Skip to contentSkip to conversation

common/rc: _fs_sysfs_dname fetch fsid using btrfs tool

PATCH Filipe Manana 2 participants 7 messages Oct 28, 2023 – Oct 30, 2023 3 messages not in the archive Btrfs file system

lore
Share & tools
lore
#2 lore raw
In reply to a message not in the archive
Quoted 18 lines · Anand Jain

Currently _fs_sysfs_dname gets fsid from the findmnt command however this command provides the metadata_uuid in the context device is mounted with temp-fsid. So instead, use btrfs filesystem show command to know the temp-fsid.

---
v2: new.

 common/rc | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
common/rc +0 -0
index 259a1ffb09b9..deeffe4228d4 100644

common/rc _fs_sysfs_dname() · line 4732

Quoted 6 lines
common/rc +2 -1
@@ -4728,7 +4728,9 @@ _fs_sysfs_dname()
 
        case "$FSTYP" in
        btrfs)
-               findmnt -n -o UUID ${dev} ;;
+               fsid=$($BTRFS_UTIL_PROG filesystem show ${dev} | grep uuid: | \
+                                                       awk '{print $NF}')

Please make the variable local.

Thanks.

Quoted 7 lines
common/rc +1 -1
+               echo $fsid ;;
        *)
                _short_dev $dev ;;
        esac
--

2.39.3

#5 lore raw

Re: [PATCH 2/2 v2] fstests: btrfs/219 cloned-device mount capability update

In reply to a message not in the archive
Quoted 38 lines · Anand Jain

This test case originally checked for failed cloned device mounts, which is no longer relevant after the commit a5b8a5f9f835 ("btrfs: support cloned-device mount capability"). So removing the obsolete part.

Additionally, add this test case back to the auto group which reverts the commit e2e7b549380a ("fstests: btrfs/219: remove it from auto group") since the previously missing kernel commit 5f58d783fd78 ("btrfs: free device in btrfs_close_devices for a single device filesystem") has already been integrated.

Add cleanups of the local variables and the _cleanup() function.

---
v2: Restores the code where it tests clone-devices.

 tests/btrfs/219 | 79 ++++++++++++++++++++++++++++++-------------------
 1 file changed, 49 insertions(+), 30 deletions(-)
tests/btrfs/219 +1 -1
index b747ce34fcc4..eb3e0487aa74 100755
@@ -12,21 +12,27 @@
 #
 
 . ./common/preamble
-_begin_fstest quick volume
+_begin_fstest auto quick volume
 
 # Override the default cleanup function.
 _cleanup()
 {
        cd /
        rm -f $tmp.*

tests/btrfs/219 line 23

Quoted 10 lines
tests/btrfs/219 +2 -8
-       if [ ! -z "$loop_mnt" ]; then
-               $UMOUNT_PROG $loop_mnt
-               rm -rf $loop_mnt
-       fi
-       [ ! -z "$loop_mnt1" ] && rm -rf $loop_mnt1
-       [ ! -z "$fs_img1" ] && rm -rf $fs_img1
-       [ ! -z "$fs_img2" ] && rm -rf $fs_img2
-       [ ! -z "$loop_dev" ] && _destroy_loop_device $loop_dev
+
+       # The variables are set before the test case can fail.

What if the _require_* statements fail?
Then the variables won't be defined...

tests/btrfs/219 line 34

Quoted 11 lines
tests/btrfs/219 +11 -0
+       $UMOUNT_PROG ${loop_mnt1} &> /dev/null
+       $UMOUNT_PROG ${loop_mnt2} &> /dev/null
+
+       _destroy_loop_device $loop_dev1 &> /dev/null
+       _destroy_loop_device $loop_dev2 &> /dev/null
+
+       rm -rf $fs_img1 &> /dev/null
+       rm -rf $fs_img2 &> /dev/null
+
+       rm -rf $loop_mnt1 &> /dev/null
+       rm -rf $loop_mnt2 &> /dev/null

Also please for simplicity and clarity don't mix this type of change with the actual purpose of the patch,
to make the test succeed on a kernel with the temp-fsid feature.

