nuttx/sched/init/nx_start.c
hujun5 508c5889d7 sched: change nxsched_islocked_global to nxsched_islocked_tcb
reason:
1 To improve efficiency, we mimic Linux's behavior where preemption disabling is only applicable to the current CPU and does not affect other CPUs.
2 In the future, we will implement "spinlock+sched_lock", and use it extensively. Under such circumstances, if preemption is still globally disabled, it will seriously impact the scheduling efficiency.
3 We have removed g_cpu_lockset and used irqcount in order to eliminate the dependency of schedlock on critical sections in the future, simplify the logic, and further enhance the performance of sched_lock.
4 We set lockcount to 1 in order to lock scheduling on all CPUs during startup, without the need to provide additional functions to disable scheduling on other CPUs.
5 Cpu1~n must wait for cpu0 to enter the idle state before enabling scheduling because it prevents CPUs1~n from competing with cpu0 for the memory manager mutex, which could cause the cpu0 idle task to enter a wait state and trigger an assert.

size nuttx
before:
   text    data     bss     dec     hex filename
 265396   51057   63646  380099   5ccc3 nuttx
after:
   text    data     bss     dec     hex filename
 265184   51057   63642  379883   5cbeb nuttx

size -216

Configuring NuttX and compile:
$ ./tools/configure.sh -l qemu-armv8a:nsh_smp
$ make
Running with qemu
$ qemu-system-aarch64 -cpu cortex-a53 -smp 4 -nographic \
   -machine virt,virtualization=on,gic-version=3 \
   -net none -chardev stdio,id=con,mux=on -serial chardev:con \
   -mon chardev=con,mode=readline -kernel ./nuttx

Signed-off-by: hujun5 <hujun5@xiaomi.com>
2024-10-10 17:57:30 +08:00

775 lines
22 KiB
C

/****************************************************************************
* sched/init/nx_start.c
*
* SPDX-License-Identifier: Apache-2.0
*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership. The
* ASF licenses this file to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance with the
* License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
* WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
* License for the specific language governing permissions and limitations
* under the License.
*
****************************************************************************/
/****************************************************************************
* Included Files
****************************************************************************/
#include <sys/types.h>
#include <stdbool.h>
#include <stdio.h>
#include <string.h>
#include <assert.h>
#include <debug.h>
#include <nuttx/arch.h>
#include <nuttx/board.h>
#include <nuttx/compiler.h>
#include <nuttx/sched.h>
#include <nuttx/fs/fs.h>
#include <nuttx/net/net.h>
#include <nuttx/mm/iob.h>
#include <nuttx/mm/kmap.h>
#include <nuttx/mm/mm.h>
#include <nuttx/kmalloc.h>
#include <nuttx/pgalloc.h>
#include <nuttx/sched_note.h>
#include <nuttx/trace.h>
#include <nuttx/binfmt/binfmt.h>
#include <nuttx/drivers/drivers.h>
#include <nuttx/init.h>
#include <nuttx/lib/math32.h>
#include "task/task.h"
#include "sched/sched.h"
#include "signal/signal.h"
#include "semaphore/semaphore.h"
#include "mqueue/mqueue.h"
#include "mqueue/msg.h"
#include "clock/clock.h"
#include "timer/timer.h"
#include "irq/irq.h"
#include "group/group.h"
#include "init/init.h"
#include "instrument/instrument.h"
#include "tls/tls.h"
/****************************************************************************
* Pre-processor Definitions
****************************************************************************/
/* This set of all CPUs */
#define SCHED_ALL_CPUS ((1 << CONFIG_SMP_NCPUS) - 1)
/****************************************************************************
* Public Data
****************************************************************************/
/* Task Lists ***************************************************************/
/* The state of a task is indicated both by the task_state field of the TCB
* and by a series of task lists. All of these tasks lists are declared
* below. Although it is not always necessary, most of these lists are
* prioritized so that common list handling logic can be used (only the
* g_readytorun, the g_pendingtasks, and the g_waitingforsemaphore lists
* need to be prioritized).
