5 files changed, 201 insertions, 106 deletions

diff --git a/firmware/export/config.h b/firmware/export/config.h
index 8f1a5e65cc..cd674b63f8 100644
--- a/firmware/export/config.h
+++ b/firmware/export/config.h
@@ -151,6 +151,10 @@
 #define USBOTG_ISP1362 1362
 #define USBOTG_M5636 5636
 
+/* Multiple cores */
+#define CPU 0
+#define COP 1
+
 /* now go and pick yours */
 #if defined(ARCHOS_PLAYER)
 #include "config-player.h"
@@ -219,6 +223,19 @@
 /* define for all cpus from PP family */
 #if (CONFIG_CPU == PP5002) || (CONFIG_CPU == PP5020) || (CONFIG_CPU == PP5024)
 #define CPU_PP
+
+/* PP family has dual cores */
+#if 0
+/* Keep it as single core until dual core support is ready */
+#define NUM_CORES 2
+#define CURRENT_CORE current_core()
+#endif
+
+#define NUM_CORES 1
+#define CURRENT_CORE 0
+#else
+#define NUM_CORES 1
+#define CURRENT_CORE 0
 #endif
 
 /* define for all cpus from ARM family */
diff --git a/firmware/export/system.h b/firmware/export/system.h
index 1b326e1b8b..d13b27c529 100644
--- a/firmware/export/system.h
+++ b/firmware/export/system.h
@@ -43,6 +43,8 @@ static inline void udelay(unsigned usecs)
     unsigned start = USEC_TIMER;
     while ((USEC_TIMER - start) < usecs);
 }
+
+unsigned int current_core(void);
 #endif
 
 struct flash_header {
@@ -88,7 +90,7 @@ void cpu_idle_mode(bool on_off);
 #define betoh32(x) swap32(x)
 #define htobe16(x) swap16(x)
 #define htobe32(x) swap32(x)
-#else 
+#else
 #define letoh16(x) swap16(x)
 #define letoh32(x) swap32(x)
 #define htole16(x) swap16(x)
@@ -178,7 +180,7 @@ static inline void coldfire_set_macsr(unsigned long flags)
 static inline unsigned long coldfire_get_macsr(void)
 {
     unsigned long m;
-    
+
     asm volatile ("move.l %%macsr, %0" : "=r" (m));
     return m;
 }
@@ -301,7 +303,7 @@ static inline void invalidate_icache(void)
                  "move.l #0x80000000,%d0\n"
                  "movec.l %d0,%cacr");
 }
-            
+
 #define CPUFREQ_DEFAULT_MULT 1
 #define CPUFREQ_DEFAULT      (CPUFREQ_DEFAULT_MULT * CPU_FREQ)
 #define CPUFREQ_NORMAL_MULT  4
@@ -315,7 +317,7 @@ static inline void invalidate_icache(void)
 
 #define CPUFREQ_DEFAULT_MULT 8
 #define CPUFREQ_DEFAULT 24000000
-#define CPUFREQ_NORMAL_MULT 10 
+#define CPUFREQ_NORMAL_MULT 10
 #define CPUFREQ_NORMAL 30000000
 #define CPUFREQ_MAX_MULT 25
 #define CPUFREQ_MAX 75000000
@@ -336,7 +338,7 @@ static inline unsigned long swap32(unsigned long value)
       result[15.. 8] = value[23..16];
       result[ 7.. 0] = value[31..24];
     */
-{   
+{
     unsigned int tmp;
 
     asm volatile (
@@ -418,7 +420,7 @@ static inline int set_irq_level(int level)
     __asm__ volatile ("clrsr ie");
   else
     __asm__ volatile ("setsr ie");
-    
+
   return result;
 }
 
diff --git a/firmware/export/thread.h b/firmware/export/thread.h
index 762e315a4c..a5034aedab 100644
--- a/firmware/export/thread.h
+++ b/firmware/export/thread.h
@@ -31,12 +31,16 @@
 
