progress indication should start at 1, not 0.

[Rockbox.git] / firmware / thread.c

/***************************************************************************
 *             __________               __   ___.
 *   Open      \______   \ ____   ____ |  | _\_ |__   _______  ___
 *   Source     |       _//  _ \_/ ___\|  |/ /| __ \ /  _ \  \/  /
 *   Jukebox    |    |   (  <_> )  \___|    < | \_\ (  <_> > <  <
 *   Firmware   |____|_  /\____/ \___  >__|_ \|___  /\____/__/\_ \
 *                     \/            \/     \/    \/            \/
 * $Id$
 *
 * Copyright (C) 2002 by Ulf Ralberg
 *
 * All files in this archive are subject to the GNU General Public License.
 * See the file COPYING in the source tree root for full license agreement.
 *
 * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
 * KIND, either express or implied.
 *
 ****************************************************************************/
#include "config.h"
#include <stdbool.h>
#include "thread.h"
#include "panic.h"
#include "sprintf.h"
#include "system.h"
#include "kernel.h"
#include "cpu.h"
#include "string.h"
#ifdef RB_PROFILE
#include <profile.h>
#endif
/****************************************************************************
 *                              ATTENTION!!                                 *
 *    See notes below on implementing processor-specific portions!          *
 ***************************************************************************/

/* Define THREAD_EXTRA_CHECKS as 1 to enable additional state checks */
#ifdef DEBUG
#define THREAD_EXTRA_CHECKS 1 /* Always 1 for DEBUG */
#else
#define THREAD_EXTRA_CHECKS 0
#endif

/**
 * General locking order to guarantee progress. Order must be observed but
 * all stages are not nescessarily obligatory. Going from 1) to 3) is
 * perfectly legal.
 *
 * 1) IRQ
 * This is first because of the likelyhood of having an interrupt occur that
 * also accesses one of the objects farther down the list. Any non-blocking
 * synchronization done may already have a lock on something during normal
 * execution and if an interrupt handler running on the same processor as
 * the one that has the resource locked were to attempt to access the
 * resource, the interrupt handler would wait forever waiting for an unlock
 * that will never happen. There is no danger if the interrupt occurs on
 * a different processor because the one that has the lock will eventually
 * unlock and the other processor's handler may proceed at that time. Not
 * nescessary when the resource in question is definitely not available to
 * interrupt handlers.
 *  
 * 2) Kernel Object
 * 1) May be needed beforehand if the kernel object allows dual-use such as
 * event queues. The kernel object must have a scheme to protect itself from
 * access by another processor and is responsible for serializing the calls
 * to block_thread(_w_tmo) and wakeup_thread both to themselves and to each
 * other. Objects' queues are also protected here.
 * 
 * 3) Thread Slot
 * This locks access to the thread's slot such that its state cannot be
 * altered by another processor when a state change is in progress such as
 * when it is in the process of going on a blocked list. An attempt to wake
 * a thread while it is still blocking will likely desync its state with
 * the other resources used for that state.
 *
 * 4) Core Lists
 * These lists are specific to a particular processor core and are accessible
 * by all processor cores and interrupt handlers. The running (rtr) list is
 * the prime example where a thread may be added by any means.
 */

/*---------------------------------------------------------------------------
 * Processor specific: core_sleep/core_wake/misc. notes
 *
 * ARM notes:
 * FIQ is not dealt with by the scheduler code and is simply restored if it
 * must by masked for some reason - because threading modifies a register
 * that FIQ may also modify and there's no way to accomplish it atomically.
 * s3c2440 is such a case.
 *
 * Audio interrupts are generally treated at a higher priority than others
 * usage of scheduler code with interrupts higher than HIGHEST_IRQ_LEVEL
 * are not in general safe. Special cases may be constructed on a per-
 * source basis and blocking operations are not available.
 *
 * core_sleep procedure to implement for any CPU to ensure an asychronous
 * wakup never results in requiring a wait until the next tick (up to
 * 10000uS!). May require assembly and careful instruction ordering.
 *
 * 1) On multicore, stay awake if directed to do so by another. If so, goto
 *    step 4.
 * 2) If processor requires, atomically reenable interrupts and perform step
 *    3.
 * 3) Sleep the CPU core. If wakeup itself enables interrupts (stop #0x2000
 *    on Coldfire) goto step 5.
 * 4) Enable interrupts.
 * 5) Exit procedure.
 *
 * core_wake and multprocessor notes for sleep/wake coordination:
 * If possible, to wake up another processor, the forcing of an interrupt on
 * the woken core by the waker core is the easiest way to ensure a non-
 * delayed wake and immediate execution of any woken threads. If that isn't
 * available then some careful non-blocking synchonization is needed (as on
 * PP targets at the moment).
 *---------------------------------------------------------------------------
 */

/* Cast to the the machine pointer size, whose size could be < 4 or > 32
 * (someday :). */
#define DEADBEEF ((uintptr_t)0xdeadbeefdeadbeefull)
struct core_entry cores[NUM_CORES] IBSS_ATTR;
struct thread_entry threads[MAXTHREADS] IBSS_ATTR;

static const char main_thread_name[] = "main";
extern uintptr_t stackbegin[];
extern uintptr_t stackend[];

static inline void core_sleep(IF_COP_VOID(unsigned int core))
        __attribute__((always_inline));

void check_tmo_threads(void)
        __attribute__((noinline));

static inline void block_thread_on_l(struct thread_entry *thread, unsigned state)
        __attribute__((always_inline));

static void add_to_list_tmo(struct thread_entry *thread)
        __attribute__((noinline));

static void core_schedule_wakeup(struct thread_entry *thread)
        __attribute__((noinline));

#if NUM_CORES > 1
static inline void run_blocking_ops(
    unsigned int core, struct thread_entry *thread)
        __attribute__((always_inline));
#endif

static void thread_stkov(struct thread_entry *thread)
        __attribute__((noinline));

static inline void store_context(void* addr)
        __attribute__((always_inline));

static inline void load_context(const void* addr)
        __attribute__((always_inline));

void switch_thread(void)
        __attribute__((noinline));

/****************************************************************************
 * Processor-specific section
 */

#ifdef MAX_PHYS_SECTOR_SIZE
/* Support a special workaround object for large-sector disks */
#define IF_NO_SKIP_YIELD(...) __VA_ARGS__
#else
#define IF_NO_SKIP_YIELD(...)
#endif

#if defined(CPU_ARM)
/*---------------------------------------------------------------------------
 * Start the thread running and terminate it if it returns
 *---------------------------------------------------------------------------
 */
static void __attribute__((naked,used)) start_thread(void)
{
    /* r0 = context */
    asm volatile (
        "ldr    sp, [r0, #32]          \n" /* Load initial sp */
        "ldr    r4, [r0, #40]          \n" /* start in r4 since it's non-volatile */
        "mov    r1, #0                 \n" /* Mark thread as running */
        "str    r1, [r0, #40]          \n"
#if NUM_CORES > 1
        "ldr    r0, =invalidate_icache \n" /* Invalidate this core's cache. */
        "mov    lr, pc                 \n" /* This could be the first entry into */
        "bx     r0                     \n" /* plugin or codec code for this core. */
#endif
        "mov    lr, pc                 \n" /* Call thread function */
        "bx     r4                     \n"
    ); /* No clobber list - new thread doesn't care */
    thread_exit();
    //asm volatile (".ltorg"); /* Dump constant pool */
}

/* For startup, place context pointer in r4 slot, start_thread pointer in r5
 * slot, and thread function pointer in context.start. See load_context for
 * what happens when thread is initially going to run. */
#define THREAD_STARTUP_INIT(core, thread, function) \
    ({ (thread)->context.r[0] = (uint32_t)&(thread)->context,  \
       (thread)->context.r[1] = (uint32_t)start_thread, \
       (thread)->context.start = (uint32_t)function; })

/*---------------------------------------------------------------------------
 * Store non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void store_context(void* addr)
{
    asm volatile(
        "stmia  %0, { r4-r11, sp, lr } \n"
        : : "r" (addr)
    );
}

/*---------------------------------------------------------------------------
 * Load non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void load_context(const void* addr)
{
    asm volatile(
        "ldr     r0, [%0, #40]          \n" /* Load start pointer */
        "cmp     r0, #0                 \n" /* Check for NULL */
        "ldmneia %0, { r0, pc }         \n" /* If not already running, jump to start */ 
        "ldmia   %0, { r4-r11, sp, lr } \n" /* Load regs r4 to r14 from context */
        : : "r" (addr) : "r0" /* only! */
    );
}

#if defined (CPU_PP)

#if NUM_CORES > 1
extern uintptr_t cpu_idlestackbegin[];
extern uintptr_t cpu_idlestackend[];
extern uintptr_t cop_idlestackbegin[];
extern uintptr_t cop_idlestackend[];
static uintptr_t * const idle_stacks[NUM_CORES] =
{
    [CPU] = cpu_idlestackbegin,
    [COP] = cop_idlestackbegin
};

#if CONFIG_CPU == PP5002
/* Bytes to emulate the PP502x mailbox bits */
struct core_semaphores
{
    volatile uint8_t intend_wake;  /* 00h */
    volatile uint8_t stay_awake;   /* 01h */
    volatile uint8_t intend_sleep; /* 02h */
    volatile uint8_t unused;       /* 03h */
};

static struct core_semaphores core_semaphores[NUM_CORES] IBSS_ATTR;
#endif /* CONFIG_CPU == PP5002 */

#endif /* NUM_CORES */

#if CONFIG_CORELOCK == SW_CORELOCK
/* Software core locks using Peterson's mutual exclusion algorithm */

/*---------------------------------------------------------------------------
 * Initialize the corelock structure.
 *---------------------------------------------------------------------------
 */
void corelock_init(struct corelock *cl)
{
    memset(cl, 0, sizeof (*cl));
}

#if 1 /* Assembly locks to minimize overhead */
/*---------------------------------------------------------------------------
 * Wait for the corelock to become free and acquire it when it does.
 *---------------------------------------------------------------------------
 */
void corelock_lock(struct corelock *cl) __attribute__((naked));
void corelock_lock(struct corelock *cl)
{
    /* Relies on the fact that core IDs are complementary bitmasks (0x55,0xaa) */
    asm volatile (
        "mov    r1, %0               \n" /* r1 = PROCESSOR_ID */
        "ldrb   r1, [r1]             \n"
        "strb   r1, [r0, r1, lsr #7] \n" /* cl->myl[core] = core */
        "eor    r2, r1, #0xff        \n" /* r2 = othercore */
        "strb   r2, [r0, #2]         \n" /* cl->turn = othercore */
    "1:                              \n"
        "ldrb   r3, [r0, r2, lsr #7] \n" /* cl->myl[othercore] == 0 ? */
        "cmp    r3, #0               \n" /* yes? lock acquired */
        "bxeq   lr                   \n"
        "ldrb   r3, [r0, #2]         \n" /* || cl->turn == core ? */
        "cmp    r3, r1               \n"
        "bxeq   lr                   \n" /* yes? lock acquired */
        "b      1b                   \n" /* keep trying */
        : : "i"(&PROCESSOR_ID)
    );
    (void)cl;
}

