Skip to contentSkip to conversation

fstests: btrfs: make nospace_cache related test cases to work with latest v2 cache

PATCH Superseded Qu Wenruo 4 participants 4 messages Nov 4, 2021 – Nov 5, 2021 Btrfs file system

v1 of 4 Superseded Latest is v4 →

lore
Share & tools
lore

Qu Wenruo

author
#0 lore raw

PATCH 1/1 fstests: btrfs: make nospace_cache related test cases to work with latest v2 cache

In the coming btrfs-progs v5.15 release, mkfs.btrfs will change to use v2 cache by default.

However nospace_cache mount option will not work with v2 cache, as it will make v2 cache out of sync with on-disk used space.

So mounting a btrfs with v2 cache using "nospace_cache" will make btrfs to reject the mount.

There are quite some test cases relying on nospace_cache to prevent v1 cache to take up data space.

For that case, we can append "clear_cache" mount option to it, so that btrfs knows we do not only want to prevent cache from being created, but also want to clear any existing v2 cache.

By this, we can keep those existing tests to do the same behavior for both v1 and v2 cache.

---
 tests/btrfs/102 | 2 +-
 tests/btrfs/140 | 2 +-
 tests/btrfs/141 | 2 +-
 tests/btrfs/142 | 2 +-
 tests/btrfs/143 | 2 +-
 tests/btrfs/151 | 2 +-
 tests/btrfs/157 | 2 +-
 tests/btrfs/158 | 2 +-
 tests/btrfs/170 | 2 +-
 tests/btrfs/199 | 5 ++++-
 tests/btrfs/215 | 2 +-
 11 files changed, 14 insertions(+), 11 deletions(-)
tests/btrfs/102 +1 -1
index e5a1b068..39c3ba37 100755
@@ -22,7 +22,7 @@ _scratch_mkfs >>$seqres.full 2>&1
 # Mount our filesystem without space caches enabled so that we do not get any
 # space used from the initial data block group that mkfs creates (space caches
 # used space from data block groups).
-_scratch_mount "-o nospace_cache"
+_scratch_mount "-o nospace_cache,clear_cache"
 
 # Need an fs with at least 2Gb to make sure mkfs.btrfs does not create an fs
 # using mixed block groups (used both for data and metadata). We really need
tests/btrfs/140 +1 -1
index 5a5f828c..e7a36927 100755
@@ -62,7 +62,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache
+_scratch_mount -o nospace_cache,clear_cache
 
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa -b 128K 0 128K" "$SCRATCH_MNT/foobar" |\
 	_filter_xfs_io_offset
tests/btrfs/141 +1 -1
index cf0979e9..bf559953 100755
@@ -61,7 +61,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache
+_scratch_mount -o nospace_cache,clear_cache
 
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa -b 128K 0 128K" "$SCRATCH_MNT/foobar" |\
 	_filter_xfs_io_offset
tests/btrfs/142 +1 -1
index 1318be0f..b2cc34a3 100755
@@ -39,7 +39,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache,nodatasum
+_scratch_mount -o nospace_cache,nodatasum,clear_cache
 
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa -b 128K 0 128K" "$SCRATCH_MNT/foobar" |\
 	_filter_xfs_io_offset
tests/btrfs/143 +1 -1
index 6736dcad..2f673ae2 100755
@@ -46,7 +46,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache,nodatasum
+_scratch_mount -o nospace_cache,nodatasum,clear_cache
 
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa -b 128K 0 128K" "$SCRATCH_MNT/foobar" |\
 	_filter_xfs_io_offset
tests/btrfs/151 +1 -1
index 099e85cc..e8bdc33c 100755
@@ -32,7 +32,7 @@ _scratch_dev_pool_get 3
 _scratch_pool_mkfs "-d raid1 -b 1G" >> $seqres.full 2>&1
 
 # we need an empty data chunk, so nospace_cache is required.
-_scratch_mount -onospace_cache
+_scratch_mount -onospace_cache,clear_cache
 
