This commit changes the invocation of the newlib malloc heap initialisation function such that it is executed during the POST_KERNEL phase instead of the APPLICATION phase. This is necessary in order to ensure that the application initialisation functions (i.e. the functions called during the APPLICATIION phase) can make use of the libc heap. Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
569 lines
13 KiB
C
569 lines
13 KiB
C
/*
|
|
* Copyright (c) 2015, Intel Corporation.
|
|
*
|
|
* SPDX-License-Identifier: Apache-2.0
|
|
*/
|
|
|
|
#include <zephyr/arch/cpu.h>
|
|
#include <errno.h>
|
|
#include <stdio.h>
|
|
#include <malloc.h>
|
|
#include <zephyr/sys/__assert.h>
|
|
#include <sys/stat.h>
|
|
#include <zephyr/linker/linker-defs.h>
|
|
#include <zephyr/sys/util.h>
|
|
#include <zephyr/sys/errno_private.h>
|
|
#include <zephyr/sys/heap_listener.h>
|
|
#include <zephyr/sys/libc-hooks.h>
|
|
#include <zephyr/syscall_handler.h>
|
|
#include <zephyr/app_memory/app_memdomain.h>
|
|
#include <zephyr/init.h>
|
|
#include <zephyr/sys/sem.h>
|
|
#include <zephyr/sys/mutex.h>
|
|
#include <zephyr/sys/mem_manage.h>
|
|
#include <sys/time.h>
|
|
|
|
#define LIBC_BSS K_APP_BMEM(z_libc_partition)
|
|
#define LIBC_DATA K_APP_DMEM(z_libc_partition)
|
|
|
|
/*
|
|
* End result of this thorny set of ifdefs is to define:
|
|
*
|
|
* - HEAP_BASE base address of the heap arena
|
|
* - MAX_HEAP_SIZE size of the heap arena
|
|
*/
|
|
|
|
#ifdef CONFIG_MMU
|
|
#ifdef CONFIG_USERSPACE
|
|
struct k_mem_partition z_malloc_partition;
|
|
#endif
|
|
|
|
LIBC_BSS static unsigned char *heap_base;
|
|
LIBC_BSS static size_t max_heap_size;
|
|
|
|
#define HEAP_BASE heap_base
|
|
#define MAX_HEAP_SIZE max_heap_size
|
|
#define USE_MALLOC_PREPARE 1
|
|
#elif CONFIG_NEWLIB_LIBC_ALIGNED_HEAP_SIZE
|
|
/* Arena size expressed in Kconfig, due to power-of-two size/align
|
|
* requirements of certain MPUs.
|
|
*
|
|
* We use an automatic memory partition instead of setting this up
|
|
* in malloc_prepare().
|
|
*/
|
|
K_APPMEM_PARTITION_DEFINE(z_malloc_partition);
|
|
#define MALLOC_BSS K_APP_BMEM(z_malloc_partition)
|
|
|
|
/* Compiler will throw an error if the provided value isn't a
|
|
* power of two
|
|
*/
|
|
MALLOC_BSS static unsigned char
|
|
__aligned(CONFIG_NEWLIB_LIBC_ALIGNED_HEAP_SIZE)
|
|
heap_base[CONFIG_NEWLIB_LIBC_ALIGNED_HEAP_SIZE];
|
|
#define MAX_HEAP_SIZE CONFIG_NEWLIB_LIBC_ALIGNED_HEAP_SIZE
|
|
#define HEAP_BASE heap_base
|
|
#else /* Not MMU or CONFIG_NEWLIB_LIBC_ALIGNED_HEAP_SIZE */
|
|
#define USED_RAM_END_ADDR POINTER_TO_UINT(&_end)
|
|
|
|
#ifdef Z_MALLOC_PARTITION_EXISTS
|
|
/* Start of malloc arena needs to be aligned per MPU