*/
/* This is the list of all tasks that are ready to run. This is a
* prioritized list with head of the list holding the highest priority
* (unassigned) task. In the non-SMP case, the head of this list is the
* currently active task and the tail of this list, the lowest priority
* task, is always the IDLE task.
*/
dq_queue_t g_readytorun;
/* In order to support SMP, the function of the g_readytorun list changes,
* The g_readytorun is still used but in the SMP case it will contain only:
*
* - Only tasks/threads that are eligible to run, but not currently running,
* and
* - Tasks/threads that have not been assigned to a CPU.
*
* Otherwise, the TCB will be retained in an assigned task list,
* g_assignedtasks. As its name suggests, on 'g_assignedtasks queue for CPU
* 'n' would contain only tasks/threads that are assigned to CPU 'n'. Tasks/
* threads would be assigned a particular CPU by one of two mechanisms:
*
* - (Semi-)permanently through an RTOS interfaces such as
* pthread_attr_setaffinity(), or
* - Temporarily through scheduling logic when a previously unassigned task
* is made to run.
*
* Tasks/threads that are assigned to a CPU via an interface like
* pthread_attr_setaffinity() would never go into the g_readytorun list, but
* would only go into the g_assignedtasks[n] list for the CPU 'n' to which
* the thread has been assigned. Hence, the g_readytorun list would hold
* only unassigned tasks/threads.
*
* Like the g_readytorun list in in non-SMP case, each g_assignedtask[] list
* is prioritized: The head of the list is the currently active task on this
* CPU. Tasks after the active task are ready-to-run and assigned to this
* CPU. The tail of this assigned task list, the lowest priority task, is
* always the CPU's IDLE task.
*/
#ifdef CONFIG_SMP
dq_queue_t g_assignedtasks[CONFIG_SMP_NCPUS];
FAR struct tcb_s *g_delivertasks[CONFIG_SMP_NCPUS];
#endif
/* g_running_tasks[] holds a references to the running task for each cpu.
* It is valid only when up_interrupt_context() returns true.
*/
FAR struct tcb_s *g_running_tasks[CONFIG_SMP_NCPUS];
/* This is the list of all tasks that are ready-to-run, but cannot be placed
* in the g_readytorun list because: (1) They are higher priority than the
* currently active task at the head of the g_readytorun list, and (2) the
* currently active task has disabled pre-emption.
*/
dq_queue_t g_pendingtasks;
/* This is the list of all tasks that are blocked waiting for a signal */
dq_queue_t g_waitingforsignal;
#ifdef CONFIG_LEGACY_PAGING
/* This is the list of all tasks that are blocking waiting for a page fill */
dq_queue_t g_waitingforfill;
#endif
#ifdef CONFIG_SIG_SIGSTOP_ACTION
/* This is the list of all tasks that have been stopped
* via SIGSTOP or SIGTSTP
*/
dq_queue_t g_stoppedtasks;
#endif
/* This list of all tasks that have been initialized, but not yet
* activated. NOTE: This is the only list that is not prioritized.
*/
dq_queue_t g_inactivetasks;
/* This is the value of the last process ID assigned to a task */
volatile pid_t g_lastpid;
/* The following hash table is used for two things:
*
* 1. This hash table greatly speeds the determination of a new unique
* process ID for a task, and
* 2. Is used to quickly map a process ID into a TCB.
*/
FAR struct tcb_s **g_pidhash;
volatile int g_npidhash;
/* This is a table of task lists. This table is indexed by the task state
* enumeration type (tstate_t) and provides a pointer to the associated
* static task list (if there is one) as well as a set of attribute flags
* indicating properties of the list, for example, if the list is an
* ordered list or not.
*/
struct tasklist_s g_tasklisttable[NUM_TASK_STATES];
/* This is the current initialization state. The level of initialization
* is only important early in the start-up sequence when certain OS or
* hardware resources may not yet be available to the kernel logic.
