esp-idf/components/vfs/vfs.c

983 lines
31 KiB
C

// Copyright 2015-2017 Espressif Systems (Shanghai) PTE LTD
//
// Licensed 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.
#include <stdlib.h>
#include <string.h>
#include <assert.h>
#include <sys/errno.h>
#include <sys/fcntl.h>
#include <sys/ioctl.h>
#include <sys/unistd.h>
#include <sys/lock.h>
#include <sys/param.h>
#include <dirent.h>
#include "freertos/FreeRTOS.h"
#include "freertos/semphr.h"
#include "esp_vfs.h"
#include "sdkconfig.h"
#ifdef CONFIG_SUPPRESS_SELECT_DEBUG_OUTPUT
#define LOG_LOCAL_LEVEL ESP_LOG_NONE
#endif //CONFIG_SUPPRESS_SELECT_DEBUG_OUTPUT
#include "esp_log.h"
static const char *TAG = "vfs";
#define VFS_MAX_COUNT 8 /* max number of VFS entries (registered filesystems) */
#define LEN_PATH_PREFIX_IGNORED SIZE_MAX /* special length value for VFS which is never recognised by open() */
#define FD_TABLE_ENTRY_UNUSED (fd_table_t) { .permanent = false, .vfs_index = -1, .local_fd = -1 }
typedef uint8_t local_fd_t;
_Static_assert((1 << (sizeof(local_fd_t)*8)) >= MAX_FDS, "file descriptor type too small");
typedef int8_t vfs_index_t;
_Static_assert((1 << (sizeof(vfs_index_t)*8)) >= VFS_MAX_COUNT, "VFS index type too small");
_Static_assert(((vfs_index_t) -1) < 0, "vfs_index_t must be a signed type");
typedef struct {
bool permanent;
vfs_index_t vfs_index;
local_fd_t local_fd;
} fd_table_t;
typedef struct vfs_entry_ {
esp_vfs_t vfs; // contains pointers to VFS functions
char path_prefix[ESP_VFS_PATH_MAX]; // path prefix mapped to this VFS
size_t path_prefix_len; // micro-optimization to avoid doing extra strlen
void* ctx; // optional pointer which can be passed to VFS
int offset; // index of this structure in s_vfs array
} vfs_entry_t;
typedef struct {
bool isset; // none or at least one bit is set in the following 3 fd sets
fd_set readfds;
fd_set writefds;
fd_set errorfds;
} fds_triple_t;
static vfs_entry_t* s_vfs[VFS_MAX_COUNT] = { 0 };
static size_t s_vfs_count = 0;
static fd_table_t s_fd_table[MAX_FDS] = { [0 ... MAX_FDS-1] = FD_TABLE_ENTRY_UNUSED };
static _lock_t s_fd_table_lock;
static esp_err_t esp_vfs_register_common(const char* base_path, size_t len, const esp_vfs_t* vfs, void* ctx, int *vfs_index)
{
if (len != LEN_PATH_PREFIX_IGNORED) {
if ((len != 0 && len < 2) || (len > ESP_VFS_PATH_MAX)) {
return ESP_ERR_INVALID_ARG;
}
if ((len > 0 && base_path[0] != '/') || base_path[len - 1] == '/') {
return ESP_ERR_INVALID_ARG;
}
}
vfs_entry_t *entry = (vfs_entry_t*) malloc(sizeof(vfs_entry_t));
if (entry == NULL) {
return ESP_ERR_NO_MEM;
}
size_t index;
for (index = 0; index < s_vfs_count; ++index) {
if (s_vfs[index] == NULL) {
break;
}
}
if (index == s_vfs_count) {
if (s_vfs_count >= VFS_MAX_COUNT) {
free(entry);
return ESP_ERR_NO_MEM;
}
++s_vfs_count;
}
s_vfs[index] = entry;
if (len != LEN_PATH_PREFIX_IGNORED) {
strcpy(entry->path_prefix, base_path); // we have already verified argument length
} else {
bzero(entry->path_prefix, sizeof(entry->path_prefix));
}
memcpy(&entry->vfs, vfs, sizeof(esp_vfs_t));
entry->path_prefix_len = len;
entry->ctx = ctx;
entry->offset = index;
if (vfs_index) {
*vfs_index = index;
}
return ESP_OK;
}