|
|
* requirements
|
|
*/
|
|
struct k_mem_partition z_malloc_partition;
|
|
|
|
#if defined(CONFIG_ARM) || defined(CONFIG_ARM64)
|
|
#define HEAP_BASE ROUND_UP(USED_RAM_END_ADDR, \
|
|
CONFIG_ARM_MPU_REGION_MIN_ALIGN_AND_SIZE)
|
|
#elif defined(CONFIG_ARC)
|
|
#define HEAP_BASE ROUND_UP(USED_RAM_END_ADDR, \
|
|
Z_ARC_MPU_ALIGN)
|
|
#else
|
|
#error "Unsupported platform"
|
|
#endif /* CONFIG_<arch> */
|
|
#define USE_MALLOC_PREPARE 1
|
|
#else
|
|
/* End of kernel image */
|
|
#define HEAP_BASE USED_RAM_END_ADDR
|
|
#endif
|
|
|
|
/* End of the malloc arena is the end of physical memory */
|
|
#if defined(CONFIG_XTENSA)
|
|
/* TODO: Why is xtensa a special case? */
|
|
extern char _heap_sentry[];
|
|
#define MAX_HEAP_SIZE (POINTER_TO_UINT(&_heap_sentry) - \
|
|
HEAP_BASE)
|
|
#else
|
|
#define MAX_HEAP_SIZE (KB(CONFIG_SRAM_SIZE) - (HEAP_BASE - \
|
|
CONFIG_SRAM_BASE_ADDRESS))
|
|
#endif /* CONFIG_XTENSA */
|
|
#endif
|
|
|
|
static int malloc_prepare(const struct device *unused)
|
|
{
|
|
ARG_UNUSED(unused);
|
|
|
|
#ifdef USE_MALLOC_PREPARE
|
|
#ifdef CONFIG_MMU
|
|
max_heap_size = MIN(CONFIG_NEWLIB_LIBC_MAX_MAPPED_REGION_SIZE,
|
|
k_mem_free_get());
|
|
|
|
if (max_heap_size != 0) {
|
|
heap_base = k_mem_map(max_heap_size, K_MEM_PERM_RW);
|
|
__ASSERT(heap_base != NULL,
|
|
"failed to allocate heap of size %zu", max_heap_size);
|
|
|
|
}
|
|
#endif /* CONFIG_MMU */
|
|
|
|
#ifdef Z_MALLOC_PARTITION_EXISTS
|
|
z_malloc_partition.start = (uintptr_t)HEAP_BASE;
|
|
z_malloc_partition.size = (size_t)MAX_HEAP_SIZE;
|
|
z_malloc_partition.attr = K_MEM_PARTITION_P_RW_U_RW;
|
|
#endif /* Z_MALLOC_PARTITION_EXISTS */
|
|
#endif /* USE_MALLOC_PREPARE */
|
|
|
|
/*
|
|
* Validate that the memory space available for the newlib heap is
|
|
* greater than the minimum required size.
|
|
*/
|
|
__ASSERT(MAX_HEAP_SIZE >= CONFIG_NEWLIB_LIBC_MIN_REQUIRED_HEAP_SIZE,
|
|
"memory space available for newlib heap is less than the "
|
|
"minimum required size specified by "
|
|
"CONFIG_NEWLIB_LIBC_MIN_REQUIRED_HEAP_SIZE");
|
|
|
|
return 0;
|
|
}
|
|
|
|
SYS_INIT(malloc_prepare, POST_KERNEL, CONFIG_KERNEL_INIT_PRIORITY_DEFAULT);
|
|
|
|
/* Current offset from HEAP_BASE of unused memory */
|
|
LIBC_BSS static size_t heap_sz;
|
|
|
|
static int _stdout_hook_default(int c)
|
|
{
|
|
(void)(c); /* Prevent warning about unused argument */
|
|
|
|
return EOF;
|
|
}
|
|
|
|
static int (*_stdout_hook)(int) = _stdout_hook_default;
|
|
|
|
void __stdout_hook_install(int (*hook)(int))
|
|
{
|
|
_stdout_hook = hook;
|
|
}
|
|
|
|
static unsigned char _stdin_hook_default(void)
|
|
{
|
|
return 0;
|
|
}
|
|
|
|
