[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Sedat Dilek]

Hi,

I wanted to test the "Xattr inode operation removal" patchset from
Andreas Gruenbacher.

So LTP has a testcase called "acl_test01" which I tried for my testing.

In LTP v20160126 and v20160510 I see...

root# cd /opt/ltp ; LC_ALL=C ./runltp -s acl_test01 | tee acl_test01.log

...
/opt/ltp/testcases/bin/acl_test01: 44: .: test.sh: not found
...

...and...

/dev/loop2 on /tmp/ltp-6pUha1I7Ey/tacl/mount-ext3 type ext3
(rw,relatime,data=ordered)
loop: can't delete device /dev/loop2: No such device or address
loop: can't delete device /dev/loop2: No such device or address
...

The 1st issue I tried to hack by...

# diff -uprN testcases/bin/acl_test01.orig testcases/bin/acl_test01
--- testcases/bin/acl_test01.orig       2016-05-30 15:11:10.771931080 +0200
+++ testcases/bin/acl_test01    2016-05-30 15:22:31.627909726 +0200
@@ -41,8 +41,6 @@ export TCID=acltest01
 export TST_TOTAL=5
 export TST_COUNT=1

-. test.sh
-
 TMP=${TMP:=/tmp}

 TEST_USER1="acltest1"
@@ -54,6 +52,8 @@ FILE_ACL="$TMP/tacl/mount-ext3/test_file
 FILE_ACL_LINK="$TMP/tacl/mount-ext3/test_file_link"
 TCbin=`pwd`

+. $TCbin/test.sh
+
 COMMAND=$(command -v "getenforce" "setenforce" |wc -l)
 if [ $COMMAND -eq 2 ]; then
        SELINUX=$(getenforce)

Not sure if this correct.

For the 2nd issue - dunno what's going on.

Beyond all this, how can I test the work of Andreas with LTP?
Any help appreciated.

Regards,
- Sedat -

[1] http://marc.info/?t=146459873600015&r=1&w=2
[2] http://marc.info/?l=linux-fsdevel&m=146461299113521&w=2
-------------- next part --------------
A non-text attachment was scrubbed...
Name: acl_test01.log
Type: application/octet-stream
Size: 8956 bytes
Desc: not available
URL: <http://lists.linux.it/pipermail/ltp/attachments/20160530/1008f519/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: LTP_RUN_ON-2016_05_30-15h_19m_49s.log
Type: application/octet-stream
Size: 430 bytes
Desc: not available
URL: <http://lists.linux.it/pipermail/ltp/attachments/20160530/1008f519/attachment-0003.obj>

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Cyril Hrubis]

Hi!
> The 1st issue I tried to hack by...
> 
> # diff -uprN testcases/bin/acl_test01.orig testcases/bin/acl_test01
> --- testcases/bin/acl_test01.orig       2016-05-30 15:11:10.771931080 +0200
> +++ testcases/bin/acl_test01    2016-05-30 15:22:31.627909726 +0200
> @@ -41,8 +41,6 @@ export TCID=acltest01
>  export TST_TOTAL=5
>  export TST_COUNT=1
> 
> -. test.sh
> -
>  TMP=${TMP:=/tmp}
> 
>  TEST_USER1="acltest1"
> @@ -54,6 +52,8 @@ FILE_ACL="$TMP/tacl/mount-ext3/test_file
>  FILE_ACL_LINK="$TMP/tacl/mount-ext3/test_file_link"
>  TCbin=`pwd`
> 
> +. $TCbin/test.sh
> +
>  COMMAND=$(command -v "getenforce" "setenforce" |wc -l)
>  if [ $COMMAND -eq 2 ]; then
>         SELINUX=$(getenforce)
> 
> Not sure if this correct.

No it's not. You have to have path to testcases/bin/ directory in $PATH
in order to run a LTP testcase (and it's even written in
documentation...).

Strangely enough the $PATH should be set correctly when you run
testcases via runltp. That is one of the first things done in setup() in
that script. And it works fine for me.

The TCbin should be removed from the test entirely.

> For the 2nd issue - dunno what's going on.

Looks like the do_cleanup is executed twice. Once via trap do_cleanup
EXIT and once at the end of the script. This should be fixed.

> Beyond all this, how can I test the work of Andreas with LTP?
> Any help appreciated.

I guess that kernel/syscalls/*xattr* should be relevant.