/*---------------------------------------------------------------------------
 * Try to aquire the corelock. If free, caller gets it, otherwise return 0.
 *---------------------------------------------------------------------------
 */
int corelock_try_lock(struct corelock *cl) __attribute__((naked));
int corelock_try_lock(struct corelock *cl)
{
    /* Relies on the fact that core IDs are complementary bitmasks (0x55,0xaa) */
    asm volatile (
        "mov    r1, %0               \n" /* r1 = PROCESSOR_ID */
        "ldrb   r1, [r1]             \n"
        "mov    r3, r0               \n"
        "strb   r1, [r0, r1, lsr #7] \n" /* cl->myl[core] = core */
        "eor    r2, r1, #0xff        \n" /* r2 = othercore */
        "strb   r2, [r0, #2]         \n" /* cl->turn = othercore */
        "ldrb   r0, [r3, r2, lsr #7] \n" /* cl->myl[othercore] == 0 ? */
        "eors   r0, r0, r2           \n" /* yes? lock acquired */
        "bxne   lr                   \n"
        "ldrb   r0, [r3, #2]         \n" /* || cl->turn == core? */
        "ands   r0, r0, r1           \n"
        "streqb r0, [r3, r1, lsr #7] \n" /* if not, cl->myl[core] = 0 */
        "bx     lr                   \n" /* return result */
        : : "i"(&PROCESSOR_ID)
    );

    return 0;
    (void)cl;
}

/*---------------------------------------------------------------------------
 * Release ownership of the corelock
 *---------------------------------------------------------------------------
 */
void corelock_unlock(struct corelock *cl) __attribute__((naked));
void corelock_unlock(struct corelock *cl)
{
    asm volatile (
        "mov    r1, %0               \n" /* r1 = PROCESSOR_ID */
        "ldrb   r1, [r1]             \n"
        "mov    r2, #0               \n" /* cl->myl[core] = 0 */
        "strb   r2, [r0, r1, lsr #7] \n"
        "bx     lr                   \n"
        : : "i"(&PROCESSOR_ID)
    );
    (void)cl;
}
#else /* C versions for reference */
/*---------------------------------------------------------------------------
 * Wait for the corelock to become free and aquire it when it does.
 *---------------------------------------------------------------------------
 */
void corelock_lock(struct corelock *cl)
{
    const unsigned int core = CURRENT_CORE;
    const unsigned int othercore = 1 - core;

    cl->myl[core] = core;
    cl->turn = othercore;

    for (;;)
    {
        if (cl->myl[othercore] == 0 || cl->turn == core)
            break;
    }
}

/*---------------------------------------------------------------------------
 * Try to aquire the corelock. If free, caller gets it, otherwise return 0.
 *---------------------------------------------------------------------------
 */
int corelock_try_lock(struct corelock *cl)
{
    const unsigned int core = CURRENT_CORE;
    const unsigned int othercore = 1 - core;

    cl->myl[core] = core;
    cl->turn = othercore;

    if (cl->myl[othercore] == 0 || cl->turn == core)
    {
        return 1;
    }

    cl->myl[core] = 0;
    return 0;
}

/*---------------------------------------------------------------------------
 * Release ownership of the corelock
 *---------------------------------------------------------------------------
 */
void corelock_unlock(struct corelock *cl)
{
    cl->myl[CURRENT_CORE] = 0;
}
#endif /* ASM / C selection */

#endif /* CONFIG_CORELOCK == SW_CORELOCK */

/*---------------------------------------------------------------------------
 * Put core in a power-saving state if waking list wasn't repopulated and if
 * no other core requested a wakeup for it to perform a task.
 *---------------------------------------------------------------------------
 */
#ifdef CPU_PP502x
#if NUM_CORES == 1
static inline void core_sleep(void)
{
    PROC_CTL(CURRENT_CORE) = PROC_SLEEP;
    nop; nop; nop;
    enable_irq();
}
#else
static inline void core_sleep(unsigned int core)
{
#if 1
    asm volatile (
        "mov    r0, #4                     \n" /* r0 = 0x4 << core */
        "mov    r0, r0, lsl %[c]           \n"
        "str    r0, [%[mbx], #4]           \n" /* signal intent to sleep */
        "ldr    r1, [%[mbx], #0]           \n" /* && !(MBX_MSG_STAT & (0x10<<core)) ? */
        "tst    r1, r0, lsl #2             \n"  
        "moveq  r1, #0x80000000            \n" /* Then sleep */
        "streq  r1, [%[ctl], %[c], lsl #2] \n"
        "moveq  r1, #0                     \n" /* Clear control reg */
        "streq  r1, [%[ctl], %[c], lsl #2] \n"
        "orr    r1, r0, r0, lsl #2         \n" /* Signal intent to wake - clear wake flag */
        "str    r1, [%[mbx], #8]           \n"
    "1:                                    \n" /* Wait for wake procedure to finish */
        "ldr    r1, [%[mbx], #0]           \n"
        "tst    r1, r0, lsr #2             \n"
        "bne    1b                         \n"
        :
        :  [ctl]"r"(&PROC_CTL(CPU)), [mbx]"r"(MBX_BASE), [c]"r"(core)
        : "r0", "r1");
#else /* C version for reference */
    /* Signal intent to sleep */
    MBX_MSG_SET = 0x4 << core;

    /* Something waking or other processor intends to wake us? */
    if ((MBX_MSG_STAT & (0x10 << core)) == 0)
    {
        PROC_CTL(core) = PROC_SLEEP; nop; /* Snooze */
        PROC_CTL(core) = 0;               /* Clear control reg */
    }

    /* Signal wake - clear wake flag */
    MBX_MSG_CLR = 0x14 << core;

    /* Wait for other processor to finish wake procedure */
    while (MBX_MSG_STAT & (0x1 << core));
#endif /* ASM/C selection */
    enable_irq();
}
#endif /* NUM_CORES */
#elif CONFIG_CPU == PP5002
#if NUM_CORES == 1
static inline void core_sleep(void)
{
    asm volatile (
        /* Sleep: PP5002 crashes if the instruction that puts it to sleep is
         * located at 0xNNNNNNN0. 4/8/C works. This sequence makes sure
         * that the correct alternative is executed. Don't change the order
         * of the next 4 instructions! */
        "tst    pc, #0x0c     \n"
        "mov    r0, #0xca     \n"
        "strne  r0, [%[ctl]]  \n"
        "streq  r0, [%[ctl]]  \n"
        "nop                  \n" /* nop's needed because of pipeline */
        "nop                  \n"
        "nop                  \n"
        :
        : [ctl]"r"(&PROC_CTL(CURRENT_CORE))
        : "r0"
    );
    enable_irq();
}
#else
/* PP5002 has no mailboxes - emulate using bytes */
static inline void core_sleep(unsigned int core)
{
#if 1
    asm volatile (
        "mov    r0, #1                     \n" /* Signal intent to sleep */
        "strb   r0, [%[sem], #2]           \n"
        "ldrb   r0, [%[sem], #1]           \n" /* && stay_awake == 0? */
        "cmp    r0, #0                     \n"
        "bne    2f                         \n"
        /* Sleep: PP5002 crashes if the instruction that puts it to sleep is
         * located at 0xNNNNNNN0. 4/8/C works. This sequence makes sure
         * that the correct alternative is executed. Don't change the order
         * of the next 4 instructions! */
        "tst    pc, #0x0c                  \n"
        "mov    r0, #0xca                  \n"
        "strne  r0, [%[ctl], %[c], lsl #2] \n"
        "streq  r0, [%[ctl], %[c], lsl #2] \n"
        "nop                               \n" /* nop's needed because of pipeline */
        "nop                               \n"
        "nop                               \n"
    "2:                                    \n"
        "mov    r0, #0                     \n" /* Clear stay_awake and sleep intent */
        "strb   r0, [%[sem], #1]           \n"
        "strb   r0, [%[sem], #2]           \n"
    "1:                                    \n" /* Wait for wake procedure to finish */
        "ldrb   r0, [%[sem], #0]           \n"
        "cmp    r0, #0                     \n"
        "bne    1b                         \n"
        :
        : [sem]"r"(&core_semaphores[core]), [c]"r"(core),
          [ctl]"r"(&PROC_CTL(CPU))
        : "r0"
        );
#else /* C version for reference */
    /* Signal intent to sleep */
    core_semaphores[core].intend_sleep = 1;

    /* Something waking or other processor intends to wake us? */
    if (core_semaphores[core].stay_awake == 0)
    {
        PROC_CTL(core) = PROC_SLEEP; /* Snooze */
        nop; nop; nop;
    }

    /* Signal wake - clear wake flag */
    core_semaphores[core].stay_awake = 0;
    core_semaphores[core].intend_sleep = 0;

    /* Wait for other processor to finish wake procedure */
    while (core_semaphores[core].intend_wake != 0);

    /* Enable IRQ */
#endif /* ASM/C selection */
    enable_irq();
}
#endif /* NUM_CORES */
#endif /* PP CPU type */

/*---------------------------------------------------------------------------
 * Wake another processor core that is sleeping or prevent it from doing so
 * if it was already destined. FIQ, IRQ should be disabled before calling.
 *---------------------------------------------------------------------------
 */
#if NUM_CORES == 1
/* Shared single-core build debugging version */
void core_wake(void)
{
    /* No wakey - core already wakey */
}
#elif defined (CPU_PP502x)
void core_wake(unsigned int othercore)
{
#if 1
    /* avoid r0 since that contains othercore */
    asm volatile (
        "mrs    r3, cpsr                    \n" /* Disable IRQ */
        "orr    r1, r3, #0x80               \n"
        "msr    cpsr_c, r1                  \n"
        "mov    r2, #0x11                   \n" /* r2 = (0x11 << othercore) */
        "mov    r2, r2, lsl %[oc]           \n" /* Signal intent to wake othercore */
        "str    r2, [%[mbx], #4]            \n"
    "1:                                     \n" /* If it intends to sleep, let it first */
        "ldr    r1, [%[mbx], #0]            \n" /* (MSG_MSG_STAT & (0x4 << othercore)) != 0 ? */
        "eor    r1, r1, #0xc                \n"
        "tst    r1, r2, lsr #2              \n"
        "ldr    r1, [%[ctl], %[oc], lsl #2] \n" /* && (PROC_CTL(othercore) & PROC_SLEEP) == 0 ? */
        "tsteq  r1, #0x80000000             \n"
        "beq    1b                          \n" /* Wait for sleep or wake */
        "tst    r1, #0x80000000             \n" /* If sleeping, wake it */
        "movne  r1, #0x0                    \n"
        "strne  r1, [%[ctl], %[oc], lsl #2] \n"
        "mov    r1, r2, lsr #4              \n"
        "str    r1, [%[mbx], #8]            \n" /* Done with wake procedure */
        "msr    cpsr_c, r3                  \n" /* Restore IRQ */
        :
        : [ctl]"r"(&PROC_CTL(CPU)), [mbx]"r"(MBX_BASE),
          [oc]"r"(othercore)
        : "r1", "r2", "r3");
#else /* C version for reference */
    /* Disable interrupts - avoid reentrancy from the tick */
    int oldlevel = disable_irq_save();

    /* Signal intent to wake other processor - set stay awake */
    MBX_MSG_SET = 0x11 << othercore;

    /* If it intends to sleep, wait until it does or aborts */
    while ((MBX_MSG_STAT & (0x4 << othercore)) != 0 &&
           (PROC_CTL(othercore) & PROC_SLEEP) == 0);

    /* If sleeping, wake it up */
    if (PROC_CTL(othercore) & PROC_SLEEP)
        PROC_CTL(othercore) = 0;

    /* Done with wake procedure */
    MBX_MSG_CLR = 0x1 << othercore;
    restore_irq(oldlevel);
#endif /* ASM/C selection */
}
#elif CONFIG_CPU == PP5002
/* PP5002 has no mailboxes - emulate using bytes */
void core_wake(unsigned int othercore)
{
#if 1
    /* avoid r0 since that contains othercore */
    asm volatile (
        "mrs    r3, cpsr                \n" /* Disable IRQ */
        "orr    r1, r3, #0x80           \n"
        "msr    cpsr_c, r1              \n"
        "mov    r1, #1                  \n" /* Signal intent to wake other core */
        "orr    r1, r1, r1, lsl #8      \n" /* and set stay_awake */
        "strh   r1, [%[sem], #0]        \n"
        "mov    r2, #0x8000             \n"
    "1:                                 \n" /* If it intends to sleep, let it first */
        "ldrb   r1, [%[sem], #2]        \n" /* intend_sleep != 0 ? */
        "cmp    r1, #1                  \n"
        "ldr    r1, [%[st]]             \n" /* && not sleeping ? */
        "tsteq  r1, r2, lsr %[oc]       \n"
        "beq    1b                      \n" /* Wait for sleep or wake */
        "tst    r1, r2, lsr %[oc]       \n"
        "ldrne  r2, =0xcf004054         \n" /* If sleeping, wake it */
        "movne  r1, #0xce               \n"
        "strne  r1, [r2, %[oc], lsl #2] \n"
        "mov    r1, #0                  \n" /* Done with wake procedure */
        "strb   r1, [%[sem], #0]        \n"
        "msr    cpsr_c, r3              \n" /* Restore IRQ */
        :
        : [sem]"r"(&core_semaphores[othercore]),
          [st]"r"(&PROC_STAT),
          [oc]"r"(othercore)
        : "r1", "r2", "r3"
    );
#else /* C version for reference */
    /* Disable interrupts - avoid reentrancy from the tick */
    int oldlevel = disable_irq_save();

