zephyrproject-rtos/zephyr
0
0
Fork 0
mirror of https://github.com/zephyrproject-rtos/zephyr synced 2025-09-02 06:02:46 +00:00
Code Issues Packages Projects Releases Wiki Activity
d6dfde36c4
zephyr/kernel/init.c
Andy Ross 15c400774e kernel: Rework SMP irq_lock() compatibility layer 
This was wrong in two ways, one subtle and one awful.

The subtle problem was that the IRQ lock isn't actually globally
recursive, it gets reset when you context switch (i.e. a _Swap()
implicitly releases and reacquires it).  So the recursive count I was
keeping needs to be per-thread or else we risk deadlock any time we
swap away from a thread holding the lock.

And because part of my brain apparently knew this, there was an
"optimization" in the code that tested the current count vs. zero
outside the lock, on the argument that if it was non-zero we must
already hold the lock.  Which would be true of a per-thread counter,
but NOT a global one: the other CPU may be holding that lock, and this
test will tell you *you* do.  The upshot is that a recursive
irq_lock() would almost always SUCCEED INCORRECTLY when there was lock
contention.  That this didn't break more things is amazing to me.

The rework is actually simpler than the original, thankfully.  Though
there are some further subtleties:

* The lock state implied by irq_lock() allows the lock to be
  implicitly released on context switch (i.e. you can _Swap() with the
  lock held at a recursion level higher than 1, which needs to allow
  other processes to run).  So return paths into threads from _Swap()
  and interrupt/exception exit need to check and restore the global
  lock state, spinning as needed.

* The idle loop design specifies a k_cpu_idle() function that is on
  common architectures expected to enable interrupts (for obvious
  reasons), but there is no place to put non-arch code to wire it into
  the global lock accounting.  So on SMP, even CPU0 needs to use the
  "dumb" spinning idle loop.

Finally this patch contains a simple bugfix too, found by inspection:
the interrupt return code used when CONFIG_SWITCH is enabled wasn't
correctly setting the active flag on the threads, opening up the
potential for a race that might result in a thread being scheduled on
two CPUs simultaneously.

