[Buildroot] [PATCH 1/1] package/nfs-utils: enable nfsv4/nfsv4.1 support

[Giulio Benetti]

Hi James,

On 7/15/21 4:13 AM, James Hilliard wrote:
> On Wed, Jul 14, 2021 at 7:42 PM Giulio Benetti
> <giulio.benetti at benettiengineering.com> wrote:
>>
>>
>> Il giorno 15 lug 2021, alle ore 03:37, Giulio Benetti <giulio.benetti at benettiengineering.com> ha scritto:
>>
>> Hi Petr, James, All,
>>
>> Il giorno 13 lug 2021, alle ore 23:51, Petr Vorel <petr.vorel at gmail.com> ha scritto:
>>
>> Hi Giulio, all,
>>
>> [Cc Thomas, Peter ]
>>
>> Il giorno 13 lug 2021, alle ore 10:19, James Hilliard <james.hilliard1 at gmail.com> ha scritto:
>>
>>
>> On Tue, Jul 13, 2021 at 2:03 AM Giulio Benetti
>>
>> <giulio.benetti at benettiengineering.com> wrote:
>>
>>
>> Hello James,
>>
>>
>> Il giorno 13 lug 2021, alle ore 05:14, James Hilliard <james.hilliard1 at gmail.com> ha scritto:
>>
>>
>> Signed-off-by: James Hilliard <james.hilliard1 at gmail.com>
>>
>> ---
>>
>> package/nfs-utils/Config.in    | 10 ++++++++++
>>
>> package/nfs-utils/nfs-utils.mk |  9 +++++++--
>>
>> 2 files changed, 17 insertions(+), 2 deletions(-)
>>
>>
>> diff --git a/package/nfs-utils/Config.in b/package/nfs-utils/Config.in
>>
>> index da563db0f1..b736f6c973 100644
>>
>> --- a/package/nfs-utils/Config.in
>>
>> +++ b/package/nfs-utils/Config.in
>>
>> @@ -14,6 +14,16 @@ config BR2_PACKAGE_NFS_UTILS
>>
>>
>> if BR2_PACKAGE_NFS_UTILS
>>
>>
>> +config BR2_PACKAGE_NFS_UTILS_NFSV4
>>
>> +    bool "NFSv4/NFSv4.1"
>>
>> +    depends on !BR2_STATIC_LIBS # keyutils, lvm2
>>
>> +    select BR2_PACKAGE_KEYUTILS
>>
>> +    select BR2_PACKAGE_LIBEVENT
>>
>> +    select BR2_PACKAGE_LVM2
>>
>> +    select BR2_PACKAGE_SQLITE
>>
>> +    help
>>
>> +      Enable NFSv4/NFSv4.1 support
>>
>> +
>>
>> config BR2_PACKAGE_NFS_UTILS_RPCDEBUG
>>
>>   bool "rpcdebug"
>>
>>   help
>>
>> diff --git a/package/nfs-utils/nfs-utils.mk b/package/nfs-utils/nfs-utils.mk
>>
>> index c384e939d8..2598f0908b 100644
>>
>> --- a/package/nfs-utils/nfs-utils.mk
>>
>> +++ b/package/nfs-utils/nfs-utils.mk
>>
>> @@ -16,8 +16,6 @@ NFS_UTILS_AUTORECONF = YES
>>
>> NFS_UTILS_CONF_ENV = knfsd_cv_bsd_signals=no
>>
>>
>> NFS_UTILS_CONF_OPTS = \
>>
>> -    --disable-nfsv4 \
>>
>> -    --disable-nfsv41 \
>>
>>
>> Is this ^^^ what prevents uuid.h build error?
>>
>>
>> No idea, have some background on that issue?
>>
>>
>> Yes, I had that issue on this commit:
>>
>>
>> http://git.linux-nfs.org/?p=steved/nfs-utils.git;a=blobdiff;f=support/export/v4root.c;h=c12a7d8562b22616fbc7840faf6719443dac64d3;hp=3654bd7c10c059d6a6ae32001ca979d671ef38f2;hb=c5528f40f9db5061e06dcf1f9b7fce5185b376c6;hpb=ed83085ff310fdc95badb06a26289efb70f99ccd
>>
>>
>> But it seems that —disable-nfsv4 avoid building it,
>>
>> so we should have no problem.
>>
>>
>> Can you be more specific how c5528f40 ("Fix NFSv4 export of tmpfs filesystems")
>> affects Buildroot? Is it a problem with -luuid
>>
>>
>> Yes, in nfs-utils —disable-uuid options doesn’t take into account v4nfs that from that commit on
>> uses uuid. It only prevents bulding libblkid support that they still declare buggy here:
>>
>> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/
>>
>>
>> Sorry wrong link, this is the right one:
>>
>> http://git.linux-nfs.org/?p=steved/nfs-utils.git;a=blob;f=configure.ac;h=93520a8052bcbc2a4f57090f89f606d8aa76c801;hb=HEAD#l169
> 
> But isn't our libblkid new enough to not have the bug?
> See the version check here:
> http://git.linux-nfs.org/?p=steved/nfs-utils.git;a=blob;f=aclocal/libblkid.m4;h=10824e9f58f1ae5164945f6baf14c000ba7949fd;hb=0ffef5f9a448013bcc551407134db84796fd2a4e#l13
> 

Aah, my bad, I've interpreted --disable-uuid bad, sorry. Yes, then what 
you're adding makes a lot of sense!

Thank you
Best regards
-- 
Giulio Benetti
Benetti Engineering sas

>>
>>
>>
>> Anyway, we disabled nfsv4/nfsv4.1 from the start, I wonder why.
>>
>>
>> As I see from James patch, he disables nfsv4 that is the part where uuid is required.
>> That is why it builds fine without uuid. If we want nfsv4 we need uuid, but then nfs-utile must be patched to make things more clear, but substituting —disable-uuid with —disable-libblkid
>> and using —disable-uuid to remove the above commit code from building.
>>
>> Anyway what is supplied by James builds then, I didn’t try but it makes sense. So it depends if someone wants to create patches to send upstream like I’ve described(if it makes sense for you all of course.
>>
>> Best regards
>> Giulio
>>
>> @Thomas, @Peter: do you remember it?
>>
>> But generally good idea, thus:
>> Reviewed-by: Petr Vorel <petr.vorel at gmail.com>
>>
>> Can you double check building with nfsv4 support and find it failing wi uuid.h? I don’t have time at the moment.
>>
>>
>> I'll try to test it this week.
>>
>> Kind regards,
>> Petr
>>
>> Thanks a lot
>>
>>
>> Giulio
>>
>> _______________________________________________
>> buildroot mailing list
>> buildroot at busybox.net
>> http://lists.busybox.net/mailman/listinfo/buildroot
> _______________________________________________
> buildroot mailing list
> buildroot at busybox.net
> http://lists.busybox.net/mailman/listinfo/buildroot
>