diff --git a/demos/Win32-MinGW/chcore.c b/demos/Win32-MinGW/chcore.c
index 9e9aba3fd..1d1a134b7 100644
--- a/demos/Win32-MinGW/chcore.c
+++ b/demos/Win32-MinGW/chcore.c
@@ -17,91 +17,76 @@
     along with this program.  If not, see <http://www.gnu.org/licenses/>.
 */
 
-/*
- * Core file for MingGW32 demo project.
+/**
+ * @addtogroup WIN32SIM_CORE
+ * @{
  */
 
-#include <windows.h>
-#include <stdio.h>
-
-#undef CDECL
-
 #include <ch.h>
 
-static LARGE_INTEGER nextcnt;
-static LARGE_INTEGER slice;
-
-void InitSimCom1(void);
-void InitSimCom2(void);
-BOOL Com1ConnInterruptSimCom(void);
-BOOL Com2ConnInterruptSimCom(void);
-BOOL Com1InInterruptSimCom(void);
-BOOL Com2InInterruptSimCom(void);
-BOOL Com1OutInterruptSimCom(void);
-BOOL Com2OutInterruptSimCom(void);
-
 /*
- * Simulated HW initialization.
+ * This file is a template of the system driver functions provided by a port.
+ * Some of the following functions may be implemented as macros in chcore.h if
+ * the implementer decides that there is an advantage in doing so, as example
+ * because performance concerns.
  */
-void InitCore(void) {
-  WSADATA wsaData;
 
-  // Initialization.
-  if (WSAStartup(2, &wsaData) != 0) {
-    printf("Unable to locate a winsock DLL\n");
-    exit(1);
-  }
-
-  printf("Win32 ChibiOS/RT simulator\n\n");
-  printf("Thread structure %d bytes\n", sizeof(Thread));
-  if (!QueryPerformanceFrequency(&slice)) {
-    printf("QueryPerformanceFrequency() error");
-    exit(1);
-  }
-  printf("Core Frequency   %u Hz\n", (int)slice.LowPart);
-  slice.QuadPart /= CH_FREQUENCY;
-  QueryPerformanceCounter(&nextcnt);
-  nextcnt.QuadPart += slice.QuadPart;
-
-  InitSimCom1();
-  InitSimCom2();
-  fflush(stdout);
-}
-
-/*
- * Interrupt simulation.
+/**
+ * Prints a message on the system console.
+ * @param msg pointer to the message
+ * @note The function is declared as a weak symbol, it is possible to redefine
+ *       it in your application code.
  */
-void ChkIntSources(void) {
-  LARGE_INTEGER n;
-
-  if (Com1InInterruptSimCom()   || Com2InInterruptSimCom()  ||
-      Com1OutInterruptSimCom()  || Com2OutInterruptSimCom() ||
-      Com1ConnInterruptSimCom() || Com2ConnInterruptSimCom()) {
-    if (chSchRescRequiredI())
-      chSchDoRescheduleI();
-    return;
-  }
-
-  // Interrupt Timer simulation (10ms interval).
-  QueryPerformanceCounter(&n);
-  if (n.QuadPart > nextcnt.QuadPart) {
-    nextcnt.QuadPart += slice.QuadPart;
-    chSysTimerHandlerI();
-    if (chSchRescRequiredI())
-      chSchDoRescheduleI();
-  }
+__attribute__((weak))
+void sys_puts(char *msg) {
 }
 
-msg_t _idle(void *p) {
+/**
+ * Performs a context switch between two threads.
+ * @param otp the thread to be switched out
+ * @param ntp the thread to be switched in
+ * @note The function is declared as a weak symbol, it is possible to redefine
+ *       it in your application code.
+ */
+__attribute__((naked, weak))
+void sys_switch(Thread *otp, Thread *ntp) {
+  register struct intctx *esp asm("esp");
 
