Weak mutex locking macros

[duke8253]

This deals with the problem brought up in the discussions under PRs #5855 and #5857. And also makes the change such that #4926 still works as intended.

[bryancall]

[approve ci autest]

[scw00]

Should this pr revert ? #5857

[duke8253]

[approve ci autest]

[bryancall]

@scw00 Unfortunately, no. These changes add 2 extra conditionals on the normal SCOPED_MUTEX_LOCK use case (the ink_release_assert on the mutex pointer in the macro and then again on conditional in the constructor).

[scw00]

It seems this pr is duplicated with #5857. So why do we still need #5857 .

[SolidWallOfCode]

@scw00 I presume this will effectively revert #5857, once we are clear on the best way to implement.

[SolidWallOfCode]

@duke8253 and I discussed the plan I put on the chat, and he's going to redo this PR using that design. Basically, create a NullMutex class with no state that always succeeds in locking, create a global singleton, and then pass the address of that to the MutexLock classes in the nullptr case for the WEAK... variants. The previously existing macros and the MutexLock classes are reverted to their previous behavior.

[SolidWallOfCode]

Well, that turned out to not work. Sigh. @duke8253's trying something else now.

[duke8253]

Not sure why it's failing on systems other than Ubuntu, I'll get some vms to test.

[duke8253]

[approve ci clang-analyzer]

[bryancall]

This PR is to fix this core dump (supplied by @duke8253):

[ 0 ] traffic_server      Mutex_lock                            ( I_Lock.h:295 )
[ 1 ] traffic_server      MutexLock                             ( I_Lock.h:363 )
[ 2 ] traffic_server      AutoStopCont::mainEvent(int, Event*)  ( traffic_server.cc:233 )
[ 3 ] traffic_server      EThread::process_event(Event*, int)   ( I_Continuation.h:167 )
[ 4 ] traffic_server      EThread::execute_regular()            ( UnixEThread.cc:249 )
[ 5 ] traffic_server      execute                               ( UnixEThread.cc:338 )
[ 6 ] traffic_server      EThread::execute()                    ( UnixEThread.cc:316 )
[ 7 ] traffic_server      spawn_thread_internal                 ( Thread.cc:92 )
[ 8 ] libpthread-2.17.so  start_thread                          

[bryancall]

👍 Looks good

[bryancall]

Cherry picked to the 9.0.x branch.