[mpich-discuss] [mvapich-discuss] Lustre 2.4 bug is back in MV2-2.1a

[Rob Latham]

On 10/01/2014 10:33 AM, Jonathan Perkins wrote:
> Hello MPICH community.  Recently we had a report of a regression on the
> MVAPICH discuss list and I've found that the regression appears in
> MPICH-3.1.2 itself.
>
> We plan on just removing the _XOPEN_SOURCE lines but thought that you'd
> like to know about this so that it could be fixed in MPICH as well.
> Thanks!

so, on solaris we needed to put back _XOPEN_SOURCE lines with 
--enable-strict in order to bring in pread/pwrite declarations, I think. 
  but i had forgotten about the caddr_t time compile issue.

thanks for the report.

==rob
>
> On Tue, Sep 30, 2014 at 07:18:45PM -0400, Jonathan Perkins wrote:
>> Thanks for the report Adam.  We'll review the fix and ticket and see
>> what we do to prevent this from appearing in future releases.
>>
>> On Tue, Sep 30, 2014 at 04:07:24PM -0700, Adam T. Moody wrote:
>>> Hello MVAPICH team,
>>> While building MV2-2.1a on our systems, I'm getting the following error:
>>>
>>> In file included from /usr/include/linux/lustre_user.h:46,
>>>                  from /usr/include/lustre/lustre_user.h:54,
>>>                  from adio/ad_lustre/ad_lustre.h:30,
>>>                  from adio/ad_lustre/ad_lustre_rwcontig.c:20:
>>> /usr/include/sys/quota.h:221: error: expected declaration specifiers or
>>> '...' before 'caddr_t'
>>> make[3]: *** [adio/ad_lustre/ad_lustre_rwcontig.lo] Error 1
>>>
>>> This happens while trying to build
>>> mvapich2-2.1a/src/mpi/romio/adio/ad_lustre/ad_lustre_rwcontig.c.
>>>
>>> This is due to an incompatibility with Lustre 2.4.  I first saw this with
>>> MV2-1.9, it was fixed in MV2-2.0, but now it's back with MV2-2.1a.
>>>
>>> The problem lines are here, which can be found starting at line 5:
>>>
>>> #ifdef _STDC_C99
>>> #define _XOPEN_SOURCE 600
>>> #else
>>> #define _XOPEN_SOURCE 500
>>> #endif
>>>
>>> The solution before was to not define _XOPEN_SOURCE in this file.  If I
>>> comment out these definitions, I've verified that MV2-2.1a builds.
>>>
>>> Here is a related MPICH ticket reported by Yan Li and Jerome Vienne:
>>>
>>> http://trac.mpich.org/projects/mpich/ticket/1973
>>> -Adam
>>> _______________________________________________
>>> mvapich-discuss mailing list
>>> mvapich-discuss at cse.ohio-state.edu
>>> http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss
>>
>> --
>> Jonathan Perkins
>

-- 
Rob Latham
Mathematics and Computer Science Division
Argonne National Lab, IL USA
_______________________________________________
discuss mailing list     discuss at mpich.org
To manage subscription options or unsubscribe:
https://lists.mpich.org/mailman/listinfo/discuss