Skip to contentSkip to conversation

Fix some tests that fail for exfat FS

PATCH Superseded Pavel Reichl 4 participants 15 messages Mar 30, 2021 – Apr 15, 2021

v2 of 4 Superseded Latest is v4 →

lore
Share & tools
lore
#0 lore raw

v2 cover Fix some tests that fail for exfat FS

Mostly just do not run test that require posix permissions and ownership.

Changes in V2 (Thank you for comments on version 1)

* Merged patch #1 and #2 (the one introducing the helper and the other actually using them)
* Fixed generic/597 by introducing _require_symlink
* Fixed generic/598 by adding _require_chmod
** changes are part of patch #1
* Introduced _format_swapon_file()

Pavel Reichl (4):
common/rc: Add _require_{chown,chmod,symlink}()
common: hide permision warning from mkswap for exfat generic/554: hide permision warning on exfat generic/003: Amend the test for exfat

 common/rc         | 48 ++++++++++++++++++++++++++++++++++++++++++++++-
 tests/generic/003 | 29 +++++++++++++++++++---------
 tests/generic/087 |  1 +
 tests/generic/088 |  1 +
 tests/generic/125 |  1 +
 tests/generic/126 |  1 +
 tests/generic/128 |  1 +
 tests/generic/193 |  1 +
 tests/generic/314 |  1 +
 tests/generic/317 |  1 +
 tests/generic/355 |  1 +
 tests/generic/554 |  3 ++-
 tests/generic/597 |  1 +
 tests/generic/598 |  1 +
 14 files changed, 80 insertions(+), 11 deletions(-)
Signature · 1 line
2.30.2
#1 lore raw

PATCH v2 1/4 common/rc: Add _require_{chown,chmod,symlink}()

Add helper functions that ensure that test is only executed on file systems that implement chown, chmod and symbolic links.

Fixed test are: generic/{87,88,125,126,128,193,314,317,355,597,598}

---
 common/rc         | 27 +++++++++++++++++++++++++++
 tests/generic/087 |  1 +
 tests/generic/088 |  1 +
 tests/generic/125 |  1 +
 tests/generic/126 |  1 +
 tests/generic/128 |  1 +
 tests/generic/193 |  1 +
 tests/generic/314 |  1 +
 tests/generic/317 |  1 +
 tests/generic/355 |  1 +
 tests/generic/597 |  1 +
 tests/generic/598 |  1 +
 12 files changed, 38 insertions(+)
common/rc +27 -0
index 0ce3cb0d..9cdfe21c 100644
@@ -2129,6 +2129,33 @@ _require_user()
     [ "$?" == "0" ] || _notrun "$qa_user cannot execute commands."
 }
 
+# check for a chown support
+#
+_require_chown()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "chown is not supported on $FSTYP"
+	fi
+}
+
+# check for a chmod support
+#
+_require_chmod()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "chmod is not supported on $FSTYP"
+	fi
+}
+
+# check for a symbolic links support
+#
+_require_symlink()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "symbolic links are not supported on $FSTYP"
+	fi
+}
+
 # check for a group on the machine, fsgqa as default
 #
 _require_group()
tests/generic/087 +1 -0
index 1f30dbf4..c3576117 100755
@@ -37,6 +37,7 @@ _cleanup()
 # real QA test starts here
 _supported_fs generic
 _require_test
+_require_chown
 
 QA_FS_PERMS=$here/src/fs_perms
 
tests/generic/088 +1 -0
index 9388a083..ad99bd7e 100755
@@ -29,6 +29,7 @@ _filter()
 # real QA test starts here
 _supported_fs generic
 _require_test
+_require_chown
 
 path=$TEST_DIR/t_access
 $here/src/t_access_root $path | tee $seqres.full | _filter
tests/generic/125 +1 -0
index e84248d3..8c8f5cd7 100755
@@ -25,6 +25,7 @@ _supported_fs generic
 _require_test
 _require_user
 _require_odirect
+_require_chmod
 
 TESTDIR=$TEST_DIR/ftrunc
 TESTFILE=$TESTDIR/ftrunc.tmp
tests/generic/126 +1 -0
index ac25d294..636ca00d 100755
@@ -27,6 +27,7 @@ _cleanup()
 # real QA test starts here
 _supported_fs generic
 _require_test
+_require_chown
 
 QA_FS_PERMS=$here/src/fs_perms
 
tests/generic/128 +1 -0
index b3e49eff..c1eae77a 100755
@@ -24,6 +24,7 @@ _supported_fs generic
 
