[LTP] [PATCH] statx04: Re-add BTRFS version check

[Richard Palethorpe]

Hello,

Cyril Hrubis <chrubis@suse.cz> writes:

> Hi!
>> Do you mean that your distribution based on older kernel ie 4.11
>> supports statx syscall but btrfs missed the btrfs patch? Also this
>> distribution doesn't update and so have no choice to backport.
>> > This is different from the
>> > other version checks which are for much newer kernels. 
>> IMO, distribution based on older kernel 4.11 still can make ext2 ext4
>> xfs supports statx because the backport looks not diffcult. So, I don't
>> think this is a difference. It depends on kernel users worked on this
>> distirbution whether have this requirement.
>> Also there could be differences in the difficulty of a backport.
>> I see xfs/btrfs code, it only fills the attributes field of stat
>> struture by parsing inode flags.
>> 
>> If you must add this check on suse distribution, I guess you just add
>> this version check for suse distribution. For centos7,8, neither of them
>> supports btrfs, but I don't know other distribution situation ie unbuntu.
>
> I just checked debian, both oldstable (4.16) and stable (5.10) have new
> enough kernels for this not to matter.
>
>> Maybe you can just add suse detection in lib/tst_kvercmp.c.
>
> I guess that this would be the cleanest solution.
>
> Actually SUSE should be detected just fine, since we parse
> /etc/os-release for ID='foo' in the test library.
>
> So this could be solved just by defining:
>
> static struct tst_kern_exv kvers[] = {
> 	{"sles", "4.13.0"}
> 	{}
> };
>
> and then doing:
>
> 	if (tst_kvercmp2(0, 0, 0, kvers) < 0)
> 		tst_brk(TCONF, "Btrfs statx() supported since 4.13");
>
>
> Also it would be a bit cleaner to add this to the tst_test structure as
> .min_kver_ex as well, but that's a different story...

After some more internal discussions. We can just filter statx04 and use
the new test which performs the feature checks. I think it would have
been better to add the checks to statx04 and add a new test without any
checks. However it is done now.

-- 
Thank you,
Richard.