-- 
Cyril Hrubis
chrubis@suse.cz

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Sedat Dilek]

On 5/30/16, Cyril Hrubis <chrubis@suse.cz> wrote:
> Hi!
>> The 1st issue I tried to hack by...
>>
>> # diff -uprN testcases/bin/acl_test01.orig testcases/bin/acl_test01
>> --- testcases/bin/acl_test01.orig       2016-05-30 15:11:10.771931080
>> +0200
>> +++ testcases/bin/acl_test01    2016-05-30 15:22:31.627909726 +0200
>> @@ -41,8 +41,6 @@ export TCID=acltest01
>>  export TST_TOTAL=5
>>  export TST_COUNT=1
>>
>> -. test.sh
>> -
>>  TMP=${TMP:=/tmp}
>>
>>  TEST_USER1="acltest1"
>> @@ -54,6 +52,8 @@ FILE_ACL="$TMP/tacl/mount-ext3/test_file
>>  FILE_ACL_LINK="$TMP/tacl/mount-ext3/test_file_link"
>>  TCbin=`pwd`
>>
>> +. $TCbin/test.sh
>> +
>>  COMMAND=$(command -v "getenforce" "setenforce" |wc -l)
>>  if [ $COMMAND -eq 2 ]; then
>>         SELINUX=$(getenforce)
>>
>> Not sure if this correct.
>
> No it's not. You have to have path to testcases/bin/ directory in $PATH
> in order to run a LTP testcase (and it's even written in
> documentation...).
>
> Strangely enough the $PATH should be set correctly when you run
> testcases via runltp. That is one of the first things done in setup() in
> that script. And it works fine for me.
>
> The TCbin should be removed from the test entirely.
>

Hi,

Sorry, I jumped to v20160510 w/o reading the new docs.

I am bit confused.
Add /path/to/testcases/bin into $PATH or should it work without this?
OOTB it does not work here w/o the hackery.

>> For the 2nd issue - dunno what's going on.
>
> Looks like the do_cleanup is executed twice. Once via trap do_cleanup
> EXIT and once at the end of the script. This should be fixed.
>

As said I use now v20160510.
Was this fixed post-v20160510?
Can you give me a hand - which commit/patch do you mean?

>> Beyond all this, how can I test the work of Andreas with LTP?
>> Any help appreciated.
>
> I guess that kernel/syscalls/*xattr* should be relevant.
>

I am only a tester - not more not less.

- Sedat -

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Cyril Hrubis]

Hi!
> I am bit confused.
> Add /path/to/testcases/bin into $PATH or should it work without this?

If you run a testcase by hand you have to add it to $PATH. If you run a
testcase via runltp script the script does that for you. And it has been
so for last few years.

> OOTB it does not work here w/o the hackery.

Even if you start the test via runltp?

I've tried that and this works fine for me.

> >> For the 2nd issue - dunno what's going on.
> >
> > Looks like the do_cleanup is executed twice. Once via trap do_cleanup
> > EXIT and once at the end of the script. This should be fixed.
> >
> 
> As said I use now v20160510.
> Was this fixed post-v20160510?
> Can you give me a hand - which commit/patch do you mean?

Should be fixed -> I'm about to push a fix, stay tuned.

The cleanup has been executefd twice at least since 2010 and possibly
since the testcase was introduced.

> >> Beyond all this, how can I test the work of Andreas with LTP?
> >> Any help appreciated.
> >
> > I guess that kernel/syscalls/*xattr* should be relevant.
> >
> 
> I am only a tester - not more not less.

And I am only a test maintainer.

But seriously I've glanced over the patches and it seems to implement
{get,set,remove}xattr via generic handlers instead of fs specific
handlers. And we have syscall tests for {get,set,remove}xattr...

-- 
Cyril Hrubis
chrubis@suse.cz

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Sedat Dilek]

On 5/30/16, Cyril Hrubis <chrubis@suse.cz> wrote:
> Hi!
>> I am bit confused.
>> Add /path/to/testcases/bin into $PATH or should it work without this?
>
> If you run a testcase by hand you have to add it to $PATH. If you run a
> testcase via runltp script the script does that for you. And it has been
> so for last few years.
>
>> OOTB it does not work here w/o the hackery.
>
> Even if you start the test via runltp?
>
> I've tried that and this works fine for me.
>

Hmmm.