    /* Signal intent to wake other processor - set stay awake */
    core_semaphores[othercore].intend_wake = 1;
    core_semaphores[othercore].stay_awake = 1;

    /* If it intends to sleep, wait until it does or aborts */
    while (core_semaphores[othercore].intend_sleep != 0 &&
           (PROC_STAT & PROC_SLEEPING(othercore)) == 0);

    /* If sleeping, wake it up */
    if (PROC_STAT & PROC_SLEEPING(othercore))
        PROC_CTL(othercore) = PROC_WAKE;

    /* Done with wake procedure */
    core_semaphores[othercore].intend_wake = 0;
    restore_irq(oldlevel);
#endif  /* ASM/C selection */
}
#endif /* CPU type */

#if NUM_CORES > 1
/*---------------------------------------------------------------------------
 * Switches to a stack that always resides in the Rockbox core.
 *
 * Needed when a thread suicides on a core other than the main CPU since the
 * stack used when idling is the stack of the last thread to run. This stack
 * may not reside in the core firmware in which case the core will continue
 * to use a stack from an unloaded module until another thread runs on it.
 *---------------------------------------------------------------------------
 */
static inline void switch_to_idle_stack(const unsigned int core)
{
    asm volatile (
        "str  sp, [%0] \n" /* save original stack pointer on idle stack */
        "mov  sp, %0   \n" /* switch stacks */
        : : "r"(&idle_stacks[core][IDLE_STACK_WORDS-1]));
    (void)core;
}

/*---------------------------------------------------------------------------
 * Perform core switch steps that need to take place inside switch_thread.
 *
 * These steps must take place while before changing the processor and after
 * having entered switch_thread since switch_thread may not do a normal return
 * because the stack being used for anything the compiler saved will not belong
 * to the thread's destination core and it may have been recycled for other
 * purposes by the time a normal context load has taken place. switch_thread
 * will also clobber anything stashed in the thread's context or stored in the
 * nonvolatile registers if it is saved there before the call since the
 * compiler's order of operations cannot be known for certain.
 */
static void core_switch_blk_op(unsigned int core, struct thread_entry *thread)
{
    /* Flush our data to ram */
    flush_icache();
    /* Stash thread in r4 slot */
    thread->context.r[0] = (uint32_t)thread;
    /* Stash restart address in r5 slot */
    thread->context.r[1] = thread->context.start;
    /* Save sp in context.sp while still running on old core */
    thread->context.sp = idle_stacks[core][IDLE_STACK_WORDS-1];
}

/*---------------------------------------------------------------------------
 * Machine-specific helper function for switching the processor a thread is
 * running on. Basically, the thread suicides on the departing core and is
 * reborn on the destination. Were it not for gcc's ill-behavior regarding
 * naked functions written in C where it actually clobbers non-volatile
 * registers before the intended prologue code, this would all be much
 * simpler.  Generic setup is done in switch_core itself.
 */

/*---------------------------------------------------------------------------
 * This actually performs the core switch.
 */
static void __attribute__((naked))
    switch_thread_core(unsigned int core, struct thread_entry *thread)
{
    /* Pure asm for this because compiler behavior isn't sufficiently predictable.
     * Stack access also isn't permitted until restoring the original stack and
     * context. */
    asm volatile (
        "stmfd  sp!, { r4-r12, lr }    \n" /* Stack all non-volatile context on current core */
        "ldr    r2, =idle_stacks       \n" /* r2 = &idle_stacks[core][IDLE_STACK_WORDS] */
        "ldr    r2, [r2, r0, lsl #2]   \n"
        "add    r2, r2, %0*4           \n"
        "stmfd  r2!, { sp }            \n" /* save original stack pointer on idle stack */
        "mov    sp, r2                 \n" /* switch stacks */
        "adr    r2, 1f                 \n" /* r2 = new core restart address */
        "str    r2, [r1, #40]          \n" /* thread->context.start = r2 */
        "ldr    pc, =switch_thread     \n" /* r0 = thread after call - see load_context */
    "1:                                \n"
        "ldr    sp, [r0, #32]          \n" /* Reload original sp from context structure */
        "mov    r1, #0                 \n" /* Clear start address */
        "str    r1, [r0, #40]          \n"
        "ldr    r0, =invalidate_icache \n" /* Invalidate new core's cache */
        "mov    lr, pc                 \n"
        "bx     r0                     \n"
        "ldmfd  sp!, { r4-r12, pc }    \n" /* Restore non-volatile context to new core and return */
        ".ltorg                        \n" /* Dump constant pool */
        : : "i"(IDLE_STACK_WORDS)
    );
    (void)core; (void)thread;
}
#endif /* NUM_CORES */

#elif CONFIG_CPU == S3C2440

/*---------------------------------------------------------------------------
 * Put core in a power-saving state if waking list wasn't repopulated.
 *---------------------------------------------------------------------------
 */
static inline void core_sleep(void)
{
    /* FIQ also changes the CLKCON register so FIQ must be disabled
       when changing it here */
    asm volatile (
        "mrs    r0, cpsr        \n"
        "orr    r2, r0, #0x40   \n" /* Disable FIQ */
        "bic    r0, r0, #0x80   \n" /* Prepare IRQ enable */
        "msr    cpsr_c, r2      \n"
        "mov    r1, #0x4c000000 \n" /* CLKCON = 0x4c00000c */
        "ldr    r2, [r1, #0xc]  \n" /* Set IDLE bit */
        "orr    r2, r2, #4      \n"
        "str    r2, [r1, #0xc]  \n"
        "msr    cpsr_c, r0      \n" /* Enable IRQ, restore FIQ */
        "mov    r2, #0          \n" /* wait for IDLE */
    "1:                         \n"
        "add    r2, r2, #1      \n"
        "cmp    r2, #10         \n"
        "bne    1b              \n"
        "orr    r2, r0, #0xc0   \n" /* Disable IRQ, FIQ */
        "msr    cpsr_c, r2      \n"
        "ldr    r2, [r1, #0xc]  \n" /* Reset IDLE bit */
        "bic    r2, r2, #4      \n"
        "str    r2, [r1, #0xc]  \n"
        "msr    cpsr_c, r0      \n" /* Enable IRQ, restore FIQ */
        :  :  : "r0", "r1", "r2");
}
#elif defined(CPU_TCC77X)
static inline void core_sleep(void)
{
    #warning TODO: Implement core_sleep
    enable_irq();
}
#elif defined(CPU_TCC780X)
static inline void core_sleep(void)
{
    /* Single core only for now. Use the generic ARMv5 wait for IRQ */
    asm volatile (
        "mov r0, #0                \n"
        "mcr p15, 0, r0, c7, c0, 4 \n" /* Wait for interrupt */
        : : : "r0"
    );
    enable_irq();
}
#elif CONFIG_CPU == IMX31L
static inline void core_sleep(void)
{
    asm volatile (
        "mov r0, #0                \n"
        "mcr p15, 0, r0, c7, c0, 4 \n" /* Wait for interrupt */
        : : : "r0"
    );
    enable_irq();
}
#else
static inline void core_sleep(void)
{
    #warning core_sleep not implemented, battery life will be decreased
    enable_irq();
}
#endif /* CONFIG_CPU == */

#elif defined(CPU_COLDFIRE)
/*---------------------------------------------------------------------------
 * Start the thread running and terminate it if it returns
 *---------------------------------------------------------------------------
 */
void start_thread(void); /* Provide C access to ASM label */
static void __attribute__((used)) __start_thread(void)
{
    /* a0=macsr, a1=context */
    asm volatile (
    "start_thread:             \n" /* Start here - no naked attribute */
        "move.l  %a0, %macsr   \n" /* Set initial mac status reg */
        "lea.l   48(%a1), %a1  \n"
        "move.l  (%a1)+, %sp   \n" /* Set initial stack */
        "move.l  (%a1), %a2    \n" /* Fetch thread function pointer */
        "clr.l   (%a1)         \n" /* Mark thread running */
        "jsr     (%a2)         \n" /* Call thread function */
    );
    thread_exit();
}

/* Set EMAC unit to fractional mode with saturation for each new thread,
 * since that's what'll be the most useful for most things which the dsp
 * will do. Codecs should still initialize their preferred modes
 * explicitly. Context pointer is placed in d2 slot and start_thread
 * pointer in d3 slot. thread function pointer is placed in context.start.
 * See load_context for what happens when thread is initially going to
 * run.
 */
#define THREAD_STARTUP_INIT(core, thread, function) \
    ({ (thread)->context.macsr = EMAC_FRACTIONAL | EMAC_SATURATE, \
       (thread)->context.d[0] = (uint32_t)&(thread)->context, \
       (thread)->context.d[1] = (uint32_t)start_thread,       \
       (thread)->context.start = (uint32_t)(function); })

/*---------------------------------------------------------------------------
 * Store non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void store_context(void* addr)
{
    asm volatile (
        "move.l  %%macsr,%%d0                  \n"
        "movem.l %%d0/%%d2-%%d7/%%a2-%%a7,(%0) \n"
        : : "a" (addr) : "d0" /* only! */
    );
}

/*---------------------------------------------------------------------------
 * Load non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void load_context(const void* addr)
{
    asm volatile (
        "move.l  52(%0), %%d0                   \n"  /* Get start address */
        "beq.b   1f                             \n"  /* NULL -> already running */
        "movem.l (%0), %%a0-%%a2                \n"  /* a0=macsr, a1=context, a2=start_thread */
        "jmp     (%%a2)                         \n"  /* Start the thread */
    "1:                                         \n"
        "movem.l (%0), %%d0/%%d2-%%d7/%%a2-%%a7 \n"  /* Load context */
        "move.l  %%d0, %%macsr                  \n"
        : : "a" (addr) : "d0" /* only! */
    );
}

/*---------------------------------------------------------------------------
 * Put core in a power-saving state if waking list wasn't repopulated.
 *---------------------------------------------------------------------------
 */
static inline void core_sleep(void)
{
    /* Supervisor mode, interrupts enabled upon wakeup */
    asm volatile ("stop #0x2000");
};

#elif CONFIG_CPU == SH7034
/*---------------------------------------------------------------------------
 * Start the thread running and terminate it if it returns
 *---------------------------------------------------------------------------
 */
void start_thread(void); /* Provide C access to ASM label */
static void __attribute__((used)) __start_thread(void)
{
    /* r8 = context */
    asm volatile (
    "_start_thread:            \n" /* Start here - no naked attribute */
        "mov.l  @(4, r8), r0   \n" /* Fetch thread function pointer */
        "mov.l  @(28, r8), r15 \n" /* Set initial sp */
        "mov    #0, r1         \n" /* Start the thread */
        "jsr    @r0            \n"
        "mov.l  r1, @(36, r8)  \n" /* Clear start address */
    );
    thread_exit();
}

/* Place context pointer in r8 slot, function pointer in r9 slot, and
 * start_thread pointer in context_start */
#define THREAD_STARTUP_INIT(core, thread, function) \
    ({ (thread)->context.r[0] = (uint32_t)&(thread)->context, \
       (thread)->context.r[1] = (uint32_t)(function),         \
       (thread)->context.start = (uint32_t)start_thread; })

/*---------------------------------------------------------------------------
 * Store non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void store_context(void* addr)
{
    asm volatile (
        "add     #36, %0   \n" /* Start at last reg. By the time routine */
        "sts.l   pr, @-%0  \n" /* is done, %0 will have the original value */
        "mov.l   r15,@-%0  \n"
        "mov.l   r14,@-%0  \n"
        "mov.l   r13,@-%0  \n"
        "mov.l   r12,@-%0  \n"
        "mov.l   r11,@-%0  \n"
        "mov.l   r10,@-%0  \n"
        "mov.l   r9, @-%0  \n"
        "mov.l   r8, @-%0  \n"
        : : "r" (addr)
    );
}

