[Openmp-dev] Five smaller patches

Hal Finkel hfinkel at anl.gov
Thu Jan 29 03:51:08 PST 2015


----- Original Message -----
> From: "Jonathan L Peyton" <jonathan.l.peyton at intel.com>
> To: openmp-dev at dcs-maillist2.engr.illinois.edu
> Sent: Wednesday, January 28, 2015 4:42:02 PM
> Subject: [Openmp-dev] Five smaller patches
> 
> Hello,
> 
> We have another set of small patches.
> 
> 
> 
> kmp_place_threads.patch – enables environment variable
> KMP_PLACE_THREADS for all non-MIC architectures.

Any reason not to make these variables int, instead of unsigned int, and thus avoid the additional casts you're adding?

> 
> proc_bind_fix.patch – Small fix that sets proc-bind-var to
> proc_bind_false if affinity is not supported.

LGTM.

> 
> kmp_get_aff_max_proc_fix.patch – fixes mistake in the
> kmp_get_affinity_max_proc() api function, also fixes misspelling
> error.

LGTM (please commit spelling fix separately)

> 
> enable_parallel_build.patch – enables parallel builds using the Top
> Level GNU Makefile. Example: $ make compiler=clang jobs=4

This seems to be doing two things:
 - Adding the jobs variable
 - Fixing the Fortran module dependencies (and some pdb file thing?)

Both of these look fine, but please commit them separately.

> 
> never_unload_windows.patch – Pins the libiomp5.dll for the lifetime
> of application (Windows Only). This is a fix for repeated loading
> and unloading of libiomp5.dll.

Why are you doing this only if GUIDEDLL_EXPORTS is defined?

 -Hal

> 
> 
> 
> Apply patches with:
> 
> patch -p0 < kmp_place_threads.patch
> 
> patch -p0 < proc_bind_fix.patch
> 
> patch -p0 < kmp_get_aff_max_proc_fix.patch
> 
> patch -p0 < enable_parallel_build.patch
> 
> patch -p0 < never_unload_windows.patch
> 
> 
> 
> -- Johnny
> 
> 
> _______________________________________________
> Openmp-dev mailing list
> Openmp-dev at dcs-maillist2.engr.illinois.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/openmp-dev
> 

-- 
Hal Finkel
Assistant Computational Scientist
Leadership Computing Facility
Argonne National Laboratory




More information about the Openmp-dev mailing list