root# echo $PATH
/opt/cmake/bin:/opt/xorg/bin:/opt/llvm/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin

root# cd /opt/ltp ; LC_ALL=C ./runltp -s acl_test01 | tee acl_test01.log

[...]
TESTING ACLS FOR USER root
Trying extended acls for files
acltest01 1 TPASS : Extended acls for files.
Trying extended acls for file links
acltest01 2 TPASS : Extended acls for links.
/opt/ltp/testcases/bin/acl_test01: 44: .: test.sh: not found

/dev/loop2 on /tmp/ltp-y9NKWCxzX3/tacl/mount-ext3 type ext3
(rw,relatime,data=ordered)
loop: can't delete device /dev/loop2: No such device or address
loop: can't delete device /dev/loop2: No such device or address
incrementing stop
...

>> >> For the 2nd issue - dunno what's going on.
>> >
>> > Looks like the do_cleanup is executed twice. Once via trap do_cleanup
>> > EXIT and once at the end of the script. This should be fixed.
>> >
>>
>> As said I use now v20160510.
>> Was this fixed post-v20160510?
>> Can you give me a hand - which commit/patch do you mean?
>
> Should be fixed -> I'm about to push a fix, stay tuned.
>
> The cleanup has been executefd twice at least since 2010 and possibly
> since the testcase was introduced.
>

OK, I am patient.
Thanks for the explanations.

>> >> Beyond all this, how can I test the work of Andreas with LTP?
>> >> Any help appreciated.
>> >
>> > I guess that kernel/syscalls/*xattr* should be relevant.
>> >
>>
>> I am only a tester - not more not less.
>
> And I am only a test maintainer.
>
> But seriously I've glanced over the patches and it seems to implement
> {get,set,remove}xattr via generic handlers instead of fs specific
> handlers. And we have syscall tests for {get,set,remove}xattr...
>

Is there a testcase around - runnable via "runltp -s $testcase_script".

Thanks in advance.

- Sedat -

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Cyril Hrubis]

Hi!
> root# echo $PATH
> /opt/cmake/bin:/opt/xorg/bin:/opt/llvm/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin

The path is added to the $PATH at the start of the runltp script, if you
want to know if it was added correctly you have to do echo $PATH at the
end of the setup() in runltp script.

> root# cd /opt/ltp ; LC_ALL=C ./runltp -s acl_test01 | tee acl_test01.log
> 
> [...]
> TESTING ACLS FOR USER root
> Trying extended acls for files
> acltest01 1 TPASS : Extended acls for files.
> Trying extended acls for file links
> acltest01 2 TPASS : Extended acls for links.
> /opt/ltp/testcases/bin/acl_test01: 44: .: test.sh: not found

This is certainly unexpected.

What is /bin/sh pointing to? Is this debian with dash?

Still this should really work.

> > Should be fixed -> I'm about to push a fix, stay tuned.
> >
> > The cleanup has been executefd twice at least since 2010 and possibly
> > since the testcase was introduced.
> >
> 
> OK, I am patient.
> Thanks for the explanations.

Fixed in:

https://github.com/linux-test-project/ltp/commit/ee2d3ee269264198837726304873195134a4232c

> > And I am only a test maintainer.
> >
> > But seriously I've glanced over the patches and it seems to implement
> > {get,set,remove}xattr via generic handlers instead of fs specific
> > handlers. And we have syscall tests for {get,set,remove}xattr...
> >
> 
> Is there a testcase around - runnable via "runltp -s $testcase_script".

These are executed as a part of syscall tests.

You can simply do ./runltp -s xattr to execute all testcases that have
xattr in their name.

-- 
Cyril Hrubis
chrubis@suse.cz

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Sedat Dilek]

On 5/30/16, Cyril Hrubis <chrubis@suse.cz> wrote:
> Hi!
>> root# echo $PATH
>> /opt/cmake/bin:/opt/xorg/bin:/opt/llvm/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
>
> The path is added to the $PATH at the start of the runltp script, if you
> want to know if it was added correctly you have to do echo $PATH at the
> end of the setup() in runltp script.
>
>> root# cd /opt/ltp ; LC_ALL=C ./runltp -s acl_test01 | tee acl_test01.log
>>
>> [...]
>> TESTING ACLS FOR USER root
>> Trying extended acls for files
>> acltest01 1 TPASS : Extended acls for files.
>> Trying extended acls for file links
>> acltest01 2 TPASS : Extended acls for links.
>> /opt/ltp/testcases/bin/acl_test01: 44: .: test.sh: not found
>
> This is certainly unexpected.
>
> What is /bin/sh pointing to? Is this debian with dash?
>
> Still this should really work.
>