 # if data chunk is empty, 'btrfs device remove' can change raid1 to
 # single.
tests/btrfs/157 +1 -1
index 0cfe3ce5..cd69cc16 100755
@@ -66,7 +66,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache
+_scratch_mount -o nospace_cache,clear_cache
 
 # [0,64K) is written to stripe 0 and [64K, 128K) is written to stripe 1
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa 0 128K" -c "fsync" \
tests/btrfs/158 +1 -1
index ad374eba..1bc73450 100755
@@ -58,7 +58,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache
+_scratch_mount -o nospace_cache,clear_cache
 
 # [0,64K) is written to stripe 0 and [64K, 128K) is written to stripe 1
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa 0 128K" -c "fsync" \
tests/btrfs/170 +1 -1
index 15382eb3..746170f4 100755
@@ -29,7 +29,7 @@ _scratch_mkfs_sized $fs_size >>$seqres.full 2>&1
 
 # Mount without space cache so that we can precisely fill all data space and
 # unallocated space later (space cache v1 uses data block groups).
-_scratch_mount "-o nospace_cache"
+_scratch_mount "-o nospace_cache,clear_cache"
 
 # Create our test file and allocate 1826.25Mb of space for it.
 # This will exhaust the existing data block group and all unallocated space on
tests/btrfs/199 +4 -1
index 6aca62f4..2369d52f 100755
@@ -70,12 +70,15 @@ mkdir -p $loop_mnt
 # - nospace_cache
 #   Since v1 cache using DATA space, it can break data extent bytenr
 #   continuousness.
+# - clear_cache
+#   v2 cache can't work with nospace_cache, so we workaround it by clearing
+#   all the space cache.
 # - nodatasum
 #   As there will be 1.5G data write, generating 1.5M csum.
 #   Disabling datasum could reduce the margin caused by metadata to minimal
 # - discard
 #   What we're testing
-_mount -o nospace_cache,nodatasum,discard $loop_dev $loop_mnt
+_mount -o nospace_cache,clear_cache,nodatasum,discard $loop_dev $loop_mnt
 
 # Craft the following extent layout:
 #         |  BG1 |      BG2        |       BG3            |
tests/btrfs/215 +1 -1
index fa622568..7f0986af 100755
@@ -30,7 +30,7 @@ _require_non_zoned_device $SCRATCH_DEV
 _scratch_mkfs > /dev/null
 # disable freespace inode to ensure file is the first thing in the data
 # blobk group
-_scratch_mount -o nospace_cache
+_scratch_mount -o nospace_cache,clear_cache
 
 pagesize=$(get_page_size)
 blocksize=$(_get_block_size $SCRATCH_MNT)
Signature · 1 line
2.33.0
Reviewed-by on v1 1/1
#1 lore raw
Quoted 6 lines · Qu Wenruo

In the coming btrfs-progs v5.15 release, mkfs.btrfs will change to use v2 cache by default.

However nospace_cache mount option will not work with v2 cache, as it will make v2 cache out of sync with on-disk used space.

will -> would

(having "will" was actually making me worry and check the code)

Quoted 7 lines

So mounting a btrfs with v2 cache using "nospace_cache" will make btrfs to reject the mount.

There are quite some test cases relying on nospace_cache to prevent v1 cache to take up data space.

Quoted 8 lines

For that case, we can append "clear_cache" mount option to it, so that btrfs knows we do not only want to prevent cache from being created, but also want to clear any existing v2 cache.

By this, we can keep those existing tests to do the same behavior for both v1 and v2 cache.

Apart from that, it looks good to me, thanks.