 int create_thread(void (*function)(void), void* stack, int stack_size,
                   const char *name);
+int create_thread_on_core(unsigned int core, void (*function)(void), void* stack, int stack_size,
+                  const char *name);
 void remove_thread(int threadnum);
+void remove_thread_on_core(unsigned int core, int threadnum);
 void switch_thread(void);
 void sleep_thread(void);
 void wake_up_thread(void);
 void init_threads(void);
 int thread_stack_usage(int threadnum);
+int thread_stack_usage_on_core(unsigned int core, int threadnum);
 #ifdef RB_PROFILE
 void profile_thread(void);
 #endif
diff --git a/firmware/system.c b/firmware/system.c
index c97b5233ec..6aee823205 100644
--- a/firmware/system.c
+++ b/firmware/system.c
@@ -91,21 +91,21 @@ bool detect_flashed_rockbox(void)
 {
     struct flash_header hdr;
     uint8_t *src = (uint8_t *)FLASH_ENTRYPOINT;
-    
+
 # ifndef BOOTLOADER
     int oldmode;
     oldmode = system_memory_guard(MEMGUARD_NONE);
 # endif
-    
+
     memcpy(&hdr, src, sizeof(struct flash_header));
-    
+
 # ifndef BOOTLOADER
     system_memory_guard(oldmode);
 # endif
-    
+
     if (hdr.magic != FLASH_MAGIC)
         return false;
-    
+
     return true;
 }
 #else
@@ -139,13 +139,13 @@ void ddma_transfer(int dir, int mem, void* intAddr, long extAddr, int num) {
     /* HW wants those two in word units. */
     num /= 2;
     externalAddress /= 2;
-    
+
     DDMACFG = (dir << 1) | (mem << 2);
     DDMAIADR = internalAddress;
     DDMAEADR = externalAddress;
     DDMANUM = num;
     DDMACOM |= 0x4; /* start */
-    
+
     ddma_wait_idle(); /* wait for completion */
     set_irq_level(irq);
 }
@@ -164,13 +164,13 @@ static void ddma_transfer_noicode(int dir, int mem, long intAddr, long extAddr,
     /* HW wants those two in word units. */
     num /= 2;
     externalAddress /= 2;
-    
+
     DDMACFG = (dir << 1) | (mem << 2);
     DDMAIADR = internalAddress;
     DDMAEADR = externalAddress;
     DDMANUM = num;
     DDMACOM |= 0x4; /* start */
-    
+
     ddma_wait_idle_noicode(); /* wait for completion */
     set_irq_level(irq);
 }
@@ -220,7 +220,7 @@ void smsc_delay() {
        Delay doesn't depend on CPU speed in Archos' firmware.
     */
     for (i = 0; i < 100; i++) {
-        
+
     }
 }
 
@@ -228,9 +228,9 @@ static void extra_init(void) {
     /* Power on stuff */
     P1 |= 0x07;
     P1CON |= 0x1f;
-    
+
     /* P5 conf
-     * lines 0, 1 & 4 are digital, other analog. : 
+     * lines 0, 1 & 4 are digital, other analog. :
      */
     P5CON = 0xec;
 
@@ -294,21 +294,21 @@ void system_init(void)
     /********
      * CPU
      */
-    
-    
+
+
     /* PLL0 (cpu osc. frequency) */
     /* set_cpu_frequency(CPU_FREQ); */
 
 
     /*******************
-     * configure S(D)RAM 
+     * configure S(D)RAM
      */
 
     /************************
      * Copy .icode section to icram
      */
     ddma_transfer_noicode(0, 0, 0x40, (long)&icodecopy, (int)&icodesize);
-   
+
 
     /***************************
      * Interrupts
@@ -472,12 +472,12 @@ void UIE (void) /* Unexpected Interrupt or Exception */
     unsigned int format_vector, pc;
     int vector;
     char str[32];
-    
+
     asm volatile ("move.l (52,%%sp),%0": "=r"(format_vector));
     asm volatile ("move.l (56,%%sp),%0": "=r"(pc));
 
     vector = (format_vector >> 18) & 0xff;
-    
+
     /* clear screen */
     lcd_clear_display ();
 #ifdef HAVE_LCD_BITMAP
@@ -488,7 +488,7 @@ void UIE (void) /* Unexpected Interrupt or Exception */
     snprintf(str,sizeof(str),"at %08x",pc);
     lcd_puts(0,1,str);
     lcd_update();
-    
+
     /* set cpu frequency to 11mhz (to prevent overheating) */
     DCR = (DCR & ~0x01ff) | 1;
     PLLCR = 0x10800000;
@@ -510,7 +510,7 @@ void UIE (void) /* Unexpected Interrupt or Exception */
 }
 
