "failed to create dm thin data device"
> +		_notrun "failed to create dm thin data device"
>  
>  	# Zap the pool metadata dev.  Explicitly fsync the zeroes to disk
>  	# because a slow-running udev running concurrently with dd can maintain
> @@ -124,16 +124,16 @@ _dmthin_init()
>  	# "start length thin-pool metadata_dev data_dev data_block_size low_water_mark"
>  	DMTHIN_POOL_TABLE="0 $data_dev_size thin-pool $DMTHIN_META_DEV $DMTHIN_DATA_DEV $cluster_size $low_water"
>  	_dmsetup_create $DMTHIN_POOL_NAME --table "$DMTHIN_POOL_TABLE" || \
> -		_fatal "failed to create dm thin pool device"
> +		_notun "failed to create dm thin pool device"
>  
>  	# Thin volume
>  	$DMSETUP_PROG message $DMTHIN_POOL_DEV 0 "create_thin $pool_id" || \
> -		_fatal "failed to message pool device"
> +		_notrun "failed to message pool device"
>  
>  	# start length thin pool_dev dev_id [external_origin_dev]
>  	DMTHIN_VOL_TABLE="0 $virtual_size thin $DMTHIN_POOL_DEV $pool_id"
>  	_dmsetup_create $DMTHIN_VOL_NAME --table "$DMTHIN_VOL_TABLE" || \
> -		_fatal "failed to create dm thin volume device"
> +		_notrun "failed to create dm thin volume device"
>  
>  }
>  
> -- 
> 2.47.2
> 
> 

Re: [PATCH 2/3] generic/081: don't hang when running on devices with protection information

[Darrick J. Wong]

On Mon, Aug 18, 2025 at 09:27:17AM +0200, Christoph Hellwig wrote:
> The dm snapshot target does not work on devices with protection
> information, and the udevadm settle command simply hangs when lvcreate
> failed because of that.  Avoid that by doing a strategic _notrun when
> lvcreate fails.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Makes sense to me
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>

--D

> ---
>  tests/generic/081 | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/tests/generic/081 b/tests/generic/081
> index 97928bf4a435..00280e9cff3b 100755
> --- a/tests/generic/081
> +++ b/tests/generic/081
> @@ -73,7 +73,8 @@ $LVM_PROG vgcreate -f $vgname $SCRATCH_DEV >>$seqres.full 2>&1 || \
>  	_notrun "LVM is too stupid for this device"
>  # We use yes pipe instead of 'lvcreate --yes' because old version of lvm
>  # (like 2.02.95 in RHEL6) don't support --yes option
> -yes | $LVM_PROG lvcreate -L ${lvsize}M -n $lvname $vgname >>$seqres.full 2>&1
> +yes | $LVM_PROG lvcreate -L ${lvsize}M -n $lvname $vgname >>$seqres.full 2>&1 \
> +	|| _notrun "Can't set up LVM"
>  _udev_wait /dev/mapper/$vgname-$lvname
>  
>  
> -- 
> 2.47.2
> 
> 

Re: [PATCH 3/3] generic/450: don't hang when running on devices with protection information

[Darrick J. Wong]

On Mon, Aug 18, 2025 at 09:27:18AM +0200, Christoph Hellwig wrote:
> The dm thinp target does not work on devices with protection information,
> and the udevadm settle command simply hangs when lvcreate failed because
> of that.  Avoid that by doing a strategic _notrun when lvcreate fails.
> 
> Signed-off-by: Christoph Hellwig <hch@lst.de>

Looks fine,
Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>

--D

> ---
>  tests/generic/459 | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/tests/generic/459 b/tests/generic/459
> index 3f5f0f2b7dc1..48520f9f4af0 100755
> --- a/tests/generic/459
> +++ b/tests/generic/459
> @@ -93,12 +93,14 @@ $LVM_PROG vgcreate -f $vgname $SCRATCH_DEV >>$seqres.full 2>&1
>  
>  $LVM_PROG lvcreate  --thinpool $poolname  --errorwhenfull y \
>  		    --zero n -L $origpsize \
> -		    --poolmetadatasize 4M $vgname >>$seqres.full 2>&1
> +		    --poolmetadatasize 4M $vgname >>$seqres.full 2>&1 || \
> +	_notrun "Can't create thin pool"
>  
>  # Create a overprovisioned 300MB dm-thin virt. device
>  $LVM_PROG lvcreate  --virtualsize $virtsize \
>  		    -T $vgname/$poolname \
> -		    -n $lvname >>$seqres.full 2>&1
> +		    -n $lvname >>$seqres.full 2>&1 || \
> +	_notrun "Can't create thin volume"
>  _udev_wait /dev/mapper/$vgname-$lvname
>  _mkfs_dev /dev/mapper/$vgname-$lvname >>$seqres.full 2>&1
>  
> -- 
> 2.47.2
> 
> 

Re: [PATCH 1/3] dmthin: _notrun when dm-thin setup fails

[Zorro Lang]

On Mon, Aug 18, 2025 at 01:51:31PM -0700, Darrick J. Wong wrote:
> On Mon, Aug 18, 2025 at 09:27:16AM +0200, Christoph Hellwig wrote:
> > dm-think does not work on device with protection information.  Instead
> > of failing the tests using it in obscure ways, _norun them.
> 
> Heh, I wish we had a thinking device mapper target. :)

Me too :) anyway I'll help to change this typo.

