Merge branch 'tensor_core' into kernels

kernel/linker/vx_link32.ld

@@ -10,6 +10,7 @@ ENTRY(_start)
 
 MEMORY {
   DRAM0    (rwx): ORIGIN = 0x80000000, LENGTH = 512M
+  DRAMARG  (rwx): ORIGIN = 0x9fff0000, LENGTH = 8K
   DRAM1    (rwx): ORIGIN = 0xa0000000, LENGTH = 16M
   DRAM2    (rwx): ORIGIN = 0xa1000000, LENGTH = 16M
 }
@@ -275,6 +276,10 @@ SECTIONS
   .gnu.attributes 0 : { KEEP (*(.gnu.attributes)) }
   /DISCARD/ : { *(.note.GNU-stack) *(.gnu_debuglink) *(.gnu.lto_*) }
 
+  .args : {
+    *(.args)
+    . += 8K;
+  }> DRAMARG
   .operand.a : {
     *(.operand.a)
     . += 32K;

kernel/src/vx_spawn.c

@@ -170,7 +170,7 @@ static void __attribute__ ((noinline)) spawn_tasks_cluster_all_cb() {
   vx_tmc_zero();
 }
 
-static void __attribute__ ((noinline)) spawn_tasks_all_cb() {  
+static void __attribute__ ((noinline)) spawn_tasks_all_cb() {
   // activate all threads
   vx_tmc(-1);
 
@@ -254,6 +254,7 @@ void vx_spawn_tasks_cluster(int num_tasks, vx_spawn_tasks_cb callback, void *arg
     vx_wspawn_wait();
   }
 
+  // TODO: this is incomplete
   // TODO: Instead of launching an additional wave just to work on remaining
   // threads, handle this in the last wave amongst other full warps.
   if (rem_threads_in_last_warp != 0 && core_id_in_cluster == 0) {