esp_err_t esp_vfs_register(const char* base_path, const esp_vfs_t* vfs, void* ctx)
{
return esp_vfs_register_common(base_path, strlen(base_path), vfs, ctx, NULL);
}
esp_err_t esp_vfs_register_fd_range(const esp_vfs_t *vfs, void *ctx, int min_fd, int max_fd)
{
if (min_fd < 0 || max_fd < 0 || min_fd > MAX_FDS || max_fd > MAX_FDS || min_fd > max_fd) {
ESP_LOGD(TAG, "Invalid arguments: esp_vfs_register_fd_range(0x%x, 0x%x, %d, %d)", (int) vfs, (int) ctx, min_fd, max_fd);
return ESP_ERR_INVALID_ARG;
}
int index = -1;
esp_err_t ret = esp_vfs_register_common("", LEN_PATH_PREFIX_IGNORED, vfs, ctx, &index);
if (ret == ESP_OK) {
_lock_acquire(&s_fd_table_lock);
for (int i = min_fd; i < max_fd; ++i) {
if (s_fd_table[i].vfs_index != -1) {
free(s_vfs[i]);
s_vfs[i] = NULL;
for (int j = min_fd; j < i; ++j) {
if (s_fd_table[j].vfs_index == index) {
s_fd_table[j] = FD_TABLE_ENTRY_UNUSED;
}
}
_lock_release(&s_fd_table_lock);
ESP_LOGD(TAG, "esp_vfs_register_fd_range cannot set fd %d (used by other VFS)", i);
return ESP_ERR_INVALID_ARG;
}
s_fd_table[i].permanent = true;
s_fd_table[i].vfs_index = index;
s_fd_table[i].local_fd = i;
}
_lock_release(&s_fd_table_lock);
}
ESP_LOGD(TAG, "esp_vfs_register_fd_range is successful for range <%d; %d) and VFS ID %d", min_fd, max_fd, index);
return ret;
}
esp_err_t esp_vfs_register_with_id(const esp_vfs_t *vfs, void *ctx, esp_vfs_id_t *vfs_id)
{
if (vfs_id == NULL) {
return ESP_ERR_INVALID_ARG;
}
*vfs_id = -1;
return esp_vfs_register_common("", LEN_PATH_PREFIX_IGNORED, vfs, ctx, vfs_id);
}
esp_err_t esp_vfs_unregister(const char* base_path)
{
for (size_t i = 0; i < s_vfs_count; ++i) {
vfs_entry_t* vfs = s_vfs[i];
if (vfs == NULL) {
continue;
}
if (memcmp(base_path, vfs->path_prefix, vfs->path_prefix_len) == 0) {
free(vfs);
s_vfs[i] = NULL;
_lock_acquire(&s_fd_table_lock);
// Delete all references from the FD lookup-table
for (int j = 0; j < MAX_FDS; ++j) {
if (s_fd_table[j].vfs_index == i) {
s_fd_table[j] = FD_TABLE_ENTRY_UNUSED;
}
}
_lock_release(&s_fd_table_lock);
return ESP_OK;
}
}
return ESP_ERR_INVALID_STATE;
}
esp_err_t esp_vfs_register_fd(esp_vfs_id_t vfs_id, int *fd)
{
if (vfs_id < 0 || vfs_id >= s_vfs_count || fd == NULL) {
ESP_LOGD(TAG, "Invalid arguments for esp_vfs_register_fd(%d, 0x%x)", vfs_id, (int) fd);
return ESP_ERR_INVALID_ARG;
}
esp_err_t ret = ESP_ERR_NO_MEM;
_lock_acquire(&s_fd_table_lock);
for (int i = 0; i < MAX_FDS; ++i) {
if (s_fd_table[i].vfs_index == -1) {
s_fd_table[i].permanent = true;
s_fd_table[i].vfs_index = vfs_id;
s_fd_table[i].local_fd = i;
*fd = i;
ret = ESP_OK;
break;
}
}
_lock_release(&s_fd_table_lock);
ESP_LOGD(TAG, "esp_vfs_register_fd(%d, 0x%x) finished with %s", vfs_id, (int) fd, esp_err_to_name(ret));
return ret;
}
esp_err_t esp_vfs_unregister_fd(esp_vfs_id_t vfs_id, int fd)
{
esp_err_t ret = ESP_ERR_INVALID_ARG;
if (vfs_id < 0 || vfs_id >= s_vfs_count || fd < 0 || fd >= MAX_FDS) {
ESP_LOGD(TAG, "Invalid arguments for esp_vfs_unregister_fd(%d, %d)", vfs_id, fd);
return ret;
}
_lock_acquire(&s_fd_table_lock);
fd_table_t *item = s_fd_table + fd;
if (item->permanent == true && item->vfs_index == vfs_id && item->local_fd == fd) {
*item = FD_TABLE_ENTRY_UNUSED;