> 
> > Signed-off-by: Christoph Hellwig <hch@lst.de>
> 
> Anyhow, failure to set up a dm-thinp volume sounds like failing to meet
> a test precondition, sooo

I'm wondering, if we choose _notrun when _dmthin_init/dmsetup_create fails,
but others _dmxxxx_init/dmsetup_create [1]  fails, we still call _fatal. Isn't
that an inconsistent behavior :)

Thanks,
Zorro

[1]
zorro@zlang-laptop:~/git/upstream/xfs/xfstests-dev$ ls common/dm*
common/dmdelay  common/dmdust  common/dmerror  common/dmflakey  common/dmhugedisk  common/dmlogwrites  common/dmthin

> 
> Reviewed-by: "Darrick J. Wong" <djwong@kernel.org>
> 
> --D
> 
> > ---
> >  common/dmthin | 10 +++++-----
> >  1 file changed, 5 insertions(+), 5 deletions(-)
> > 
> > diff --git a/common/dmthin b/common/dmthin
> > index 453e43a0499e..a1e1fb8763c0 100644
> > --- a/common/dmthin
> > +++ b/common/dmthin
> > @@ -105,13 +105,13 @@ _dmthin_init()
> >  	# Metadata device
> >  	DMTHIN_META_TABLE="0 $meta_dev_size linear $dm_backing_dev $meta_dev_offset"
> >  	_dmsetup_create $DMTHIN_META_NAME --table "$DMTHIN_META_TABLE" || \
> > -		_fatal "failed to create dm thin meta device"
> > +		_notrun "failed to create dm thin meta device"
> >  
> >  	# Data device
> >  	local data_dev_offset=$((meta_dev_offset + $meta_dev_size))
> >  	DMTHIN_DATA_TABLE="0 $data_dev_size linear $dm_backing_dev $data_dev_offset"
> >  	_dmsetup_create $DMTHIN_DATA_NAME --table "$DMTHIN_DATA_TABLE" || \
> > -		_fatal "failed to create dm thin data device"
> > +		_notrun "failed to create dm thin data device"
> >  
> >  	# Zap the pool metadata dev.  Explicitly fsync the zeroes to disk
> >  	# because a slow-running udev running concurrently with dd can maintain
> > @@ -124,16 +124,16 @@ _dmthin_init()
> >  	# "start length thin-pool metadata_dev data_dev data_block_size low_water_mark"
> >  	DMTHIN_POOL_TABLE="0 $data_dev_size thin-pool $DMTHIN_META_DEV $DMTHIN_DATA_DEV $cluster_size $low_water"
> >  	_dmsetup_create $DMTHIN_POOL_NAME --table "$DMTHIN_POOL_TABLE" || \
> > -		_fatal "failed to create dm thin pool device"
> > +		_notun "failed to create dm thin pool device"
> >  
> >  	# Thin volume
> >  	$DMSETUP_PROG message $DMTHIN_POOL_DEV 0 "create_thin $pool_id" || \
> > -		_fatal "failed to message pool device"
> > +		_notrun "failed to message pool device"
> >  
> >  	# start length thin pool_dev dev_id [external_origin_dev]
> >  	DMTHIN_VOL_TABLE="0 $virtual_size thin $DMTHIN_POOL_DEV $pool_id"
> >  	_dmsetup_create $DMTHIN_VOL_NAME --table "$DMTHIN_VOL_TABLE" || \
> > -		_fatal "failed to create dm thin volume device"
> > +		_notrun "failed to create dm thin volume device"
> >  
> >  }
> >  
> > -- 
> > 2.47.2
> > 
> > 
> 

Re: [PATCH 1/3] dmthin: _notrun when dm-thin setup fails

[Christoph Hellwig]

On Mon, Aug 25, 2025 at 11:44:10PM +0800, Zorro Lang wrote:
> > Anyhow, failure to set up a dm-thinp volume sounds like failing to meet
> > a test precondition, sooo
> 
> I'm wondering, if we choose _notrun when _dmthin_init/dmsetup_create fails,
> but others _dmxxxx_init/dmsetup_create [1]  fails, we still call _fatal. Isn't
> that an inconsistent behavior :)

I think more notrun would be fine.  These are the ones I ran into
with a totally normal setup that has PI enabled on the test and
scratch devices, so they were my priority.

Re: [PATCH 1/3] dmthin: _notrun when dm-thin setup fails

[Zorro Lang]

On Tue, Aug 26, 2025 at 03:22:30PM +0200, Christoph Hellwig wrote:
> On Mon, Aug 25, 2025 at 11:44:10PM +0800, Zorro Lang wrote:
> > > Anyhow, failure to set up a dm-thinp volume sounds like failing to meet
> > > a test precondition, sooo
> > 
> > I'm wondering, if we choose _notrun when _dmthin_init/dmsetup_create fails,
> > but others _dmxxxx_init/dmsetup_create [1]  fails, we still call _fatal. Isn't
> > that an inconsistent behavior :)
> 
> I think more notrun would be fine.  These are the ones I ran into
> with a totally normal setup that has PI enabled on the test and
> scratch devices, so they were my priority.

OK, the patchset has been merged into patches-in-queue branch, will be pushed
if no regression from it. If PI enabled devices don't support thinp related
tests, you can exclude "thin" group. (But we might need to make sure that
the "thin" group is added into all cases use _dmthin_init.)

Thanks,
Zorro

>