 /* reset vectors are handled in crt0.S */
-void (* const vbr[]) (void) __attribute__ ((section (".vectors"))) = 
+void (* const vbr[]) (void) __attribute__ ((section (".vectors"))) =
 {
     UIE,UIE,UIE,UIE,UIE,UIE,
     UIE,UIE,UIE,UIE,UIE,UIE,UIE,UIE,
@@ -520,7 +520,7 @@ void (* const vbr[]) (void) __attribute__ ((section (".vectors"))) =
 
     TRAP0,TRAP1,TRAP2,TRAP3,TRAP4,TRAP5,TRAP6,TRAP7,
     TRAP8,TRAP9,TRAP10,TRAP11,TRAP12,TRAP13,TRAP14,TRAP15,
-    
+
     SWT,UIE,UIE,I2C,UART1,UART2,DMA0,DMA1,
     DMA2,DMA3,QSPI,UIE,UIE,UIE,UIE,UIE,
     PDIR1FULL,PDIR2FULL,EBUTXEMPTY,IIS2TXEMPTY,
@@ -741,7 +741,7 @@ static const char* const irqname[] = {
 
 asm (
 
-/* Vector table. 
+/* Vector table.
  * Handled in asm because gcc 4.x doesn't allow weak aliases to symbols
  * defined in an asm block -- silly.
  * Reset vectors (0..3) are handled in crt0.S */
@@ -854,7 +854,7 @@ asm (
     RESERVE_INTERRUPT (        108)
     DEFAULT_INTERRUPT (ADITI,  109)
 
-/* UIE# block. 
+/* UIE# block.
  * Must go into the same section as the UIE() handler */
 
     "\t.text\n"
@@ -979,7 +979,7 @@ void UIE (unsigned int pc) /* Unexpected Interrupt or Exception */
     char str[32];
 
     asm volatile ("sts\tpr,%0" : "=r"(n));
-    
+
     /* clear screen */
     lcd_clear_display ();
 #ifdef HAVE_LCD_BITMAP
@@ -1002,7 +1002,7 @@ void UIE (unsigned int pc) /* Unexpected Interrupt or Exception */
         volatile int i;
         led (state);
         state = !state;
-        
+
         for (i = 0; i < 240000; ++i);
 #endif
 
@@ -1018,7 +1018,7 @@ void UIE (unsigned int pc) /* Unexpected Interrupt or Exception */
 #elif CONFIG_KEYPAD == ONDIO_PAD
         if (!(PCDR & 0x0008))
 #endif
-        {   
+        {
             /* enable the watchguard timer, but don't service it */
             RSTCSR_W = 0x5a40; /* Reset enabled, power-on reset */
             TCSR_W = 0xa560; /* Watchdog timer mode, timer enabled, sysclk/2 */
@@ -1086,7 +1086,7 @@ int system_memory_guard(int newmode)
 
     int oldmode = MEMGUARD_NONE;
     int i;
-    
+
     /* figure out the old mode from what is in the UBC regs. If the register
        values don't match any mode, assume MEMGUARD_NONE */
     for (i = MEMGUARD_NONE; i < MAXMEMGUARD; i++)
@@ -1098,7 +1098,7 @@ int system_memory_guard(int newmode)
             break;
         }
     }
-    
+
     if (newmode == MEMGUARD_KEEP)
         newmode = oldmode;
 