 _require_scratch
 _require_user
+_require_chmod
 
 _scratch_mkfs >/dev/null 2>&1
 _scratch_mount "-o nosuid"
tests/generic/193 +1 -0
index 3125efdd..fd0ebbf6 100755
@@ -56,6 +56,7 @@ _supported_fs generic
 
 _require_test
 _require_user
+_require_chown
 
 test_root=$TEST_DIR/$seq.$$.root
 test_user=$TEST_DIR/$seq.$$.user
tests/generic/314 +1 -0
index 03df81ce..540f0feb 100755
@@ -29,6 +29,7 @@ _cleanup()
 _supported_fs generic
 _require_test
 _require_user
+_require_chown
 
 rm -rf $TEST_DIR/$seq-dir
 
tests/generic/317 +1 -0
index 29c37a57..289dfabe 100755
@@ -45,6 +45,7 @@ _require_scratch
 _require_user
 _require_ugid_map
 _require_userns
+_require_chown
 qa_user_id=`id -u $qa_user`
 
 _filter_output()
tests/generic/355 +1 -0
index 161dd042..74fba0f9 100755
@@ -32,6 +32,7 @@ _supported_fs generic
 _require_test
 _require_user
 _require_odirect
+_require_chown
 
 testfile=$TEST_DIR/$seq.test
 rm -f $testfile
tests/generic/597 +1 -0
index ba769d73..f596406c 100755
@@ -43,6 +43,7 @@ _require_sysctl_variable fs.protected_hardlinks
 _require_user fsgqa2
 # Do this SECOND so that qa_user is fsgqa, and _user_do uses that account
 _require_user fsgqa
+_require_symlink
 
 OWNER=fsgqa2
 OTHER=fsgqa
tests/generic/598 +1 -0
index 6b765275..230c3ac7 100755
@@ -43,6 +43,7 @@ _require_sysctl_variable fs.protected_fifos
 _require_user fsgqa2
 # Do this SECOND so that qa_user is fsgqa, and _user_do uses that account
 _require_user fsgqa
+_require_chmod
 
 USER1=fsgqa2
 USER2=fsgqa
Signature · 1 line
2.30.2
#2 lore raw

Re: [PATCH v2 1/4] common/rc: Add _require_{chown,chmod,symlink}()

Quoted 45 lines · Pavel Reichl

Add helper functions that ensure that test is only executed on file systems that implement chown, chmod and symbolic links.

Fixed test are: generic/{87,88,125,126,128,193,314,317,355,597,598}

---
 common/rc         | 27 +++++++++++++++++++++++++++
 tests/generic/087 |  1 +
 tests/generic/088 |  1 +
 tests/generic/125 |  1 +
 tests/generic/126 |  1 +
 tests/generic/128 |  1 +
 tests/generic/193 |  1 +
 tests/generic/314 |  1 +
 tests/generic/317 |  1 +
 tests/generic/355 |  1 +
 tests/generic/597 |  1 +
 tests/generic/598 |  1 +
 12 files changed, 38 insertions(+)
common/rc +16 -0
index 0ce3cb0d..9cdfe21c 100644
@@ -2129,6 +2129,33 @@ _require_user()
     [ "$?" == "0" ] || _notrun "$qa_user cannot execute commands."
 }
 
+# check for a chown support
+#
+_require_chown()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "chown is not supported on $FSTYP"
+	fi
+}
+
+# check for a chmod support
+#
+_require_chmod()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "chmod is not supported on $FSTYP"
+	fi

common/rc _require_user() · line 2157

Quoted 10 lines
common/rc +10 -0
+}
+
+# check for a symbolic links support
+#
+_require_symlink()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "symbolic links are not supported on $FSTYP"
+	fi
+}

There's _require_symlinks() in common/rc, I remember Eric used it for some exfat errors before. Does it work for you?