Signed-off-by: Andy Ross <andrew.j.ross@intel.com>
2018-05-02 10:00:17 -07:00

461 lines
12 KiB
C
Raw Blame History

/*
* Copyright (c) 2010-2014 Wind River Systems, Inc.
*
* SPDX-License-Identifier: Apache-2.0
*/
/**
* @file
* @brief Kernel initialization module
*
* This module contains routines that are used to initialize the kernel.
*/
#include <zephyr.h>
#include <offsets_short.h>
#include <kernel.h>
#include <misc/printk.h>
#include <misc/stack.h>
#include <random/rand32.h>
#include <linker/sections.h>
#include <toolchain.h>
#include <kernel_structs.h>
#include <device.h>
#include <init.h>
#include <linker/linker-defs.h>
#include <ksched.h>
#include <version.h>
#include <string.h>
#include <misc/dlist.h>
#include <kernel_internal.h>
#include <kswap.h>
/* kernel build timestamp items */
#define BUILD_TIMESTAMP "BUILD: " __DATE__ " " __TIME__
/* boot banner items */
#if defined(CONFIG_BOOT_DELAY) && CONFIG_BOOT_DELAY > 0
#define BOOT_DELAY_BANNER " (delayed boot " \
STRINGIFY(CONFIG_BOOT_DELAY) "ms)"
static const unsigned int boot_delay = CONFIG_BOOT_DELAY;
#else
#define BOOT_DELAY_BANNER ""
static const unsigned int boot_delay;
#endif
#ifdef BUILD_VERSION
#define BOOT_BANNER "Booting Zephyr OS " \
STRINGIFY(BUILD_VERSION) BOOT_DELAY_BANNER
#else
#define BOOT_BANNER "Booting Zephyr OS " \
KERNEL_VERSION_STRING BOOT_DELAY_BANNER
#endif
#if !defined(CONFIG_BOOT_BANNER)
#define PRINT_BOOT_BANNER() do { } while (0)
#else
#define PRINT_BOOT_BANNER() printk("***** " BOOT_BANNER " *****\n")
#endif
/* boot time measurement items */
#ifdef CONFIG_BOOT_TIME_MEASUREMENT
u64_t __noinit __start_time_stamp; /* timestamp when kernel starts */
u64_t __noinit __main_time_stamp; /* timestamp when main task starts */
u64_t __noinit __idle_time_stamp; /* timestamp when CPU goes idle */
#endif
/* init/main and idle threads */
#define IDLE_STACK_SIZE CONFIG_IDLE_STACK_SIZE
#if CONFIG_MAIN_STACK_SIZE & (STACK_ALIGN - 1)
#error "MAIN_STACK_SIZE must be a multiple of the stack alignment"
#endif
#if IDLE_STACK_SIZE & (STACK_ALIGN - 1)
#error "IDLE_STACK_SIZE must be a multiple of the stack alignment"
#endif
#define MAIN_STACK_SIZE CONFIG_MAIN_STACK_SIZE
K_THREAD_STACK_DEFINE(_main_stack, MAIN_STACK_SIZE);
K_THREAD_STACK_DEFINE(_idle_stack, IDLE_STACK_SIZE);
static struct k_thread _main_thread_s;
static struct k_thread _idle_thread_s;
k_tid_t const _main_thread = (k_tid_t)&_main_thread_s;
k_tid_t const _idle_thread = (k_tid_t)&_idle_thread_s;
/*
* storage space for the interrupt stack
*
* Note: This area is used as the system stack during kernel initialization,
* since the kernel hasn't yet set up its own stack areas. The dual purposing
* of this area is safe since interrupts are disabled until the kernel context
* switches to the init thread.
*/
#if CONFIG_ISR_STACK_SIZE & (STACK_ALIGN - 1)
#error "ISR_STACK_SIZE must be a multiple of the stack alignment"
#endif
K_THREAD_STACK_DEFINE(_interrupt_stack, CONFIG_ISR_STACK_SIZE);
/*
* Similar idle thread & interrupt stack definitions for the
* auxiliary CPUs. The declaration macros aren't set up to define an
* array, so do it with a simple test for up to 4 processors. Should
* clean this up in the future.
*/
#if defined(CONFIG_SMP) && CONFIG_MP_NUM_CPUS > 1
K_THREAD_STACK_DEFINE(_idle_stack1, IDLE_STACK_SIZE);
static struct k_thread _idle_thread1_s;
k_tid_t const _idle_thread1 = (k_tid_t)&_idle_thread1_s;
K_THREAD_STACK_DEFINE(_interrupt_stack1, CONFIG_ISR_STACK_SIZE);
#endif
#if defined(CONFIG_SMP) && CONFIG_MP_NUM_CPUS > 2
K_THREAD_STACK_DEFINE(_idle_stack2, IDLE_STACK_SIZE);
static struct k_thread _idle_thread2_s;
k_tid_t const _idle_thread2 = (k_tid_t)&_idle_thread2_s;
K_THREAD_STACK_DEFINE(_interrupt_stack2, CONFIG_ISR_STACK_SIZE);
#endif
#if defined(CONFIG_SMP) && CONFIG_MP_NUM_CPUS > 3
K_THREAD_STACK_DEFINE(_idle_stack3, IDLE_STACK_SIZE);
static struct k_thread _idle_thread3_s;
k_tid_t const _idle_thread3 = (k_tid_t)&_idle_thread3_s;
K_THREAD_STACK_DEFINE(_interrupt_stack3, CONFIG_ISR_STACK_SIZE);
#endif
#ifdef CONFIG_SYS_CLOCK_EXISTS
#define initialize_timeouts() do { \
