Boot Linux faster!

Check our new training course

Boot Linux faster!

Check our new training course
and Creative Commons CC-BY-SA
lecture and lab materials

Bootlin logo

Elixir Cross Referencer

  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
/*
 * Copyright (c) 2013-2014 Wind River Systems, Inc.
 *
 * SPDX-License-Identifier: Apache-2.0
 */

/**
 * @file
 * @brief Kernel fatal error handler
 */

#include <toolchain.h>
#include <linker/sections.h>

#include <kernel.h>
#include <kernel_structs.h>
#include <drivers/interrupt_controller/sysapic.h>
#include <arch/x86/ia32/segmentation.h>
#include <ia32/exception.h>
#include <inttypes.h>
#include <exc_handle.h>
#include <logging/log_ctrl.h>

__weak void z_debug_fatal_hook(const z_arch_esf_t *esf) { ARG_UNUSED(esf); }

#ifdef CONFIG_THREAD_STACK_INFO
/**
 * @brief Check if a memory address range falls within the stack
 *
 * Given a memory address range, ensure that it falls within the bounds
 * of the faulting context's stack.
 *
 * @param addr Starting address
 * @param size Size of the region, or 0 if we just want to see if addr is
 *             in bounds
 * @param cs Code segment of faulting context
 * @return true if addr/size region is not within the thread stack
 */
static bool check_stack_bounds(u32_t addr, size_t size, u16_t cs)
{
	u32_t start, end;

	if (z_is_in_isr()) {
		/* We were servicing an interrupt */
		start = (u32_t)Z_ARCH_THREAD_STACK_BUFFER(_interrupt_stack);
		end = start + CONFIG_ISR_STACK_SIZE;
	} else if ((cs & 0x3U) != 0U ||
		   (_current->base.user_options & K_USER) == 0) {
		/* Thread was in user mode, or is not a user mode thread.
		 * The normal stack buffer is what we will check.
		 */
		start = _current->stack_info.start;
		end = STACK_ROUND_DOWN(_current->stack_info.start +
				       _current->stack_info.size);
	} else {
		/* User thread was doing a syscall, check kernel stack bounds */
		start = _current->stack_info.start - MMU_PAGE_SIZE;
		end = _current->stack_info.start;
	}

	return (addr <= start) || (addr + size > end);
}
#endif

#if defined(CONFIG_EXCEPTION_STACK_TRACE)
struct stack_frame {
	u32_t next;
	u32_t ret_addr;
	u32_t args;
};

#define MAX_STACK_FRAMES 8

static void unwind_stack(u32_t base_ptr, u16_t cs)
{
	struct stack_frame *frame;
	int i;

	if (base_ptr == 0U) {
		z_fatal_print("NULL base ptr");
		return;
	}

	for (i = 0; i < MAX_STACK_FRAMES; i++) {
		if (base_ptr % sizeof(base_ptr) != 0U) {
			z_fatal_print("unaligned frame ptr");
			return;
		}

		frame = (struct stack_frame *)base_ptr;
		if (frame == NULL) {
			break;
		}

#ifdef CONFIG_THREAD_STACK_INFO
		/* Ensure the stack frame is within the faulting context's
		 * stack buffer
		 */
		if (check_stack_bounds((u32_t)frame, sizeof(*frame), cs)) {
			z_fatal_print("     corrupted? (bp=%p)", frame);
			break;
		}
#endif

		if (frame->ret_addr == 0U) {
			break;
		}
#ifdef CONFIG_X86_IAMCU
		z_fatal_print("     0x%08x", frame->ret_addr);
#else
		z_fatal_print("     0x%08x (0x%x)", frame->ret_addr,
			      frame->args);
#endif
		base_ptr = frame->next;
	}
}
#endif /* CONFIG_EXCEPTION_STACK_TRACE */

#ifdef CONFIG_BOARD_QEMU_X86
FUNC_NORETURN void z_arch_system_halt(unsigned int reason)
{
	ARG_UNUSED(reason);

	/* Causes QEMU to exit. We passed the following on the command line:
	 * -device isa-debug-exit,iobase=0xf4,iosize=0x04
	 */
	sys_out32(0, 0xf4);
	CODE_UNREACHABLE;
}
#endif