Thanks,
Zorro

Quoted 139 lines
common/rc +1 -0
+
 # check for a group on the machine, fsgqa as default
 #
 _require_group()
tests/generic/087 +1 -0
index 1f30dbf4..c3576117 100755
@@ -37,6 +37,7 @@ _cleanup()
 # real QA test starts here
 _supported_fs generic
 _require_test
+_require_chown
 
 QA_FS_PERMS=$here/src/fs_perms
 
tests/generic/088 +1 -0
index 9388a083..ad99bd7e 100755
@@ -29,6 +29,7 @@ _filter()
 # real QA test starts here
 _supported_fs generic
 _require_test
+_require_chown
 
 path=$TEST_DIR/t_access
 $here/src/t_access_root $path | tee $seqres.full | _filter
tests/generic/125 +1 -0
index e84248d3..8c8f5cd7 100755
@@ -25,6 +25,7 @@ _supported_fs generic
 _require_test
 _require_user
 _require_odirect
+_require_chmod
 
 TESTDIR=$TEST_DIR/ftrunc
 TESTFILE=$TESTDIR/ftrunc.tmp
tests/generic/126 +1 -0
index ac25d294..636ca00d 100755
@@ -27,6 +27,7 @@ _cleanup()
 # real QA test starts here
 _supported_fs generic
 _require_test
+_require_chown
 
 QA_FS_PERMS=$here/src/fs_perms
 
tests/generic/128 +1 -0
index b3e49eff..c1eae77a 100755
@@ -24,6 +24,7 @@ _supported_fs generic
 
 _require_scratch
 _require_user
+_require_chmod
 
 _scratch_mkfs >/dev/null 2>&1
 _scratch_mount "-o nosuid"
tests/generic/193 +1 -0
index 3125efdd..fd0ebbf6 100755
@@ -56,6 +56,7 @@ _supported_fs generic
 
 _require_test
 _require_user
+_require_chown
 
 test_root=$TEST_DIR/$seq.$$.root
 test_user=$TEST_DIR/$seq.$$.user
tests/generic/314 +1 -0
index 03df81ce..540f0feb 100755
@@ -29,6 +29,7 @@ _cleanup()
 _supported_fs generic
 _require_test
 _require_user
+_require_chown
 
 rm -rf $TEST_DIR/$seq-dir
 
tests/generic/317 +1 -0
index 29c37a57..289dfabe 100755
@@ -45,6 +45,7 @@ _require_scratch
 _require_user
 _require_ugid_map
 _require_userns
+_require_chown
 qa_user_id=`id -u $qa_user`
 
 _filter_output()
tests/generic/355 +1 -0
index 161dd042..74fba0f9 100755
@@ -32,6 +32,7 @@ _supported_fs generic
 _require_test
 _require_user
 _require_odirect
+_require_chown
 
 testfile=$TEST_DIR/$seq.test
 rm -f $testfile
tests/generic/597 +1 -0
index ba769d73..f596406c 100755
@@ -43,6 +43,7 @@ _require_sysctl_variable fs.protected_hardlinks
 _require_user fsgqa2
 # Do this SECOND so that qa_user is fsgqa, and _user_do uses that account
 _require_user fsgqa
+_require_symlink
 
 OWNER=fsgqa2
 OTHER=fsgqa
tests/generic/598 +1 -0
index 6b765275..230c3ac7 100755
@@ -43,6 +43,7 @@ _require_sysctl_variable fs.protected_fifos
 _require_user fsgqa2
 # Do this SECOND so that qa_user is fsgqa, and _user_do uses that account
 _require_user fsgqa
+_require_chmod
 
 USER1=fsgqa2
 USER2=fsgqa
Signature · 1 line
2.30.2
#3 lore raw

Re: [PATCH v2 1/4] common/rc: Add _require_{chown,chmod,symlink}()

Quoted 4 lines · Zorro Lang
Quoted 3 lines · Pavel Reichl

Add helper functions that ensure that test is only executed on file systems that implement chown, chmod and symbolic links.

Quoted 13 lines

common/rc _require_user() · line 2157

Quoted 9 lines
common/rc +9 -0
+
+# check for a symbolic links support
+#
+_require_symlink()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "symbolic links are not supported on $FSTYP"
+	fi
+}

There's _require_symlinks() in common/rc, I remember Eric used it for some exfat errors before. Does it work for you?

Hi,

Thanks for the info! Yes, it works. I will fix it in the next version.

I'll just wait for more comments a few days...I guess.

Thanks!

Bye.

#4 lore raw

Re: [PATCH v2 1/4] common/rc: Add _require_{chown,chmod,symlink}()

In reply to Pavel Reichl
Quoted 37 lines · Pavel Reichl

Add helper functions that ensure that test is only executed on file systems that implement chown, chmod and symbolic links.