sys_dlist_init(&_timeout_q); \
} while ((0))
#else
#define initialize_timeouts() do { } while ((0))
#endif
extern void idle(void *unused1, void *unused2, void *unused3);
#if defined(CONFIG_INIT_STACKS) && defined(CONFIG_PRINTK)
extern K_THREAD_STACK_DEFINE(sys_work_q_stack,
CONFIG_SYSTEM_WORKQUEUE_STACK_SIZE);
void k_call_stacks_analyze(void)
{
printk("Kernel stacks:\n");
STACK_ANALYZE("main ", _main_stack);
STACK_ANALYZE("idle ", _idle_stack);
STACK_ANALYZE("interrupt", _interrupt_stack);
STACK_ANALYZE("workqueue", sys_work_q_stack);
}
#else
void k_call_stacks_analyze(void) { }
#endif
/**
*
* @brief Clear BSS
*
* This routine clears the BSS region, so all bytes are 0.
*
* @return N/A
*/
void _bss_zero(void)
{
memset(&__bss_start, 0,
((u32_t) &__bss_end - (u32_t) &__bss_start));
#ifdef CONFIG_CCM_BASE_ADDRESS
memset(&__ccm_bss_start, 0,
((u32_t) &__ccm_bss_end - (u32_t) &__ccm_bss_start));
#endif
#ifdef CONFIG_APPLICATION_MEMORY
memset(&__app_bss_start, 0,
((u32_t) &__app_bss_end - (u32_t) &__app_bss_start));
#endif
}
#ifdef CONFIG_XIP
/**
*
* @brief Copy the data section from ROM to RAM
*
* This routine copies the data section from ROM to RAM.
*
* @return N/A
*/
void _data_copy(void)
{
memcpy(&__data_ram_start, &__data_rom_start,
((u32_t) &__data_ram_end - (u32_t) &__data_ram_start));
#ifdef CONFIG_CCM_BASE_ADDRESS
memcpy(&__ccm_data_start, &__ccm_data_rom_start,
((u32_t) &__ccm_data_end - (u32_t) &__ccm_data_start));
#endif
#ifdef CONFIG_APPLICATION_MEMORY
memcpy(&__app_data_ram_start, &__app_data_rom_start,
((u32_t) &__app_data_ram_end - (u32_t) &__app_data_ram_start));
#endif
}
#endif
/**
*
* @brief Mainline for kernel's background thread
*
* This routine completes kernel initialization by invoking the remaining
* init functions, then invokes application's main() routine.
*
* @return N/A
*/
static void bg_thread_main(void *unused1, void *unused2, void *unused3)
{
ARG_UNUSED(unused1);
ARG_UNUSED(unused2);
ARG_UNUSED(unused3);
_sys_device_do_config_level(_SYS_INIT_LEVEL_POST_KERNEL);
if (boot_delay > 0) {
printk("***** delaying boot " STRINGIFY(CONFIG_BOOT_DELAY)
"ms (per build configuration) *****\n");
k_busy_wait(CONFIG_BOOT_DELAY * USEC_PER_MSEC);
}
PRINT_BOOT_BANNER();
/* Final init level before app starts */
_sys_device_do_config_level(_SYS_INIT_LEVEL_APPLICATION);
#ifdef CONFIG_CPLUSPLUS
/* Process the .ctors and .init_array sections */
extern void __do_global_ctors_aux(void);
extern void __do_init_array_aux(void);
__do_global_ctors_aux();
__do_init_array_aux();
#endif
_init_static_threads();
#ifdef CONFIG_SMP
smp_init();
#endif
#ifdef CONFIG_BOOT_TIME_MEASUREMENT
/* record timestamp for kernel's _main() function */
extern u64_t __main_time_stamp;
__main_time_stamp = (u64_t)k_cycle_get_32();
#endif
extern void main(void);
main();
/* Terminate thread normally since it has no more work to do */
_main_thread->base.user_options &= ~K_ESSENTIAL;
}
void __weak main(void)
{
/* NOP default main() if the application does not provide one. */
}
#if defined(CONFIG_MULTITHREADING)
static void init_idle_thread(struct k_thread *thr, k_thread_stack_t *stack)
{
#ifdef CONFIG_SMP
thr->base.is_idle = 1;
#endif
_setup_new_thread(thr, stack,
IDLE_STACK_SIZE, idle, NULL, NULL, NULL,
K_LOWEST_THREAD_PRIO, K_ESSENTIAL);
_mark_thread_as_started(thr);
_ready_thread(thr);
}
#endif
/**
*
* @brief Initializes kernel data structures
*
* This routine initializes various kernel data structures, including
* the init and idle threads and any architecture-specific initialization.
*
* Note that all fields of "_kernel" are set to zero on entry, which may
* be all the initialization many of them require.
*
* @return N/A
*/
static void prepare_multithreading(struct k_thread *dummy_thread)
{
#ifdef CONFIG_ARCH_HAS_CUSTOM_SWAP_TO_MAIN
ARG_UNUSED(dummy_thread);
#else
/*
* Initialize the current execution thread to permit a level of
* debugging output if an exception should happen during kernel
* initialization. However, don't waste effort initializing the
* fields of the dummy thread beyond those needed to identify it as a
* dummy thread.
*/