static unsigned char (*_stdin_hook)(void) = _stdin_hook_default;
|
|
|
|
void __stdin_hook_install(unsigned char (*hook)(void))
|
|
{
|
|
_stdin_hook = hook;
|
|
}
|
|
|
|
int z_impl_zephyr_read_stdin(char *buf, int nbytes)
|
|
{
|
|
int i = 0;
|
|
|
|
for (i = 0; i < nbytes; i++) {
|
|
*(buf + i) = _stdin_hook();
|
|
if ((*(buf + i) == '\n') || (*(buf + i) == '\r')) {
|
|
i++;
|
|
break;
|
|
}
|
|
}
|
|
return i;
|
|
}
|
|
|
|
#ifdef CONFIG_USERSPACE
|
|
static inline int z_vrfy_zephyr_read_stdin(char *buf, int nbytes)
|
|
{
|
|
Z_OOPS(Z_SYSCALL_MEMORY_WRITE(buf, nbytes));
|
|
return z_impl_zephyr_read_stdin((char *)buf, nbytes);
|
|
}
|
|
#include <syscalls/zephyr_read_stdin_mrsh.c>
|
|
#endif
|
|
|
|
int z_impl_zephyr_write_stdout(const void *buffer, int nbytes)
|
|
{
|
|
const char *buf = buffer;
|
|
int i;
|
|
|
|
for (i = 0; i < nbytes; i++) {
|
|
if (*(buf + i) == '\n') {
|
|
_stdout_hook('\r');
|
|
}
|
|
_stdout_hook(*(buf + i));
|
|
}
|
|
return nbytes;
|
|
}
|
|
|
|
#ifdef CONFIG_USERSPACE
|
|
static inline int z_vrfy_zephyr_write_stdout(const void *buf, int nbytes)
|
|
{
|
|
Z_OOPS(Z_SYSCALL_MEMORY_READ(buf, nbytes));
|
|
return z_impl_zephyr_write_stdout((const void *)buf, nbytes);
|
|
}
|
|
#include <syscalls/zephyr_write_stdout_mrsh.c>
|
|
#endif
|
|
|
|
#ifndef CONFIG_POSIX_API
|
|
int _read(int fd, char *buf, int nbytes)
|
|
{
|
|
ARG_UNUSED(fd);
|
|
|
|
return zephyr_read_stdin(buf, nbytes);
|
|
}
|
|
__weak FUNC_ALIAS(_read, read, int);
|
|
|
|
int _write(int fd, const void *buf, int nbytes)
|
|
{
|
|
ARG_UNUSED(fd);
|
|
|
|
return zephyr_write_stdout(buf, nbytes);
|
|
}
|
|
__weak FUNC_ALIAS(_write, write, int);
|
|
|
|
int _open(const char *name, int mode)
|
|
{
|
|
return -1;
|
|
}
|
|
__weak FUNC_ALIAS(_open, open, int);
|
|
|
|
int _close(int file)
|
|
{
|
|
return -1;
|
|
}
|
|
__weak FUNC_ALIAS(_close, close, int);
|
|
|
|
int _lseek(int file, int ptr, int dir)
|
|
{
|
|
return 0;
|
|
}
|
|
__weak FUNC_ALIAS(_lseek, lseek, int);
|
|
#else
|
|
extern ssize_t write(int file, const char *buffer, size_t count);
|
|
#define _write write
|
|
#endif
|
|
|
|
int _isatty(int file)
|
|
{
|
|
return file <= 2;
|
|
}
|
|
__weak FUNC_ALIAS(_isatty, isatty, int);
|
|
|
|
int _kill(int i, int j)
|
|
{
|
|
return 0;
|
|
}
|
|
__weak FUNC_ALIAS(_kill, kill, int);
|
|
|
|
int _getpid(void)
|
|
{
|
|
return 0;
|
|
}
|
|
__weak FUNC_ALIAS(_getpid, getpid, int);
|
|
|
|
int _fstat(int file, struct stat *st)
|
|
{
|
|
st->st_mode = S_IFCHR;
|
|
return 0;
|
|
}
|
|
__weak FUNC_ALIAS(_fstat, fstat, int);
|
|
|
|
__weak void _exit(int status)
|
|
{
|
|
_write(1, "exit\n", 5);
|
|
while (1) {
|
|
;
|
|
