Hi all.
This seems a bit silly to throw into the bugzilla, but 1.5.91 doesn''t
build for me on amd64 with gcc 4.0.2 and gcc 3.4.5 (Ubuntu Breezy).
lov_ea.c:198: error: conflicting types for ''lovea_off2le''
lov_internal.h:246: error: previous declaration of
''lovea_off2le'' was here
A quick google shows this in some buffalo reports as well, so I
suppose the relevant trivial fix is applied by now.
Additionally, I really recommend the developers to try building it
with gcc4 and fix the massive amounts of warnings produced by
printf format strings not matching the variable types used as
argument. It simply can''t be healthy in the current state.
I''m really hoping for this to get fixed and gcc4 be used on a standard
basis in the development process, since we would really like to be
able to use Lustre on Ubuntu Dapper LTS, which only has gcc 4...
/Nikke
--
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Niklas Edmundsson, Admin @ {acc,hpc2n}.umu.se | nikke@hpc2n.umu.se
---------------------------------------------------------------------------
1st Law of Thermodynamics: Go to class!!
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
I had the same error re lov_internal.h. This patch fixes the problem:
<patch>
--- ./lustre/lov/lov_internal.h~ 2006-06-30 15:08:00.000000000 +0000
+++ ./lustre/lov/lov_internal.h 2006-06-30 17:44:16.000000000 +0000
@@ -243,7 +243,7 @@
struct lov_mds_md *lmm);
struct lov_extent *lovea_idx2le(struct lov_stripe_md *lsm, int stripe_no);
-struct lov_extent *lovea_off2le(struct lov_stripe_md *lsm, obd_off lov_off);
+struct lov_extent *lovea_off2le(struct lov_stripe_md *lsm, __u64 lov_off);
int lovea_destroy_object(struct lov_obd *lov, struct lov_stripe_md *lsm,
struct obdo *oa, void *data);
/* lproc_lov.c */
</patch>
Also, I had stability problems with gcc 4.0. I''ve downgraded to using
gcc 3.4.4 and it works great.
eser
On 7/3/06, Niklas Edmundsson <Niklas.Edmundsson@hpc2n.umu.se>
wrote:>
> Hi all.
>
> This seems a bit silly to throw into the bugzilla, but 1.5.91
doesn''t
> build for me on amd64 with gcc 4.0.2 and gcc 3.4.5 (Ubuntu Breezy).
>
> lov_ea.c:198: error: conflicting types for ''lovea_off2le''
> lov_internal.h:246: error: previous declaration of
''lovea_off2le'' was here
>
> A quick google shows this in some buffalo reports as well, so I
> suppose the relevant trivial fix is applied by now.
>
> Additionally, I really recommend the developers to try building it
> with gcc4 and fix the massive amounts of warnings produced by
> printf format strings not matching the variable types used as
> argument. It simply can''t be healthy in the current state.
>
> I''m really hoping for this to get fixed and gcc4 be used on a
standard
> basis in the development process, since we would really like to be
> able to use Lustre on Ubuntu Dapper LTS, which only has gcc 4...
>
> /Nikke
> --
> -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
> Niklas Edmundsson, Admin @ {acc,hpc2n}.umu.se | nikke@hpc2n.umu.se
> ---------------------------------------------------------------------------
> 1st Law of Thermodynamics: Go to class!!
>
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=->
_______________________________________________
> Lustre-devel mailing list
> Lustre-devel@clusterfs.com
> https://mail.clusterfs.com/mailman/listinfo/lustre-devel
>
>
On Mon, 3 Jul 2006, EKC wrote:> I had the same error re lov_internal.h. This patch fixes the problem: > > <patch> > --- ./lustre/lov/lov_internal.h~ 2006-06-30 15:08:00.000000000 +0000 > +++ ./lustre/lov/lov_internal.h 2006-06-30 17:44:16.000000000 +0000 > @@ -243,7 +243,7 @@ > struct lov_mds_md *lmm); > > struct lov_extent *lovea_idx2le(struct lov_stripe_md *lsm, int stripe_no); > -struct lov_extent *lovea_off2le(struct lov_stripe_md *lsm, obd_off lov_off); > +struct lov_extent *lovea_off2le(struct lov_stripe_md *lsm, __u64 lov_off); > int lovea_destroy_object(struct lov_obd *lov, struct lov_stripe_md *lsm, > struct obdo *oa, void *data); > /* lproc_lov.c */ > </patch>I would do this the other way round, since obd_off seems to be the correct type and __u64 seems to be a lazy programmer.> Also, I had stability problems with gcc 4.0. I''ve downgraded to using > gcc 3.4.4 and it works great.Yeah, that''s a viable workaround for systems where GCC 3.x is available. Just telling people to use an older compiler isn''t a viable long term solution though. /Nikke -- -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Niklas Edmundsson, Admin @ {acc,hpc2n}.umu.se | nikke@hpc2n.umu.se --------------------------------------------------------------------------- "It''s been exactly one week since we copulated." - Anya =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
Hi Niklas - I think we have seen some serious problems with gcc 4 that are unresolved. I am trying to get a complete story on this issue and then we will reply. Thanks! - Peter -> -----Original Message----- > From: lustre-devel-bounces@clusterfs.com > [mailto:lustre-devel-bounces@clusterfs.com] On Behalf Of > Niklas Edmundsson > Sent: Monday, July 03, 2006 8:56 AM > To: lustre-devel@clusterfs.com > Subject: Re: [Lustre-devel] Lustre 1.6b4 (1.5.91) build errors > > On Mon, 3 Jul 2006, EKC wrote: > > > I had the same error re lov_internal.h. This patch fixes > the problem: > > > > <patch> > > --- ./lustre/lov/lov_internal.h~ 2006-06-30 > 15:08:00.000000000 +0000 > > +++ ./lustre/lov/lov_internal.h 2006-06-30 17:44:16.000000000 +0000 > > @@ -243,7 +243,7 @@ > > struct lov_mds_md *lmm); > > > > struct lov_extent *lovea_idx2le(struct lov_stripe_md *lsm, int > > stripe_no); -struct lov_extent *lovea_off2le(struct lov_stripe_md > > *lsm, obd_off lov_off); > > +struct lov_extent *lovea_off2le(struct lov_stripe_md *lsm, __u64 > > +lov_off); > > int lovea_destroy_object(struct lov_obd *lov, struct > lov_stripe_md *lsm, > > struct obdo *oa, void *data); > > /* lproc_lov.c */ > > </patch> > > I would do this the other way round, since obd_off seems to > be the correct type and __u64 seems to be a lazy programmer. > > > Also, I had stability problems with gcc 4.0. I''ve > downgraded to using > > gcc 3.4.4 and it works great. > > Yeah, that''s a viable workaround for systems where GCC 3.x is > available. Just telling people to use an older compiler isn''t > a viable long term solution though. > > /Nikke > -- > -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-> -=-=-=-=-=-=- > Niklas Edmundsson, Admin @ {acc,hpc2n}.umu.se | > nikke@hpc2n.umu.se > -------------------------------------------------------------- > ------------- > "It''s been exactly one week since we copulated." - Anya > =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- > =-=-=-=-=-=-> _______________________________________________ > Lustre-devel mailing list > Lustre-devel@clusterfs.com > https://mail.clusterfs.com/mailman/listinfo/lustre-devel > >