*/
volatile uint8_t g_nx_initstate; /* See enum nx_initstate_e */
/****************************************************************************
* Private Data
****************************************************************************/
/* This is an array of task control block (TCB) for the IDLE thread of each
* CPU. For the non-SMP case, this is a a single TCB; For the SMP case,
* there is one TCB per CPU. NOTE: The system boots on CPU0 into the IDLE
* task. The IDLE task later starts the other CPUs and spawns the user
* initialization task. That user initialization task is responsible for
* bringing up the rest of the system.
*/
struct tcb_s g_idletcb[CONFIG_SMP_NCPUS];
/* This is the name of the idle task */
#if CONFIG_TASK_NAME_SIZE > 0 && !defined(CONFIG_SMP)
static const char g_idlename[] = "Idle_Task";
#endif
/****************************************************************************
* Private Functions
****************************************************************************/
/****************************************************************************
* Name: tasklist_initialize
*
* Description:
* Initialization of table of task lists.This table is indexed by the
* task state enumeration type (tstate_t) and provides a pointer to
* the associated static task list (if there is one) as well as a set
* of attribute flags indicating properties of the list, for example,
* if the list is an ordered list or not.
*
****************************************************************************/
static void tasklist_initialize(void)
{
FAR struct tasklist_s *tlist = (FAR void *)&g_tasklisttable;
/* TSTATE_TASK_INVALID */
tlist[TSTATE_TASK_INVALID].list = NULL;
tlist[TSTATE_TASK_INVALID].attr = 0;
/* TSTATE_TASK_PENDING */
tlist[TSTATE_TASK_PENDING].list = list_pendingtasks();
tlist[TSTATE_TASK_PENDING].attr = TLIST_ATTR_PRIORITIZED;
#ifdef CONFIG_SMP
/* TSTATE_TASK_READYTORUN */
tlist[TSTATE_TASK_READYTORUN].list = list_readytorun();
tlist[TSTATE_TASK_READYTORUN].attr = TLIST_ATTR_PRIORITIZED;
/* TSTATE_TASK_ASSIGNED */
tlist[TSTATE_TASK_ASSIGNED].list = list_assignedtasks(0);
tlist[TSTATE_TASK_ASSIGNED].attr = TLIST_ATTR_PRIORITIZED |
TLIST_ATTR_INDEXED |
TLIST_ATTR_RUNNABLE;
/* TSTATE_TASK_RUNNING */
tlist[TSTATE_TASK_RUNNING].list = list_assignedtasks(0);
tlist[TSTATE_TASK_RUNNING].attr = TLIST_ATTR_PRIORITIZED |
TLIST_ATTR_INDEXED |
TLIST_ATTR_RUNNABLE;
#else
/* TSTATE_TASK_READYTORUN */
tlist[TSTATE_TASK_READYTORUN].list = list_readytorun();
tlist[TSTATE_TASK_READYTORUN].attr = TLIST_ATTR_PRIORITIZED |
TLIST_ATTR_RUNNABLE;
/* TSTATE_TASK_RUNNING */
tlist[TSTATE_TASK_RUNNING].list = list_readytorun();
tlist[TSTATE_TASK_RUNNING].attr = TLIST_ATTR_PRIORITIZED |
TLIST_ATTR_RUNNABLE;
#endif
/* TSTATE_TASK_INACTIVE */
tlist[TSTATE_TASK_INACTIVE].list = list_inactivetasks();
tlist[TSTATE_TASK_INACTIVE].attr = 0;
/* TSTATE_WAIT_SEM */
tlist[TSTATE_WAIT_SEM].list = (FAR void *)offsetof(sem_t, waitlist);
tlist[TSTATE_WAIT_SEM].attr = TLIST_ATTR_PRIORITIZED |
TLIST_ATTR_OFFSET;
/* TSTATE_WAIT_SIG */
tlist[TSTATE_WAIT_SIG].list = list_waitingforsignal();