@@ -1109,7 +1109,7 @@ int system_memory_guard(int newmode)
     BAR  = modes[newmode].addr;
     BAMR = modes[newmode].mask;
     BBR  = modes[newmode].bbr;
-    
+
     return oldmode;
 }
 #elif defined(CPU_ARM)
@@ -1124,7 +1124,7 @@ static const char* const uiename[] = {
 void UIE(unsigned int pc, unsigned int num)
 {
     char str[32];
-    
+
     lcd_clear_display();
 #ifdef HAVE_LCD_BITMAP
     lcd_setfont(FONT_SYSFIXED);
@@ -1133,7 +1133,7 @@ void UIE(unsigned int pc, unsigned int num)
     snprintf(str, sizeof(str), "at %08x", pc);
     lcd_puts(0, 1, str);
     lcd_update();
-    
+
     while (1)
     {
         /* TODO: perhaps add button handling in here when we get a polling
@@ -1160,7 +1160,7 @@ void irq(void)
     else if (CPU_INT_STAT & TIMER2_MASK)
         TIMER2();
     else if (CPU_HI_INT_STAT & GPIO_MASK)
-        ipod_mini_button_int();        
+        ipod_mini_button_int();
 }
 #elif (defined IRIVER_H10) || (defined IRIVER_H10_5GB)
 /* TODO: this should really be in the target tree, but moving it there caused
@@ -1170,14 +1170,14 @@ void irq(void)
     if (CPU_INT_STAT & TIMER1_MASK)
         TIMER1();
     else if (CPU_INT_STAT & TIMER2_MASK)
-        TIMER2();        
+        TIMER2();
 }
 #else
 extern void ipod_4g_button_int(void);
 
 void irq(void)
 {
-    if (CPU_INT_STAT & TIMER1_MASK) 
+    if (CPU_INT_STAT & TIMER1_MASK)
         TIMER1();
     else if (CPU_INT_STAT & TIMER2_MASK)
         TIMER2();
@@ -1187,11 +1187,21 @@ void irq(void)
 #endif
 #endif /* BOOTLOADER */
 
+unsigned int current_core(void)
+{
+    if(((*(volatile unsigned long *)(0x60000000)) & 0xff) == 0x55)
+    {
+        return CPU;
+    }
+    return COP;
+}
+
+
 /* TODO: The following two function have been lifted straight from IPL, and
    hence have a lot of numeric addresses used straight. I'd like to use
    #defines for these, but don't know what most of them are for or even what
    they should be named. Because of this I also have no way of knowing how
-   to extend the funtions to do alternate cache configurations and/or 
+   to extend the funtions to do alternate cache configurations and/or
    some other CPU frequency scaling. */
 
 #ifndef BOOTLOADER
@@ -1251,7 +1261,7 @@ void set_cpu_frequency(long frequency)
 #if defined(IPOD_COLOR) || defined(IPOD_4G) || defined(IPOD_MINI)
     /* We don't know why the timer interrupt gets disabled on the PP5020
        based ipods, but without the following line, the 4Gs will freeze
-       when CPU frequency changing is enabled.  
+       when CPU frequency changing is enabled.
 
        Note also that a simple "CPU_INT_EN = TIMER1_MASK;" (as used
        elsewhere to enable interrupts) doesn't work, we need "|=".
@@ -1321,19 +1331,29 @@ extern void TIMER2(void);
 
 void irq(void)
 {
-    if (CPU_INT_STAT & TIMER1_MASK) 
+    if (CPU_INT_STAT & TIMER1_MASK)
         TIMER1();
-    else if (CPU_INT_STAT & TIMER2_MASK) 
+    else if (CPU_INT_STAT & TIMER2_MASK)
         TIMER2();
 }
 
 #endif
 
+unsigned int current_core(void)
+{
+    if(((*(volatile unsigned long *)(0xc4000000)) & 0xff) == 0x55)
+    {
+        return CPU;
+    }
+    return COP;
+}
+
+
 /* TODO: The following two function have been lifted straight from IPL, and
    hence have a lot of numeric addresses used straight. I'd like to use
    #defines for these, but don't know what most of them are for or even what
    they should be named. Because of this I also have no way of knowing how
-   to extend the funtions to do alternate cache configurations and/or 
+   to extend the funtions to do alternate cache configurations and/or
    some other CPU frequency scaling. */
 
