fixed loader script stack setup

runtime/include/vx_intrinsics.h

@@ -20,21 +20,24 @@ void vx_split(int predicate);
 // Join
 void vx_join();
 
-// Return the warp's unique thread id 
+// Return active warp's thread id 
 int vx_thread_id();
 
-// Return the core's unique warp id
-int vx_warp_id();
-
-// Return processsor unique core id
-int vx_core_id();
+// Return active core's local thread id
+int vx_thread_lid();
 
 // Return processsor global thread id
 int vx_thread_gid();
 
-// Return processsor global warp id
+// Return active core's local warp id
+int vx_warp_id();
+
+// Return processsor's global warp id
 int vx_warp_gid();
 
+// Return processsor core id
+int vx_core_id();
+
 // Return the number of threads in a warp
 int vx_num_threads();
 

runtime/linker/vx_link.ld

@@ -9,7 +9,6 @@ OUTPUT_ARCH(riscv)
 ENTRY(_start)
 SECTIONS
 {
-  PROVIDE(__stack_top = 0x6ffff000);
   . = 0x80000000;
   .interp         : { *(.interp) }
   .note.gnu.build-id  : { *(.note.gnu.build-id) }
@@ -191,10 +190,24 @@ SECTIONS
   . = SEGMENT_START("ldata-segment", .);
   . = ALIGN(32 / 8);
   __BSS_END__ = .;
-    __global_pointer$ = MIN(__SDATA_BEGIN__ + 0x800,
+    __global_pointer = MIN(__SDATA_BEGIN__ + 0x800,
                 MAX(__DATA_BEGIN__ + 0x800, __BSS_END__ - 0x800));
   _end = .; PROVIDE (end = .);
-  . = DATA_SEGMENT_END (.);
+  . = DATA_SEGMENT_END (.);  
+  
+  /* .stack_dummy section doesn't contains any symbols. It is only
+	 * used for linker to calculate size of stack sections, and assign
+	 * values to stack symbols later */
+	.stack_dummy (COPY):
+	{
+		KEEP(*(.stack*))
+	}
+  __stack_usage = SIZEOF(.stack_dummy);
+  PROVIDE(__stack_top  = 0xFF000000);
+  PROVIDE(__stack_size = 0x400);
+	PROVIDE(__stack = __stack_top);
+	ASSERT(__stack_usage <= __stack_size, "stack overflow")
+
   /* Stabs debugging sections.  */
   .stab          0 : { *(.stab) }
   .stabstr       0 : { *(.stabstr) }

runtime/src/vx_intrinsics.S

@@ -47,6 +47,12 @@ vx_warp_gid:
 .type vx_thread_id, @function
 .global vx_thread_id
 vx_thread_id:
+	csrr a0, CSR_WTID
+	ret
+
+.type vx_thread_lid, @function
+.global vx_thread_lid
+vx_thread_lid:
 	csrr a0, CSR_LTID
 	ret
 

runtime/src/vx_spawn.c

@@ -6,7 +6,7 @@
 extern "C" {
 #endif
 
-#define NUM_CORES_MAX 8
+#define NUM_CORES_MAX 16
 
 typedef struct {
 	func_t function;

runtime/src/vx_start.S

@@ -57,18 +57,20 @@ vx_set_sp:
   # set global pointer register
   .option push
   .option norelax
-  la gp, __global_pointer$
+  la gp, __global_pointer
   .option pop
 
   # allocate stack region for a threads on the processor 
   # set stack pointer
-  csrr a1, CSR_GTID    # get global thread id
-  slli a1, a1, 10      # multiply by 1024
-  csrr a2, CSR_LTID    # get local thread id  
-  slli a2, a2, 2       # multiply by 4
-  la sp, __stack_top$  # load stack base address
+  la sp, __stack_top   # load stack base address
+  la a1, __stack_size  # stack size
+  #if SM_ENABLE
+  csrr a2, CSR_LTID    # get lobal thread id
+  #else
+  csrr a2, CSR_GTID    # get global thread id
+  #endif
+  mul  a1, a1, a2
   sub  sp, sp, a1      # sub thread block
-  add  sp, sp, a2      # reduce addr collision for perf
 
   # disable active warps except warp0
   csrr a3, CSR_LWID    # get local wid

runtime/tests/dev/vx_dev_main.c

@@ -1,14 +1,13 @@
-
-#include <vx_intrinsics.h>
-
-
-// #include <utlist.h>
 #include <string.h>
 #include <unistd.h>
 #include <stdint.h>
 #include <stdbool.h>
 #include <stdlib.h>
 
+#include <vx_intrinsics.h>
+#include <vx_print.h>
+#include <vx_spawn.h>
+
 typedef struct
 {
 	unsigned * x;

runtime/tests/nlTest/vx_nl_main.c

@@ -1,16 +1,5 @@
-
 #include <vx_intrinsics.h>
-
-// #include <utlist.h>
-#include <string.h>
-#include <unistd.h>
-#include <stdint.h>
-#include <stdbool.h>
-#include <stdlib.h>
-
-
-// Newlib
-#include <stdio.h>
+#include <vx_print.h>
 
 int main()
 {