tlist[TSTATE_WAIT_SIG].attr = 0;
#ifndef CONFIG_DISABLE_MQUEUE
/* TSTATE_WAIT_MQNOTEMPTY */
tlist[TSTATE_WAIT_MQNOTEMPTY].list =
(FAR void *)offsetof(struct mqueue_inode_s, cmn.waitfornotempty);
tlist[TSTATE_WAIT_MQNOTEMPTY].attr = TLIST_ATTR_PRIORITIZED |
TLIST_ATTR_OFFSET;
/* TSTATE_WAIT_MQNOTFULL */
tlist[TSTATE_WAIT_MQNOTFULL].list =
(FAR void *)offsetof(struct mqueue_inode_s, cmn.waitfornotfull);
tlist[TSTATE_WAIT_MQNOTFULL].attr = TLIST_ATTR_PRIORITIZED |
TLIST_ATTR_OFFSET;
#endif
#ifdef CONFIG_LEGACY_PAGING
/* TSTATE_WAIT_PAGEFILL */
tlist[TSTATE_WAIT_PAGEFILL].list = list_waitingforfill();
tlist[TSTATE_WAIT_PAGEFILL].attr = TLIST_ATTR_PRIORITIZED;
#endif
#ifdef CONFIG_SIG_SIGSTOP_ACTION
/* TSTATE_TASK_STOPPED */
tlist[TSTATE_TASK_STOPPED].list = list_stoppedtasks();
tlist[TSTATE_TASK_STOPPED].attr = 0;
#endif
}
/****************************************************************************
* Name: idle_task_initialize
*
* Description:
* IDLE Task Initialization
*
****************************************************************************/
static void idle_task_initialize(void)
{
FAR struct tcb_s *tcb;
FAR dq_queue_t *tasklist;
int i;
memset(g_idletcb, 0, sizeof(g_idletcb));
for (i = 0; i < CONFIG_SMP_NCPUS; i++)
{
tcb = &g_idletcb[i];
/* Initialize a TCB for this thread of execution. NOTE: The default
* value for most components of the g_idletcb are zero. The entire
* structure is set to zero. Then only the (potentially) non-zero
* elements are initialized. NOTE: The idle task is the only task in
* that has pid == 0 and sched_priority == 0.
*/
tcb->pid = i;
tcb->task_state = TSTATE_TASK_RUNNING;
tcb->lockcount = 1;
/* Set the entry point. This is only for debug purposes. NOTE: that
* the start_t entry point is not saved. That is acceptable, however,
* because it can be used only for restarting a task: The IDLE task
* cannot be restarted.
*/
#ifdef CONFIG_SMP
if (i > 0)
{
tcb->start = nx_idle_trampoline;
tcb->entry.main = (main_t)nx_idle_trampoline;
}
else
#endif
{
tcb->start = nx_start;
tcb->entry.main = (main_t)nx_start;
}
/* Set the task flags to indicate that this is a kernel thread and, if
* configured for SMP, that this task is locked to this CPU.
*/
#ifdef CONFIG_SMP
tcb->flags = (TCB_FLAG_TTYPE_KERNEL | TCB_FLAG_CPU_LOCKED);
tcb->cpu = i;
/* Set the affinity mask to allow the thread to run on all CPUs. No,
* this IDLE thread can only run on its assigned CPU. That is
* enforced by the TCB_FLAG_CPU_LOCKED which overrides the affinity
* mask. This is essential because all tasks inherit the affinity
* mask from their parent and, ultimately, the parent of all tasks is
* the IDLE task.
*/
tcb->affinity =
(cpu_set_t)(CONFIG_SMP_DEFAULT_CPUSET & SCHED_ALL_CPUS);
#else
tcb->flags = TCB_FLAG_TTYPE_KERNEL;
#endif
#if CONFIG_TASK_NAME_SIZE > 0
/* Set the IDLE task name */
# ifdef CONFIG_SMP
snprintf(tcb->name, CONFIG_TASK_NAME_SIZE, "CPU%d IDLE", i);
# else
strlcpy(tcb->name, g_idlename, CONFIG_TASK_NAME_SIZE);
# endif
#endif /* CONFIG_TASK_NAME_SIZE */
/* Then add the idle task's TCB to the head of the current ready to
* run list.
