Text archives Help
- From: bigler@sci.utah.edu
- To: manta@sci.utah.edu
- Subject: [Manta] r1779 - trunk/Engine/Control
- Date: Tue, 16 Oct 2007 17:11:28 -0600 (MDT)
Author: bigler
Date: Tue Oct 16 17:11:27 2007
New Revision: 1779
Modified:
trunk/Engine/Control/RTRT.cc
Log:
Engine/Control/RTRT.cc
I'm sick of Manta dying. I've commited a hack that keeps the code
from crashing. The hack will stay until I can get the other
pipeline debugged.
The problem was thread 0 reallocating the thread_storage pointers
for only the threads that were going to be around. The threads
dying tried to call allocateStorage with a dead pointer and it blew
up. The real fix would be to have the threads leave before getting
in this chuck of code. Also, new threads should start off with this
chunk of code as there may be important thread setup stuff here.
Modified: trunk/Engine/Control/RTRT.cc
==============================================================================
--- trunk/Engine/Control/RTRT.cc (original)
+++ trunk/Engine/Control/RTRT.cc Tue Oct 16 17:11:27 2007
@@ -750,7 +750,10 @@
barrier3.wait(numProcs);
// Allocate thread local storage for each thread.
- thread_storage->allocateStorage( proc );
+ if(proc < workersRendering) {
+ // Crappy hack just to keep it from dying.
+ thread_storage->allocateStorage( proc );
+ }
for(size_t index = 0;index < channels.size();index++){
Channel* channel = channels[index];
- [Manta] r1779 - trunk/Engine/Control, bigler, 10/16/2007
Archive powered by MHonArc 2.6.16.