}
|
|
}
|
|
|
|
void *_sbrk(intptr_t count)
|
|
{
|
|
void *ret, *ptr;
|
|
|
|
ptr = ((char *)HEAP_BASE) + heap_sz;
|
|
|
|
if ((heap_sz + count) < MAX_HEAP_SIZE) {
|
|
heap_sz += count;
|
|
ret = ptr;
|
|
|
|
#ifdef CONFIG_NEWLIB_LIBC_HEAP_LISTENER
|
|
heap_listener_notify_resize(HEAP_ID_LIBC, ptr, (char *)ptr + count);
|
|
#endif
|
|
} else {
|
|
ret = (void *)-1;
|
|
}
|
|
|
|
return ret;
|
|
}
|
|
__weak FUNC_ALIAS(_sbrk, sbrk, void *);
|
|
|
|
#ifdef CONFIG_MULTITHREADING
|
|
/*
|
|
* Newlib Retargetable Locking Interface Implementation
|
|
*
|
|
* When multithreading is enabled, the newlib retargetable locking interface is
|
|
* defined below to override the default void implementation and provide the
|
|
* Zephyr-side locks.
|
|
*
|
|
* NOTE: `k_mutex` and `k_sem` are used instead of `sys_mutex` and `sys_sem`
|
|
* because the latter do not support dynamic allocation for now.
|
|
*/
|
|
|
|
/* Static locks */
|
|
K_MUTEX_DEFINE(__lock___sinit_recursive_mutex);
|
|
K_MUTEX_DEFINE(__lock___sfp_recursive_mutex);
|
|
K_MUTEX_DEFINE(__lock___atexit_recursive_mutex);
|
|
K_MUTEX_DEFINE(__lock___malloc_recursive_mutex);
|
|
K_MUTEX_DEFINE(__lock___env_recursive_mutex);
|
|
K_SEM_DEFINE(__lock___at_quick_exit_mutex, 1, 1);
|
|
K_SEM_DEFINE(__lock___tz_mutex, 1, 1);
|
|
K_SEM_DEFINE(__lock___dd_hash_mutex, 1, 1);
|
|
K_SEM_DEFINE(__lock___arc4random_mutex, 1, 1);
|
|
|
|
#ifdef CONFIG_USERSPACE
|
|
/* Grant public access to all static locks after boot */
|
|
static int newlib_locks_prepare(const struct device *unused)
|
|
{
|
|
ARG_UNUSED(unused);
|
|
|
|
/* Initialise recursive locks */
|
|
k_object_access_all_grant(&__lock___sinit_recursive_mutex);
|
|
k_object_access_all_grant(&__lock___sfp_recursive_mutex);
|
|
k_object_access_all_grant(&__lock___atexit_recursive_mutex);
|
|
k_object_access_all_grant(&__lock___malloc_recursive_mutex);
|
|
k_object_access_all_grant(&__lock___env_recursive_mutex);
|
|
|
|
/* Initialise non-recursive locks */
|
|
k_object_access_all_grant(&__lock___at_quick_exit_mutex);
|
|
k_object_access_all_grant(&__lock___tz_mutex);
|
|
k_object_access_all_grant(&__lock___dd_hash_mutex);
|
|
k_object_access_all_grant(&__lock___arc4random_mutex);
|
|
|
|
return 0;
|
|
}
|
|
|
|
SYS_INIT(newlib_locks_prepare, POST_KERNEL,
|
|
CONFIG_KERNEL_INIT_PRIORITY_DEFAULT);
|
|
#endif /* CONFIG_USERSPACE */
|
|
|
|
/* Create a new dynamic non-recursive lock */
|
|
void __retarget_lock_init(_LOCK_T *lock)
|
|
{
|
|
__ASSERT_NO_MSG(lock != NULL);
|
|
|
|
/* Allocate semaphore object */
|
|
#ifndef CONFIG_USERSPACE
|
|
*lock = malloc(sizeof(struct k_sem));
|
|
#else
|
|
*lock = k_object_alloc(K_OBJ_SEM);