-  while (TRUE) {
-
-    ChkIntSources();
-    Sleep(0);
-  }
+  asm volatile ("push    %ebp                                   \n\t" \
+                "push    %esi                                   \n\t" \
+                "push    %edi                                   \n\t" \
+                "push    %ebx");
+  otp->p_ctx.esp = esp;
+  esp = ntp->p_ctx.esp;
+  asm volatile ("pop     %ebx                                   \n\t" \
+                "pop     %edi                                   \n\t" \
+                "pop     %esi                                   \n\t" \
+                "pop     %ebp                                   \n\t" \
+                "ret" : : "r" (esp));
 }
 
-__attribute__((fastcall)) void chSysHalt(void) {
+/**
+ * Halts the system. In this implementation it just exits the simulation.
+ * @note The function is declared as a weak symbol, it is possible to redefine
+ *       it in your application code.
+ */
+__attribute__((weak))
+void sys_halt(void) {
 
   exit(2);
 }
+
+/**
+ * Threads return point, it just invokes @p chThdExit().
+ * @note The function is declared as a weak symbol, it is possible to redefine
+ *       it in your application code.
+ */
+__attribute__((naked, weak))
+void threadexit(void) {
+
+  asm volatile ("push    %eax                                   \n\t" \
+                "call    _chThdExit");
+}
+
+}
+/** @} */
diff --git a/demos/Win32-MinGW/chcore.h b/demos/Win32-MinGW/chcore.h
index d1f209a68..6c9bf1f7e 100644
--- a/demos/Win32-MinGW/chcore.h
+++ b/demos/Win32-MinGW/chcore.h
@@ -17,27 +17,40 @@
     along with this program.  If not, see <http://www.gnu.org/licenses/>.
 */
 
-/*
- * Core file for MingGW32 demo project.
+/**
+ * @addtogroup WIN32SIM_CORE
+ * @{
  */
 
 #ifndef _CHCORE_H_
 #define _CHCORE_H_
 
-/*
- * Unique macro for the implemented architecture.
+/**
+ * Macro defining the a simulated architecture into Win32.
  */
 #define CH_ARCHITECTURE_WIN32SIM
 
-/*
- * Base type for stack alignment.
+/**
+ * 32 bit stack alignment.
  */
 typedef uint32_t stkalign_t;
 
+/**
+ * Generic x86 register.
+ */
 typedef void *regx86;
 
-/*
- * Stack saved context.
+/**
+ * Interrupt saved context.
+ * This structure represents the stack frame saved during a preemption-capable
+ * interrupt handler.
+ */
+struct extctx {
+};
+
+/**
+ * System saved context.
+ * @note In this demo the floating point registers are not saved.
  */
 struct intctx {
   regx86  ebx;
@@ -47,17 +60,26 @@ struct intctx {
   regx86  eip;
 };
 
+/**
+ * Platform dependent part of the @p Thread structure.
+ * This structure usually contains just the saved stack pointer defined as a
+ * pointer to a @p intctx structure.
+ */
 typedef struct {
   struct intctx *esp;
 } Context;
 
 #define APUSH(p, a) (p) -= sizeof(void *), *(void **)(p) = (void*)(a)
 
-#define SETUP_CONTEXT(workspace, wsize, pf, arg)                        \
-{                                                                       \
+/**
+ * Platform dependent part of the @p chThdInit() API.
+ * This code usually setup the context switching frame represented by a
+ * @p intctx structure.
+ */
+#define SETUP_CONTEXT(workspace, wsize, pf, arg) {                      \
   uint8_t *esp = (uint8_t *)workspace + wsize;                          \
   APUSH(esp, arg);                                                      \
-  APUSH(esp, threadstart);                                              \
+  APUSH(esp, threadexit);                                               \
   esp -= sizeof(struct intctx);                                         \
   ((struct intctx *)esp)->eip = pf;                                     \
   ((struct intctx *)esp)->ebx = 0;                                      \
@@ -67,33 +89,96 @@ typedef struct {
   tp->p_ctx.esp = (struct intctx *)esp;                                 \
 }
 
-#define chSysLock()
-#define chSysUnlock()
-#define chSysEnable()
-#define chSysPuts(msg) {}
-#define chSysIRQEnterI()
-#define chSysIRQExitI()
+/**
+ * Stack size for the system idle thread.
+ */
+#ifndef IDLE_THREAD_STACK_SIZE
+#define IDLE_THREAD_STACK_SIZE 256
+#endif
 