*/
#ifdef CONFIG_SMP
tasklist = TLIST_HEAD(tcb, i);
#else
tasklist = TLIST_HEAD(tcb);
#endif
dq_addfirst((FAR dq_entry_t *)tcb, tasklist);
/* Mark the idle task as the running task */
g_running_tasks[i] = tcb;
}
}
/****************************************************************************
* Name: idle_group_initialize
*
* Description:
* IDLE Group Initialization
*
****************************************************************************/
static void idle_group_initialize(void)
{
FAR struct tcb_s *tcb;
int hashndx;
int i;
/* Assign the process ID(s) of ZERO to the idle task(s) */
for (i = 0; i < CONFIG_SMP_NCPUS; i++)
{
tcb = &g_idletcb[i];
hashndx = PIDHASH(i);
g_pidhash[hashndx] = tcb;
/* Allocate the IDLE group */
DEBUGVERIFY(
group_initialize((FAR struct task_tcb_s *)tcb, tcb->flags));
/* Initialize the task join */
nxtask_joininit(tcb);
#ifdef CONFIG_SMP
/* Create a stack for all CPU IDLE threads (except CPU0 which already
* has a stack).
*/
if (i > 0)
{
DEBUGVERIFY(up_cpu_idlestack(i, tcb, CONFIG_IDLETHREAD_STACKSIZE));
}
#endif
/* Initialize the processor-specific portion of the TCB */
up_initial_state(tcb);
/* Initialize the thread local storage */
tls_init_info(tcb);
/* Complete initialization of the IDLE group. Suppress retention
* of child status in the IDLE group.
*/
group_postinitialize((FAR struct task_tcb_s *)tcb);
tcb->group->tg_flags = GROUP_FLAG_NOCLDWAIT | GROUP_FLAG_PRIVILEGED;
}
}
/****************************************************************************
* Public Functions
****************************************************************************/
/****************************************************************************
* Name: nx_start
*
* Description:
* This function is called to initialize the operating system and to spawn
* the user initialization thread of execution. This is the initial entry
* point into NuttX.
*
* Input Parameters:
* None
*
* Returned Value:
* Does not return.
*
****************************************************************************/
void nx_start(void)
{
int i;
sinfo("Entry\n");
/* Boot up is complete */
g_nx_initstate = OSINIT_BOOT;
/* Initialize task list table *********************************************/
tasklist_initialize();
/* Initialize the IDLE task TCB *******************************************/
idle_task_initialize();
/* Task lists are initialized */
g_nx_initstate = OSINIT_TASKLISTS;
/* Initialize RTOS Data ***************************************************/
drivers_early_initialize();
sched_trace_begin();
/* Initialize RTOS facilities *********************************************/
/* Initialize the semaphore facility. This has to be done very early
* because many subsystems depend upon fully functional semaphores.
*/
nxsem_initialize();
#if defined(MM_KERNEL_USRHEAP_INIT) || defined(CONFIG_MM_KERNEL_HEAP) || \
defined(CONFIG_MM_PGALLOC)
/* Initialize the memory manager */
{
FAR void *heap_start;
size_t heap_size;
#ifdef MM_KERNEL_USRHEAP_INIT
/* Get the user-mode heap from the platform specific code and configure
* the user-mode memory allocator.
*/
up_allocate_heap(&heap_start, &heap_size);
kumm_initialize(heap_start, heap_size);
#endif
#ifdef CONFIG_MM_KERNEL_HEAP
/* Get the kernel-mode heap from the platform specific code and
* configure the kernel-mode memory allocator.
*/
up_allocate_kheap(&heap_start, &heap_size);
kmm_initialize(heap_start, heap_size);
#endif
#ifdef CONFIG_MM_PGALLOC
/* If there is a page allocator in the configuration, then get the page
* heap information from the platform-specific code and configure the
* page allocator.