/*---------------------------------------------------------------------------
 * Load non-volatile context.
 *---------------------------------------------------------------------------
 */
static inline void load_context(const void* addr)
{
    asm volatile (
        "mov.l  @(36, %0), r0 \n" /* Get start address */
        "tst    r0, r0        \n"
        "bt     .running      \n" /* NULL -> already running */
        "jmp    @r0           \n" /* r8 = context */
    ".running:                \n"
        "mov.l  @%0+, r8      \n" /* Executes in delay slot and outside it */
        "mov.l  @%0+, r9      \n"
        "mov.l  @%0+, r10     \n"
        "mov.l  @%0+, r11     \n"
        "mov.l  @%0+, r12     \n"
        "mov.l  @%0+, r13     \n"
        "mov.l  @%0+, r14     \n"
        "mov.l  @%0+, r15     \n"
        "lds.l  @%0+, pr      \n"
        : : "r" (addr) : "r0" /* only! */
    );
}

/*---------------------------------------------------------------------------
 * Put core in a power-saving state.
 *---------------------------------------------------------------------------
 */
static inline void core_sleep(void)
{
    asm volatile (
        "and.b  #0x7f, @(r0, gbr) \n" /* Clear SBY (bit 7) in SBYCR */
        "mov    #0, r1            \n" /* Enable interrupts */
        "ldc    r1, sr            \n" /* Following instruction cannot be interrupted */
        "sleep                    \n" /* Execute standby */
        : : "z"(&SBYCR-GBR) : "r1");
}

#endif /* CONFIG_CPU == */

/*
 * End Processor-specific section
 ***************************************************************************/

#if THREAD_EXTRA_CHECKS
static void thread_panicf(const char *msg, struct thread_entry *thread)
{
    IF_COP( const unsigned int core = thread->core; )
    static char name[32];
    thread_get_name(name, 32, thread);
    panicf ("%s %s" IF_COP(" (%d)"), msg, name IF_COP(, core));
}
static void thread_stkov(struct thread_entry *thread)
{
    thread_panicf("Stkov", thread);
}
#define THREAD_PANICF(msg, thread) \
    thread_panicf(msg, thread)
#define THREAD_ASSERT(exp, msg, thread) \
    ({ if (!({ exp; })) thread_panicf((msg), (thread)); })
#else
static void thread_stkov(struct thread_entry *thread)
{
    IF_COP( const unsigned int core = thread->core; )
    static char name[32];
    thread_get_name(name, 32, thread);
    panicf("Stkov %s" IF_COP(" (%d)"), name IF_COP(, core));
}
#define THREAD_PANICF(msg, thread)
#define THREAD_ASSERT(exp, msg, thread)
#endif /* THREAD_EXTRA_CHECKS */

/* Thread locking */
#if NUM_CORES > 1
#define LOCK_THREAD(thread) \
    ({ corelock_lock(&(thread)->slot_cl); })
#define TRY_LOCK_THREAD(thread) \
    ({ corelock_try_lock(&thread->slot_cl); })
#define UNLOCK_THREAD(thread) \
    ({ corelock_unlock(&(thread)->slot_cl); })
#define UNLOCK_THREAD_AT_TASK_SWITCH(thread) \
    ({ unsigned int _core = (thread)->core; \
       cores[_core].blk_ops.flags |= TBOP_UNLOCK_CORELOCK; \
       cores[_core].blk_ops.cl_p = &(thread)->slot_cl; })
#else
#define LOCK_THREAD(thread) \
    ({ })
#define TRY_LOCK_THREAD(thread) \
    ({ })
#define UNLOCK_THREAD(thread) \
    ({ })
#define UNLOCK_THREAD_AT_TASK_SWITCH(thread) \
    ({ })
#endif

/* RTR list */
#define RTR_LOCK(core) \
    ({ corelock_lock(&cores[core].rtr_cl); })
#define RTR_UNLOCK(core) \
    ({ corelock_unlock(&cores[core].rtr_cl); })

#ifdef HAVE_PRIORITY_SCHEDULING
#define rtr_add_entry(core, priority) \
    prio_add_entry(&cores[core].rtr, (priority))

#define rtr_subtract_entry(core, priority) \
    prio_subtract_entry(&cores[core].rtr, (priority))

#define rtr_move_entry(core, from, to) \
    prio_move_entry(&cores[core].rtr, (from), (to))
#else
#define rtr_add_entry(core, priority)
#define rtr_add_entry_inl(core, priority)
#define rtr_subtract_entry(core, priority)
#define rtr_subtract_entry_inl(core, priotity)
#define rtr_move_entry(core, from, to)
#define rtr_move_entry_inl(core, from, to)
#endif

/*---------------------------------------------------------------------------
 * Thread list structure - circular:
 *    +------------------------------+
 *    |                              |
 *    +--+---+<-+---+<-+---+<-+---+<-+
 * Head->| T |  | T |  | T |  | T |
 *    +->+---+->+---+->+---+->+---+--+
 *    |                              |
 *    +------------------------------+
 *---------------------------------------------------------------------------
 */

/*---------------------------------------------------------------------------
 * Adds a thread to a list of threads using "insert last". Uses the "l"
 * links.
 *---------------------------------------------------------------------------
 */
static void add_to_list_l(struct thread_entry **list,
                          struct thread_entry *thread)
{
    struct thread_entry *l = *list;

    if (l == NULL)
    {
        /* Insert into unoccupied list */
        thread->l.prev = thread;
        thread->l.next = thread;
        *list = thread;
        return;
    }

    /* Insert last */
    thread->l.prev = l->l.prev;
    thread->l.next = l;
    l->l.prev->l.next = thread;
    l->l.prev = thread;
}

/*---------------------------------------------------------------------------
 * Removes a thread from a list of threads. Uses the "l" links.
 *---------------------------------------------------------------------------
 */
static void remove_from_list_l(struct thread_entry **list,
                               struct thread_entry *thread)
{
    struct thread_entry *prev, *next;

    next = thread->l.next;

    if (thread == next)
    {
        /* The only item */
        *list = NULL;
        return;
    }

    if (thread == *list)
    {
        /* List becomes next item */
        *list = next;
    }

    prev = thread->l.prev;
    
    /* Fix links to jump over the removed entry. */
    next->l.prev = prev;
    prev->l.next = next;
}

/*---------------------------------------------------------------------------
 * Timeout list structure - circular reverse (to make "remove item" O(1)),
 * NULL-terminated forward (to ease the far more common forward traversal):
 *    +------------------------------+
 *    |                              |
 *    +--+---+<-+---+<-+---+<-+---+<-+
 * Head->| T |  | T |  | T |  | T |
 *       +---+->+---+->+---+->+---+-X
 *---------------------------------------------------------------------------
 */

/*---------------------------------------------------------------------------
 * Add a thread from the core's timout list by linking the pointers in its
 * tmo structure.
 *---------------------------------------------------------------------------
 */
static void add_to_list_tmo(struct thread_entry *thread)
{
    struct thread_entry *tmo = cores[IF_COP_CORE(thread->core)].timeout;
    THREAD_ASSERT(thread->tmo.prev == NULL,
                  "add_to_list_tmo->already listed", thread);

    thread->tmo.next = NULL;

    if (tmo == NULL)
    {
        /* Insert into unoccupied list */
        thread->tmo.prev = thread;
        cores[IF_COP_CORE(thread->core)].timeout = thread;
        return;
    }

    /* Insert Last */
    thread->tmo.prev = tmo->tmo.prev;
    tmo->tmo.prev->tmo.next = thread;
    tmo->tmo.prev = thread;
}

/*---------------------------------------------------------------------------
 * Remove a thread from the core's timout list by unlinking the pointers in
 * its tmo structure. Sets thread->tmo.prev to NULL to indicate the timeout
 * is cancelled.
 *---------------------------------------------------------------------------
 */
static void remove_from_list_tmo(struct thread_entry *thread)
{
    struct thread_entry **list = &cores[IF_COP_CORE(thread->core)].timeout;
    struct thread_entry *prev = thread->tmo.prev;
    struct thread_entry *next = thread->tmo.next;

    THREAD_ASSERT(prev != NULL, "remove_from_list_tmo->not listed", thread);

    if (next != NULL)
        next->tmo.prev = prev;

    if (thread == *list)
    {
        /* List becomes next item and empty if next == NULL */
        *list = next;
        /* Mark as unlisted */
        thread->tmo.prev = NULL;
    }
    else
    {
        if (next == NULL)
            (*list)->tmo.prev = prev;
        prev->tmo.next = next;
        /* Mark as unlisted */
        thread->tmo.prev = NULL;
    }
}


#ifdef HAVE_PRIORITY_SCHEDULING
/*---------------------------------------------------------------------------
 * Priority distribution structure (one category for each possible priority):
 *
 *       +----+----+----+ ... +-----+
 * hist: | F0 | F1 | F2 |     | F31 |
 *       +----+----+----+ ... +-----+
 * mask: | b0 | b1 | b2 |     | b31 |
 *       +----+----+----+ ... +-----+
 *
 * F = count of threads at priority category n (frequency)
 * b = bitmask of non-zero priority categories (occupancy)
 *
 *        / if H[n] != 0 : 1
 * b[n] = |
 *        \ else         : 0 
 *
 *---------------------------------------------------------------------------
 * Basic priority inheritance priotocol (PIP):
 *
 * Mn = mutex n, Tn = thread n
 *
 * A lower priority thread inherits the priority of the highest priority
 * thread blocked waiting for it to complete an action (such as release a
 * mutex or respond to a message via queue_send):
 *
 * 1) T2->M1->T1
 *
 * T1 owns M1, T2 is waiting for M1 to realease M1. If T2 has a higher
 * priority than T1 then T1 inherits the priority of T2.
 *
 * 2) T3
 *    \/
 *    T2->M1->T1
 *
 * Situation is like 1) but T2 and T3 are both queued waiting for M1 and so
 * T1 inherits the higher of T2 and T3.
 *
 * 3) T3->M2->T2->M1->T1
 *
 * T1 owns M1, T2 owns M2. If T3 has a higher priority than both T1 and T2,
 * then T1 inherits the priority of T3 through T2.
 *
 * Blocking chains can grow arbitrarily complex (though it's best that they
 * not form at all very often :) and build-up from these units.
 *---------------------------------------------------------------------------
 */

/*---------------------------------------------------------------------------
 * Increment frequency at category "priority"
 *---------------------------------------------------------------------------
 */
static inline unsigned int prio_add_entry(
    struct priority_distribution *pd, int priority)
{
    unsigned int count;
    /* Enough size/instruction count difference for ARM makes it worth it to
     * use different code (192 bytes for ARM). Only thing better is ASM. */
#ifdef CPU_ARM
    count = pd->hist[priority];
    if (++count == 1)
        pd->mask |= 1 << priority;
    pd->hist[priority] = count;
#else /* This one's better for Coldfire */
    if ((count = ++pd->hist[priority]) == 1)
        pd->mask |= 1 << priority;
#endif

    return count;
}

/*---------------------------------------------------------------------------
 * Decrement frequency at category "priority"
 *---------------------------------------------------------------------------
 */
static inline unsigned int prio_subtract_entry(
    struct priority_distribution *pd, int priority)
{
    unsigned int count;

#ifdef CPU_ARM
    count = pd->hist[priority];
    if (--count == 0)
        pd->mask &= ~(1 << priority);
    pd->hist[priority] = count;
#else
    if ((count = --pd->hist[priority]) == 0)
        pd->mask &= ~(1 << priority);
#endif

    return count;
}

/*---------------------------------------------------------------------------
 * Remove from one category and add to another
 *---------------------------------------------------------------------------
 */
static inline void prio_move_entry(
    struct priority_distribution *pd, int from, int to)
{
    uint32_t mask = pd->mask;

#ifdef CPU_ARM
    unsigned int count;

    count = pd->hist[from];
    if (--count == 0)
        mask &= ~(1 << from);
    pd->hist[from] = count;

    count = pd->hist[to];
    if (++count == 1)
        mask |= 1 << to;
    pd->hist[to] = count;
#else
    if (--pd->hist[from] == 0)
        mask &= ~(1 << from);

    if (++pd->hist[to] == 1)
        mask |= 1 << to;
#endif

    pd->mask = mask;
}

/*---------------------------------------------------------------------------
 * Change the priority and rtr entry for a running thread
 *---------------------------------------------------------------------------
 */
static inline void set_running_thread_priority(
    struct thread_entry *thread, int priority)
{
    const unsigned int core = IF_COP_CORE(thread->core);
    RTR_LOCK(core);
    rtr_move_entry(core, thread->priority, priority);
    thread->priority = priority;
    RTR_UNLOCK(core);
}