Fixed test are: generic/{87,88,125,126,128,193,314,317,355,597,598}

---
 common/rc         | 27 +++++++++++++++++++++++++++
 tests/generic/087 |  1 +
 tests/generic/088 |  1 +
 tests/generic/125 |  1 +
 tests/generic/126 |  1 +
 tests/generic/128 |  1 +
 tests/generic/193 |  1 +
 tests/generic/314 |  1 +
 tests/generic/317 |  1 +
 tests/generic/355 |  1 +
 tests/generic/597 |  1 +
 tests/generic/598 |  1 +
 12 files changed, 38 insertions(+)
common/rc +8 -0
index 0ce3cb0d..9cdfe21c 100644
@@ -2129,6 +2129,33 @@ _require_user()
     [ "$?" == "0" ] || _notrun "$qa_user cannot execute commands."
 }
 
+# check for a chown support
+#
+_require_chown()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "chown is not supported on $FSTYP"
+	fi
+}

common/rc _require_user() · line 2149

Quoted 10 lines
common/rc +10 -0
+
+# check for a chmod support
+#
+_require_chmod()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "chmod is not supported on $FSTYP"
+	fi
+}
+

Does chown/chmod fail on exfat? Like the existing _require_symlink() implementation and many other _require rules, we try to actually do the action on $TEST_DIR, and check if command succeeds to see if the action is supported by current $FSTYP. Is it possible for exfat to do the same check?

We only use whitelist if it's impossible to do such check.

Thanks,
Eryu

Quoted 146 lines
common/rc +9 -0
+# check for a symbolic links support
+#
+_require_symlink()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "symbolic links are not supported on $FSTYP"
+	fi
+}
+
 # check for a group on the machine, fsgqa as default
 #
 _require_group()
tests/generic/087 +1 -0
index 1f30dbf4..c3576117 100755
@@ -37,6 +37,7 @@ _cleanup()
 # real QA test starts here
 _supported_fs generic
 _require_test
+_require_chown
 
 QA_FS_PERMS=$here/src/fs_perms
 
tests/generic/088 +1 -0
index 9388a083..ad99bd7e 100755
@@ -29,6 +29,7 @@ _filter()
 # real QA test starts here
 _supported_fs generic
 _require_test
+_require_chown
 
 path=$TEST_DIR/t_access
 $here/src/t_access_root $path | tee $seqres.full | _filter
tests/generic/125 +1 -0
index e84248d3..8c8f5cd7 100755
@@ -25,6 +25,7 @@ _supported_fs generic
 _require_test
 _require_user
 _require_odirect
+_require_chmod
 
 TESTDIR=$TEST_DIR/ftrunc
 TESTFILE=$TESTDIR/ftrunc.tmp
tests/generic/126 +1 -0
index ac25d294..636ca00d 100755
@@ -27,6 +27,7 @@ _cleanup()
 # real QA test starts here
 _supported_fs generic
 _require_test
+_require_chown
 
 QA_FS_PERMS=$here/src/fs_perms
 
tests/generic/128 +1 -0
index b3e49eff..c1eae77a 100755
@@ -24,6 +24,7 @@ _supported_fs generic
 
 _require_scratch
 _require_user
+_require_chmod
 
 _scratch_mkfs >/dev/null 2>&1
 _scratch_mount "-o nosuid"
tests/generic/193 +1 -0
index 3125efdd..fd0ebbf6 100755
@@ -56,6 +56,7 @@ _supported_fs generic
 
 _require_test
 _require_user
+_require_chown
 
 test_root=$TEST_DIR/$seq.$$.root
 test_user=$TEST_DIR/$seq.$$.user
tests/generic/314 +1 -0
index 03df81ce..540f0feb 100755
@@ -29,6 +29,7 @@ _cleanup()
 _supported_fs generic
 _require_test
 _require_user
+_require_chown
 
 rm -rf $TEST_DIR/$seq-dir
 
tests/generic/317 +1 -0
index 29c37a57..289dfabe 100755
@@ -45,6 +45,7 @@ _require_scratch
 _require_user
 _require_ugid_map
 _require_userns
+_require_chown
 qa_user_id=`id -u $qa_user`
 
 _filter_output()
tests/generic/355 +1 -0
index 161dd042..74fba0f9 100755
@@ -32,6 +32,7 @@ _supported_fs generic
 _require_test
 _require_user
 _require_odirect
+_require_chown
 
 testfile=$TEST_DIR/$seq.test
 rm -f $testfile
tests/generic/597 +1 -0
index ba769d73..f596406c 100755
@@ -43,6 +43,7 @@ _require_sysctl_variable fs.protected_hardlinks
 _require_user fsgqa2
 # Do this SECOND so that qa_user is fsgqa, and _user_do uses that account
 _require_user fsgqa
+_require_symlink
 