You're mixing 3 different changes in the same patch...

Quoted 11 lines
tests/btrfs/219 +2 -0
+
        _btrfs_rescan_devices
 }
 
@@ -38,55 +44,68 @@ _cleanup()
 _supported_fs btrfs
 _require_test
 _require_loop
+_require_btrfs_fs_sysfs
 _require_btrfs_forget_or_module_loadable
 _fixed_by_kernel_commit 5f58d783fd78 \

tests/btrfs/219 _cleanup() · line 55

Quoted 10 lines
tests/btrfs/219 +4 -4
        "btrfs: free device in btrfs_close_devices for a single device filesystem"
 
-loop_mnt=$TEST_DIR/$seq.mnt
-loop_mnt1=$TEST_DIR/$seq.mnt1
-fs_img1=$TEST_DIR/$seq.img1
-fs_img2=$TEST_DIR/$seq.img2
+loop_mnt1=$TEST_DIR/$seq/mnt1
+loop_mnt2=$TEST_DIR/$seq/mnt2
+fs_img1=$TEST_DIR/$seq/img1
+fs_img2=$TEST_DIR/$seq/img2

So this is the other unrelated change, renaming all these variables... This is making the diff larger to follow as this has nothing to do with the goal of making the test succeed on a kernel with the temp-fsid feature.

Quoted 56 lines
tests/btrfs/219 +16 -16
-mkdir $loop_mnt
-mkdir $loop_mnt1
+mkdir -p $loop_mnt1
+mkdir -p $loop_mnt2
 
 $XFS_IO_PROG -f -c "truncate 256m" $fs_img1 >>$seqres.full 2>&1
 
 _mkfs_dev $fs_img1 >>$seqres.full 2>&1
 cp $fs_img1 $fs_img2
 
+loop_dev1=`_create_loop_device $fs_img1`
+loop_dev2=`_create_loop_device $fs_img2`
+
 # Normal single device case, should pass just fine
-_mount -o loop $fs_img1 $loop_mnt > /dev/null  2>&1 || \
+_mount $loop_dev1 $loop_mnt1 > /dev/null  2>&1 || \
        _fail "Couldn't do initial mount"
-$UMOUNT_PROG $loop_mnt
+$UMOUNT_PROG $loop_mnt1
 
 _btrfs_forget_or_module_reload
 
 # Now mount the new version again to get the higher generation cached, umount
 # and try to mount the old version.  Mount the new version again just for good
 # measure.
-loop_dev=`_create_loop_device $fs_img1`
-
-_mount $loop_dev $loop_mnt > /dev/null 2>&1 || \
+_mount $loop_dev1 $loop_mnt1 > /dev/null 2>&1 || \
        _fail "Failed to mount the second time"
-$UMOUNT_PROG $loop_mnt
+$UMOUNT_PROG $loop_mnt1
 
-_mount -o loop $fs_img2 $loop_mnt > /dev/null 2>&1 || \
+_mount $loop_dev2 $loop_mnt2 > /dev/null 2>&1 || \
        _fail "We couldn't mount the old generation"
-$UMOUNT_PROG $loop_mnt
+$UMOUNT_PROG $loop_mnt2
 
-_mount $loop_dev $loop_mnt > /dev/null 2>&1 || \
+_mount $loop_dev1 $loop_mnt1 > /dev/null 2>&1 || \
        _fail "Failed to mount the second time"
-$UMOUNT_PROG $loop_mnt
+$UMOUNT_PROG $loop_mnt1
 
-# Now we definitely can't mount them at the same time, because we're still tied
-# to the limitation of one fs_devices per fsid.
+# Now try mount them at the same time, if kernel does not support
+# temp-fsid feature then mount should fail.
 _btrfs_forget_or_module_reload
 