_current = dummy_thread;
dummy_thread->base.user_options = K_ESSENTIAL;
dummy_thread->base.thread_state = _THREAD_DUMMY;
#ifdef CONFIG_THREAD_STACK_INFO
dummy_thread->stack_info.start = 0;
dummy_thread->stack_info.size = 0;
#endif
#ifdef CONFIG_USERSPACE
dummy_thread->mem_domain_info.mem_domain = 0;
#endif
#endif
/* _kernel.ready_q is all zeroes */
/*
* The interrupt library needs to be initialized early since a series
* of handlers are installed into the interrupt table to catch
* spurious interrupts. This must be performed before other kernel
* subsystems install bonafide handlers, or before hardware device
* drivers are initialized.
*/
_IntLibInit();
/* ready the init/main and idle threads */
for (int ii = 0; ii < K_NUM_PRIORITIES; ii++) {
sys_dlist_init(&_ready_q.q[ii]);
}
#ifndef CONFIG_SMP
/*
* prime the cache with the main thread since:
*
* - the cache can never be NULL
* - the main thread will be the one to run first
* - no other thread is initialized yet and thus their priority fields
* contain garbage, which would prevent the cache loading algorithm
* to work as intended
*/
_ready_q.cache = _main_thread;
#endif
_setup_new_thread(_main_thread, _main_stack,
MAIN_STACK_SIZE, bg_thread_main,
NULL, NULL, NULL,
CONFIG_MAIN_THREAD_PRIORITY, K_ESSENTIAL);
_mark_thread_as_started(_main_thread);
_ready_thread(_main_thread);
#ifdef CONFIG_MULTITHREADING
init_idle_thread(_idle_thread, _idle_stack);
#endif
#if defined(CONFIG_SMP) && CONFIG_MP_NUM_CPUS > 1
init_idle_thread(_idle_thread1, _idle_stack1);
_kernel.cpus[1].id = 1;
_kernel.cpus[1].irq_stack = K_THREAD_STACK_BUFFER(_interrupt_stack1)
+ CONFIG_ISR_STACK_SIZE;
#endif
#if defined(CONFIG_SMP) && CONFIG_MP_NUM_CPUS > 2
init_idle_thread(_idle_thread2, _idle_stack2);
_kernel.cpus[2].id = 2;
_kernel.cpus[2].irq_stack = K_THREAD_STACK_BUFFER(_interrupt_stack2)
+ CONFIG_ISR_STACK_SIZE;
#endif
#if defined(CONFIG_SMP) && CONFIG_MP_NUM_CPUS > 3
init_idle_thread(_idle_thread3, _idle_stack3);
_kernel.cpus[3].id = 3;
_kernel.cpus[3].irq_stack = K_THREAD_STACK_BUFFER(_interrupt_stack3)
+ CONFIG_ISR_STACK_SIZE;
#endif
initialize_timeouts();
/* perform any architecture-specific initialization */
kernel_arch_init();
}
static void switch_to_main_thread(void)
{
#ifdef CONFIG_ARCH_HAS_CUSTOM_SWAP_TO_MAIN
_arch_switch_to_main_thread(_main_thread, _main_stack, MAIN_STACK_SIZE,
bg_thread_main);
#else
/*
* Context switch to main task (entry function is _main()): the
* current fake thread is not on a wait queue or ready queue, so it
* will never be rescheduled in.
*/
_Swap(irq_lock());
#endif
}
#ifdef CONFIG_STACK_CANARIES
extern void *__stack_chk_guard;
#endif
/**
*
* @brief Initialize kernel
*
* This routine is invoked when the system is ready to run C code. The
* processor must be running in 32-bit mode, and the BSS must have been
* cleared/zeroed.
*
* @return Does not return
*/
FUNC_NORETURN void _Cstart(void)
{
#ifdef CONFIG_ARCH_HAS_CUSTOM_SWAP_TO_MAIN
struct k_thread *dummy_thread = NULL;
#else
/* Normally, kernel objects are not allowed on the stack, special case
* here since this is just being used to bootstrap the first _Swap()
*/
char dummy_thread_memory[sizeof(struct k_thread)];
struct k_thread *dummy_thread = (struct k_thread *)&dummy_thread_memory;
memset(dummy_thread_memory, 0, sizeof(dummy_thread_memory));
#endif
/*
* Initialize kernel data structures. This step includes
* initializing the interrupt subsystem, which must be performed
* before the hardware initialization phase.
*/
prepare_multithreading(dummy_thread);
/* perform basic hardware initialization */
_sys_device_do_config_level(_SYS_INIT_LEVEL_PRE_KERNEL_1);
_sys_device_do_config_level(_SYS_INIT_LEVEL_PRE_KERNEL_2);
/* initialize stack canaries */
#ifdef CONFIG_STACK_CANARIES
__stack_chk_guard = (void *)sys_rand32_get();
#endif
/* display boot banner */
switch_to_main_thread();
/*
* Compiler can't tell that the above routines won't return and issues
* a warning unless we explicitly tell it that control never gets this
* far.
*/
CODE_UNREACHABLE;
}
Reference in New Issue View Git Blame Copy Permalink