FUNC_NORETURN void z_x86_fatal_error(unsigned int reason, const z_arch_esf_t *esf)
{
	if (esf != NULL) {
		z_fatal_print("eax: 0x%08x, ebx: 0x%08x, ecx: 0x%08x, edx: 0x%08x",
			      esf->eax, esf->ebx, esf->ecx, esf->edx);
		z_fatal_print("esi: 0x%08x, edi: 0x%08x, ebp: 0x%08x, esp: 0x%08x",
			      esf->esi, esf->edi, esf->ebp, esf->esp);
		z_fatal_print("eflags: 0x%08x cs: 0x%04x cr3: %p", esf->eflags,
			      esf->cs & 0xFFFFU, z_x86_page_tables_get());

#ifdef CONFIG_EXCEPTION_STACK_TRACE
		z_fatal_print("call trace:");
#endif
		z_fatal_print("eip: 0x%08x", esf->eip);
#ifdef CONFIG_EXCEPTION_STACK_TRACE
		unwind_stack(esf->ebp, esf->cs);
#endif
	}

	z_fatal_error(reason, esf);
	CODE_UNREACHABLE;
}

void z_x86_spurious_irq(const z_arch_esf_t *esf)
{
	int vector = z_irq_controller_isr_vector_get();

	if (vector >= 0) {
		z_fatal_print("IRQ vector: %d", vector);
	}

	z_x86_fatal_error(K_ERR_SPURIOUS_IRQ, esf);
}

void z_arch_syscall_oops(void *ssf_ptr)
{
	struct _x86_syscall_stack_frame *ssf =
		(struct _x86_syscall_stack_frame *)ssf_ptr;
	z_arch_esf_t oops = {
		.eip = ssf->eip,
		.cs = ssf->cs,
		.eflags = ssf->eflags
	};

	if (oops.cs == USER_CODE_SEG) {
		oops.esp = ssf->esp;
	}

	z_x86_fatal_error(K_ERR_KERNEL_OOPS, &oops);
}

#ifdef CONFIG_X86_KERNEL_OOPS
void z_do_kernel_oops(const z_arch_esf_t *esf)
{
	u32_t *stack_ptr = (u32_t *)esf->esp;
	u32_t reason = *stack_ptr;

#ifdef CONFIG_USERSPACE
	/* User mode is only allowed to induce oopses and stack check
	 * failures via this software interrupt
	 */
	if (esf->cs == USER_CODE_SEG && !(reason == K_ERR_KERNEL_OOPS ||
					  reason == K_ERR_STACK_CHK_FAIL)) {
		reason = K_ERR_KERNEL_OOPS;
	}
#endif

	z_x86_fatal_error(reason, esf);
}

extern void (*_kernel_oops_handler)(void);
NANO_CPU_INT_REGISTER(_kernel_oops_handler, NANO_SOFT_IRQ,
		      CONFIG_X86_KERNEL_OOPS_VECTOR / 16,
		      CONFIG_X86_KERNEL_OOPS_VECTOR, 3);
#endif

#if CONFIG_EXCEPTION_DEBUG

FUNC_NORETURN static void generic_exc_handle(unsigned int vector,
					     const z_arch_esf_t *pEsf)
{
	switch (vector) {
	case IV_GENERAL_PROTECTION:
		z_fatal_print("General Protection Fault");
		break;
	case IV_DEVICE_NOT_AVAILABLE:
		z_fatal_print("Floating point unit not enabled");
		break;
	default:
		z_fatal_print("CPU exception %d", vector);
		break;
	}
	if ((BIT(vector) & _EXC_ERROR_CODE_FAULTS) != 0) {
		z_fatal_print("Exception code: 0x%x", pEsf->errorCode);
	}
	z_x86_fatal_error(K_ERR_CPU_EXCEPTION, pEsf);
}

#define _EXC_FUNC(vector) \
FUNC_NORETURN void handle_exc_##vector(const z_arch_esf_t *pEsf) \
{ \
	generic_exc_handle(vector, pEsf); \
}

#define Z_EXC_FUNC_CODE(vector) \
	_EXC_FUNC(vector) \
	_EXCEPTION_CONNECT_CODE(handle_exc_##vector, vector)

#define Z_EXC_FUNC_NOCODE(vector) \
	_EXC_FUNC(vector) \
	_EXCEPTION_CONNECT_NOCODE(handle_exc_##vector, vector)

/* Necessary indirection to ensure 'vector' is expanded before we expand
 * the handle_exc_##vector
 */
#define EXC_FUNC_NOCODE(vector) \
	Z_EXC_FUNC_NOCODE(vector)

#define EXC_FUNC_CODE(vector) \
	Z_EXC_FUNC_CODE(vector)