-_mount $loop_dev $loop_mnt > /dev/null 2>&1 || \
+_mount $loop_dev1 $loop_mnt1 > /dev/null 2>&1 || \
        _fail "Failed to mount the third time"
-_mount -o loop $fs_img2 $loop_mnt1 > /dev/null 2>&1 && \

tests/btrfs/219 _cleanup() · line 103

Quoted 10 lines
tests/btrfs/219 +9 -1
-       _fail "We were allowed to mount when we should have failed"
+_mount $loop_dev2 $loop_mnt2 > /dev/null 2>&1
+
+if [ $? == 0 ]; then
+       # On temp-fsid supported kernels, mounting cloned device will be successfull.
+       if _has_fs_sysfs_attr $loop_dev2 temp_fsid; then
+               temp_fsid=$(_get_fs_sysfs_attr ${loop_dev2} temp_fsid)
+               if [ $temp_fsid == 0 ]; then
+                       _fail "Cloned devices mounted without temp_fsid"
+               fi

This is too complex. Why not just surround the existing code in an if statement like this:

if "sysfs-file-for-temp-fsid does not exist" then
run this code that fails with a temp-fsid enabled kernel
fi

Thanks.

Quoted 10 lines
tests/btrfs/219 +4 -1
+       else
+               _fail "We were allowed to mount when we should have failed"
+       fi
+fi
 
 _btrfs_rescan_devices
 # success, all done
--

2.39.3

#6 lore raw

Re: [PATCH 2/2 v2] fstests: btrfs/219 cloned-device mount capability update

Quoted 8 lines

Review

Quoted 5 lines
(quoted diff) +2 -0
+
+       # The variables are set before the test case can fail.

What if the _require_* statements fail?
Then the variables won't be defined...

Oh, yes indeed. It was fixed in v3 by moving the variable definition before the 'require' statement.

Quoted 18 lines

Review

Quoted 11 lines
(quoted diff) +11 -0
+       $UMOUNT_PROG ${loop_mnt1} &> /dev/null
+       $UMOUNT_PROG ${loop_mnt2} &> /dev/null
+
+       _destroy_loop_device $loop_dev1 &> /dev/null
+       _destroy_loop_device $loop_dev2 &> /dev/null
+
+       rm -rf $fs_img1 &> /dev/null
+       rm -rf $fs_img2 &> /dev/null
+
+       rm -rf $loop_mnt1 &> /dev/null
+       rm -rf $loop_mnt2 &> /dev/null

Also please for simplicity and clarity don't mix this type of change with the actual purpose of the patch,
to make the test succeed on a kernel with the temp-fsid feature.

You're mixing 3 different changes in the same patch...

Quoted 13 lines

Review

Quoted 8 lines
(quoted diff) +4 -4
-loop_mnt=$TEST_DIR/$seq.mnt
-loop_mnt1=$TEST_DIR/$seq.mnt1
-fs_img1=$TEST_DIR/$seq.img1
-fs_img2=$TEST_DIR/$seq.img2
+loop_mnt1=$TEST_DIR/$seq/mnt1
+loop_mnt2=$TEST_DIR/$seq/mnt2
+fs_img1=$TEST_DIR/$seq/img1
+fs_img2=$TEST_DIR/$seq/img2

So this is the other unrelated change, renaming all these variables... This is making the diff larger to follow as this has nothing to do with the goal of making the test succeed on a kernel with the temp-fsid feature.

Sure. I'll move these changes to a new patch in v3.

Quoted 10 lines

Review

Quoted 9 lines
(quoted diff) +9 -0
+_mount $loop_dev2 $loop_mnt2 > /dev/null 2>&1
+
+if [ $? == 0 ]; then
+       # On temp-fsid supported kernels, mounting cloned device will be successfull.
+       if _has_fs_sysfs_attr $loop_dev2 temp_fsid; then
+               temp_fsid=$(_get_fs_sysfs_attr ${loop_dev2} temp_fsid)
+               if [ $temp_fsid == 0 ]; then
+                       _fail "Cloned devices mounted without temp_fsid"
+               fi

Hmm. It is exactly testing what happens with and without the temp-fsid feature.