-#define INT_REQUIRED_STACK 0
+/**
+ * Per-thread stack overhead for interrupts servicing, it is used in the
+ * calculation of the correct working area size.
+ * It requires stack space because the simulated "interrupt handlers" invoke
+ * Win32 APIs inside so it better have a lot of space.
+ */
+#ifndef INT_REQUIRED_STACK
+#define INT_REQUIRED_STACK 16384
+#endif
 
+/**
+ * Enforces a correct alignment for a stack area size value.
+ */
 #define STACK_ALIGN(n) ((((n) - 1) | sizeof(stkalign_t)) + 1)
 
+ /**
+  * Computes the thread working area global size.
+  */
 #define THD_WA_SIZE(n) STACK_ALIGN(sizeof(Thread) +                     \
                                    sizeof(void *) * 2 +                 \
                                    sizeof(struct intctx) +              \
-                                   (n) +                                \
-                                   INT_REQUIRED_STACK)
+                                   sizeof(struct extctx) +              \
+                                  (n) + (INT_REQUIRED_STACK))
 
+/**
+ * Macro used to allocate a thread working area aligned as both position and
+ * size.
+ */
 #define WORKING_AREA(s, n) stkalign_t s[THD_WA_SIZE(n) / sizeof(stkalign_t)];
 
-/*
- * Stack size for the system idle thread.
+/**
+ * IRQ prologue code, inserted at the start of all IRQ handlers enabled to
+ * invoke system APIs.
  */
-#define IDLE_THREAD_STACK_SIZE 16384
+#define SYS_IRQ_PROLOGUE()
 
-msg_t _idle(void *p);
-__attribute__((fastcall)) void chSysHalt(void);
-__attribute__((fastcall)) void chSysSwitchI(Thread *otp, Thread *ntp);
-__attribute__((fastcall)) void threadstart(void);
+/**
+ * IRQ epilogue code, inserted at the end of all IRQ handlers enabled to
+ * invoke system APIs.
+ */
+#define SYS_IRQ_EPILOGUE()
+
+/**
+ * Does nothing in this simulator.
+ */
+#define sys_disable()
+
+/**
+ * Does nothing in this simulator.
+ */
+#define sys_enable()
+
+/**
+ * Does nothing in this simulator.
+ */
+#define sys_disable_from_isr()
+
+/**
+ * Does nothing in this simulator.
+ */
+#define sys_enable_from_isr()
+
+/**
+ * Does nothing in this simulator.
+ */
+#define sys_wait_for_interrupt()
+
+/**
+ * IRQ handler function modifier.
+ */
+#define SYS_IRQ_HANDLER
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+  __attribute__((fastcall)) void sys_puts(char *msg);
+  __attribute__((fastcall)) void sys_switch(Thread *otp, Thread *ntp);
+  __attribute__((fastcall)) void sys_halt(void);
+  void threadexit(void);
+#ifdef __cplusplus
+}
+#endif
 
 #endif /* _CHCORE_H_ */
+
+/** @} */
diff --git a/ports/AVR/chcore.h b/ports/AVR/chcore.h
index e41ec673f..90755cb4e 100644
--- a/ports/AVR/chcore.h
+++ b/ports/AVR/chcore.h
@@ -28,7 +28,6 @@
 #include <avr/io.h>
 #include <avr/interrupt.h>
 
-
 /*
  * Port-related configuration parameters.
  */
@@ -226,6 +225,7 @@ extern "C" {
   void sys_puts(char *msg);
   void sys_switch(Thread *otp, Thread *ntp);
   void sys_halt(void);
+  void threadstart(void);
 #ifdef __cplusplus
 }
 #endif
diff --git a/ports/MSP430/chcore.h b/ports/MSP430/chcore.h
index 42f0c2f71..8510aa76c 100644
--- a/ports/MSP430/chcore.h
+++ b/ports/MSP430/chcore.h
@@ -193,6 +193,7 @@ extern "C" {
   void sys_puts(char *msg);
   void sys_switch(Thread *otp, Thread *ntp);
   void sys_halt(void);
+  void threadstart(void);
 #ifdef __cplusplus
 }
 #endif