|
|
#endif /* !CONFIG_USERSPACE */
|
|
__ASSERT(*lock != NULL, "non-recursive lock allocation failed");
|
|
|
|
k_sem_init((struct k_sem *)*lock, 1, 1);
|
|
}
|
|
|
|
/* Create a new dynamic recursive lock */
|
|
void __retarget_lock_init_recursive(_LOCK_T *lock)
|
|
{
|
|
__ASSERT_NO_MSG(lock != NULL);
|
|
|
|
/* Allocate mutex object */
|
|
#ifndef CONFIG_USERSPACE
|
|
*lock = malloc(sizeof(struct k_mutex));
|
|
#else
|
|
*lock = k_object_alloc(K_OBJ_MUTEX);
|
|
#endif /* !CONFIG_USERSPACE */
|
|
__ASSERT(*lock != NULL, "recursive lock allocation failed");
|
|
|
|
k_mutex_init((struct k_mutex *)*lock);
|
|
}
|
|
|
|
/* Close dynamic non-recursive lock */
|
|
void __retarget_lock_close(_LOCK_T lock)
|
|
{
|
|
__ASSERT_NO_MSG(lock != NULL);
|
|
#ifndef CONFIG_USERSPACE
|
|
free(lock);
|
|
#else
|
|
k_object_release(lock);
|
|
#endif /* !CONFIG_USERSPACE */
|
|
}
|
|
|
|
/* Close dynamic recursive lock */
|
|
void __retarget_lock_close_recursive(_LOCK_T lock)
|
|
{
|
|
__ASSERT_NO_MSG(lock != NULL);
|
|
#ifndef CONFIG_USERSPACE
|
|
free(lock);
|
|
#else
|
|
k_object_release(lock);
|
|
#endif /* !CONFIG_USERSPACE */
|
|
}
|
|
|
|
/* Acquiure non-recursive lock */
|
|
void __retarget_lock_acquire(_LOCK_T lock)
|
|
{
|
|
__ASSERT_NO_MSG(lock != NULL);
|
|
k_sem_take((struct k_sem *)lock, K_FOREVER);
|
|
}
|
|
|
|
/* Acquiure recursive lock */
|
|
void __retarget_lock_acquire_recursive(_LOCK_T lock)
|
|
{
|
|
__ASSERT_NO_MSG(lock != NULL);
|
|
k_mutex_lock((struct k_mutex *)lock, K_FOREVER);
|
|
}
|
|
|
|
/* Try acquiring non-recursive lock */
|
|
int __retarget_lock_try_acquire(_LOCK_T lock)
|
|
{
|
|
__ASSERT_NO_MSG(lock != NULL);
|
|
return !k_sem_take((struct k_sem *)lock, K_NO_WAIT);
|
|
}
|
|
|
|
/* Try acquiring recursive lock */
|
|
int __retarget_lock_try_acquire_recursive(_LOCK_T lock)
|
|
{
|
|
__ASSERT_NO_MSG(lock != NULL);
|
|
return !k_mutex_lock((struct k_mutex *)lock, K_NO_WAIT);
|
|
}
|
|
|
|
/* Release non-recursive lock */
|
|
void __retarget_lock_release(_LOCK_T lock)
|
|
{
|
|
__ASSERT_NO_MSG(lock != NULL);
|
|
k_sem_give((struct k_sem *)lock);
|
|
}
|
|
|
|
/* Release recursive lock */
|
|
void __retarget_lock_release_recursive(_LOCK_T lock)
|
|
{
|
|
__ASSERT_NO_MSG(lock != NULL);
|
|
k_mutex_unlock((struct k_mutex *)lock);
|
|
}
|
|
#endif /* CONFIG_MULTITHREADING */
|
|
|
|
__weak int *__errno(void)
|
|
{
|
|
return z_errno();
|
|
}
|
|
|
|
/* This function gets called if static buffer overflow detection is enabled
|
|
* on stdlib side (Newlib here), in case such an overflow is detected. Newlib
|
|
* provides an implementation not suitable for us, so we override it here.