EXC_FUNC_NOCODE(IV_DIVIDE_ERROR);
EXC_FUNC_NOCODE(IV_NON_MASKABLE_INTERRUPT);
EXC_FUNC_NOCODE(IV_OVERFLOW);
EXC_FUNC_NOCODE(IV_BOUND_RANGE);
EXC_FUNC_NOCODE(IV_INVALID_OPCODE);
EXC_FUNC_NOCODE(IV_DEVICE_NOT_AVAILABLE);
#ifndef CONFIG_X86_ENABLE_TSS
EXC_FUNC_NOCODE(IV_DOUBLE_FAULT);
#endif
EXC_FUNC_CODE(IV_INVALID_TSS);
EXC_FUNC_CODE(IV_SEGMENT_NOT_PRESENT);
EXC_FUNC_CODE(IV_STACK_FAULT);
EXC_FUNC_CODE(IV_GENERAL_PROTECTION);
EXC_FUNC_NOCODE(IV_X87_FPU_FP_ERROR);
EXC_FUNC_CODE(IV_ALIGNMENT_CHECK);
EXC_FUNC_NOCODE(IV_MACHINE_CHECK);

/* Page fault error code flags */
#define PRESENT	BIT(0)
#define WR	BIT(1)
#define US	BIT(2)
#define RSVD	BIT(3)
#define ID	BIT(4)
#define PK	BIT(5)
#define SGX	BIT(15)

#ifdef CONFIG_X86_MMU
static void dump_entry_flags(const char *name, x86_page_entry_data_t flags)
{
	z_fatal_print("%s: 0x%x%x %s, %s, %s, %s", name, (u32_t)(flags>>32),
	       (u32_t)(flags),
	       flags & (x86_page_entry_data_t)MMU_ENTRY_PRESENT ?
	       "Present" : "Non-present",
	       flags & (x86_page_entry_data_t)MMU_ENTRY_WRITE ?
	       "Writable" : "Read-only",
	       flags & (x86_page_entry_data_t)MMU_ENTRY_USER ?
	       "User" : "Supervisor",
	       flags & (x86_page_entry_data_t)MMU_ENTRY_EXECUTE_DISABLE ?
	       "Execute Disable" : "Execute Enabled");
}

static void dump_mmu_flags(struct x86_mmu_pdpt *pdpt, void *addr)
{
	x86_page_entry_data_t pde_flags, pte_flags;

	z_x86_mmu_get_flags(pdpt, addr, &pde_flags, &pte_flags);

	dump_entry_flags("PDE", pde_flags);
	dump_entry_flags("PTE", pte_flags);
}
#endif /* CONFIG_X86_MMU */

static void dump_page_fault(z_arch_esf_t *esf)
{
	u32_t err, cr2;

	/* See Section 6.15 of the IA32 Software Developer's Manual vol 3 */
	__asm__ ("mov %%cr2, %0" : "=r" (cr2));

	err = esf->errorCode;
	z_fatal_print("***** CPU Page Fault (error code 0x%08x)", err);

	z_fatal_print("%s thread %s address 0x%08x",
		      (err & US) != 0U ? "User" : "Supervisor",
		      (err & ID) != 0U ? "executed" : ((err & WR) != 0U ?
						       "wrote" :
						       "read"), cr2);

#ifdef CONFIG_X86_MMU
#ifdef CONFIG_X86_KPTI
	if (err & US) {
		dump_mmu_flags(&z_x86_user_pdpt, (void *)cr2);
		return;
	}
#endif
	dump_mmu_flags(&z_x86_kernel_pdpt, (void *)cr2);
#endif
}
#endif /* CONFIG_EXCEPTION_DEBUG */

#ifdef CONFIG_USERSPACE
Z_EXC_DECLARE(z_arch_user_string_nlen);

static const struct z_exc_handle exceptions[] = {
	Z_EXC_HANDLE(z_arch_user_string_nlen)
};
#endif

void page_fault_handler(z_arch_esf_t *esf)
{
#ifdef CONFIG_USERSPACE
	int i;

	for (i = 0; i < ARRAY_SIZE(exceptions); i++) {
		if ((void *)esf->eip >= exceptions[i].start &&
		    (void *)esf->eip < exceptions[i].end) {
			esf->eip = (unsigned int)(exceptions[i].fixup);
			return;
		}
	}
#endif
#ifdef CONFIG_EXCEPTION_DEBUG
	dump_page_fault(esf);
#endif
#ifdef CONFIG_THREAD_STACK_INFO
	if (check_stack_bounds(esf->esp, 0, esf->cs)) {
		z_x86_fatal_error(K_ERR_STACK_CHK_FAIL, esf);
	}
#endif
	z_x86_fatal_error(K_ERR_CPU_EXCEPTION, esf);
	CODE_UNREACHABLE;
}
_EXCEPTION_CONNECT_CODE(page_fault_handler, IV_PAGE_FAULT);