Hi,

# echo $PATH
/opt/cmake/bin:/opt/xorg/bin:/opt/llvm/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin

# cd /opt/ltp ; LC_ALL=C ./runltp -s acl_test01 | tee acl_test01.log
...
TESTING ACLS FOR USER root
Trying extended acls for files
acltest01 1 TPASS : Extended acls for files.
Trying extended acls for file links
acltest01 2 TPASS : Extended acls for links.
/opt/ltp/testcases/bin/acl_test01: 44: .: test.sh: not found

loop: can't delete device /dev/loop2: No such device or address
incrementing stop
<<<execution_status>>>
...

# echo $PATH
/opt/cmake/bin:/opt/xorg/bin:/opt/llvm/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin

No change in $PATH after executing runltp.
The loop-thingie fails once now.

YES, dash is the default-shell on Ubuntu/precise.

# LC_ALL=C ls -l /bin/sh
lrwxrwxrwx 1 root root 4 Mar 29  2012 /bin/sh -> dash

>> > Should be fixed -> I'm about to push a fix, stay tuned.
>> >
>> > The cleanup has been executefd twice at least since 2010 and possibly
>> > since the testcase was introduced.
>> >
>>
>> OK, I am patient.
>> Thanks for the explanations.
>
> Fixed in:
>
> https://github.com/linux-test-project/ltp/commit/ee2d3ee269264198837726304873195134a4232c
>

Yupp, I tested with this new version of acl_test01.

>> > And I am only a test maintainer.
>> >
>> > But seriously I've glanced over the patches and it seems to implement
>> > {get,set,remove}xattr via generic handlers instead of fs specific
>> > handlers. And we have syscall tests for {get,set,remove}xattr...
>> >
>>
>> Is there a testcase around - runnable via "runltp -s $testcase_script".
>
> These are executed as a part of syscall tests.
>
> You can simply do ./runltp -s xattr to execute all testcases that have
> xattr in their name.
>

Thanks.

This looks good - only the cgroup_xattr FAILS.
Andreas might have a look.

# cat results/LTP_RUN_ON-2016_05_30-17h_49m_41s.log
Test Start Time: Mon May 30 17:49:42 2016
-----------------------------------------
Testcase                       Result     Exit Value
--------                       ------     ----------
getxattr01                     PASS       0
getxattr02                     PASS       0
getxattr03                     PASS       0
llistxattr01                   PASS       0
llistxattr02                   PASS       0
llistxattr03                   PASS       0
removexattr01                  PASS       0
removexattr02                  PASS       0
setxattr01                     PASS       0
setxattr02                     PASS       0
setxattr03                     PASS       0
cgroup_xattr                   FAIL       1

-----------------------------------------------
Total Tests: 12
Total Skipped Tests: 0
Total Failures: 1
Kernel Version: 4.7.0-rc1-2-iniza-small
Machine Architecture: x86_64
Hostname: fambox

- Sedat -
-------------- next part --------------
A non-text attachment was scrubbed...
Name: LTP_RUN_ON-2016_05_30-17h_49m_41s.log
Type: application/octet-stream
Size: 959 bytes
Desc: not available
URL: <http://lists.linux.it/pipermail/ltp/attachments/20160530/bf3d9ddb/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: xattr.log
Type: application/octet-stream
Size: 205062 bytes
Desc: not available
URL: <http://lists.linux.it/pipermail/ltp/attachments/20160530/bf3d9ddb/attachment-0003.obj>

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Cyril Hrubis]

Hi!
> # echo $PATH
> /opt/cmake/bin:/opt/xorg/bin:/opt/llvm/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
> 
> # cd /opt/ltp ; LC_ALL=C ./runltp -s acl_test01 | tee acl_test01.log
> ...
> TESTING ACLS FOR USER root
> Trying extended acls for files
> acltest01 1 TPASS : Extended acls for files.
> Trying extended acls for file links
> acltest01 2 TPASS : Extended acls for links.
> /opt/ltp/testcases/bin/acl_test01: 44: .: test.sh: not found
> 
> loop: can't delete device /dev/loop2: No such device or address
> incrementing stop
> <<<execution_status>>>
> ...
> 
> # echo $PATH
> /opt/cmake/bin:/opt/xorg/bin:/opt/llvm/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
> 
> No change in $PATH after executing runltp.

