adds some hooks for SMT startup

env/start.S

@@ -9,18 +9,35 @@ _start:
 .option push
 .option norelax
 .option norvc
-//#ifdef WITH_SIGNATURE
+#ifdef WITH_SIGNATURE
 	j 1f
 	.2byte	0x4e4d
 	.2byte	0x5352
 	.4byte  0x669
 1:
-//#endif
+#endif
     la gp, __global_pointer$
 .option pop
-	la sp, _sp
+    csrr  t0, mhartid   // Get hart ID
+    la t1, __stack_size // stack size
+    la sp, _sp          // Base stack address
+	/* this loop is to avoid multiplication which is not available on all ISAs */
+1:
+	beqz t0, 2f
+    sub sp, sp, t1      // Subtract __stack_size to get per-hart stack base
+	addi t0, t0, -1
+	j 1b
+2:
     la t0, trap_entry
     csrw mtvec, t0
+#ifdef HARTX_WAIT4WFI
+	/* block other cores until hart 0 has finished initialization */
+	csrr  t0, mhartid 
+	beqz t0, hart0_init
+	wfi
+	j hartx_start
+hart0_init:
+#endif
 	/* Load data section */
 	la a0, _data_lma
 	la a1, _data
@@ -50,6 +67,10 @@ _start:
 	call atexit
 	call __libc_init_array
 #endif
+	la a0, mtx 
+	li t0, 1
+	sw t0, 0(a0)
+hartx_start:
 #ifndef __riscv_float_abi_soft
 	/* Enable FPU */
 	li t0, MSTATUS_FS
@@ -66,3 +87,8 @@ _start:
 	li a1, 0
 	call main
 	tail _exit
+
+	
+.data
+mtx: 
+    .word 0        # 32-bit value