 OWNER=fsgqa2
 OTHER=fsgqa
tests/generic/598 +1 -0
index 6b765275..230c3ac7 100755
@@ -43,6 +43,7 @@ _require_sysctl_variable fs.protected_fifos
 _require_user fsgqa2
 # Do this SECOND so that qa_user is fsgqa, and _user_do uses that account
 _require_user fsgqa
+_require_chmod
 
 USER1=fsgqa2
 USER2=fsgqa
Signature · 1 line
2.30.2
#5 lore raw

Re: [PATCH v2 1/4] common/rc: Add _require_{chown,chmod,symlink}()

Quoted 55 lines · Eryu Guan
Quoted 37 lines · Pavel Reichl

Add helper functions that ensure that test is only executed on file systems that implement chown, chmod and symbolic links.

Fixed test are: generic/{87,88,125,126,128,193,314,317,355,597,598}

---
 common/rc         | 27 +++++++++++++++++++++++++++
 tests/generic/087 |  1 +
 tests/generic/088 |  1 +
 tests/generic/125 |  1 +
 tests/generic/126 |  1 +
 tests/generic/128 |  1 +
 tests/generic/193 |  1 +
 tests/generic/314 |  1 +
 tests/generic/317 |  1 +
 tests/generic/355 |  1 +
 tests/generic/597 |  1 +
 tests/generic/598 |  1 +
 12 files changed, 38 insertions(+)
common/rc +8 -0
index 0ce3cb0d..9cdfe21c 100644
@@ -2129,6 +2129,33 @@ _require_user()
     [ "$?" == "0" ] || _notrun "$qa_user cannot execute commands."
 }
 
+# check for a chown support
+#
+_require_chown()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "chown is not supported on $FSTYP"
+	fi
+}

common/rc _require_user() · line 2149

Quoted 10 lines
common/rc +10 -0
+
+# check for a chmod support
+#
+_require_chmod()
+{
+	if [ "$FSTYP" = "exfat" ]; then
+		_notrun "chmod is not supported on $FSTYP"
+	fi
+}
+

Does chown/chmod fail on exfat? Like the existing _require_symlink() implementation and many other _require rules, we try to actually do the action on $TEST_DIR, and check if command succeeds to see if the action is supported by current $FSTYP. Is it possible for exfat to do the same check?

Quoted 5 lines

We only use whitelist if it's impossible to do such check.

Thanks,
Eryu

Hi,

it does fail. It was actually my original intention to write the _require*() so it would check if the command succeeds as you are suggesting.

However, Eric and Dave were worried that adding more _require*() through the tests would lead to further slowing test execution. This worry actually makes sense to me.

Is there a significant benefit of testing the support vs. maintaining check based on FSTYP variable? I guess doing the check saves us from the need to update the code when new file-system is added, however actually doing the check increases time of test execution (but I haven't done any measurements yet - it's just my guess).

I really don't mind doing it either way and I'm happy to change the code - I'm just trying to explain :-)

Thanks for the comment.

Have a nice day.

#6 lore raw

Re: [PATCH v2 1/4] common/rc: Add _require_{chown,chmod,symlink}()

Quoted 7 lines · Pavel Reichl
Quoted 5 lines

We only use whitelist if it's impossible to do such check.

Thanks,
Eryu

Hi,

Quoted 7 lines

it does fail. It was actually my original intention to write the _require*() so it would check if the command succeeds as you are suggesting.

However, Eric and Dave were worried that adding more _require*() through the tests would lead to further slowing test execution. This worry actually makes sense to me.

Is there a significant benefit of testing the support vs. maintaining check based on FSTYP variable? I guess doing the check saves us from the need to update the code when new file-system is added, however actually doing the check increases time of test execution (but I haven't done any measurements yet - it's just my guess).

I really don't mind doing it either way and I'm happy to change the code - I'm just trying to explain :-)

that's my fault, sorry. Dave had expressed some concern about exfat changes slowing down testing for every other fs, and we talked about a whitelist. But now that I think about it, I'm not sure this functiona test takes any real extra time.

Pavel, maybe you can just evaluate whether there really is any significant time difference, and if there is not, go back to the functional test.

Sorry for the hassle.

-Eric

#7 lore raw

PATCH v2 2/4 common: hide permision warning from mkswap for exfat

In reply to Pavel Reichl

exfat does not support posix file permisions, so warning from mkswap is inavitable. This patch hides the warning message so the test won't fail.