Once again the $PATH is set _ONLY_ inside of the runltp script and is
inherited for everything that is executed from it not the other way
around. You _HAVE_ to print the $PATH from the runltp script after it
was initialized.

And I've reproduced it on Debian 7.2, looks like the "su $TEST_USER1"
replaces $PATH completly on Debian while it seems to retain the added
path to /opt/ltp/testcases/bin on the rest of the systems I've tried.

-- 
Cyril Hrubis
chrubis@suse.cz

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Sedat Dilek]

[...]
>> Is there a testcase around - runnable via "runltp -s $testcase_script".
>
> These are executed as a part of syscall tests.
>
> You can simply do ./runltp -s xattr to execute all testcases that have
> xattr in their name.
>

Running xattr testcase has shown a different issue.
In [1] Andreas explains why cgroup_xattr tests fails...

"The test seems to try getting/setting an xattr named "trusted." (no
suffix). That's an invalid name; such invalid names are rejected since
commit 98e9cb571 "vfs: Distinguish between full xattr names and proper
prefixes" (v4.5-rc1).

Could you try removing that from the LTP test and rerun it?"

I followed the advice and the attached diff fixes the problem here.
Xattr testcase runs successfully,

See attached logs for details.

- Sedat -

[1] http://marc.info/?l=linux-fsdevel&m=146463254520809&w=2
-------------- next part --------------
--- ltp-full-20160510.orig/testcases/kernel/controllers/cgroup_xattr/cgroup_xattr.c
+++ ltp-full-20160510/testcases/kernel/controllers/cgroup_xattr/cgroup_xattr.c
@@ -66,7 +66,9 @@ struct tst_key {
 static struct tst_key tkeys[] = {
 	{ .name = "security.",		.good = 1,	},
 	{ .name = "trusted.test",	.good = 1,	},
+#if 0
 	{ .name = "trusted.",		.good = 1,	},
+#endif
 	{ .name = "user.",		.good = 0,	},
 	{ .name = "system.",		.good = 0,	},
 };
-------------- next part --------------
A non-text attachment was scrubbed...
Name: LTP_RUN_ON-2016_05_30-21h_15m_19s.log
Type: application/octet-stream
Size: 959 bytes
Desc: not available
URL: <http://lists.linux.it/pipermail/ltp/attachments/20160530/d5e0f9df/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: xattr.log
Type: application/octet-stream
Size: 163378 bytes
Desc: not available
URL: <http://lists.linux.it/pipermail/ltp/attachments/20160530/d5e0f9df/attachment-0003.obj>

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Sedat Dilek]

On 5/30/16, Cyril Hrubis <chrubis@suse.cz> wrote:
> Hi!
>> # echo $PATH
>> /opt/cmake/bin:/opt/xorg/bin:/opt/llvm/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
>>
>> # cd /opt/ltp ; LC_ALL=C ./runltp -s acl_test01 | tee acl_test01.log
>> ...
>> TESTING ACLS FOR USER root
>> Trying extended acls for files
>> acltest01 1 TPASS : Extended acls for files.
>> Trying extended acls for file links
>> acltest01 2 TPASS : Extended acls for links.
>> /opt/ltp/testcases/bin/acl_test01: 44: .: test.sh: not found
>>
>> loop: can't delete device /dev/loop2: No such device or address
>> incrementing stop
>> <<<execution_status>>>
>> ...
>>
>> # echo $PATH
>> /opt/cmake/bin:/opt/xorg/bin:/opt/llvm/bin:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
>>
>> No change in $PATH after executing runltp.
>
> Once again the $PATH is set _ONLY_ inside of the runltp script and is
> inherited for everything that is executed from it not the other way
> around. You _HAVE_ to print the $PATH from the runltp script after it
> was initialized.
>
> And I've reproduced it on Debian 7.2, looks like the "su $TEST_USER1"
> replaces $PATH completly on Debian while it seems to retain the added
> path to /opt/ltp/testcases/bin on the rest of the systems I've tried.
>

Thanks for testing with Debian.