|
|
*/
|
|
__weak FUNC_NORETURN void __chk_fail(void)
|
|
{
|
|
static const char chk_fail_msg[] = "* buffer overflow detected *\n";
|
|
_write(2, chk_fail_msg, sizeof(chk_fail_msg) - 1);
|
|
k_oops();
|
|
CODE_UNREACHABLE;
|
|
}
|
|
|
|
#if CONFIG_XTENSA
|
|
extern int _read(int fd, char *buf, int nbytes);
|
|
extern int _open(const char *name, int mode);
|
|
extern int _close(int file);
|
|
extern int _lseek(int file, int ptr, int dir);
|
|
|
|
/* The Newlib in xtensa toolchain has a few missing functions for the
|
|
* reentrant versions of the syscalls.
|
|
*/
|
|
_ssize_t _read_r(struct _reent *r, int fd, void *buf, size_t nbytes)
|
|
{
|
|
ARG_UNUSED(r);
|
|
|
|
return _read(fd, (char *)buf, nbytes);
|
|
}
|
|
|
|
_ssize_t _write_r(struct _reent *r, int fd, const void *buf, size_t nbytes)
|
|
{
|
|
ARG_UNUSED(r);
|
|
|
|
return _write(fd, buf, nbytes);
|
|
}
|
|
|
|
int _open_r(struct _reent *r, const char *name, int flags, int mode)
|
|
{
|
|
ARG_UNUSED(r);
|
|
ARG_UNUSED(flags);
|
|
|
|
return _open(name, mode);
|
|
}
|
|
|
|
int _close_r(struct _reent *r, int file)
|
|
{
|
|
ARG_UNUSED(r);
|
|
|
|
return _close(file);
|
|
}
|
|
|
|
_off_t _lseek_r(struct _reent *r, int file, _off_t ptr, int dir)
|
|
{
|
|
ARG_UNUSED(r);
|
|
|
|
return _lseek(file, ptr, dir);
|
|
}
|
|
|
|
int _isatty_r(struct _reent *r, int file)
|
|
{
|
|
ARG_UNUSED(r);
|
|
|
|
return _isatty(file);
|
|
}
|
|
|
|
int _kill_r(struct _reent *r, int i, int j)
|
|
{
|
|
ARG_UNUSED(r);
|
|
|
|
return _kill(i, j);
|
|
}
|
|
|
|
int _getpid_r(struct _reent *r)
|
|
{
|
|
ARG_UNUSED(r);
|
|
|
|
return _getpid();
|
|
}
|
|
|
|
int _fstat_r(struct _reent *r, int file, struct stat *st)
|
|
{
|
|
ARG_UNUSED(r);
|
|
|
|
return _fstat(file, st);
|
|
}
|
|
|
|
void _exit_r(struct _reent *r, int status)
|
|
{
|
|
ARG_UNUSED(r);
|
|
|
|
_exit(status);
|
|
}
|
|
|
|
void *_sbrk_r(struct _reent *r, int count)
|
|
{
|
|
ARG_UNUSED(r);
|
|
|
|
return _sbrk(count);
|
|
}
|
|
#endif /* CONFIG_XTENSA */
|
|
|
|
int _gettimeofday(struct timeval *__tp, void *__tzp)
|
|
{
|
|
#ifdef CONFIG_POSIX_API
|
|
return gettimeofday(__tp, __tzp);
|
|
#else
|
|
/* Non-posix systems should not call gettimeofday() here as it will
|
|
* result in a recursive call loop and result in a stack overflow.
|
|
*/
|
|
return -1;
|
|
#endif
|
|
}
|