IIRC, the concern was with where the thread safety should reside. Some classes (e.g., opal_list) were littered with thread locks for every operation. So if someone implemented thread protection at a higher level (e.g., protecting the list while cycling thru it), then all these lower-level lock/unlock operations were just a waste of cycles.

However, some people felt that there were places where it helped to have the locking down below. So this was the compromise - use the version that fits your situation.

Personally, I'm not wild about it, but I can live with it. I'd prefer to see no lock/unlock calls in the classes themselves as they are too atomistic, and would have opted for providing a macro version of the function that included the appropriate lock/unlocks around the function.

Anyway, that was the thinking at the meeting last June.

On Oct 9, 2013, at 1:40 AM, George Bosilca <bosilca@icl.utk.edu> wrote:

My concern is that increasing the number of interfaces will not make the code thread safe, as in most cases thread safety is not only a matter of using a _mt version of the basic class object but a matter of a careful manipulation of higher level concepts.

We can hardly use the lack of the _MT function as a reason for not having thread safety in the code. We did have the thread safety a while back without the support of _MT version of all the basic classes. 

So I really wonder what is the rationale behind such an intrusive change of the codebase?


On Oct 8, 2013, at 18:14 , Ralph Castain <rhc@open-mpi.org> wrote:

Hi folks

This was one item from the last devel meeting that can be done independent of other things:

resolution: all opal and orte (and possibly ompi) classes need to have a thread safe and thread-free interface
_st suffix: single thread (i.e., not thread safe variant)
_mt suffix: multi thread (i.e., thread safe variant)
for functions that have both st/mt, they will *both* have suffixes
other functions (that do not have st/mt versions) will be naked names
need to rename all classes that have locking enabled already (e.g., opal_free_list)
so today, we go rename all the functions (e.g., opal_free_list functions get _mt suffix) throughout the code base
as someone needs the _st version, they go create it and use it as they want to
Ralph will do the orte classes
Aurelien will do this for the ompi classes

I believe some of these have been done - I will take care of the ORTE classes this week, so consider this a "heads up" for that change.

devel mailing list

devel mailing list