ret = ESP_OK;
}
_lock_release(&s_fd_table_lock);
ESP_LOGD(TAG, "esp_vfs_unregister_fd(%d, %d) finished with %s", vfs_id, fd, esp_err_to_name(ret));
return ret;
}
static inline const vfs_entry_t *get_vfs_for_index(int index)
{
if (index < 0 || index >= s_vfs_count) {
return NULL;
} else {
return s_vfs[index];
}
}
static inline bool fd_valid(int fd)
{
return (fd < MAX_FDS) && (fd >= 0);
}
static const vfs_entry_t *get_vfs_for_fd(int fd)
{
const vfs_entry_t *vfs = NULL;
if (fd_valid(fd)) {
const int index = s_fd_table[fd].vfs_index; // single read -> no locking is required
vfs = get_vfs_for_index(index);
}
return vfs;
}
static inline int get_local_fd(const vfs_entry_t *vfs, int fd)
{
int local_fd = -1;
if (vfs && fd_valid(fd)) {
local_fd = s_fd_table[fd].local_fd; // single read -> no locking is required
}
return local_fd;
}
static const char* translate_path(const vfs_entry_t* vfs, const char* src_path)
{
assert(strncmp(src_path, vfs->path_prefix, vfs->path_prefix_len) == 0);
if (strlen(src_path) == vfs->path_prefix_len) {
// special case when src_path matches the path prefix exactly
return "/";
}
return src_path + vfs->path_prefix_len;
}
static const vfs_entry_t* get_vfs_for_path(const char* path)
{
const vfs_entry_t* best_match = NULL;
ssize_t best_match_prefix_len = -1;
size_t len = strlen(path);
for (size_t i = 0; i < s_vfs_count; ++i) {
const vfs_entry_t* vfs = s_vfs[i];
if (!vfs || vfs->path_prefix_len == LEN_PATH_PREFIX_IGNORED) {
continue;
}
// match path prefix
if (len < vfs->path_prefix_len ||
memcmp(path, vfs->path_prefix, vfs->path_prefix_len) != 0) {
continue;
}
// this is the default VFS and we don't have a better match yet.
if (vfs->path_prefix_len == 0 && !best_match) {
best_match = vfs;
continue;
}
// if path is not equal to the prefix, expect to see a path separator
// i.e. don't match "/data" prefix for "/data1/foo.txt" path
if (len > vfs->path_prefix_len &&
path[vfs->path_prefix_len] != '/') {
continue;
}
// Out of all matching path prefixes, select the longest one;
// i.e. if "/dev" and "/dev/uart" both match, for "/dev/uart/1" path,
// choose "/dev/uart",
// This causes all s_vfs_count VFS entries to be scanned when opening
// a file by name. This can be optimized by introducing a table for
// FS search order, sorted so that longer prefixes are checked first.
if (best_match_prefix_len < (ssize_t) vfs->path_prefix_len) {
best_match_prefix_len = (ssize_t) vfs->path_prefix_len;
best_match = vfs;
}
}
return best_match;
}
/*
* Using huge multi-line macros is never nice, but in this case
* the only alternative is to repeat this chunk of code (with different function names)
* for each syscall being implemented. Given that this define is contained within a single
* file, this looks like a good tradeoff.
*
* First we check if syscall is implemented by VFS (corresponding member is not NULL),
* then call the right flavor of the method (e.g. open or open_p) depending on
* ESP_VFS_FLAG_CONTEXT_PTR flag. If ESP_VFS_FLAG_CONTEXT_PTR is set, context is passed
* in as first argument and _p variant is used for the call.
* It is enough to check just one of them for NULL, as both variants are part of a union.