 #ifndef BOOTLOADER
@@ -1370,7 +1390,7 @@ void set_cpu_frequency(long frequency)
     else
         postmult = CPUFREQ_DEFAULT_MULT;
     cpu_frequency = frequency;
-    
+
     outl(0x02, 0xcf005008);
     outl(0x55, 0xcf00500c);
     outl(0x6000, 0xcf005010);
@@ -1501,7 +1521,7 @@ void system_init(void)
     IRQ_WRITE_WAIT(0, 0, v == 0);
     IRQ_WRITE_WAIT(4, 0, v == 0);
     */
-    
+
     for (i = 0; i < 0x1c; i++)
     {
         IRQ_WRITE_WAIT(0x404 + i * 4, 0x1e000001, (v & 0x3010f) == 1);
diff --git a/firmware/thread.c b/firmware/thread.c
index 0013d49519..c194b2694d 100644
--- a/firmware/thread.c
+++ b/firmware/thread.c
@@ -41,7 +41,7 @@ struct regs
     void         *pr;    /* Procedure register */
     void         *start; /* Thread start address, or NULL when started */
 };
-#elif defined(CPU_ARM) 
+#elif defined(CPU_ARM)
 struct regs
 {
     unsigned int r[8];   /* Registers r4-r11 */
@@ -62,18 +62,31 @@ struct regs
 /* Cast to the the machine int type, whose size could be < 4. */
 
 
-int num_threads;
-static volatile int num_sleepers;
-static int current_thread;
-static struct regs thread_contexts[MAXTHREADS] IBSS_ATTR;
-const char *thread_name[MAXTHREADS];
-void *thread_stack[MAXTHREADS];
-int thread_stack_size[MAXTHREADS];
+int num_threads[NUM_CORES];
+static volatile int num_sleepers[NUM_CORES];
+static int current_thread[NUM_CORES];
+static struct regs thread_contexts[NUM_CORES][MAXTHREADS] IBSS_ATTR;
+const char *thread_name[NUM_CORES][MAXTHREADS];
+void *thread_stack[NUM_CORES][MAXTHREADS];
+int thread_stack_size[NUM_CORES][MAXTHREADS];
 static const char main_thread_name[] = "main";
 
 extern int stackbegin[];
 extern int stackend[];
 
+#ifdef CPU_PP
+#ifndef BOOTLOADER
+extern int cop_stackbegin[];
+extern int cop_stackend[];
+#else
+/* The coprocessor stack is not set up in the bootloader code, but the
+   threading is.  No threads are run on the coprocessor, so set up some dummy
+   stack */
+int *cop_stackbegin = stackbegin;
+int *cop_stackend = stackend;
+#endif
+#endif
+
 void switch_thread(void) ICODE_ATTR;
 static inline void store_context(void* addr) __attribute__ ((always_inline));
 static inline void load_context(const void* addr) __attribute__ ((always_inline));
@@ -86,7 +99,7 @@ void profile_thread(void) {
 #endif
 
 #if defined(CPU_ARM)
-/*--------------------------------------------------------------------------- 
+/*---------------------------------------------------------------------------
  * Store non-volatile context.
  *---------------------------------------------------------------------------
  */
@@ -116,7 +129,7 @@ static inline void load_context(const void* addr)
 }
 
 #elif defined(CPU_COLDFIRE)
-/*--------------------------------------------------------------------------- 
+/*---------------------------------------------------------------------------
  * Store non-volatile context.
  *---------------------------------------------------------------------------
  */
@@ -199,7 +212,7 @@ static inline void load_context(const void* addr)
 }
 
 #elif CONFIG_CPU == TCC730
-/*--------------------------------------------------------------------------- 
+/*---------------------------------------------------------------------------
  * Store non-volatile context.
  *---------------------------------------------------------------------------
  */
@@ -215,7 +228,7 @@ static inline void load_context(const void* addr)
         "push a14\n\t"                                  \
         "ldw @[%0+0], a15\n\t" : : "a" (addr) );
 