---
 common/rc | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
common/rc +7 -1
index 9cdfe21c..2d658711 100644
@@ -2392,7 +2392,13 @@ _format_swapfile() {
 	# Swap files must be nocow on Btrfs.
 	$CHATTR_PROG +C "$fname" > /dev/null 2>&1
 	_pwrite_byte 0x61 0 "$sz" "$fname" >> $seqres.full
-	$MKSWAP_PROG "$fname" >> $seqres.full
+	if [ "$FSTYP" = "exfat" ]; then
+		# exfat does not support posix file permisions, so warning is
+		# to be expected
+		$MKSWAP_PROG "$fname" 2>&1 | grep -v 'insecure permission' >> $seqres.full
+	else
+		$MKSWAP_PROG "$fname" >> $seqres.full
+	fi
 }
 
 # Check that the filesystem supports swapfiles
Signature · 1 line
2.30.2
#8 lore raw

Re: [PATCH v2 2/4] common: hide permision warning from mkswap for exfat

Quoted 4 lines · Pavel Reichl

exfat does not support posix file permisions, so warning from mkswap is inavitable. This patch hides the warning message so the test won't fail.

hah, I never even thought about swap on exfat ;)

Assuming it actually works other than this warning, maybe you don't need to special case it for exfat? If so just filter that warning for everyone (in the same way that the CHATTR isn't special cased for btrfs; we just ignore failures there, because they don't matter).

If there's a real need to special case exfat that's fine, but it might be better to not litter special cases around if they aren't material to the test.

also, maybe there is some risk to 2>&1 >> seqres "eating" any other potential errors? Or maybe my bash fu is weak. ;)

I /think/ this will do the trick but give it a shot. It should filter only the one string from stderr, and send all of stdout to $seqres.full

# Ignore permission complaints on filesystems that don't support perms $MKSWAP_PROG "$fname" 2> >(grep -v 'insecure permission' >&2) >> $seqres.full

-Eric

Quoted 24 lines
---
 common/rc | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
common/rc +7 -1
index 9cdfe21c..2d658711 100644
@@ -2392,7 +2392,13 @@ _format_swapfile() {
 	# Swap files must be nocow on Btrfs.
 	$CHATTR_PROG +C "$fname" > /dev/null 2>&1
 	_pwrite_byte 0x61 0 "$sz" "$fname" >> $seqres.full
-	$MKSWAP_PROG "$fname" >> $seqres.full
+	if [ "$FSTYP" = "exfat" ]; then
+		# exfat does not support posix file permisions, so warning is
+		# to be expected
+		$MKSWAP_PROG "$fname" 2>&1 | grep -v 'insecure permission' >> $seqres.full
+	else
+		$MKSWAP_PROG "$fname" >> $seqres.full
+	fi
 }
 
 # Check that the filesystem supports swapfiles
#9 lore raw

Re: [PATCH v2 2/4] common: hide permision warning from mkswap for exfat

Quoted 17 lines · Eric Sandeen
Quoted 4 lines · Pavel Reichl

exfat does not support posix file permisions, so warning from mkswap is inavitable. This patch hides the warning message so the test won't fail.

hah, I never even thought about swap on exfat ;)

Assuming it actually works other than this warning, maybe you don't need to special case it for exfat? If so just filter that warning for everyone (in the same way that the CHATTR isn't special cased for btrfs; we just ignore failures there, because they don't matter).

If there's a real need to special case exfat that's fine, but it might be better to not litter special cases around if they aren't material to the test.

Quoted 8 lines

also, maybe there is some risk to 2>&1 >> seqres "eating" any other potential errors? Or maybe my bash fu is weak. ;)

I /think/ this will do the trick but give it a shot. It should filter only the one string from stderr, and send all of stdout to $seqres.full

# Ignore permission complaints on filesystems that don't support perms $MKSWAP_PROG "$fname" 2> >(grep -v 'insecure permission' >&2) >> $seqres.full

Works nice, thanks for the hint.

I also applied these comments on patch #3.

