Hi Dominique, Thanks for trying this out so quickly and finding this rather fundamental problem.
In pr30617, DTIO is obliquely described as recursive IO. It isn't really, is it? I'll contact Jerry on #gfortran to see what we can do about this. Cheers Paul On 22 August 2016 at 17:09, Dominique d'Humières <domi...@lps.ens.fr> wrote: > The executable hangs with > > Call graph: > 2699 Thread_25555235 DispatchQueue_1: com.apple.main-thread (serial) > 2699 get_external_unit (in libgfortran.3.dylib) + 254 [0x10de297be] > gthr-default.h:748 > 2699 _pthread_mutex_lock_wait (in libsystem_pthread.dylib) + 89 > [0x7fff96686e4a] > 2699 __psynch_mutexwait (in libsystem_kernel.dylib) + 10 > [0x7fff875bfde6] > > I have no idea about the default thread support in Fedora. > > Dominique > >> Le 22 août 2016 à 17:03, Paul Richard Thomas <paul.richard.tho...@gmail.com> >> a écrit : >> >> Dear Dominique, >> >> It is related to pr30617. When I compile the dtio testcases with >> -lpthread, they hang as you say (as does the testcase of pr30617). >> >> Blast! Both Jerry and I are using Fedora. Do I gather that the use of >> threads is not the default there? >> >> Cheers >> >> Paul >> >> On 22 August 2016 at 15:57, Dominique d'Humières <domi...@lps.ens.fr> wrote: >>> Dear Paul, >>> >>> I have tried the patch, but all the tests hang at run time (expect >>> dtio_10.f90). This is on x86_64-apple-darwin15. Could it be related to >>> pr30617? >>> >>> Cheers, >>> >>> Dominique >>> >> >> >> >> -- >> The difference between genius and stupidity is; genius has its limits. >> >> Albert Einstein > -- The difference between genius and stupidity is; genius has its limits. Albert Einstein
