Re: [PATCH] common/rc: check 'chattr +/-x' on dax device.

[Shiyang Ruan <ruansy.fnst@cn.fujitsu.com>]

On 7/30/19 10:47 PM, Darrick J. Wong wrote:
> On Tue, Jul 30, 2019 at 04:40:09PM +0800, Shiyang Ruan wrote:
>> 'chattr +/-x' only works on a dax device.  When checking if the 'x'
>> attribute is supported by XFS_IO_PROG:
>>      _require_xfs_io_command "chattr" "x"    (called by xfs/260)
>> it's better to do the check on a dax device mounted with dax option.
>>
>> Signed-off-by: Shiyang Ruan <ruansy.fnst@cn.fujitsu.com>
>> ---
>>   common/rc | 11 +++++++++++
>>   1 file changed, 11 insertions(+)
>>
>> diff --git a/common/rc b/common/rc
>> index e0b087c1..73ee5563 100644
>> --- a/common/rc
>> +++ b/common/rc
>> @@ -2094,11 +2094,22 @@ _require_xfs_io_command()
>>   		if [ -z "$param" ]; then
>>   			param=s
>>   		fi
>> +
>> +		# Attribute "x" should be tested on a dax device
>> +		if [ "$param" == "x" ]; then
>> +			_scratch_mount "-o dax"
>> +			testfile=$SCRATCH_MNT/$$.xfs_io
> 
> NAK, the dax mount option is not intended to remain as a long-term
> interface.
> 
> Also, "==" is a bashism (which probably is fine for fstests)
> 
> Also, there's no _require_scratch which means this can totally blow up
> if the user doesn't specify a scratch device.

Yes, and _require_scratch_dax seems to be a better choice.

And I found that the "chattr +/-x" check could also pass on a dax device 
mounted without "-o dax".


-- 
Thanks,
Shiyang Ruan.
> 
> --D
> 
>> +		fi
>> +
>>   		# Test xfs_io chattr support AND
>>   		# filesystem FS_IOC_FSSETXATTR support
>>   		testio=`$XFS_IO_PROG -F -f -c "chattr +$param" $testfile 2>&1`
>>   		$XFS_IO_PROG -F -f -r -c "chattr -$param" $testfile 2>&1
>>   		param_checked="+$param"
>> +
>> +		if [ "$param" == "x" ]; then
>> +			_scratch_unmount
>> +		fi
>>   		;;
>>   	"chproj")
>>   		testio=`$XFS_IO_PROG -F -f -c "chproj 0" $testfile 2>&1`
>> -- 
>> 2.17.0
>>
>>
>>
> 
>