Cleanup pthread locks in ofed RDMA verbs

[sean.lim_dell.com]

Summary

On FreeBSD pthread mutex, cond, and spinlocks allocate memory. On Linux based systems, these calls do not allocate memory. So there was a safe assumption that the ofed RDMA verb calls do not need to explicitly destroy the pthread locks. This assumption is false on FreeBSD. So let us rearrange the code to cleanup the pthread locks.

Test Plan

I was able to test and exercise the ibverbs, mlx4 and mlx5 libraries by call the specific calls and then calling free, while using a memory leak detector tool.

I was able to do this for the simpler calls in those libraries. All other _ex version of calls I was unable to do. The cxgb4 ones I was not able to test.

[vangyzen]

In light of @hselasky's untimely departure, would someone on network like to review this, or suggest another reviewer?

contrib/ofed/libibverbs/cmd.c

728

This comment made sense in the OneFS tree, but not here. Nuke it.

959

This comment made sense in the OneFS tree, but not here. Nuke it.

1210

This comment made sense in the OneFS tree, but not here. Nuke it.

contrib/ofed/libibverbs/device.c

286

This comment made sense in the OneFS tree, but not here. Nuke it.

contrib/ofed/libibverbs/verbs.h

2171

This comment made sense in the OneFS tree, but not here. Nuke it.

contrib/ofed/libmlx4/mlx4.c

264

Check indentation. It should be one tab.

269

Check indentation. It should be one tab.

contrib/ofed/libmlx5/mlx5.c

962

Check indentation.

1007

Check indentation.

[sean.lim_dell.com]

Ping. Any reviewers?

[jlduran]

In D41105#953123, @sean.lim_dell.com wrote:

Ping. Any reviewers?

Maybe @np ?

[delphij]

The change LGTM. We should probably try to upstream the change...

[sean.lim_dell.com]

What is supposed to happen next? Do I need more reviews or does someone need to merge this code in?