Quoted 168 lines
---
 tests/btrfs/102 | 2 +-
 tests/btrfs/140 | 2 +-
 tests/btrfs/141 | 2 +-
 tests/btrfs/142 | 2 +-
 tests/btrfs/143 | 2 +-
 tests/btrfs/151 | 2 +-
 tests/btrfs/157 | 2 +-
 tests/btrfs/158 | 2 +-
 tests/btrfs/170 | 2 +-
 tests/btrfs/199 | 5 ++++-
 tests/btrfs/215 | 2 +-
 11 files changed, 14 insertions(+), 11 deletions(-)
tests/btrfs/102 +1 -1
index e5a1b068..39c3ba37 100755
@@ -22,7 +22,7 @@ _scratch_mkfs >>$seqres.full 2>&1
 # Mount our filesystem without space caches enabled so that we do not get any
 # space used from the initial data block group that mkfs creates (space caches
 # used space from data block groups).
-_scratch_mount "-o nospace_cache"
+_scratch_mount "-o nospace_cache,clear_cache"
 
 # Need an fs with at least 2Gb to make sure mkfs.btrfs does not create an fs
 # using mixed block groups (used both for data and metadata). We really need
tests/btrfs/140 +1 -1
index 5a5f828c..e7a36927 100755
@@ -62,7 +62,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache
+_scratch_mount -o nospace_cache,clear_cache
 
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa -b 128K 0 128K" "$SCRATCH_MNT/foobar" |\
        _filter_xfs_io_offset
tests/btrfs/141 +1 -1
index cf0979e9..bf559953 100755
@@ -61,7 +61,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache
+_scratch_mount -o nospace_cache,clear_cache
 
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa -b 128K 0 128K" "$SCRATCH_MNT/foobar" |\
        _filter_xfs_io_offset
tests/btrfs/142 +1 -1
index 1318be0f..b2cc34a3 100755
@@ -39,7 +39,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache,nodatasum
+_scratch_mount -o nospace_cache,nodatasum,clear_cache
 
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa -b 128K 0 128K" "$SCRATCH_MNT/foobar" |\
        _filter_xfs_io_offset
tests/btrfs/143 +1 -1
index 6736dcad..2f673ae2 100755
@@ -46,7 +46,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache,nodatasum
+_scratch_mount -o nospace_cache,nodatasum,clear_cache
 
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa -b 128K 0 128K" "$SCRATCH_MNT/foobar" |\
        _filter_xfs_io_offset
tests/btrfs/151 +1 -1
index 099e85cc..e8bdc33c 100755
@@ -32,7 +32,7 @@ _scratch_dev_pool_get 3
 _scratch_pool_mkfs "-d raid1 -b 1G" >> $seqres.full 2>&1
 
 # we need an empty data chunk, so nospace_cache is required.
-_scratch_mount -onospace_cache
+_scratch_mount -onospace_cache,clear_cache
 
 # if data chunk is empty, 'btrfs device remove' can change raid1 to
 # single.
tests/btrfs/157 +1 -1
index 0cfe3ce5..cd69cc16 100755
@@ -66,7 +66,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache
+_scratch_mount -o nospace_cache,clear_cache
 
 # [0,64K) is written to stripe 0 and [64K, 128K) is written to stripe 1
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa 0 128K" -c "fsync" \
tests/btrfs/158 +1 -1
index ad374eba..1bc73450 100755
@@ -58,7 +58,7 @@ _scratch_pool_mkfs $mkfs_opts >>$seqres.full 2>&1
 
 # -o nospace_cache makes sure data is written to the start position of the data
 # chunk
-_scratch_mount -o nospace_cache
+_scratch_mount -o nospace_cache,clear_cache
 
 # [0,64K) is written to stripe 0 and [64K, 128K) is written to stripe 1
 $XFS_IO_PROG -f -d -c "pwrite -S 0xaa 0 128K" -c "fsync" \
tests/btrfs/170 +1 -1
index 15382eb3..746170f4 100755
@@ -29,7 +29,7 @@ _scratch_mkfs_sized $fs_size >>$seqres.full 2>&1
 
 # Mount without space cache so that we can precisely fill all data space and
 # unallocated space later (space cache v1 uses data block groups).
-_scratch_mount "-o nospace_cache"
+_scratch_mount "-o nospace_cache,clear_cache"
 
