restG4 gets stuck at the time of writing geometry

RestG4
1

I started to experience some problems when launching a restG4 simulation.

The problem was solved when I modified line 487 from int stat_val; to int stat_val = 0.

I dont understand very well why this stat_val is given by reference to the wait function, and why it was not initiliazed?

Was this bug? What should be the value of stat_val and why this trick of forking the process was introduced?

470     if (pid == 0) {
471         // writing the geometry object
472         freopen("/dev/null", "w", stdout);
473         freopen("/dev/null", "w", stderr);
474         Console::CompatibilityMode = true;
475     
476         TFile* f1 = new TFile(Filename, "update");
477         TGeoManager* geo2 = gdml->CreateGeoM();
478         
479         f1->cd();
480         geo2->SetName("Geometry");
481         geo2->Write();
482         f1->Close();
483         exit(0);
484     }
485     // father process
486     else {
487         int stat_val = 0;
488         pid_t child_pid;
489     
490         printf("Writing geometry ... \n");
491     
492         child_pid = wait(&stat_val);
493     
494         printf("Geometry writting process exited, pid = %d, Code %d\n", child_pid, WEXITSTATUS(stat_val));
495         if (WEXITSTATUS(stat_val) != 0) printf("REST Error: geometry writting is abnormal!\n");
496     }
2

Indeed it is a bug because it is not expected to happen. By defining int stat_val it is not initialized. It will have a random value. This is true. I am not very familiar with multi process of c. Maybe we must input the reference of int with vaue 0 to wait() method.

It was introduced to bypass the occasional seg.fault: