summaryrefslogtreecommitdiffstats
path: root/src/core/hle/kernel/thread.cpp
blob: f9d821a8035441a923bbafbb12779bcb34acc594 (plain) (blame)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
// Copyright 2014 Citra Emulator Project / PPSSPP Project
// Licensed under GPLv2 or any later version
// Refer to the license.txt file included.

#include <algorithm>
#include <list>
#include <vector>
#include "common/assert.h"
#include "common/common_types.h"
#include "common/logging/log.h"
#include "common/math_util.h"
#include "common/thread_queue_list.h"
#include "core/arm/arm_interface.h"
#include "core/core.h"
#include "core/core_timing.h"
#include "core/hle/kernel/errors.h"
#include "core/hle/kernel/handle_table.h"
#include "core/hle/kernel/kernel.h"
#include "core/hle/kernel/memory.h"
#include "core/hle/kernel/mutex.h"
#include "core/hle/kernel/process.h"
#include "core/hle/kernel/thread.h"
#include "core/hle/result.h"
#include "core/memory.h"

namespace Kernel {

/// Event type for the thread wake up event
static CoreTiming::EventType* ThreadWakeupEventType = nullptr;

bool Thread::ShouldWait(Thread* thread) const {
    return status != THREADSTATUS_DEAD;
}

void Thread::Acquire(Thread* thread) {
    ASSERT_MSG(!ShouldWait(thread), "object unavailable!");
}

// TODO(yuriks): This can be removed if Thread objects are explicitly pooled in the future, allowing
//               us to simply use a pool index or similar.
static Kernel::HandleTable wakeup_callback_handle_table;

// Lists all thread ids that aren't deleted/etc.
static std::vector<SharedPtr<Thread>> thread_list;

// Lists only ready thread ids.
static Common::ThreadQueueList<Thread*, THREADPRIO_LOWEST + 1> ready_queue;

static SharedPtr<Thread> current_thread;

// The first available thread id at startup
static u32 next_thread_id;

/**
 * Creates a new thread ID
 * @return The new thread ID
 */
inline static u32 const NewThreadId() {
    return next_thread_id++;
}

Thread::Thread() {}
Thread::~Thread() {}

Thread* GetCurrentThread() {
    return current_thread.get();
}

/**
 * Check if the specified thread is waiting on the specified address to be arbitrated
 * @param thread The thread to test
 * @param wait_address The address to test against
 * @return True if the thread is waiting, false otherwise
 */
static bool CheckWait_AddressArbiter(const Thread* thread, VAddr wait_address) {
    return thread->status == THREADSTATUS_WAIT_ARB && wait_address == thread->wait_address;
}

void Thread::Stop() {
    // Cancel any outstanding wakeup events for this thread
    CoreTiming::UnscheduleEvent(ThreadWakeupEventType, callback_handle);
    wakeup_callback_handle_table.Close(callback_handle);
    callback_handle = 0;

    // Clean up thread from ready queue
    // This is only needed when the thread is termintated forcefully (SVC TerminateProcess)
    if (status == THREADSTATUS_READY) {
        ready_queue.remove(current_priority, this);
    }

    status = THREADSTATUS_DEAD;

    WakeupAllWaitingThreads();

    // Clean up any dangling references in objects that this thread was waiting for
    for (auto& wait_object : wait_objects) {
        wait_object->RemoveWaitingThread(this);
    }
    wait_objects.clear();

    // Release all the mutexes that this thread holds
    ReleaseThreadMutexes(this);

    // Mark the TLS slot in the thread's page as free.
    u64 tls_page = (tls_address - Memory::TLS_AREA_VADDR) / Memory::PAGE_SIZE;
    u64 tls_slot =
        ((tls_address - Memory::TLS_AREA_VADDR) % Memory::PAGE_SIZE) / Memory::TLS_ENTRY_SIZE;
    Kernel::g_current_process->tls_slots[tls_page].reset(tls_slot);
}

Thread* ArbitrateHighestPriorityThread(u32 address) {
    Thread* highest_priority_thread = nullptr;
    u32 priority = THREADPRIO_LOWEST;

    // Iterate through threads, find highest priority thread that is waiting to be arbitrated...
    for (auto& thread : thread_list) {
        if (!CheckWait_AddressArbiter(thread.get(), address))
            continue;

        if (thread == nullptr)
            continue;

        if (thread->current_priority <= priority) {
            highest_priority_thread = thread.get();
            priority = thread->current_priority;
        }
    }

    // If a thread was arbitrated, resume it
    if (nullptr != highest_priority_thread) {
        highest_priority_thread->ResumeFromWait();
    }

    return highest_priority_thread;
}

void ArbitrateAllThreads(u32 address) {
    // Resume all threads found to be waiting on the address
    for (auto& thread : thread_list) {
        if (CheckWait_AddressArbiter(thread.get(), address))
            thread->ResumeFromWait();
    }
}

/**
 * Switches the CPU's active thread context to that of the specified thread
 * @param new_thread The thread to switch to
 */
static void SwitchContext(Thread* new_thread) {
    Thread* previous_thread = GetCurrentThread();

    // Save context for previous thread
    if (previous_thread) {
        previous_thread->last_running_ticks = CoreTiming::GetTicks();
        Core::CPU().SaveContext(previous_thread->context);

        if (previous_thread->status == THREADSTATUS_RUNNING) {
            // This is only the case when a reschedule is triggered without the current thread
            // yielding execution (i.e. an event triggered, system core time-sliced, etc)
            ready_queue.push_front(previous_thread->current_priority, previous_thread);
            previous_thread->status = THREADSTATUS_READY;
        }
    }

    // Load context of new thread
    if (new_thread) {
        ASSERT_MSG(new_thread->status == THREADSTATUS_READY,
                   "Thread must be ready to become running.");

        // Cancel any outstanding wakeup events for this thread
        CoreTiming::UnscheduleEvent(ThreadWakeupEventType, new_thread->callback_handle);

        auto previous_process = Kernel::g_current_process;

        current_thread = new_thread;

        ready_queue.remove(new_thread->current_priority, new_thread);
        new_thread->status = THREADSTATUS_RUNNING;

        if (previous_process != current_thread->owner_process) {
            Kernel::g_current_process = current_thread->owner_process;
            SetCurrentPageTable(&Kernel::g_current_process->vm_manager.page_table);
        }

        Core::CPU().LoadContext(new_thread->context);
        Core::CPU().SetTlsAddress(new_thread->GetTLSAddress());
    } else {
        current_thread = nullptr;
        // Note: We do not reset the current process and current page table when idling because
        // technically we haven't changed processes, our threads are just paused.
    }
}

/**
 * Pops and returns the next thread from the thread queue
 * @return A pointer to the next ready thread
 */
static Thread* PopNextReadyThread() {
    Thread* next;
    Thread* thread = GetCurrentThread();

    if (thread && thread->status == THREADSTATUS_RUNNING) {
        // We have to do better than the current thread.
        // This call returns null when that's not possible.
        next = ready_queue.pop_first_better(thread->current_priority);
        if (!next) {
            // Otherwise just keep going with the current thread
            next = thread;
        }
    } else {
        next = ready_queue.pop_first();
    }

    return next;
}

void WaitCurrentThread_Sleep() {
    Thread* thread = GetCurrentThread();
    thread->status = THREADSTATUS_WAIT_SLEEP;
}

void WaitCurrentThread_ArbitrateAddress(VAddr wait_address) {
    Thread* thread = GetCurrentThread();
    thread->wait_address = wait_address;
    thread->status = THREADSTATUS_WAIT_ARB;
}

void ExitCurrentThread() {
    Thread* thread = GetCurrentThread();
    thread->Stop();
    thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread),
                      thread_list.end());
}

/**
 * Callback that will wake up the thread it was scheduled for
 * @param thread_handle The handle of the thread that's been awoken
 * @param cycles_late The number of CPU cycles that have passed since the desired wakeup time
 */
static void ThreadWakeupCallback(u64 thread_handle, int cycles_late) {
    SharedPtr<Thread> thread = wakeup_callback_handle_table.Get<Thread>((Handle)thread_handle);
    if (thread == nullptr) {
        LOG_CRITICAL(Kernel, "Callback fired for invalid thread %08X", (Handle)thread_handle);
        return;
    }

    if (thread->status == THREADSTATUS_WAIT_SYNCH_ANY ||
        thread->status == THREADSTATUS_WAIT_SYNCH_ALL || thread->status == THREADSTATUS_WAIT_ARB) {

        // Invoke the wakeup callback before clearing the wait objects
        if (thread->wakeup_callback)
            thread->wakeup_callback(ThreadWakeupReason::Timeout, thread, nullptr);

        // Remove the thread from each of its waiting objects' waitlists
        for (auto& object : thread->wait_objects)
            object->RemoveWaitingThread(thread.get());
        thread->wait_objects.clear();
    }

    thread->ResumeFromWait();
}

void Thread::WakeAfterDelay(s64 nanoseconds) {
    // Don't schedule a wakeup if the thread wants to wait forever
    if (nanoseconds == -1)
        return;

    CoreTiming::ScheduleEvent(nsToCycles(nanoseconds), ThreadWakeupEventType, callback_handle);
}

void Thread::ResumeFromWait() {
    ASSERT_MSG(wait_objects.empty(), "Thread is waking up while waiting for objects");

    switch (status) {
    case THREADSTATUS_WAIT_SYNCH_ALL:
    case THREADSTATUS_WAIT_SYNCH_ANY:
    case THREADSTATUS_WAIT_ARB:
    case THREADSTATUS_WAIT_SLEEP:
        break;

    case THREADSTATUS_READY:
        // The thread's wakeup callback must have already been cleared when the thread was first
        // awoken.
        ASSERT(wakeup_callback == nullptr);
        // If the thread is waiting on multiple wait objects, it might be awoken more than once
        // before actually resuming. We can ignore subsequent wakeups if the thread status has
        // already been set to THREADSTATUS_READY.
        return;

    case THREADSTATUS_RUNNING:
        DEBUG_ASSERT_MSG(false, "Thread with object id %u has already resumed.", GetObjectId());
        return;
    case THREADSTATUS_DEAD:
        // This should never happen, as threads must complete before being stopped.
        DEBUG_ASSERT_MSG(false, "Thread with object id %u cannot be resumed because it's DEAD.",
                         GetObjectId());
        return;
    }

    wakeup_callback = nullptr;

    ready_queue.push_back(current_priority, this);
    status = THREADSTATUS_READY;
    Core::System::GetInstance().PrepareReschedule();
}

/**
 * Prints the thread queue for debugging purposes
 */
static void DebugThreadQueue() {
    Thread* thread = GetCurrentThread();
    if (!thread) {
        LOG_DEBUG(Kernel, "Current: NO CURRENT THREAD");
    } else {
        LOG_DEBUG(Kernel, "0x%02X %u (current)", thread->current_priority,
                  GetCurrentThread()->GetObjectId());
    }

    for (auto& t : thread_list) {
        u32 priority = ready_queue.contains(t.get());
        if (priority != -1) {
            LOG_DEBUG(Kernel, "0x%02X %u", priority, t->GetObjectId());
        }
    }
}

/**
 * Finds a free location for the TLS section of a thread.
 * @param tls_slots The TLS page array of the thread's owner process.
 * Returns a tuple of (page, slot, alloc_needed) where:
 * page: The index of the first allocated TLS page that has free slots.
 * slot: The index of the first free slot in the indicated page.
 * alloc_needed: Whether there's a need to allocate a new TLS page (All pages are full).
 */
std::tuple<u32, u32, bool> GetFreeThreadLocalSlot(std::vector<std::bitset<8>>& tls_slots) {
    // Iterate over all the allocated pages, and try to find one where not all slots are used.
    for (unsigned page = 0; page < tls_slots.size(); ++page) {
        const auto& page_tls_slots = tls_slots[page];
        if (!page_tls_slots.all()) {
            // We found a page with at least one free slot, find which slot it is
            for (unsigned slot = 0; slot < page_tls_slots.size(); ++slot) {
                if (!page_tls_slots.test(slot)) {
                    return std::make_tuple(page, slot, false);
                }
            }
        }
    }

    return std::make_tuple(0, 0, true);
}

/**
 * Resets a thread context, making it ready to be scheduled and run by the CPU
 * @param context Thread context to reset
 * @param stack_top Address of the top of the stack
 * @param entry_point Address of entry point for execution
 * @param arg User argument for thread
 */
static void ResetThreadContext(ARM_Interface::ThreadContext& context, VAddr stack_top,
                               VAddr entry_point, u64 arg) {
    memset(&context, 0, sizeof(ARM_Interface::ThreadContext));

    context.cpu_registers[0] = arg;
    context.pc = entry_point;
    context.sp = stack_top;
    context.cpsr = 0;
    context.fpscr = 0;
}

ResultVal<SharedPtr<Thread>> Thread::Create(std::string name, VAddr entry_point, u32 priority,
                                            u64 arg, s32 processor_id, VAddr stack_top,
                                            SharedPtr<Process> owner_process) {
    // Check if priority is in ranged. Lowest priority -> highest priority id.
    if (priority > THREADPRIO_LOWEST) {
        LOG_ERROR(Kernel_SVC, "Invalid thread priority: %d", priority);
        return ERR_OUT_OF_RANGE;
    }

    if (processor_id > THREADPROCESSORID_MAX) {
        LOG_ERROR(Kernel_SVC, "Invalid processor id: %d", processor_id);
        return ERR_OUT_OF_RANGE_KERNEL;
    }

    // TODO(yuriks): Other checks, returning 0xD9001BEA

    if (!Memory::IsValidVirtualAddress(*owner_process, entry_point)) {
        LOG_ERROR(Kernel_SVC, "(name=%s): invalid entry %08x", name.c_str(), entry_point);
        // TODO (bunnei): Find the correct error code to use here
        return ResultCode(-1);
    }

    SharedPtr<Thread> thread(new Thread);

    thread_list.push_back(thread);
    ready_queue.prepare(priority);

    thread->thread_id = NewThreadId();
    thread->status = THREADSTATUS_DORMANT;
    thread->entry_point = entry_point;
    thread->stack_top = stack_top;
    thread->nominal_priority = thread->current_priority = priority;
    thread->last_running_ticks = CoreTiming::GetTicks();
    thread->processor_id = processor_id;
    thread->wait_objects.clear();
    thread->wait_address = 0;
    thread->name = std::move(name);
    thread->callback_handle = wakeup_callback_handle_table.Create(thread).Unwrap();
    thread->owner_process = owner_process;

    // Find the next available TLS index, and mark it as used
    auto& tls_slots = owner_process->tls_slots;
    bool needs_allocation = true;
    u32 available_page; // Which allocated page has free space
    u32 available_slot; // Which slot within the page is free

    std::tie(available_page, available_slot, needs_allocation) = GetFreeThreadLocalSlot(tls_slots);

    if (needs_allocation) {
        // There are no already-allocated pages with free slots, lets allocate a new one.
        // TLS pages are allocated from the BASE region in the linear heap.
        MemoryRegionInfo* memory_region = GetMemoryRegion(MemoryRegion::BASE);
        auto& linheap_memory = memory_region->linear_heap_memory;

        if (linheap_memory->size() + Memory::PAGE_SIZE > memory_region->size) {
            LOG_ERROR(Kernel_SVC,
                      "Not enough space in region to allocate a new TLS page for thread");
            return ERR_OUT_OF_MEMORY;
        }

        size_t offset = linheap_memory->size();

        // Allocate some memory from the end of the linear heap for this region.
        linheap_memory->insert(linheap_memory->end(), Memory::PAGE_SIZE, 0);
        memory_region->used += Memory::PAGE_SIZE;
        owner_process->linear_heap_used += Memory::PAGE_SIZE;

        tls_slots.emplace_back(0); // The page is completely available at the start
        available_page = static_cast<u32>(tls_slots.size() - 1);
        available_slot = 0; // Use the first slot in the new page

        auto& vm_manager = owner_process->vm_manager;
        vm_manager.RefreshMemoryBlockMappings(linheap_memory.get());

        // Map the page to the current process' address space.
        // TODO(Subv): Find the correct MemoryState for this region.
        vm_manager.MapMemoryBlock(Memory::TLS_AREA_VADDR + available_page * Memory::PAGE_SIZE,
                                  linheap_memory, offset, Memory::PAGE_SIZE, MemoryState::ThreadLocalStorage);
    }

    // Mark the slot as used
    tls_slots[available_page].set(available_slot);
    thread->tls_address = Memory::TLS_AREA_VADDR + available_page * Memory::PAGE_SIZE +
                          available_slot * Memory::TLS_ENTRY_SIZE;

    // TODO(peachum): move to ScheduleThread() when scheduler is added so selected core is used
    // to initialize the context
    ResetThreadContext(thread->context, stack_top, entry_point, arg);

    return MakeResult<SharedPtr<Thread>>(std::move(thread));
}

void Thread::SetPriority(u32 priority) {
    ASSERT_MSG(priority <= THREADPRIO_LOWEST && priority >= THREADPRIO_HIGHEST,
               "Invalid priority value.");
    // If thread was ready, adjust queues
    if (status == THREADSTATUS_READY)
        ready_queue.move(this, current_priority, priority);
    else
        ready_queue.prepare(priority);

    nominal_priority = current_priority = priority;
}

void Thread::UpdatePriority() {
    u32 best_priority = nominal_priority;
    for (auto& mutex : held_mutexes) {
        if (mutex->priority < best_priority)
            best_priority = mutex->priority;
    }
    BoostPriority(best_priority);
}

void Thread::BoostPriority(u32 priority) {
    // If thread was ready, adjust queues
    if (status == THREADSTATUS_READY)
        ready_queue.move(this, current_priority, priority);
    else
        ready_queue.prepare(priority);
    current_priority = priority;
}

SharedPtr<Thread> SetupMainThread(VAddr entry_point, u32 priority,
                                  SharedPtr<Process> owner_process) {
    // Setup page table so we can write to memory
    SetCurrentPageTable(&Kernel::g_current_process->vm_manager.page_table);

    // Initialize new "main" thread
    auto thread_res = Thread::Create("main", entry_point, priority, 0, THREADPROCESSORID_0,
                                     Memory::HEAP_VADDR_END, owner_process);

    SharedPtr<Thread> thread = std::move(thread_res).Unwrap();

    // Register 1 must be a handle to the main thread
    thread->guest_handle = Kernel::g_handle_table.Create(thread).Unwrap();;
    thread->context.cpu_registers[1] = thread->guest_handle;

    // Threads by default are dormant, wake up the main thread so it runs when the scheduler fires
    thread->ResumeFromWait();

    return thread;
}

bool HaveReadyThreads() {
    return ready_queue.get_first() != nullptr;
}

void Reschedule() {
    Thread* cur = GetCurrentThread();
    Thread* next = PopNextReadyThread();

    if (cur && next) {
        LOG_TRACE(Kernel, "context switch %u -> %u", cur->GetObjectId(), next->GetObjectId());
    } else if (cur) {
        LOG_TRACE(Kernel, "context switch %u -> idle", cur->GetObjectId());
    } else if (next) {
        LOG_TRACE(Kernel, "context switch idle -> %u", next->GetObjectId());
    }

    SwitchContext(next);
}

void Thread::SetWaitSynchronizationResult(ResultCode result) {
    context.cpu_registers[0] = result.raw;
}

void Thread::SetWaitSynchronizationOutput(s32 output) {
    context.cpu_registers[1] = output;
}

s32 Thread::GetWaitObjectIndex(WaitObject* object) const {
    ASSERT_MSG(!wait_objects.empty(), "Thread is not waiting for anything");
    auto match = std::find(wait_objects.rbegin(), wait_objects.rend(), object);
    return static_cast<s32>(std::distance(match, wait_objects.rend()) - 1);
}

VAddr Thread::GetCommandBufferAddress() const {
    // Offset from the start of TLS at which the IPC command buffer begins.
    static constexpr int CommandHeaderOffset = 0x80;
    return GetTLSAddress() + CommandHeaderOffset;
}

////////////////////////////////////////////////////////////////////////////////////////////////////

void ThreadingInit() {
    ThreadWakeupEventType = CoreTiming::RegisterEvent("ThreadWakeupCallback", ThreadWakeupCallback);

    current_thread = nullptr;
    next_thread_id = 1;
}

void ThreadingShutdown() {
    current_thread = nullptr;

    for (auto& t : thread_list) {
        t->Stop();
    }
    thread_list.clear();
    ready_queue.clear();
}

const std::vector<SharedPtr<Thread>>& GetThreadList() {
    return thread_list;
}

} // namespace