*/
#define CHECK_AND_CALL(ret, r, pvfs, func, ...) \
if (pvfs->vfs.func == NULL) { \
__errno_r(r) = ENOSYS; \
return -1; \
} \
if (pvfs->vfs.flags & ESP_VFS_FLAG_CONTEXT_PTR) { \
ret = (*pvfs->vfs.func ## _p)(pvfs->ctx, __VA_ARGS__); \
} else { \
ret = (*pvfs->vfs.func)(__VA_ARGS__);\
}
#define CHECK_AND_CALLV(r, pvfs, func, ...) \
if (pvfs->vfs.func == NULL) { \
__errno_r(r) = ENOSYS; \
return; \
} \
if (pvfs->vfs.flags & ESP_VFS_FLAG_CONTEXT_PTR) { \
(*pvfs->vfs.func ## _p)(pvfs->ctx, __VA_ARGS__); \
} else { \
(*pvfs->vfs.func)(__VA_ARGS__);\
}
#define CHECK_AND_CALLP(ret, r, pvfs, func, ...) \
if (pvfs->vfs.func == NULL) { \
__errno_r(r) = ENOSYS; \
return NULL; \
} \
if (pvfs->vfs.flags & ESP_VFS_FLAG_CONTEXT_PTR) { \
ret = (*pvfs->vfs.func ## _p)(pvfs->ctx, __VA_ARGS__); \
} else { \
ret = (*pvfs->vfs.func)(__VA_ARGS__);\
}
int esp_vfs_open(struct _reent *r, const char * path, int flags, int mode)
{
const vfs_entry_t *vfs = get_vfs_for_path(path);
if (vfs == NULL) {
__errno_r(r) = ENOENT;
return -1;
}
const char *path_within_vfs = translate_path(vfs, path);
int fd_within_vfs;
CHECK_AND_CALL(fd_within_vfs, r, vfs, open, path_within_vfs, flags, mode);
if (fd_within_vfs >= 0) {
_lock_acquire(&s_fd_table_lock);
for (int i = 0; i < MAX_FDS; ++i) {
if (s_fd_table[i].vfs_index == -1) {
s_fd_table[i].permanent = false;
s_fd_table[i].vfs_index = vfs->offset;
s_fd_table[i].local_fd = fd_within_vfs;
_lock_release(&s_fd_table_lock);
return i;
}
}
_lock_release(&s_fd_table_lock);
int ret;
CHECK_AND_CALL(ret, r, vfs, close, fd_within_vfs);
(void) ret; // remove "set but not used" warning
__errno_r(r) = ENOMEM;
return -1;
}
__errno_r(r) = ENOENT;
return -1;
}
ssize_t esp_vfs_write(struct _reent *r, int fd, const void * data, size_t size)
{
const vfs_entry_t* vfs = get_vfs_for_fd(fd);
const int local_fd = get_local_fd(vfs, fd);
if (vfs == NULL || local_fd < 0) {
__errno_r(r) = EBADF;
return -1;
}
ssize_t ret;
CHECK_AND_CALL(ret, r, vfs, write, local_fd, data, size);
return ret;
}
off_t esp_vfs_lseek(struct _reent *r, int fd, off_t size, int mode)
{
const vfs_entry_t* vfs = get_vfs_for_fd(fd);
const int local_fd = get_local_fd(vfs, fd);
if (vfs == NULL || local_fd < 0) {
__errno_r(r) = EBADF;
return -1;
}
off_t ret;
CHECK_AND_CALL(ret, r, vfs, lseek, local_fd, size, mode);
return ret;
}
ssize_t esp_vfs_read(struct _reent *r, int fd, void * dst, size_t size)
{
const vfs_entry_t* vfs = get_vfs_for_fd(fd);
const int local_fd = get_local_fd(vfs, fd);
if (vfs == NULL || local_fd < 0) {
__errno_r(r) = EBADF;
return -1;
}
ssize_t ret;
CHECK_AND_CALL(ret, r, vfs, read, local_fd, dst, size);
return ret;
}
int esp_vfs_close(struct _reent *r, int fd)
{
const vfs_entry_t* vfs = get_vfs_for_fd(fd);
const int local_fd = get_local_fd(vfs, fd);
if (vfs == NULL || local_fd < 0) {
__errno_r(r) = EBADF;
return -1;
}
int ret;
CHECK_AND_CALL(ret, r, vfs, close, local_fd);
_lock_acquire(&s_fd_table_lock);
if (!s_fd_table[fd].permanent) {
s_fd_table[fd] = FD_TABLE_ENTRY_UNUSED;
}
_lock_release(&s_fd_table_lock);
return ret;
}
int esp_vfs_fstat(struct _reent *r, int fd, struct stat * st)
{
const vfs_entry_t* vfs = get_vfs_for_fd(fd);