*/
up_allocate_pgheap(&heap_start, &heap_size);
mm_pginitialize(heap_start, heap_size);
#endif
}
#endif
#ifdef CONFIG_MM_KMAP
/* Initialize the kernel dynamic mapping module */
kmm_map_initialize();
#endif
#ifdef CONFIG_ARCH_HAVE_EXTRA_HEAPS
/* Initialize any extra heap. */
up_extraheaps_init();
#endif
#ifdef CONFIG_MM_IOB
/* Initialize IO buffering */
iob_initialize();
#endif
/* Initialize the logic that determine unique process IDs. */
g_npidhash = 1 << LOG2_CEIL(CONFIG_PID_INITIAL_COUNT);
while (g_npidhash <= CONFIG_SMP_NCPUS)
{
g_npidhash <<= 1;
}
g_pidhash = kmm_zalloc(sizeof(*g_pidhash) * g_npidhash);
DEBUGASSERT(g_pidhash);
/* IDLE Group Initialization **********************************************/
idle_group_initialize();
g_lastpid = CONFIG_SMP_NCPUS - 1;
/* The memory manager is available */
g_nx_initstate = OSINIT_MEMORY;
/* Initialize tasking data structures */
task_initialize();
/* Initialize the instrument function */
instrument_initialize();
/* Initialize the file system (needed to support device drivers) */
fs_initialize();
/* Initialize the interrupt handling subsystem (if included) */
irq_initialize();
/* Initialize the POSIX timer facility (if included in the link) */
clock_initialize();
#ifndef CONFIG_DISABLE_POSIX_TIMERS
timer_initialize();
#endif
/* Initialize the signal facility (if in link) */
nxsig_initialize();
#if !defined(CONFIG_DISABLE_MQUEUE) || !defined(CONFIG_DISABLE_MQUEUE_SYSV)
/* Initialize the named message queue facility (if in link) */
nxmq_initialize();
#endif
#ifdef CONFIG_NET
/* Initialize the networking system */
net_initialize();
#endif
#ifndef CONFIG_BINFMT_DISABLE
/* Initialize the binfmt system */
binfmt_initialize();
#endif
/* Initialize Hardware Facilities *****************************************/
/* The processor specific details of running the operating system
* will be handled here. Such things as setting up interrupt
* service routines and starting the clock are some of the things
* that are different for each processor and hardware platform.
*/
up_initialize();
/* Initialize common drivers */
drivers_initialize();
#ifdef CONFIG_BOARD_EARLY_INITIALIZE
/* Call the board-specific up_initialize() extension to support
* early initialization of board-specific drivers and resources
* that cannot wait until board_late_initialize.
*/
board_early_initialize();
#endif
/* Hardware resources are now available */
g_nx_initstate = OSINIT_HARDWARE;
/* Setup for Multi-Tasking ************************************************/
/* Announce that the CPU0 IDLE task has started */
sched_note_start(&g_idletcb[0]);
/* Initialize stdio for the IDLE task of each CPU */
for (i = 0; i < CONFIG_SMP_NCPUS; i++)
{
if (i > 0)
{
/* Clone stdout, stderr, stdin from the CPU0 IDLE task. */
DEBUGVERIFY(group_setuptaskfiles(
(FAR struct task_tcb_s *)&g_idletcb[i], NULL, true));
}
else
{
/* Create stdout, stderr, stdin on the CPU0 IDLE task. These
* will be inherited by all of the threads created by the CPU0
* IDLE task.
*/
DEBUGVERIFY(group_setupidlefiles());
}
}
#ifdef CONFIG_SMP
/* Start all CPUs *********************************************************/
/* A few basic sanity checks */
DEBUGASSERT(this_cpu() == 0);
/* Then start the other CPUs */
DEBUGVERIFY(nx_smp_start());
#endif /* CONFIG_SMP */
/* Bring Up the System ****************************************************/
/* The OS is fully initialized and we are beginning multi-tasking */
g_nx_initstate = OSINIT_OSREADY;
/* Create initial tasks and bring-up the system */
DEBUGVERIFY(nx_bringup());
/* Enter to idleloop */
g_nx_initstate = OSINIT_IDLELOOP;
/* Let other threads have access to the memory manager */
sched_trace_end();
sched_unlock();
/* The IDLE Loop **********************************************************/
/* When control is return to this point, the system is idle. */
sinfo("CPU0: Beginning Idle Loop\n");
#ifndef CONFIG_DISABLE_IDLE_LOOP
for (; ; )
{
/* Perform any processor-specific idle state operations */
up_idle();
}
#endif
}