Integration change. Bug fixes from AMD-CMU trace gen branch.

[git-p4: depot-paths = "//depot/gpgpu_sim_research/fermi/distribution/": change = 8389]

3 files changed, 7 insertions, 2 deletions

diff --git a/src/gpgpu-sim/gpu-sim.cc b/src/gpgpu-sim/gpu-sim.cc
index 05a9b55..1526ef3 100644
--- a/src/gpgpu-sim/gpu-sim.cc
+++ b/src/gpgpu-sim/gpu-sim.cc
@@ -711,6 +711,7 @@ void shader_core_ctx::issue_block2core( kernel_info_t &kernel )
         m_thread[i].m_cta_id = free_cta_hw_id;
         unsigned warp_id = i/m_config->warp_size;
         nthreads_in_block += ptx_sim_init_thread(kernel,&m_thread[i].m_functional_model_thread_state,m_sid,i,cta_size-(i-start_thread),m_config->n_thread_per_shader,this,free_cta_hw_id,warp_id,m_cluster->get_gpu());
+        m_thread[i].m_active = true; 
         warps.set( warp_id );
     }
     assert( nthreads_in_block > 0 && nthreads_in_block <= m_config->n_thread_per_shader); // should be at least one, but less than max
diff --git a/src/gpgpu-sim/shader.cc b/src/gpgpu-sim/shader.cc
index 2380bcf..2233566 100644
--- a/src/gpgpu-sim/shader.cc
+++ b/src/gpgpu-sim/shader.cc
@@ -135,6 +135,7 @@ shader_core_ctx::shader_core_ctx( class gpgpu_sim *gpu,
    for (unsigned i = 0; i<config->n_thread_per_shader; i++) {
       m_thread[i].m_functional_model_thread_state = NULL;
       m_thread[i].m_cta_id = -1;
+      m_thread[i].m_active = false;
    }
    
    m_icnt = new shader_memory_interface(this,cluster);
@@ -519,12 +520,13 @@ void shader_core_ctx::fetch()
                 bool did_exit=false;
                 for( unsigned t=0; t<m_config->warp_size;t++) {
                     unsigned tid=warp_id*m_config->warp_size+t;
-                    if( m_thread[tid].m_functional_model_thread_state ) {
+                    if( m_thread[tid].m_active == true ) {
+                        m_thread[tid].m_active = false; 
                         unsigned cta_id = m_warp[warp_id].get_cta_id();
                         register_cta_thread_exit(cta_id);
                         m_not_completed -= 1;
                         m_active_threads.reset(tid);
-                        m_thread[tid].m_functional_model_thread_state=NULL;
+                        assert( m_thread[tid].m_functional_model_thread_state != NULL );
                         did_exit=true;
                     }
                 }
diff --git a/src/gpgpu-sim/shader.h b/src/gpgpu-sim/shader.h
index 2b834a6..ee0269f 100644
--- a/src/gpgpu-sim/shader.h
+++ b/src/gpgpu-sim/shader.h
@@ -120,6 +120,8 @@ public:
    unsigned n_l1_mis_ac;
    unsigned n_l1_mrghit_ac;
    unsigned n_l1_access_ac; 
+
+   bool m_active; 
 };
 
 class shd_warp_t {