 # Create our test file and allocate 1826.25Mb of space for it.
 # This will exhaust the existing data block group and all unallocated space on
tests/btrfs/199 +4 -1
index 6aca62f4..2369d52f 100755
@@ -70,12 +70,15 @@ mkdir -p $loop_mnt
 # - nospace_cache
 #   Since v1 cache using DATA space, it can break data extent bytenr
 #   continuousness.
+# - clear_cache
+#   v2 cache can't work with nospace_cache, so we workaround it by clearing
+#   all the space cache.
 # - nodatasum
 #   As there will be 1.5G data write, generating 1.5M csum.
 #   Disabling datasum could reduce the margin caused by metadata to minimal
 # - discard
 #   What we're testing
-_mount -o nospace_cache,nodatasum,discard $loop_dev $loop_mnt
+_mount -o nospace_cache,clear_cache,nodatasum,discard $loop_dev $loop_mnt
 
 # Craft the following extent layout:
 #         |  BG1 |      BG2        |       BG3            |
tests/btrfs/215 +1 -2
index fa622568..7f0986af 100755
@@ -30,7 +30,7 @@ _require_non_zoned_device $SCRATCH_DEV
 _scratch_mkfs > /dev/null
 # disable freespace inode to ensure file is the first thing in the data
 # blobk group
-_scratch_mount -o nospace_cache
+_scratch_mount -o nospace_cache,clear_cache
 
 pagesize=$(get_page_size)
 blocksize=$(_get_block_size $SCRATCH_MNT)
--

2.33.0

Signature · 3 lines
Filipe David Manana,

“Whether you think you can, or you think you can't — you're right.”
#2 lore raw
In reply to Qu Wenruo
Quoted 12 lines · Qu Wenruo

In the coming btrfs-progs v5.15 release, mkfs.btrfs will change to use v2 cache by default.

However nospace_cache mount option will not work with v2 cache, as it will make v2 cache out of sync with on-disk used space.

So mounting a btrfs with v2 cache using "nospace_cache" will make btrfs to reject the mount.

There are quite some test cases relying on nospace_cache to prevent v1 cache to take up data space.

Quoted 8 lines

For that case, we can append "clear_cache" mount option to it, so that btrfs knows we do not only want to prevent cache from being created, but also want to clear any existing v2 cache.

By this, we can keep those existing tests to do the same behavior for both v1 and v2 cache.

This will re-generate the free space tree needlessly, and in fact won't be allowed for extent-tree-v2, so we'll just have to mess with it again. Instead add a helper to get the options if they're needed, something like

_btrfs_nocache_opt() {
$BTRFS_UTIL_PROG inspect-internal $SCRATCH_DEV | grep FREE_SPACE_TREE if [ $? -eq 0 ]; then
return "-o nospace_cache"
fi
return ""
}

or something like that. Thanks,

Josef

#3 lore raw
Quoted 26 lines · Josef Bacik
Quoted 12 lines · Qu Wenruo

In the coming btrfs-progs v5.15 release, mkfs.btrfs will change to use v2 cache by default.

However nospace_cache mount option will not work with v2 cache, as it will make v2 cache out of sync with on-disk used space.

So mounting a btrfs with v2 cache using "nospace_cache" will make btrfs to reject the mount.

There are quite some test cases relying on nospace_cache to prevent v1 cache to take up data space.

Quoted 8 lines

For that case, we can append "clear_cache" mount option to it, so that btrfs knows we do not only want to prevent cache from being created, but also want to clear any existing v2 cache.

By this, we can keep those existing tests to do the same behavior for both v1 and v2 cache.

This will re-generate the free space tree needlessly, and in fact won't be allowed for extent-tree-v2, so we'll just have to mess with it again. Instead add a helper to get the options if they're needed, something like

Quoted 7 lines

_btrfs_nocache_opt() {
$BTRFS_UTIL_PROG inspect-internal $SCRATCH_DEV | grep FREE_SPACE_TREE if [ $? -eq 0 ]; then
return "-o nospace_cache"
fi
return ""
}

OK, this indeed sounds better.

Thanks,
Qu

Quoted 5 lines

or something like that. Thanks,

Josef