-/*--------------------------------------------------------------------------- 
+/*---------------------------------------------------------------------------
  * Load non-volatile context.
  *---------------------------------------------------------------------------
  */
@@ -260,7 +273,7 @@ void switch_thread(void)
 #ifdef SIMULATOR
     /* Do nothing */
 #else
-    while (num_sleepers == num_threads)
+    while (num_sleepers[CURRENT_CORE] == num_threads[CURRENT_CORE])
     {
         /* Enter sleep mode, woken up on interrupt */
 #ifdef CPU_COLDFIRE
@@ -284,21 +297,21 @@ void switch_thread(void)
 #endif
     }
 #endif
-    current = current_thread;
-    store_context(&thread_contexts[current]);
-    
+    current = current_thread[CURRENT_CORE];
+    store_context(&thread_contexts[CURRENT_CORE][current]);
+
 #if CONFIG_CPU != TCC730
     /* Check if the current thread stack is overflown */
-    stackptr = thread_stack[current];
+    stackptr = thread_stack[CURRENT_CORE][current];
     if(stackptr[0] != DEADBEEF)
-       panicf("Stkov %s", thread_name[current]);
+       panicf("Stkov %s", thread_name[CURRENT_CORE][current]);
 #endif
 
-    if (++current >= num_threads)
+    if (++current >= num_threads[CURRENT_CORE])
         current = 0;
 
-    current_thread = current;
-    load_context(&thread_contexts[current]);
+    current_thread[CURRENT_CORE] = current;
+    load_context(&thread_contexts[CURRENT_CORE][current]);
 #ifdef RB_PROFILE
     profile_thread_started(current_thread);
 #endif
@@ -306,29 +319,42 @@ void switch_thread(void)
 
 void sleep_thread(void)
 {
-    ++num_sleepers;
+    ++num_sleepers[CURRENT_CORE];
     switch_thread();
 }
 
 void wake_up_thread(void)
 {
-    num_sleepers = 0;
+    num_sleepers[CURRENT_CORE] = 0;
 }
 
-/*--------------------------------------------------------------------------- 
- * Create thread.
+
+/*---------------------------------------------------------------------------
+ * Create thread on the current core.
  * Return ID if context area could be allocated, else -1.
  *---------------------------------------------------------------------------
  */
 int create_thread(void (*function)(void), void* stack, int stack_size,
                   const char *name)
 {
+    return create_thread_on_core(CURRENT_CORE, function, stack, stack_size,
+                  name);
+}
+
+/*---------------------------------------------------------------------------
+ * Create thread on a specific core.
+ * Return ID if context area could be allocated, else -1.
+ *---------------------------------------------------------------------------
+ */
+int create_thread_on_core(unsigned int core, void (*function)(void), void* stack, int stack_size,
+                  const char *name)
+{
     unsigned int i;
     unsigned int stacklen;
     unsigned int *stackptr;
     struct regs *regs;
 
-    if (num_threads >= MAXTHREADS)
+    if (num_threads[core] >= MAXTHREADS)
         return -1;
 
     /* Munge the stack to make it easy to spot stack overflows */
@@ -340,10 +366,10 @@ int create_thread(void (*function)(void), void* stack, int stack_size,
     }
 
     /* Store interesting information */
-    thread_name[num_threads] = name;
-    thread_stack[num_threads] = stack;
-    thread_stack_size[num_threads] = stack_size;
-    regs = &thread_contexts[num_threads];
+    thread_name[core][num_threads[core]] = name;
+    thread_stack[core][num_threads[core]] = stack;
+    thread_stack_size[core][num_threads[core]] = stack_size;
+    regs = &thread_contexts[core][num_threads[core]];
 #if defined(CPU_COLDFIRE) || (CONFIG_CPU == SH7034) || defined(CPU_ARM)
     /* Align stack to an even 32 bit boundary */
     regs->sp = (void*)(((unsigned int)stack + stack_size) & ~3);
@@ -355,65 +381,91 @@ int create_thread(void (*function)(void), void* stack, int stack_size,
     regs->start = (void*)function;
 
     wake_up_thread();
-    return num_threads++; /* return the current ID, e.g for remove_thread() */
+    return num_threads[core]++; /* return the current ID, e.g for remove_thread() */
 }
 
-/*--------------------------------------------------------------------------- 
- * Remove a thread from the scheduler.
+/*---------------------------------------------------------------------------
+ * Remove a thread on the current core from the scheduler.
  * Parameter is the ID as returned from create_thread().
  *---------------------------------------------------------------------------
  */
 void remove_thread(int threadnum)
 {
+    remove_thread_on_core(CURRENT_CORE, threadnum);
+}
+
+/*---------------------------------------------------------------------------
+ * Remove a thread on the specified core from the scheduler.
+ * Parameters are the core and the ID as returned from create_thread().
+ *---------------------------------------------------------------------------
+ */
+void remove_thread_on_core(unsigned int core, int threadnum)
+{
     int i;
 