Quoted 28 lines

-Eric

Quoted 24 lines
---
 common/rc | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
common/rc +7 -1
index 9cdfe21c..2d658711 100644
@@ -2392,7 +2392,13 @@ _format_swapfile() {
 	# Swap files must be nocow on Btrfs.
 	$CHATTR_PROG +C "$fname" > /dev/null 2>&1
 	_pwrite_byte 0x61 0 "$sz" "$fname" >> $seqres.full
-	$MKSWAP_PROG "$fname" >> $seqres.full
+	if [ "$FSTYP" = "exfat" ]; then
+		# exfat does not support posix file permisions, so warning is
+		# to be expected
+		$MKSWAP_PROG "$fname" 2>&1 | grep -v 'insecure permission' >> $seqres.full
+	else
+		$MKSWAP_PROG "$fname" >> $seqres.full
+	fi
 }
 
 # Check that the filesystem supports swapfiles
#10 lore raw

PATCH v2 3/4 generic/554: hide permision warning on exfat

In reply to Pavel Reichl
---
 common/rc         | 13 +++++++++++++
 tests/generic/554 |  3 ++-
 2 files changed, 15 insertions(+), 1 deletion(-)
common/rc +13 -0
index 2d658711..379140ea 100644
@@ -2401,6 +2401,19 @@ _format_swapfile() {
 	fi
 }
 
+_format_swapon_file() {
+	local fname="$1"
+
+	if [ "$FSTYP" = "exfat" ]; then
+		# exfat does not support posix file permisions, so warning is
+		# to be expected
+		swapon "$fname" |& grep -v "insecure permissions"
+	else
+		swapon "$fname"
+	fi
+
+}
+
 # Check that the filesystem supports swapfiles
 _require_scratch_swapfile()
 {
tests/generic/554 +2 -1
index fa4f97d2..c2f9eee8 100755
@@ -46,7 +46,8 @@ $XFS_IO_PROG -f -c "pwrite -S 0x61 0 128k" $SCRATCH_MNT/file >> $seqres.full 2>&
 
 echo swap files return ETXTBUSY
 _format_swapfile $SCRATCH_MNT/swapfile 16m
-swapon $SCRATCH_MNT/swapfile
+_format_swapon_file $SCRATCH_MNT/swapfile
+
 $XFS_IO_PROG -f -c "copy_range -l 32k $SCRATCH_MNT/file" $SCRATCH_MNT/swapfile
 swapoff $SCRATCH_MNT/swapfile
 
Signature · 1 line
2.30.2
#11 lore raw

Re: [PATCH v2 3/4] generic/554: hide permision warning on exfat

Quoted 11 lines · Pavel Reichl
---
 common/rc         | 13 +++++++++++++
 tests/generic/554 |  3 ++-
 2 files changed, 15 insertions(+), 1 deletion(-)
common/rc +0 -0
index 2d658711..379140ea 100644

common/rc _format_swapfile() { · line 2404

Quoted 4 lines
common/rc +1 -0
@@ -2401,6 +2401,19 @@ _format_swapfile() {
 	fi
 }
 
+_format_swapon_file() {

It's not a "format" action anymore, the _format prefix seems misleading. I think _swapon_file seems fine.

Thanks,
Eryu

Quoted 31 lines
common/rc +12 -0
+	local fname="$1"
+
+	if [ "$FSTYP" = "exfat" ]; then
+		# exfat does not support posix file permisions, so warning is
+		# to be expected
+		swapon "$fname" |& grep -v "insecure permissions"
+	else
+		swapon "$fname"
+	fi
+
+}
+
 # Check that the filesystem supports swapfiles
 _require_scratch_swapfile()
 {
tests/generic/554 +2 -1
index fa4f97d2..c2f9eee8 100755
@@ -46,7 +46,8 @@ $XFS_IO_PROG -f -c "pwrite -S 0x61 0 128k" $SCRATCH_MNT/file >> $seqres.full 2>&
 
 echo swap files return ETXTBUSY
 _format_swapfile $SCRATCH_MNT/swapfile 16m
-swapon $SCRATCH_MNT/swapfile
+_format_swapon_file $SCRATCH_MNT/swapfile
+
 $XFS_IO_PROG -f -c "copy_range -l 32k $SCRATCH_MNT/file" $SCRATCH_MNT/swapfile
 swapoff $SCRATCH_MNT/swapfile
 
Signature · 1 line
2.30.2
#12 lore raw

PATCH v2 4/4 generic/003: Amend the test for exfat

In reply to Pavel Reichl

Update the test so it can be run even for exfat which has 2 seconds granularity for access_time and does not have a timestamp for metadata change.