- Sedat -

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Sedat Dilek]

On 5/30/16, Sedat Dilek <sedat.dilek@gmail.com> wrote:
> [...]
>>> Is there a testcase around - runnable via "runltp -s $testcase_script".
>>
>> These are executed as a part of syscall tests.
>>
>> You can simply do ./runltp -s xattr to execute all testcases that have
>> xattr in their name.
>>
>
> Running xattr testcase has shown a different issue.
> In [1] Andreas explains why cgroup_xattr tests fails...
>
> "The test seems to try getting/setting an xattr named "trusted." (no
> suffix). That's an invalid name; such invalid names are rejected since
> commit 98e9cb571 "vfs: Distinguish between full xattr names and proper
> prefixes" (v4.5-rc1).
>
> Could you try removing that from the LTP test and rerun it?"
>
> I followed the advice and the attached diff fixes the problem here.
> Xattr testcase runs successfully,
>
> See attached logs for details.
>
> - Sedat -
>
> [1] http://marc.info/?l=linux-fsdevel&m=146463254520809&w=2
>

Andreas send a patch [1] which works fine.

- Sedat -

[1] http://marc.info/?l=linux-fsdevel&m=146463658521851&w=2

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Cyril Hrubis]

Hi!
> > Once again the $PATH is set _ONLY_ inside of the runltp script and is
> > inherited for everything that is executed from it not the other way
> > around. You _HAVE_ to print the $PATH from the runltp script after it
> > was initialized.
> >
> > And I've reproduced it on Debian 7.2, looks like the "su $TEST_USER1"
> > replaces $PATH completly on Debian while it seems to retain the added
> > path to /opt/ltp/testcases/bin on the rest of the systems I've tried.
> >
> 
> Thanks for testing with Debian.

I've fixed the problems in three patches.

The message about loop device that could not be deleted turned out to be
harmless, the cleanup attempted to detach the device twice. Anyway the
test should be all fixed now.

-- 
Cyril Hrubis
chrubis@suse.cz

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Cyril Hrubis]

Hi!
> Andreas send a patch [1] which works fine.
> 
> - Sedat -
> 
> [1] http://marc.info/?l=linux-fsdevel&m=146463658521851&w=2

Andreas, please at least CC ltp ML (ltp@lists.linux.it) with patches for
LTP testcases, otherwise I cannot apply them.

And if you think that testcase is useless or plain wrong just provide
reasoning why and I will get rid of it.

-- 
Cyril Hrubis
chrubis@suse.cz

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Andreas Gruenbacher]

Cyril,

On Tue, May 31, 2016 at 5:21 PM, Cyril Hrubis <chrubis@suse.cz> wrote:
> Andreas, please at least CC ltp ML (ltp@lists.linux.it) with patches for
> LTP testcases, otherwise I cannot apply them.

here's the fix again; it at least makes the test work on 4.5 and later kernels.

> And if you think that testcase is useless or plain wrong just provide
> reasoning why and I will get rid of it.

The test seems rather weird. Tu check for (the lack of) "user.*"
support, it should at least use a "user.*" xattr name that would be
valid on ther filesystems (any non-empty suffix would do). Likewise
for system; it could check for "system.posix_acl_access" support, for
example, but only checking if invalid names are properly rejected
isn't all that helpful. I don't know what the real idea behind this
test case was though; the documentation doesn't say much.

Thanks,
Andreas
-------------- next part --------------
A non-text attachment was scrubbed...
Name: ltp-xattr.patch
Type: text/x-patch
Size: 1629 bytes
Desc: not available
URL: <http://lists.linux.it/pipermail/ltp/attachments/20160601/1932a180/attachment.bin>

[LTP] [LTP-20160510] acl_test01: 44: .: test.sh: not found

[Cyril Hrubis]

Hi!
> > Andreas, please at least CC ltp ML (ltp@lists.linux.it) with patches for
> > LTP testcases, otherwise I cannot apply them.
> 
> here's the fix again; it at least makes the test work on 4.5 and later kernels.

Applied with your signed-of-by.

> > And if you think that testcase is useless or plain wrong just provide
> > reasoning why and I will get rid of it.
> 
> The test seems rather weird. Tu check for (the lack of) "user.*"
> support, it should at least use a "user.*" xattr name that would be
> valid on ther filesystems (any non-empty suffix would do). Likewise
> for system; it could check for "system.posix_acl_access" support, for
> example, but only checking if invalid names are properly rejected
> isn't all that helpful. I don't know what the real idea behind this
> test case was though; the documentation doesn't say much.

Sounds reasonably to me. Feel free to submit a patch that would change
the attrbutes to a valid values.

Also CCing Alexey who seems to be the original author of the test (at
least git log says so).

-- 
Cyril Hrubis
chrubis@suse.cz