/*---------------------------------------------------------------------------
 * Finds the highest priority thread in a list of threads. If the list is
 * empty, the PRIORITY_IDLE is returned.
 *
 * It is possible to use the struct priority_distribution within an object
 * instead of scanning the remaining threads in the list but as a compromise,
 * the resulting per-object memory overhead is saved at a slight speed
 * penalty under high contention.
 *---------------------------------------------------------------------------
 */
static int find_highest_priority_in_list_l(
    struct thread_entry * const thread)
{
    if (thread != NULL)
    {
        /* Go though list until the ending up at the initial thread */
        int highest_priority = thread->priority;
        struct thread_entry *curr = thread;

        do
        {
            int priority = curr->priority;

            if (priority < highest_priority)
                highest_priority = priority;

            curr = curr->l.next;
        }
        while (curr != thread);

        return highest_priority;
    }

    return PRIORITY_IDLE;
}

/*---------------------------------------------------------------------------
 * Register priority with blocking system and bubble it down the chain if
 * any until we reach the end or something is already equal or higher.
 *
 * NOTE: A simultaneous circular wait could spin deadlock on multiprocessor
 * targets but that same action also guarantees a circular block anyway and
 * those are prevented, right? :-)
 *---------------------------------------------------------------------------
 */
static struct thread_entry *
    blocker_inherit_priority(struct thread_entry *current)
{
    const int priority = current->priority;
    struct blocker *bl = current->blocker;
    struct thread_entry * const tstart = current;
    struct thread_entry *bl_t = bl->thread;

    /* Blocker cannot change since the object protection is held */
    LOCK_THREAD(bl_t);

    for (;;)
    {
        struct thread_entry *next;
        int bl_pr = bl->priority;

        if (priority >= bl_pr)
            break; /* Object priority already high enough */

        bl->priority = priority;

        /* Add this one */
        prio_add_entry(&bl_t->pdist, priority);

        if (bl_pr < PRIORITY_IDLE)
        {
            /* Not first waiter - subtract old one */
            prio_subtract_entry(&bl_t->pdist, bl_pr);
        }

        if (priority >= bl_t->priority)
            break; /* Thread priority high enough */

        if (bl_t->state == STATE_RUNNING)
        {
            /* Blocking thread is a running thread therefore there are no
             * further blockers. Change the "run queue" on which it
             * resides. */
            set_running_thread_priority(bl_t, priority);
            break;
        }

        bl_t->priority = priority;

        /* If blocking thread has a blocker, apply transitive inheritance */
        bl = bl_t->blocker;

        if (bl == NULL)
            break; /* End of chain or object doesn't support inheritance */

        next = bl->thread;

        if (next == tstart)
            break; /* Full-circle - deadlock! */

        UNLOCK_THREAD(current);

#if NUM_CORES > 1
        for (;;)
        {
            LOCK_THREAD(next);

            /* Blocker could change - retest condition */
            if (bl->thread == next)
                break;

            UNLOCK_THREAD(next);
            next = bl->thread;
        }
#endif
        current = bl_t;
        bl_t = next;
    }

    UNLOCK_THREAD(bl_t);

    return current;
}

/*---------------------------------------------------------------------------
 * Readjust priorities when waking a thread blocked waiting for another
 * in essence "releasing" the thread's effect on the object owner. Can be
 * performed from any context.
 *---------------------------------------------------------------------------
 */
struct thread_entry *
    wakeup_priority_protocol_release(struct thread_entry *thread)
{
    const int priority = thread->priority;
    struct blocker *bl = thread->blocker;
    struct thread_entry * const tstart = thread;
    struct thread_entry *bl_t = bl->thread;

    /* Blocker cannot change since object will be locked */
    LOCK_THREAD(bl_t);

    thread->blocker = NULL; /* Thread not blocked */

    for (;;)
    {
        struct thread_entry *next;
        int bl_pr = bl->priority;

        if (priority > bl_pr)
            break; /* Object priority higher */

        next = *thread->bqp;

        if (next == NULL)
        {
            /* No more threads in queue */
            prio_subtract_entry(&bl_t->pdist, bl_pr);
            bl->priority = PRIORITY_IDLE;
        }
        else
        {
            /* Check list for highest remaining priority */
            int queue_pr = find_highest_priority_in_list_l(next);

            if (queue_pr == bl_pr)
                break; /* Object priority not changing */

            /* Change queue priority */
            prio_move_entry(&bl_t->pdist, bl_pr, queue_pr);
            bl->priority = queue_pr;
        }

        if (bl_pr > bl_t->priority)
            break; /* thread priority is higher */

        bl_pr = find_first_set_bit(bl_t->pdist.mask);

        if (bl_pr == bl_t->priority)
            break; /* Thread priority not changing */

        if (bl_t->state == STATE_RUNNING)
        {
            /* No further blockers */
            set_running_thread_priority(bl_t, bl_pr);
            break;
        }

        bl_t->priority = bl_pr;

        /* If blocking thread has a blocker, apply transitive inheritance */
        bl = bl_t->blocker;

        if (bl == NULL)
            break; /* End of chain or object doesn't support inheritance */

        next = bl->thread;

        if (next == tstart)
            break; /* Full-circle - deadlock! */

        UNLOCK_THREAD(thread);

#if NUM_CORES > 1
        for (;;)
        {
            LOCK_THREAD(next);

            /* Blocker could change - retest condition */
            if (bl->thread == next)
                break;

            UNLOCK_THREAD(next);
            next = bl->thread;
        }
#endif
        thread = bl_t;
        bl_t = next;
    }

    UNLOCK_THREAD(bl_t);

#if NUM_CORES > 1
    if (thread != tstart)
    {
        /* Relock original if it changed */
        LOCK_THREAD(tstart);
    }
#endif

    return cores[CURRENT_CORE].running;
}

/*---------------------------------------------------------------------------
 * Transfer ownership to a thread waiting for an objects and transfer
 * inherited priority boost from other waiters. This algorithm knows that
 * blocking chains may only unblock from the very end.
 *
 * Only the owning thread itself may call this and so the assumption that
 * it is the running thread is made.
 *---------------------------------------------------------------------------
 */
struct thread_entry *
    wakeup_priority_protocol_transfer(struct thread_entry *thread)
{
    /* Waking thread inherits priority boost from object owner */
    struct blocker *bl = thread->blocker;
    struct thread_entry *bl_t = bl->thread;
    struct thread_entry *next;
    int bl_pr;

    THREAD_ASSERT(thread_get_current() == bl_t,
                  "UPPT->wrong thread", thread_get_current());

    LOCK_THREAD(bl_t);

    bl_pr = bl->priority;

    /* Remove the object's boost from the owning thread */
    if (prio_subtract_entry(&bl_t->pdist, bl_pr) == 0 &&
        bl_pr <= bl_t->priority)
    {
        /* No more threads at this priority are waiting and the old level is
         * at least the thread level */
        int priority = find_first_set_bit(bl_t->pdist.mask);

        if (priority != bl_t->priority)
        {
            /* Adjust this thread's priority */
            set_running_thread_priority(bl_t, priority);
        }
    }

    next = *thread->bqp;

    if (next == NULL)
    {
        /* Expected shortcut - no more waiters */
        bl_pr = PRIORITY_IDLE;
    }
    else
    {
        if (thread->priority <= bl_pr)
        {
            /* Need to scan threads remaining in queue */
            bl_pr = find_highest_priority_in_list_l(next);
        }

        if (prio_add_entry(&thread->pdist, bl_pr) == 1 &&
            bl_pr < thread->priority)
        {
            /* Thread priority must be raised */
            thread->priority = bl_pr;
        }
    }

    bl->thread = thread;    /* This thread pwns */
    bl->priority = bl_pr;   /* Save highest blocked priority */
    thread->blocker = NULL; /* Thread not blocked */

    UNLOCK_THREAD(bl_t);

    return bl_t;
}

/*---------------------------------------------------------------------------
 * No threads must be blocked waiting for this thread except for it to exit.
 * The alternative is more elaborate cleanup and object registration code.
 * Check this for risk of silent data corruption when objects with
 * inheritable blocking are abandoned by the owner - not precise but may
 * catch something.
 *---------------------------------------------------------------------------
 */
void check_for_obj_waiters(const char *function, struct thread_entry *thread)
{
    /* Only one bit in the mask should be set with a frequency on 1 which
     * represents the thread's own base priority */
    uint32_t mask = thread->pdist.mask;
    if ((mask & (mask - 1)) != 0 ||
        thread->pdist.hist[find_first_set_bit(mask)] > 1)
    {
        unsigned char name[32];
        thread_get_name(name, 32, thread);
        panicf("%s->%s with obj. waiters", function, name);
    }
}
#endif /* HAVE_PRIORITY_SCHEDULING */

/*---------------------------------------------------------------------------
 * Move a thread back to a running state on its core.
 *---------------------------------------------------------------------------
 */
static void core_schedule_wakeup(struct thread_entry *thread)
{
    const unsigned int core = IF_COP_CORE(thread->core);

    RTR_LOCK(core);

    thread->state = STATE_RUNNING;

    add_to_list_l(&cores[core].running, thread);
    rtr_add_entry(core, thread->priority);

    RTR_UNLOCK(core);

#if NUM_CORES > 1
    if (core != CURRENT_CORE)
        core_wake(core);
#endif
}

/*---------------------------------------------------------------------------
 * Check the core's timeout list when at least one thread is due to wake.
 * Filtering for the condition is done before making the call. Resets the
 * tick when the next check will occur.
 *---------------------------------------------------------------------------
 */
void check_tmo_threads(void)
{
    const unsigned int core = CURRENT_CORE;
    const long tick = current_tick; /* snapshot the current tick */
    long next_tmo_check = tick + 60*HZ; /* minimum duration: once/minute */
    struct thread_entry *next = cores[core].timeout;

    /* If there are no processes waiting for a timeout, just keep the check
       tick from falling into the past. */

    /* Break the loop once we have walked through the list of all
     * sleeping processes or have removed them all. */
    while (next != NULL)
    {
        /* Check sleeping threads. Allow interrupts between checks. */
        enable_irq();

        struct thread_entry *curr = next;

        next = curr->tmo.next;

        /* Lock thread slot against explicit wakeup */
        disable_irq();
        LOCK_THREAD(curr);

        unsigned state = curr->state;

        if (state < TIMEOUT_STATE_FIRST)
        {
            /* Cleanup threads no longer on a timeout but still on the
             * list. */
            remove_from_list_tmo(curr);
        }
        else if (TIME_BEFORE(tick, curr->tmo_tick))
        {
            /* Timeout still pending - this will be the usual case */
            if (TIME_BEFORE(curr->tmo_tick, next_tmo_check))
            {
                /* Earliest timeout found so far - move the next check up
                   to its time */
                next_tmo_check = curr->tmo_tick;
            }
        }
        else
        {
            /* Sleep timeout has been reached so bring the thread back to
             * life again. */
            if (state == STATE_BLOCKED_W_TMO)
            {
#if NUM_CORES > 1
                /* Lock the waiting thread's kernel object */
                struct corelock *ocl = curr->obj_cl;

                if (corelock_try_lock(ocl) == 0)
                {
                    /* Need to retry in the correct order though the need is
                     * unlikely */
                    UNLOCK_THREAD(curr);
                    corelock_lock(ocl);
                    LOCK_THREAD(curr);

                    if (curr->state != STATE_BLOCKED_W_TMO)
                    {
                        /* Thread was woken or removed explicitely while slot
                         * was unlocked */
                        corelock_unlock(ocl);
                        remove_from_list_tmo(curr);
                        UNLOCK_THREAD(curr);
                        continue;
                    }
                }
#endif /* NUM_CORES */

                remove_from_list_l(curr->bqp, curr);

#ifdef HAVE_WAKEUP_EXT_CB
                if (curr->wakeup_ext_cb != NULL)
                    curr->wakeup_ext_cb(curr);
#endif

#ifdef HAVE_PRIORITY_SCHEDULING
                if (curr->blocker != NULL)
                    wakeup_priority_protocol_release(curr);
#endif
                corelock_unlock(ocl);
            }
            /* else state == STATE_SLEEPING */

            remove_from_list_tmo(curr);