#ifdef CONFIG_X86_ENABLE_TSS
static __noinit volatile z_arch_esf_t _df_esf;

/* Very tiny stack; just enough for the bogus error code pushed by the CPU
 * and a frame pointer push by the compiler. All df_handler_top does is
 * shuffle some data around with 'mov' statements and then 'iret'.
 */
static __noinit char _df_stack[8];

static FUNC_NORETURN __used void df_handler_top(void);

#ifdef CONFIG_X86_KPTI
extern char z_trampoline_stack_end[];
#endif

Z_GENERIC_SECTION(.tss)
struct task_state_segment _main_tss = {
	.ss0 = DATA_SEG,
#ifdef CONFIG_X86_KPTI
	/* Stack to land on when we get a soft/hard IRQ in user mode.
	 * In a special kernel page that, unlike all other kernel pages,
	 * is marked present in the user page table.
	 */
	.esp0 = (u32_t)&z_trampoline_stack_end
#endif
};

/* Special TSS for handling double-faults with a known good stack */
Z_GENERIC_SECTION(.tss)
struct task_state_segment _df_tss = {
	.esp = (u32_t)(_df_stack + sizeof(_df_stack)),
	.cs = CODE_SEG,
	.ds = DATA_SEG,
	.es = DATA_SEG,
	.ss = DATA_SEG,
	.eip = (u32_t)df_handler_top,
	.cr3 = (u32_t)&z_x86_kernel_pdpt
};

static __used void df_handler_bottom(void)
{
	/* We're back in the main hardware task on the interrupt stack */
	int reason = K_ERR_CPU_EXCEPTION;

	/* Restore the top half so it is runnable again */
	_df_tss.esp = (u32_t)(_df_stack + sizeof(_df_stack));
	_df_tss.eip = (u32_t)df_handler_top;

	z_fatal_print("Double Fault");
#ifdef CONFIG_THREAD_STACK_INFO
	if (check_stack_bounds(_df_esf.esp, 0, _df_esf.cs)) {
		reason = K_ERR_STACK_CHK_FAIL;
	}
#endif
	z_x86_fatal_error(reason, (z_arch_esf_t *)&_df_esf);
}

static FUNC_NORETURN __used void df_handler_top(void)
{
	/* State of the system when the double-fault forced a task switch
	 * will be in _main_tss. Set up a z_arch_esf_t and copy system state into
	 * it
	 */
	_df_esf.esp = _main_tss.esp;
	_df_esf.ebp = _main_tss.ebp;
	_df_esf.ebx = _main_tss.ebx;
	_df_esf.esi = _main_tss.esi;
	_df_esf.edi = _main_tss.edi;
	_df_esf.edx = _main_tss.edx;
	_df_esf.eax = _main_tss.eax;
	_df_esf.ecx = _main_tss.ecx;
	_df_esf.errorCode = 0;
	_df_esf.eip = _main_tss.eip;
	_df_esf.cs = _main_tss.cs;
	_df_esf.eflags = _main_tss.eflags;

	/* Restore the main IA task to a runnable state */
	_main_tss.esp = (u32_t)(Z_ARCH_THREAD_STACK_BUFFER(_interrupt_stack) +
				CONFIG_ISR_STACK_SIZE);
	_main_tss.cs = CODE_SEG;
	_main_tss.ds = DATA_SEG;
	_main_tss.es = DATA_SEG;
	_main_tss.ss = DATA_SEG;
	_main_tss.eip = (u32_t)df_handler_bottom;
	_main_tss.cr3 = (u32_t)&z_x86_kernel_pdpt;
	_main_tss.eflags = 0U;

	/* NT bit is set in EFLAGS so we will task switch back to _main_tss
	 * and run df_handler_bottom
	 */
	__asm__ volatile ("iret");
	CODE_UNREACHABLE;
}

/* Configure a task gate descriptor in the IDT for the double fault
 * exception
 */
_X86_IDT_TSS_REGISTER(DF_TSS, -1, -1, IV_DOUBLE_FAULT, 0);

#endif /* CONFIG_X86_ENABLE_TSS */