---
 tests/generic/003 | 29 ++++++++++++++++++++---------
 1 file changed, 20 insertions(+), 9 deletions(-)
tests/generic/003 +20 -9
index ec4fdfc1..39596665 100755
@@ -37,6 +37,13 @@ _require_relatime
 
 rm -f $seqres.full
 
+if [ "$FSTYP" = "exfat" ]; then
+	# exfat's timestamp for access_time has double seconds granularity
+	access_delay=2
+else
+	access_delay=1
+fi
+
 _stat() {
 	stat -c "%x;%y;%z" $1
 }
@@ -79,14 +86,14 @@ echo "aaa" > $TPATH/dir1/file1
 file1_stat_before_first_access=`_stat $TPATH/dir1/file1`
 
 # Accessing file1 the first time
-sleep 1
+sleep $access_delay
 cat $TPATH/dir1/file1 > /dev/null
 file1_stat_after_first_access=`_stat $TPATH/dir1/file1`
 _compare_stat_times YNN "$file1_stat_before_first_access" \
 	"$file1_stat_after_first_access" "after accessing file1 first time"
 
 # Accessing file1 a second time
-sleep 1
+sleep $access_delay
 cat $TPATH/dir1/file1 > /dev/null
 file1_stat_after_second_access=`_stat $TPATH/dir1/file1`
 _compare_stat_times NNN "$file1_stat_after_first_access" \
@@ -109,7 +116,7 @@ _compare_stat_times NYY "$dir2_stat_before_file_creation" \
 
 # Accessing file2
 file2_stat_before_first_access=`_stat $TPATH/dir2/file2`
-sleep 1
+sleep $access_delay
 cat $TPATH/dir2/file2 > /dev/null
 file2_stat_after_first_access=`_stat $TPATH/dir2/file2`
 _compare_stat_times YNN "$file2_stat_before_first_access" \
@@ -135,11 +142,15 @@ echo "xyz" > $TPATH/dir1/file1
 file1_stat_after_modify=`_stat $TPATH/dir1/file1`
 _compare_stat_times NYY "$file1_stat_before_modify" \
 	"$file1_stat_after_modify" "after modifying file1"
-sleep 1
-mv $TPATH/dir1/file1 $TPATH/dir1/file1_renamed
-file1_stat_after_change=`_stat $TPATH/dir1/file1_renamed`
-_compare_stat_times NNY "$file1_stat_after_modify" \
-	"$file1_stat_after_change" "after changing file1"
+
+# exfat does not support last metadata change timestamp
+if [ "$FSTYP" != "exfat" ]; then
+	sleep 1
+	mv $TPATH/dir1/file1 $TPATH/dir1/file1_renamed
+	file1_stat_after_change=`_stat $TPATH/dir1/file1_renamed`
+	_compare_stat_times NNY "$file1_stat_after_modify" \
+		"$file1_stat_after_change" "after changing file1"
+fi
 
 # Mounting with strictatime option and
 # accessing a previously created file twice
@@ -148,7 +159,7 @@ cat $TPATH/dir2/file3 > /dev/null
 file3_stat_after_second_access=`_stat $TPATH/dir2/file3`
 _compare_stat_times YNN "$file3_stat_after_first_access" \
 	"$file3_stat_after_second_access" "after accessing file3 second time"
-sleep 1
+sleep $access_delay
 cat $TPATH/dir2/file3 > /dev/null
 file3_stat_after_third_access=`_stat $TPATH/dir2/file3`
 _compare_stat_times YNN "$file3_stat_after_second_access" \
Signature · 1 line
2.30.2
#13 lore raw

Re: [PATCH v2 4/4] generic/003: Amend the test for exfat

Quoted 5 lines · Pavel Reichl

Update the test so it can be run even for exfat which has 2 seconds granularity for access_time and does not have a timestamp for metadata change.

This looks a whole lot like what I was playing with earlier. :)

Can you run it say 20 times in a row, and make sure that you don't get spurious failures by sleeping for /exactly/ the atime granularity?

Thanks,
-Eric

#14 lore raw

Re: [PATCH v2 4/4] generic/003: Amend the test for exfat

Quoted 8 lines · Eric Sandeen
Quoted 5 lines · Pavel Reichl

Update the test so it can be run even for exfat which has 2 seconds granularity for access_time and does not have a timestamp for metadata change.

This looks a whole lot like what I was playing with earlier. :)

good :-)

Quoted 3 lines

Can you run it say 20 times in a row, and make sure that you don't get spurious failures by sleeping for /exactly/ the atime granularity?

I tried that for tens of times and no failure. However, since sleep() does not require integer as the parameter, in the next revision I'll propose something as a 2.1 instead of the 2 to be on the safe side, OK?

Quoted 4 lines

Thanks,
-Eric