            RTR_LOCK(core);

            curr->state = STATE_RUNNING;

            add_to_list_l(&cores[core].running, curr);
            rtr_add_entry(core, curr->priority);

            RTR_UNLOCK(core);
        }

        UNLOCK_THREAD(curr);
    }

    cores[core].next_tmo_check = next_tmo_check;
}

/*---------------------------------------------------------------------------
 * Performs operations that must be done before blocking a thread but after
 * the state is saved.
 *---------------------------------------------------------------------------
 */
#if NUM_CORES > 1
static inline void run_blocking_ops(
    unsigned int core, struct thread_entry *thread)
{
    struct thread_blk_ops *ops = &cores[core].blk_ops;
    const unsigned flags = ops->flags;

    if (flags == TBOP_CLEAR)
        return;

    switch (flags)
    {
    case TBOP_SWITCH_CORE:
        core_switch_blk_op(core, thread);
        /* Fall-through */
    case TBOP_UNLOCK_CORELOCK:
        corelock_unlock(ops->cl_p);
        break;
    }

    ops->flags = TBOP_CLEAR;
}
#endif /* NUM_CORES > 1 */

#ifdef RB_PROFILE
void profile_thread(void)
{
    profstart(cores[CURRENT_CORE].running - threads);
}
#endif

/*---------------------------------------------------------------------------
 * Prepares a thread to block on an object's list and/or for a specified
 * duration - expects object and slot to be appropriately locked if needed
 * and interrupts to be masked.
 *---------------------------------------------------------------------------
 */
static inline void block_thread_on_l(struct thread_entry *thread,
                                     unsigned state)
{
    /* If inlined, unreachable branches will be pruned with no size penalty
       because state is passed as a constant parameter. */
    const unsigned int core = IF_COP_CORE(thread->core);

    /* Remove the thread from the list of running threads. */
    RTR_LOCK(core);
    remove_from_list_l(&cores[core].running, thread);
    rtr_subtract_entry(core, thread->priority);
    RTR_UNLOCK(core);

    /* Add a timeout to the block if not infinite */
    switch (state)
    {
    case STATE_BLOCKED:
    case STATE_BLOCKED_W_TMO:
        /* Put the thread into a new list of inactive threads. */
        add_to_list_l(thread->bqp, thread);

        if (state == STATE_BLOCKED)
            break;

        /* Fall-through */
    case STATE_SLEEPING:
        /* If this thread times out sooner than any other thread, update
           next_tmo_check to its timeout */
        if (TIME_BEFORE(thread->tmo_tick, cores[core].next_tmo_check))
        {
            cores[core].next_tmo_check = thread->tmo_tick;
        }

        if (thread->tmo.prev == NULL)
        {
            add_to_list_tmo(thread);
        }
        /* else thread was never removed from list - just keep it there */
        break;
    }

    /* Remember the the next thread about to block. */
    cores[core].block_task = thread;

    /* Report new state. */
    thread->state = state;
}

/*---------------------------------------------------------------------------
 * Switch thread in round robin fashion for any given priority. Any thread
 * that removed itself from the running list first must specify itself in
 * the paramter.
 *
 * INTERNAL: Intended for use by kernel and not for programs.
 *---------------------------------------------------------------------------
 */
void switch_thread(void)
{
    const unsigned int core = CURRENT_CORE;
    struct thread_entry *block = cores[core].block_task;
    struct thread_entry *thread = cores[core].running;

    /* Get context to save - next thread to run is unknown until all wakeups
     * are evaluated */
    if (block != NULL)
    {
        cores[core].block_task = NULL;

#if NUM_CORES > 1
        if (thread == block)
        {
            /* This was the last thread running and another core woke us before
             * reaching here. Force next thread selection to give tmo threads or
             * other threads woken before this block a first chance. */
            block = NULL;
        }
        else
#endif
        {
            /* Blocking task is the old one */
            thread = block;
        }
    }

#ifdef RB_PROFILE
    profile_thread_stopped(thread - threads);
#endif

    /* Begin task switching by saving our current context so that we can
     * restore the state of the current thread later to the point prior
     * to this call. */
    store_context(&thread->context);

    /* Check if the current thread stack is overflown */
    if (thread->stack[0] != DEADBEEF)
        thread_stkov(thread);

#if NUM_CORES > 1
    /* Run any blocking operations requested before switching/sleeping */
    run_blocking_ops(core, thread);
#endif

#ifdef HAVE_PRIORITY_SCHEDULING
    IF_NO_SKIP_YIELD( if (thread->skip_count != -1) )
    /* Reset the value of thread's skip count */
        thread->skip_count = 0;
#endif

    for (;;)
    {
        /* If there are threads on a timeout and the earliest wakeup is due,
         * check the list and wake any threads that need to start running
         * again. */
        if (!TIME_BEFORE(current_tick, cores[core].next_tmo_check))
        {
            check_tmo_threads();
        }

        disable_irq();
        RTR_LOCK(core);

        thread = cores[core].running;

        if (thread == NULL)
        {
            /* Enter sleep mode to reduce power usage - woken up on interrupt
             * or wakeup request from another core - expected to enable
             * interrupts. */
            RTR_UNLOCK(core);
            core_sleep(IF_COP(core));
        }
        else
        {
#ifdef HAVE_PRIORITY_SCHEDULING
            /* Select the new task based on priorities and the last time a
             * process got CPU time relative to the highest priority runnable
             * task. */
            struct priority_distribution *pd = &cores[core].rtr;
            int max = find_first_set_bit(pd->mask);

            if (block == NULL)
            {
                /* Not switching on a block, tentatively select next thread */
                thread = thread->l.next;
            }

            for (;;)
            {
                int priority = thread->priority;
                int diff;

                /* This ridiculously simple method of aging seems to work
                 * suspiciously well. It does tend to reward CPU hogs (under
                 * yielding) but that's generally not desirable at all. On the
                 * plus side, it, relatively to other threads, penalizes excess
                 * yielding which is good if some high priority thread is
                 * performing no useful work such as polling for a device to be
                 * ready. Of course, aging is only employed when higher and lower
                 * priority threads are runnable. The highest priority runnable
                 * thread(s) are never skipped. */
                if (priority <= max ||
                    IF_NO_SKIP_YIELD( thread->skip_count == -1 || )
                    (diff = priority - max, ++thread->skip_count > diff*diff))
                {
                    cores[core].running = thread;
                    break;
                }

                thread = thread->l.next;
            }
#else
            /* Without priority use a simple FCFS algorithm */
            if (block == NULL)
            {
                /* Not switching on a block, select next thread */
                thread = thread->l.next;
                cores[core].running = thread;
            }
#endif /* HAVE_PRIORITY_SCHEDULING */

            RTR_UNLOCK(core);
            enable_irq();
            break;
        }
    }

    /* And finally give control to the next thread. */
    load_context(&thread->context);

#ifdef RB_PROFILE
    profile_thread_started(thread - threads);
#endif
}

/*---------------------------------------------------------------------------
 * Sleeps a thread for at least a specified number of ticks with zero being
 * a wait until the next tick.
 *
 * INTERNAL: Intended for use by kernel and not for programs.
 *---------------------------------------------------------------------------
 */
void sleep_thread(int ticks)
{
    struct thread_entry *current = cores[CURRENT_CORE].running;

    LOCK_THREAD(current);

    /* Set our timeout, remove from run list and join timeout list. */
    current->tmo_tick = current_tick + ticks + 1;
    block_thread_on_l(current, STATE_SLEEPING);

    UNLOCK_THREAD(current);
}

/*---------------------------------------------------------------------------
 * Indefinitely block a thread on a blocking queue for explicit wakeup.
 *
 * INTERNAL: Intended for use by kernel objects and not for programs.
 *---------------------------------------------------------------------------
 */
void block_thread(struct thread_entry *current)
{
    /* Set the state to blocked and take us off of the run queue until we
     * are explicitly woken */
    LOCK_THREAD(current);

    /* Set the list for explicit wakeup */
    block_thread_on_l(current, STATE_BLOCKED);

#ifdef HAVE_PRIORITY_SCHEDULING
    if (current->blocker != NULL)
    {
        /* Object supports PIP */
        current = blocker_inherit_priority(current);
    }
#endif

    UNLOCK_THREAD(current);
}

/*---------------------------------------------------------------------------
 * Block a thread on a blocking queue for a specified time interval or until
 * explicitly woken - whichever happens first.
 *
 * INTERNAL: Intended for use by kernel objects and not for programs.
 *---------------------------------------------------------------------------
 */
void block_thread_w_tmo(struct thread_entry *current, int timeout)
{
    /* Get the entry for the current running thread. */
    LOCK_THREAD(current);

    /* Set the state to blocked with the specified timeout */
    current->tmo_tick = current_tick + timeout;

    /* Set the list for explicit wakeup */
    block_thread_on_l(current, STATE_BLOCKED_W_TMO);

#ifdef HAVE_PRIORITY_SCHEDULING
    if (current->blocker != NULL)
    {
        /* Object supports PIP */
        current = blocker_inherit_priority(current);
    }
#endif

    UNLOCK_THREAD(current);
}

/*---------------------------------------------------------------------------
 * Explicitly wakeup a thread on a blocking queue. Only effects threads of
 * STATE_BLOCKED and STATE_BLOCKED_W_TMO.
 *
 * This code should be considered a critical section by the caller meaning
 * that the object's corelock should be held.
 *
 * INTERNAL: Intended for use by kernel objects and not for programs.
 *---------------------------------------------------------------------------
 */
unsigned int wakeup_thread(struct thread_entry **list)
{
    struct thread_entry *thread = *list;
    unsigned int result = THREAD_NONE;

    /* Check if there is a blocked thread at all. */
    if (thread == NULL)
        return result;

    LOCK_THREAD(thread);

    /* Determine thread's current state. */
    switch (thread->state)
    {
    case STATE_BLOCKED:
    case STATE_BLOCKED_W_TMO:
        remove_from_list_l(list, thread);

        result = THREAD_OK;

#ifdef HAVE_PRIORITY_SCHEDULING
        struct thread_entry *current;
        struct blocker *bl = thread->blocker;

        if (bl == NULL)
        {
            /* No inheritance - just boost the thread by aging */
            IF_NO_SKIP_YIELD( if (thread->skip_count != -1) )
                thread->skip_count = thread->priority;
            current = cores[CURRENT_CORE].running;
        }
        else
        {
            /* Call the specified unblocking PIP */
            current = bl->wakeup_protocol(thread);
        }

        if (current != NULL && thread->priority < current->priority
            IF_COP( && thread->core == current->core ))
        {
            /* Woken thread is higher priority and exists on the same CPU core;
             * recommend a task switch. Knowing if this is an interrupt call
             * would be helpful here. */
            result |= THREAD_SWITCH;
        }
#endif /* HAVE_PRIORITY_SCHEDULING */

        core_schedule_wakeup(thread);
        break;

    /* Nothing to do. State is not blocked. */
#if THREAD_EXTRA_CHECKS
    default:
        THREAD_PANICF("wakeup_thread->block invalid", thread);
    case STATE_RUNNING:
    case STATE_KILLED:
        break;
#endif
    }

    UNLOCK_THREAD(thread);
    return result;
}

/*---------------------------------------------------------------------------
 * Wakeup an entire queue of threads - returns bitwise-or of return bitmask
 * from each operation or THREAD_NONE of nothing was awakened. Object owning
 * the queue must be locked first.
 *
 * INTERNAL: Intended for use by kernel objects and not for programs.
 *---------------------------------------------------------------------------
 */
unsigned int thread_queue_wake(struct thread_entry **list)
{
    unsigned result = THREAD_NONE;

    for (;;)
    {
        unsigned int rc = wakeup_thread(list);

        if (rc == THREAD_NONE)
            break; /* No more threads */

        result |= rc;
    }

    return result;
}

/*---------------------------------------------------------------------------
 * Find an empty thread slot or MAXTHREADS if none found. The slot returned
 * will be locked on multicore.
 *---------------------------------------------------------------------------
 */
static struct thread_entry * find_empty_thread_slot(void)
{
    /* Any slot could be on an interrupt-accessible list */
    IF_COP( int oldlevel = disable_irq_save(); )
    struct thread_entry *thread = NULL;
    int n;

    for (n = 0; n < MAXTHREADS; n++)
    {
        /* Obtain current slot state - lock it on multicore */
        struct thread_entry *t = &threads[n];
        LOCK_THREAD(t);

        if (t->state == STATE_KILLED IF_COP( && t->name != THREAD_DESTRUCT ))
        {
            /* Slot is empty - leave it locked and caller will unlock */
            thread = t;
            break;
        }

