Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] Trunk borked
From: Jeff Squyres (jsquyres_at_[hidden])
Date: 2008-01-28 09:38:36


Doh -- sorry about that. r17282 removes the erroneous #if.

On Jan 28, 2008, at 9:26 AM, Ralph H Castain wrote:

> We seem to have a problem on the trunk this morning. I am building
> on a
> platform with the following configuration:
>
> with_threads=no
> enable_dlopen=no
> enable_pty_support=no
> with_tm=/opt/PBS
> LDFLAGS=-L/opt/PBS/lib64
> with_openib=/opt/ofed
> with_memory_manager=no
> enable_mem_debug=yes
> enable_mem_profile=no
> enable_debug_symbols=yes
> enable_binaries=yes
> with_devel_headers=yes
> enable_heterogeneous=no
> enable_picky=yes
>
> The compile errors out in the OpenIB BTL with the following error:
>
> btl_openib_proc.c: In function `mca_btl_openib_proc_create':
> btl_openib_proc.c:159: error: `i' undeclared (first use in this
> function)
> btl_openib_proc.c:159: error: (Each undeclared identifier is
> reported only
> once
> btl_openib_proc.c:159: error: for each function it appears in.)
> make[2]: *** [btl_openib_proc.lo] Error 1
> make[1]: *** [all-recursive] Error 1
> make: *** [all-recursive] Error 1
>
> When I look at the code, the problem is the following #if:
>
> #if !defined(WORDS_BIGENDIAN) && OMPI_ENABLE_HETEROGENEOUS_SUPPORT
> size_t i;
> #endif
>
> Yet the code will ALWAYS use that variable to unpack all the ports. I
> removed the #if to clear the problem, but before committing the
> change, I
> wanted to ask why someone thought this test needed to be in the code.
>
> Should the entire loop unpacking all the ports be similarly
> protected, or
> was the protection around the variable declaration simply an error?
>
> Thanks
> Ralph
>
>
> _______________________________________________
> devel mailing list
> devel_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/devel

-- 
Jeff Squyres
Cisco Systems