Quoted 7 lines

This is too complex. Why not just surround the existing code in an if statement like this:

if "sysfs-file-for-temp-fsid does not exist" then
run this code that fails with a temp-fsid enabled kernel
fi

Your suggestion will test what to expect when there is no 'temp-fsid.'

However, if 'temp-fsid' is present, we expect its sysfs temp_fsid to be set to 1, which I believe is a straightforward verification.

And do you think adding more comments will make it simpler? Such as:
# If the second mount is successful, then check if 'temp-fsid'

# is in the kernel. If it's present, verify that it outputs 1.

Or, Are you suggesting we postpone confirming the correct operation of 'temp-fsid' for another test case? However, I believe in thoroughly verifying it wherever it's used, as this approach promotes a better understanding of what to anticipate. No?

I will await your response before sending v3. Thx.

Thanks, Anand

#3 lore raw
In reply to Filipe Manana
Quoted 11 lines

Review _fs_sysfs_dname() · line 4732

Quoted 9 lines
(quoted diff) +2 -1
@@ -4728,7 +4728,9 @@ _fs_sysfs_dname()
 
         case "$FSTYP" in
         btrfs)
-               findmnt -n -o UUID ${dev} ;;
+               fsid=$($BTRFS_UTIL_PROG filesystem show ${dev} | grep uuid: | \
+                                                       awk '{print $NF}')

Please make the variable local.

Added in v3. Thx, Anand

#7 lore raw

Re: [PATCH 2/2 v2] fstests: btrfs/219 cloned-device mount capability update

In reply to Anand Jain
Quoted 4 lines · Anand Jain
Quoted 8 lines
Quoted 5 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+
+       # The variables are set before the test case can fail.

What if the _require_* statements fail?
Then the variables won't be defined...

Oh, yes indeed. It was fixed in v3 by moving the variable definition before the 'require' statement.

Is it really worth doing this type of change?
I mean it doesn't change the correctness of the test, doesn't make it more readable or
maintainable, or even shorter... It seems pointless to me, no clear benefit of any sort.

Quoted 68 lines
Quoted 20 lines

Review

Quoted 13 lines
(quoted diff) +11 -0
+       $UMOUNT_PROG ${loop_mnt1} &> /dev/null
+       $UMOUNT_PROG ${loop_mnt2} &> /dev/null
+
+       _destroy_loop_device $loop_dev1 &> /dev/null
+       _destroy_loop_device $loop_dev2 &> /dev/null
+
+       rm -rf $fs_img1 &> /dev/null
+       rm -rf $fs_img2 &> /dev/null
+
+       rm -rf $loop_mnt1 &> /dev/null
+       rm -rf $loop_mnt2 &> /dev/null

Also please for simplicity and clarity don't mix this type of change with the actual purpose of the patch,
to make the test succeed on a kernel with the temp-fsid feature.

You're mixing 3 different changes in the same patch...

Quoted 13 lines

Review

Quoted 8 lines
(quoted diff) +4 -4
-loop_mnt=$TEST_DIR/$seq.mnt
-loop_mnt1=$TEST_DIR/$seq.mnt1
-fs_img1=$TEST_DIR/$seq.img1
-fs_img2=$TEST_DIR/$seq.img2
+loop_mnt1=$TEST_DIR/$seq/mnt1
+loop_mnt2=$TEST_DIR/$seq/mnt2
+fs_img1=$TEST_DIR/$seq/img1
+fs_img2=$TEST_DIR/$seq/img2

So this is the other unrelated change, renaming all these variables... This is making the diff larger to follow as this has nothing to do with the goal of making the test succeed on a kernel with the temp-fsid feature.