        /* Finished examining slot - no longer busy - unlock on multicore */
        UNLOCK_THREAD(t);
    }

    IF_COP( restore_irq(oldlevel); ) /* Reenable interrups - this slot is
                                          not accesible to them yet */
    return thread;
}


/*---------------------------------------------------------------------------
 * Place the current core in idle mode - woken up on interrupt or wake
 * request from another core.
 *---------------------------------------------------------------------------
 */
void core_idle(void)
{
    IF_COP( const unsigned int core = CURRENT_CORE; )
    disable_irq();
    core_sleep(IF_COP(core));
}

/*---------------------------------------------------------------------------
 * Create a thread. If using a dual core architecture, specify which core to
 * start the thread on.
 *
 * Return ID if context area could be allocated, else NULL.
 *---------------------------------------------------------------------------
 */
struct thread_entry* 
    create_thread(void (*function)(void), void* stack, size_t stack_size,
                  unsigned flags, const char *name
                  IF_PRIO(, int priority)
                  IF_COP(, unsigned int core))
{
    unsigned int i;
    unsigned int stack_words;
    uintptr_t stackptr, stackend;
    struct thread_entry *thread;
    unsigned state;
    int oldlevel;

    thread = find_empty_thread_slot();
    if (thread == NULL)
    {
        return NULL;
    }

    oldlevel = disable_irq_save();

    /* Munge the stack to make it easy to spot stack overflows */
    stackptr = ALIGN_UP((uintptr_t)stack, sizeof (uintptr_t));
    stackend = ALIGN_DOWN((uintptr_t)stack + stack_size, sizeof (uintptr_t));
    stack_size = stackend - stackptr;
    stack_words = stack_size / sizeof (uintptr_t);

    for (i = 0; i < stack_words; i++)
    {
        ((uintptr_t *)stackptr)[i] = DEADBEEF;
    }

    /* Store interesting information */
    thread->name = name;
    thread->stack = (uintptr_t *)stackptr;
    thread->stack_size = stack_size;
    thread->queue = NULL;
#ifdef HAVE_WAKEUP_EXT_CB
    thread->wakeup_ext_cb = NULL;
#endif
#ifdef HAVE_SCHEDULER_BOOSTCTRL
    thread->cpu_boost = 0;
#endif
#ifdef HAVE_PRIORITY_SCHEDULING
    memset(&thread->pdist, 0, sizeof(thread->pdist));
    thread->blocker = NULL;
    thread->base_priority = priority;
    thread->priority = priority;
    thread->skip_count = priority;
    prio_add_entry(&thread->pdist, priority);
#endif

#if NUM_CORES > 1
    thread->core = core;

    /* Writeback stack munging or anything else before starting */
    if (core != CURRENT_CORE)
    {
        flush_icache();
    }
#endif

    /* Thread is not on any timeout list but be a bit paranoid */
    thread->tmo.prev = NULL;

    state = (flags & CREATE_THREAD_FROZEN) ?
        STATE_FROZEN : STATE_RUNNING;
    
    thread->context.sp = (typeof (thread->context.sp))stackend;

    /* Load the thread's context structure with needed startup information */
    THREAD_STARTUP_INIT(core, thread, function);

    thread->state = state;

    if (state == STATE_RUNNING)
        core_schedule_wakeup(thread);

    UNLOCK_THREAD(thread);

    restore_irq(oldlevel);

    return thread;
}

#ifdef HAVE_SCHEDULER_BOOSTCTRL
/*---------------------------------------------------------------------------
 * Change the boost state of a thread boosting or unboosting the CPU
 * as required.
 *---------------------------------------------------------------------------
 */
static inline void boost_thread(struct thread_entry *thread, bool boost)
{
    if ((thread->cpu_boost != 0) != boost)
    {
        thread->cpu_boost = boost;
        cpu_boost(boost);
    }
}

void trigger_cpu_boost(void)
{
    struct thread_entry *current = cores[CURRENT_CORE].running;
    boost_thread(current, true);
}

void cancel_cpu_boost(void)
{
    struct thread_entry *current = cores[CURRENT_CORE].running;
    boost_thread(current, false);
}
#endif /* HAVE_SCHEDULER_BOOSTCTRL */

/*---------------------------------------------------------------------------
 * Block the current thread until another thread terminates. A thread may
 * wait on itself to terminate which prevents it from running again and it
 * will need to be killed externally.
 * Parameter is the ID as returned from create_thread().
 *---------------------------------------------------------------------------
 */
void thread_wait(struct thread_entry *thread)
{
    struct thread_entry *current = cores[CURRENT_CORE].running;

    if (thread == NULL)
        thread = current;

    /* Lock thread-as-waitable-object lock */
    corelock_lock(&thread->waiter_cl);

    /* Be sure it hasn't been killed yet */
    if (thread->state != STATE_KILLED)
    {
        IF_COP( current->obj_cl = &thread->waiter_cl; )
        current->bqp = &thread->queue;

        disable_irq();
        block_thread(current);

        corelock_unlock(&thread->waiter_cl);

        switch_thread();
        return;
    }

    corelock_unlock(&thread->waiter_cl);
}

/*---------------------------------------------------------------------------
 * Exit the current thread. The Right Way to Do Things (TM).
 *---------------------------------------------------------------------------
 */
void thread_exit(void)
{
    const unsigned int core = CURRENT_CORE;
    struct thread_entry *current = cores[core].running;

    /* Cancel CPU boost if any */
    cancel_cpu_boost();

    disable_irq();

    corelock_lock(&current->waiter_cl);
    LOCK_THREAD(current);

#if defined (ALLOW_REMOVE_THREAD) && NUM_CORES > 1
    if (current->name == THREAD_DESTRUCT)
    {
        /* Thread being killed - become a waiter */
        UNLOCK_THREAD(current);
        corelock_unlock(&current->waiter_cl);
        thread_wait(current);
        THREAD_PANICF("thread_exit->WK:*R", current);
    }
#endif

#ifdef HAVE_PRIORITY_SCHEDULING
    check_for_obj_waiters("thread_exit", current);
#endif

    if (current->tmo.prev != NULL)
    {
        /* Cancel pending timeout list removal */
        remove_from_list_tmo(current);
    }

    /* Switch tasks and never return */
    block_thread_on_l(current, STATE_KILLED);

#if NUM_CORES > 1
    /* Switch to the idle stack if not on the main core (where "main"
     * runs) - we can hope gcc doesn't need the old stack beyond this
     * point. */
    if (core != CPU)
    {
        switch_to_idle_stack(core);
    }

    flush_icache();
#endif
    current->name = NULL;

    /* Signal this thread */
    thread_queue_wake(&current->queue);
    corelock_unlock(&current->waiter_cl);
    /* Slot must be unusable until thread is really gone */
    UNLOCK_THREAD_AT_TASK_SWITCH(current);
    switch_thread();
    /* This should never and must never be reached - if it is, the
     * state is corrupted */
    THREAD_PANICF("thread_exit->K:*R", current);
}

#ifdef ALLOW_REMOVE_THREAD
/*---------------------------------------------------------------------------
 * Remove a thread from the scheduler. Not The Right Way to Do Things in
 * normal programs.
 *
 * Parameter is the ID as returned from create_thread().
 *
 * Use with care on threads that are not under careful control as this may
 * leave various objects in an undefined state.
 *---------------------------------------------------------------------------
 */
void remove_thread(struct thread_entry *thread)
{
#if NUM_CORES > 1
    /* core is not constant here because of core switching */
    unsigned int core = CURRENT_CORE;
    unsigned int old_core = NUM_CORES;
    struct corelock *ocl = NULL;
#else
    const unsigned int core = CURRENT_CORE;
#endif
    struct thread_entry *current = cores[core].running;

    unsigned state;
    int oldlevel;

    if (thread == NULL)
        thread = current;

    if (thread == current)
        thread_exit(); /* Current thread - do normal exit */

    oldlevel = disable_irq_save();

    corelock_lock(&thread->waiter_cl);
    LOCK_THREAD(thread);

    state = thread->state;

    if (state == STATE_KILLED)
    {
        goto thread_killed;
    }

#if NUM_CORES > 1
    if (thread->name == THREAD_DESTRUCT)
    {
        /* Thread being killed - become a waiter */
        UNLOCK_THREAD(thread);
        corelock_unlock(&thread->waiter_cl);
        restore_irq(oldlevel);
        thread_wait(thread);
        return;
    }

    thread->name = THREAD_DESTRUCT; /* Slot can't be used for now */

#ifdef HAVE_PRIORITY_SCHEDULING
    check_for_obj_waiters("remove_thread", thread);
#endif

    if (thread->core != core)
    {
        /* Switch cores and safely extract the thread there */
        /* Slot HAS to be unlocked or a deadlock could occur which means other
         * threads have to be guided into becoming thread waiters if they
         * attempt to remove it. */
        unsigned int new_core = thread->core;

        corelock_unlock(&thread->waiter_cl);

        UNLOCK_THREAD(thread);
        restore_irq(oldlevel);

        old_core = switch_core(new_core);

        oldlevel = disable_irq_save();

        corelock_lock(&thread->waiter_cl);
        LOCK_THREAD(thread);

        state = thread->state;
        core = new_core;
        /* Perform the extraction and switch ourselves back to the original
           processor */
    }
#endif /* NUM_CORES > 1 */

    if (thread->tmo.prev != NULL)
    {
        /* Clean thread off the timeout list if a timeout check hasn't
         * run yet */
        remove_from_list_tmo(thread);
    }

#ifdef HAVE_SCHEDULER_BOOSTCTRL
    /* Cancel CPU boost if any */
    boost_thread(thread, false);
#endif

IF_COP( retry_state: )

    switch (state)
    {
    case STATE_RUNNING:
        RTR_LOCK(core);
        /* Remove thread from ready to run tasks */
        remove_from_list_l(&cores[core].running, thread);
        rtr_subtract_entry(core, thread->priority);
        RTR_UNLOCK(core);
        break;
    case STATE_BLOCKED:
    case STATE_BLOCKED_W_TMO:
        /* Remove thread from the queue it's blocked on - including its
         * own if waiting there */
#if NUM_CORES > 1
        if (&thread->waiter_cl != thread->obj_cl)
        {
            ocl = thread->obj_cl;

            if (corelock_try_lock(ocl) == 0)
            {
                UNLOCK_THREAD(thread);
                corelock_lock(ocl);
                LOCK_THREAD(thread);

                if (thread->state != state)
                {
                    /* Something woke the thread */
                    state = thread->state;
                    corelock_unlock(ocl);
                    goto retry_state;
                }
            }
        }
#endif
        remove_from_list_l(thread->bqp, thread);

#ifdef HAVE_WAKEUP_EXT_CB
        if (thread->wakeup_ext_cb != NULL)
            thread->wakeup_ext_cb(thread);
#endif

#ifdef HAVE_PRIORITY_SCHEDULING
        if (thread->blocker != NULL)
        {
            /* Remove thread's priority influence from its chain */
            wakeup_priority_protocol_release(thread);
        }
#endif

#if NUM_CORES > 1
        if (ocl != NULL)
            corelock_unlock(ocl);
#endif
        break;
    /* Otherwise thread is frozen and hasn't run yet */
    }

    thread->state = STATE_KILLED;

    /* If thread was waiting on itself, it will have been removed above.
     * The wrong order would result in waking the thread first and deadlocking
     * since the slot is already locked. */
    thread_queue_wake(&thread->queue);

    thread->name = NULL;

thread_killed: /* Thread was already killed */
    /* Removal complete - safe to unlock and reenable interrupts */
    corelock_unlock(&thread->waiter_cl);
    UNLOCK_THREAD(thread);
    restore_irq(oldlevel);