-    if(threadnum >= num_threads)
+    if(threadnum >= num_threads[core])
        return;
 
-    num_threads--;
-    for (i=threadnum; i<num_threads-1; i++)
+    num_threads[core]--;
+    for (i=threadnum; i<num_threads[core]-1; i++)
     {   /* move all entries which are behind */
-        thread_name[i]       = thread_name[i+1];
-        thread_stack[i]      = thread_stack[i+1];
-        thread_stack_size[i] = thread_stack_size[i+1];
-        thread_contexts[i]   = thread_contexts[i+1];
+        thread_name[core][i]       = thread_name[core][i+1];
+        thread_stack[core][i]      = thread_stack[core][i+1];
+        thread_stack_size[core][i] = thread_stack_size[core][i+1];
+        thread_contexts[core][i]   = thread_contexts[core][i+1];
     }
 
-    if (current_thread == threadnum) /* deleting the current one? */
-        current_thread = num_threads; /* set beyond last, avoid store harm */
-    else if (current_thread > threadnum) /* within the moved positions? */
-        current_thread--; /* adjust it, point to same context again */
+    if (current_thread[core] == threadnum) /* deleting the current one? */
+        current_thread[core] = num_threads[core]; /* set beyond last, avoid store harm */
+    else if (current_thread[core] > threadnum) /* within the moved positions? */
+        current_thread[core]--; /* adjust it, point to same context again */
 }
 
 void init_threads(void)
 {
-    num_threads = 1; /* We have 1 thread to begin with */
-    current_thread = 0; /* The current thread is number 0 */
-    thread_name[0] = main_thread_name;
-    thread_stack[0] = stackbegin;
-    thread_stack_size[0] = (int)stackend - (int)stackbegin;
+    unsigned int core = CURRENT_CORE;
+
+    num_threads[core] = 1; /* We have 1 thread to begin with */
+    current_thread[core] = 0; /* The current thread is number 0 */
+    thread_name[core][0] = main_thread_name;
+/* In multiple core setups, each core has a different stack.  There is probably
+   a much better way to do this. */
+    if(core == CPU)
+    {
+        thread_stack[CPU][0] = stackbegin;
+        thread_stack_size[CPU][0] = (int)stackend - (int)stackbegin;
+    } else {
+#if NUM_CORES > 1  /* This code path will not be run on single core targets */
+        thread_stack[COP][0] = cop_stackbegin;
+        thread_stack_size[COP][0] = (int)cop_stackend - (int)cop_stackbegin;
+#endif
+    }
 #if CONFIG_CPU == TCC730
-    thread_contexts[0].started = 1;
+    thread_contexts[core][0].started = 1;
 #else
-    thread_contexts[0].start = 0; /* thread 0 already running */
+    thread_contexts[core][0].start = 0; /* thread 0 already running */
 #endif
-    num_sleepers = 0;
+    num_sleepers[core] = 0;
+}
+
+int thread_stack_usage(int threadnum){
+    return thread_stack_usage_on_core(CURRENT_CORE, threadnum);
 }
 
-int thread_stack_usage(int threadnum)
+int thread_stack_usage_on_core(unsigned int core, int threadnum)
 {
     unsigned int i;
-    unsigned int *stackptr = thread_stack[threadnum];
+    unsigned int *stackptr = thread_stack[core][threadnum];
 
-    if(threadnum >= num_threads)
+    if(threadnum >= num_threads[core])
         return -1;
 
-    for(i = 0;i < thread_stack_size[threadnum]/sizeof(int);i++)
+    for(i = 0;i < thread_stack_size[core][threadnum]/sizeof(int);i++)
     {
         if(stackptr[i] != DEADBEEF)
             break;
     }
 
-    return ((thread_stack_size[threadnum] - i * sizeof(int)) * 100) /
-        thread_stack_size[threadnum];
-}       
+    return ((thread_stack_size[core][threadnum] - i * sizeof(int)) * 100) /
+        thread_stack_size[core][threadnum];
+}