Sure. I'll move these changes to a new patch in v3.

Quoted 10 lines

Review

Quoted 9 lines
(quoted diff) +9 -0
+_mount $loop_dev2 $loop_mnt2 > /dev/null 2>&1
+
+if [ $? == 0 ]; then
+       # On temp-fsid supported kernels, mounting cloned device will be successfull.
+       if _has_fs_sysfs_attr $loop_dev2 temp_fsid; then
+               temp_fsid=$(_get_fs_sysfs_attr ${loop_dev2} temp_fsid)
+               if [ $temp_fsid == 0 ]; then
+                       _fail "Cloned devices mounted without temp_fsid"
+               fi

Hmm. It is exactly testing what happens with and without the temp-fsid feature.

Quoted 7 lines

This is too complex. Why not just surround the existing code in an if statement like this:

if "sysfs-file-for-temp-fsid does not exist" then
run this code that fails with a temp-fsid enabled kernel
fi

Your suggestion will test what to expect when there is no 'temp-fsid.'

However, if 'temp-fsid' is present, we expect its sysfs temp_fsid to be set to 1, which I believe is a straightforward verification.

And do you think adding more comments will make it simpler? Such as:
# If the second mount is successful, then check if 'temp-fsid'

Quoted 6 lines
# is in the kernel. If it's present, verify that it outputs 1.

Or, Are you suggesting we postpone confirming the correct operation of 'temp-fsid' for another test case? However, I believe in thoroughly verifying it wherever it's used, as this approach promotes a better understanding of what to anticipate. No?

I'm not suggesting a new test case.

Remember the code you removed in v1?
My suggestion was to instead of removing it, just surround it in the body of an if statement:

if temp-fsid-feature-not-abailable; then
run that code you tried to remove in v1
fi

Isn't that a lot simpler and clear?

Thanks.

Quoted 5 lines

I will await your response before sending v3. Thx.

Thanks, Anand

#8 lore raw

Re: [PATCH 2/2 v2] fstests: btrfs/219 cloned-device mount capability update

Quoted 7 lines

Is it really worth doing this type of change?
I mean it doesn't change the correctness of the test, doesn't make it more readable or
maintainable, or even shorter... It seems pointless to me, no clear benefit of any sort.

It fixes the cleanup bug that, when a test case failed, it failed to remove the 2nd loop device.

Quoted 11 lines

I'm not suggesting a new test case.

Remember the code you removed in v1?
My suggestion was to instead of removing it, just surround it in the body of an if statement:

if temp-fsid-feature-not-abailable; then
run that code you tried to remove in v1
fi

Isn't that a lot simpler and clear?

Alright, I'll maintain the simplicity as mentioned above. Nevertheless, implementing a thorough check for temp-fsid doesn't add much complexity either.

#9 lore raw

Re: [PATCH 2/2 v2] fstests: btrfs/219 cloned-device mount capability update

Quoted 11 lines · Anand Jain
Quoted 8 lines

Is it really worth doing this type of change?
I mean it doesn't change the correctness of the test, doesn't make it more readable or
maintainable, or even shorter... It seems pointless to me, no clear benefit of any sort.

It fixes the cleanup bug that, when a test case failed, it failed to remove the 2nd loop device.

Then that's a very strong reason to make it a separate patch and explain the bug, instead
of sneaking it in with other different changes without being explicit about it.

Thanks.

Quoted 19 lines
Quoted 7 lines

I'm not suggesting a new test case.

Remember the code you removed in v1?
My suggestion was to instead of removing it, just surround it in the body of an if statement:

Quoted 5 lines

if temp-fsid-feature-not-abailable; then
run that code you tried to remove in v1
fi

Isn't that a lot simpler and clear?

Alright, I'll maintain the simplicity as mentioned above. Nevertheless, implementing a thorough check for temp-fsid doesn't add much complexity either.