#if NUM_CORES > 1
    if (old_core < NUM_CORES)
    {
        /* Did a removal on another processor's thread - switch back to
           native core */
        switch_core(old_core);
    }
#endif
}
#endif /* ALLOW_REMOVE_THREAD */

#ifdef HAVE_PRIORITY_SCHEDULING
/*---------------------------------------------------------------------------
 * Sets the thread's relative base priority for the core it runs on. Any
 * needed inheritance changes also may happen.
 *---------------------------------------------------------------------------
 */
int thread_set_priority(struct thread_entry *thread, int priority)
{
    int old_base_priority = -1;

    /* A little safety measure */
    if (priority < HIGHEST_PRIORITY || priority > LOWEST_PRIORITY)
        return -1;

    if (thread == NULL)
        thread = cores[CURRENT_CORE].running;

    /* Thread could be on any list and therefore on an interrupt accessible
       one - disable interrupts */
    int oldlevel = disable_irq_save();

    LOCK_THREAD(thread);

    /* Make sure it's not killed */
    if (thread->state != STATE_KILLED)
    {
        int old_priority = thread->priority;

        old_base_priority = thread->base_priority;
        thread->base_priority = priority;

        prio_move_entry(&thread->pdist, old_base_priority, priority);
        priority = find_first_set_bit(thread->pdist.mask);

        if (old_priority == priority)
        {
            /* No priority change - do nothing */
        }
        else if (thread->state == STATE_RUNNING)
        {
            /* This thread is running - change location on the run
             * queue. No transitive inheritance needed. */
            set_running_thread_priority(thread, priority);
        }
        else
        {
            thread->priority = priority;

            if (thread->blocker != NULL)
            {
                /* Bubble new priority down the chain */
                struct blocker *bl = thread->blocker;   /* Blocker struct */
                struct thread_entry *bl_t = bl->thread; /* Blocking thread */
                struct thread_entry * const tstart = thread;   /* Initial thread */
                const int highest = MIN(priority, old_priority); /* Higher of new or old */

                for (;;)
                {
                    struct thread_entry *next; /* Next thread to check */
                    int bl_pr;    /* Highest blocked thread */
                    int queue_pr; /* New highest blocked thread */
#if NUM_CORES > 1
                    /* Owner can change but thread cannot be dislodged - thread
                     * may not be the first in the queue which allows other
                     * threads ahead in the list to be given ownership during the
                     * operation. If thread is next then the waker will have to
                     * wait for us and the owner of the object will remain fixed.
                     * If we successfully grab the owner -- which at some point
                     * is guaranteed -- then the queue remains fixed until we
                     * pass by. */
                    for (;;)
                    {
                        LOCK_THREAD(bl_t);

                        /* Double-check the owner - retry if it changed */
                        if (bl->thread == bl_t)
                            break;

                        UNLOCK_THREAD(bl_t);
                        bl_t = bl->thread;
                    }
#endif
                    bl_pr = bl->priority;

                    if (highest > bl_pr)
                        break; /* Object priority won't change */

                    /* This will include the thread being set */
                    queue_pr = find_highest_priority_in_list_l(*thread->bqp);

                    if (queue_pr == bl_pr)
                        break; /* Object priority not changing */

                    /* Update thread boost for this object */
                    bl->priority = queue_pr;
                    prio_move_entry(&bl_t->pdist, bl_pr, queue_pr);
                    bl_pr = find_first_set_bit(bl_t->pdist.mask);

                    if (bl_t->priority == bl_pr)
                        break; /* Blocking thread priority not changing */

                    if (bl_t->state == STATE_RUNNING)
                    {
                        /* Thread not blocked - we're done */
                        set_running_thread_priority(bl_t, bl_pr);
                        break;
                    }

                    bl_t->priority = bl_pr;
                    bl = bl_t->blocker; /* Blocking thread has a blocker? */

                    if (bl == NULL)
                        break; /* End of chain */

                    next = bl->thread;

                    if (next == tstart)
                        break; /* Full-circle */

                    UNLOCK_THREAD(thread);

                    thread = bl_t;
                    bl_t = next;
                } /* for (;;) */

                UNLOCK_THREAD(bl_t);
            }
        }
    }

    UNLOCK_THREAD(thread);

    restore_irq(oldlevel);

    return old_base_priority;
}

/*---------------------------------------------------------------------------
 * Returns the current base priority for a thread.
 *---------------------------------------------------------------------------
 */
int thread_get_priority(struct thread_entry *thread)
{
    /* Simple, quick probe. */
    if (thread == NULL)
        thread = cores[CURRENT_CORE].running;

    return thread->base_priority;
}
#endif /* HAVE_PRIORITY_SCHEDULING */

/*---------------------------------------------------------------------------
 * Starts a frozen thread - similar semantics to wakeup_thread except that
 * the thread is on no scheduler or wakeup queue at all. It exists simply by
 * virtue of the slot having a state of STATE_FROZEN.
 *---------------------------------------------------------------------------
 */
void thread_thaw(struct thread_entry *thread)
{
    int oldlevel = disable_irq_save();
    LOCK_THREAD(thread);

    if (thread->state == STATE_FROZEN)
        core_schedule_wakeup(thread);

    UNLOCK_THREAD(thread);
    restore_irq(oldlevel);
}

/*---------------------------------------------------------------------------
 * Return the ID of the currently executing thread.
 *---------------------------------------------------------------------------
 */
struct thread_entry * thread_get_current(void)
{
    return cores[CURRENT_CORE].running;
}

#if NUM_CORES > 1
/*---------------------------------------------------------------------------
 * Switch the processor that the currently executing thread runs on.
 *---------------------------------------------------------------------------
 */
unsigned int switch_core(unsigned int new_core)
{
    const unsigned int core = CURRENT_CORE;
    struct thread_entry *current = cores[core].running;

    if (core == new_core)
    {
        /* No change - just return same core */
        return core;
    }

    int oldlevel = disable_irq_save();
    LOCK_THREAD(current);

    if (current->name == THREAD_DESTRUCT)
    {
        /* Thread being killed - deactivate and let process complete */
        UNLOCK_THREAD(current);
        restore_irq(oldlevel);
        thread_wait(current);
        /* Should never be reached */
        THREAD_PANICF("switch_core->D:*R", current);
    }

    /* Get us off the running list for the current core */
    RTR_LOCK(core);
    remove_from_list_l(&cores[core].running, current);
    rtr_subtract_entry(core, current->priority);
    RTR_UNLOCK(core);

    /* Stash return value (old core) in a safe place */
    current->retval = core;

    /* If a timeout hadn't yet been cleaned-up it must be removed now or
     * the other core will likely attempt a removal from the wrong list! */
    if (current->tmo.prev != NULL)
    {
        remove_from_list_tmo(current);
    }

    /* Change the core number for this thread slot */
    current->core = new_core;

    /* Do not use core_schedule_wakeup here since this will result in
     * the thread starting to run on the other core before being finished on
     * this one. Delay the  list unlock to keep the other core stuck
     * until this thread is ready. */
    RTR_LOCK(new_core);

    rtr_add_entry(new_core, current->priority);
    add_to_list_l(&cores[new_core].running, current);

    /* Make a callback into device-specific code, unlock the wakeup list so
     * that execution may resume on the new core, unlock our slot and finally
     * restore the interrupt level */
    cores[core].blk_ops.flags = TBOP_SWITCH_CORE;
    cores[core].blk_ops.cl_p  = &cores[new_core].rtr_cl;
    cores[core].block_task    = current;

    UNLOCK_THREAD(current);

    /* Alert other core to activity */
    core_wake(new_core);

    /* Do the stack switching, cache_maintenence and switch_thread call -
       requires native code */
    switch_thread_core(core, current);

    /* Finally return the old core to caller */
    return current->retval;
}
#endif /* NUM_CORES > 1 */

/*---------------------------------------------------------------------------
 * Initialize threading API. This assumes interrupts are not yet enabled. On
 * multicore setups, no core is allowed to proceed until create_thread calls
 * are safe to perform.
 *---------------------------------------------------------------------------
 */
void init_threads(void)
{
    const unsigned int core = CURRENT_CORE;
    struct thread_entry *thread;

    /* CPU will initialize first and then sleep */
    thread = find_empty_thread_slot();

    if (thread == NULL)
    {
        /* WTF? There really must be a slot available at this stage.
         * This can fail if, for example, .bss isn't zero'ed out by the loader
         * or threads is in the wrong section. */
        THREAD_PANICF("init_threads->no slot", NULL);
    }

    /* Initialize initially non-zero members of core */
    cores[core].next_tmo_check = current_tick; /* Something not in the past */

    /* Initialize initially non-zero members of slot */
    UNLOCK_THREAD(thread); /* No sync worries yet */
    thread->name = main_thread_name;
    thread->state = STATE_RUNNING;
    IF_COP( thread->core = core; )
#ifdef HAVE_PRIORITY_SCHEDULING
    corelock_init(&cores[core].rtr_cl);
    thread->base_priority = PRIORITY_USER_INTERFACE;
    prio_add_entry(&thread->pdist, PRIORITY_USER_INTERFACE);
    thread->priority = PRIORITY_USER_INTERFACE;
    rtr_add_entry(core, PRIORITY_USER_INTERFACE);
#endif
    corelock_init(&thread->waiter_cl);
    corelock_init(&thread->slot_cl);

    add_to_list_l(&cores[core].running, thread);

    if (core == CPU)
    {
        thread->stack = stackbegin;
        thread->stack_size = (uintptr_t)stackend - (uintptr_t)stackbegin;
#if NUM_CORES > 1  /* This code path will not be run on single core targets */
        /* TODO: HAL interface for this */
        /* Wake up coprocessor and let it initialize kernel and threads */
#ifdef CPU_PP502x
        MBX_MSG_CLR = 0x3f;
#endif
        COP_CTL = PROC_WAKE;
        /* Sleep until finished */
        CPU_CTL = PROC_SLEEP;
        nop; nop; nop; nop;
    } 
    else
    {
        /* Initial stack is the COP idle stack */
        thread->stack = cop_idlestackbegin;
        thread->stack_size = IDLE_STACK_SIZE;
        /* Get COP safely primed inside switch_thread where it will remain
         * until a thread actually exists on it */
        CPU_CTL = PROC_WAKE;
        thread_exit();
#endif /* NUM_CORES */
    }
}

/* Shared stack scan helper for thread_stack_usage and idle_stack_usage */
#if NUM_CORES == 1
static inline int stack_usage(uintptr_t *stackptr, size_t stack_size)
#else
static int stack_usage(uintptr_t *stackptr, size_t stack_size)
#endif
{
    unsigned int stack_words = stack_size / sizeof (uintptr_t);
    unsigned int i;
    int usage = 0;

    for (i = 0; i < stack_words; i++)
    {
        if (stackptr[i] != DEADBEEF)
        {
            usage = ((stack_words - i) * 100) / stack_words;
            break;
        }
    }

    return usage;
}

/*---------------------------------------------------------------------------
 * Returns the maximum percentage of stack a thread ever used while running.
 * NOTE: Some large buffer allocations that don't use enough the buffer to
 * overwrite stackptr[0] will not be seen.
 *---------------------------------------------------------------------------
 */
int thread_stack_usage(const struct thread_entry *thread)
{
    return stack_usage(thread->stack, thread->stack_size);
}

#if NUM_CORES > 1
/*---------------------------------------------------------------------------
 * Returns the maximum percentage of the core's idle stack ever used during
 * runtime.
 *---------------------------------------------------------------------------
 */
int idle_stack_usage(unsigned int core)
{
    return stack_usage(idle_stacks[core], IDLE_STACK_SIZE);
}
#endif

/*---------------------------------------------------------------------------
 * Fills in the buffer with the specified thread's name. If the name is NULL,
 * empty, or the thread is in destruct state a formatted ID is written
 * instead.
 *---------------------------------------------------------------------------
 */
void thread_get_name(char *buffer, int size,
                     struct thread_entry *thread)
{
    if (size <= 0)
        return;

    *buffer = '\0';

    if (thread)
    {
        /* Display thread name if one or ID if none */
        const char *name = thread->name;
        const char *fmt = "%s";
        if (name == NULL IF_COP(|| name == THREAD_DESTRUCT) || *name == '\0')
        {
            name = (const char *)thread;
            fmt = "%08lX";
        }
        snprintf(buffer, size, fmt, name);
    }
}