const int local_fd = get_local_fd(vfs, fd);
if (vfs == NULL || local_fd < 0) {
__errno_r(r) = EBADF;
return -1;
}
int ret;
CHECK_AND_CALL(ret, r, vfs, fstat, local_fd, st);
return ret;
}
int esp_vfs_stat(struct _reent *r, const char * path, struct stat * st)
{
const vfs_entry_t* vfs = get_vfs_for_path(path);
if (vfs == NULL) {
__errno_r(r) = ENOENT;
return -1;
}
const char* path_within_vfs = translate_path(vfs, path);
int ret;
CHECK_AND_CALL(ret, r, vfs, stat, path_within_vfs, st);
return ret;
}
int esp_vfs_link(struct _reent *r, const char* n1, const char* n2)
{
const vfs_entry_t* vfs = get_vfs_for_path(n1);
if (vfs == NULL) {
__errno_r(r) = ENOENT;
return -1;
}
const vfs_entry_t* vfs2 = get_vfs_for_path(n2);
if (vfs != vfs2) {
__errno_r(r) = EXDEV;
return -1;
}
const char* path1_within_vfs = translate_path(vfs, n1);
const char* path2_within_vfs = translate_path(vfs, n2);
int ret;
CHECK_AND_CALL(ret, r, vfs, link, path1_within_vfs, path2_within_vfs);
return ret;
}
int esp_vfs_unlink(struct _reent *r, const char *path)
{
const vfs_entry_t* vfs = get_vfs_for_path(path);
if (vfs == NULL) {
__errno_r(r) = ENOENT;
return -1;
}
const char* path_within_vfs = translate_path(vfs, path);
int ret;
CHECK_AND_CALL(ret, r, vfs, unlink, path_within_vfs);
return ret;
}
int esp_vfs_rename(struct _reent *r, const char *src, const char *dst)
{
const vfs_entry_t* vfs = get_vfs_for_path(src);
if (vfs == NULL) {
__errno_r(r) = ENOENT;
return -1;
}
const vfs_entry_t* vfs_dst = get_vfs_for_path(dst);
if (vfs != vfs_dst) {
__errno_r(r) = EXDEV;
return -1;
}
const char* src_within_vfs = translate_path(vfs, src);
const char* dst_within_vfs = translate_path(vfs, dst);
int ret;
CHECK_AND_CALL(ret, r, vfs, rename, src_within_vfs, dst_within_vfs);
return ret;
}
DIR* opendir(const char* name)
{
const vfs_entry_t* vfs = get_vfs_for_path(name);
struct _reent* r = __getreent();
if (vfs == NULL) {
__errno_r(r) = ENOENT;
return NULL;
}
const char* path_within_vfs = translate_path(vfs, name);
DIR* ret;
CHECK_AND_CALLP(ret, r, vfs, opendir, path_within_vfs);
if (ret != NULL) {
ret->dd_vfs_idx = vfs->offset;
}
return ret;
}
struct dirent* readdir(DIR* pdir)
{
const vfs_entry_t* vfs = get_vfs_for_index(pdir->dd_vfs_idx);
struct _reent* r = __getreent();
if (vfs == NULL) {
__errno_r(r) = EBADF;
return NULL;
}
struct dirent* ret;
CHECK_AND_CALLP(ret, r, vfs, readdir, pdir);
return ret;
}
int readdir_r(DIR* pdir, struct dirent* entry, struct dirent** out_dirent)
{
const vfs_entry_t* vfs = get_vfs_for_index(pdir->dd_vfs_idx);
struct _reent* r = __getreent();
if (vfs == NULL) {
errno = EBADF;
return -1;
}
int ret;
CHECK_AND_CALL(ret, r, vfs, readdir_r, pdir, entry, out_dirent);
return ret;
}
long telldir(DIR* pdir)
{
const vfs_entry_t* vfs = get_vfs_for_index(pdir->dd_vfs_idx);
struct _reent* r = __getreent();
if (vfs == NULL) {
errno = EBADF;
return -1;
}
long ret;
CHECK_AND_CALL(ret, r, vfs, telldir, pdir);
return ret;
}
void seekdir(DIR* pdir, long loc)
{
const vfs_entry_t* vfs = get_vfs_for_index(pdir->dd_vfs_idx);
struct _reent* r = __getreent();
if (vfs == NULL) {
errno = EBADF;
return;
}
CHECK_AND_CALLV(r, vfs, seekdir, pdir, loc);
}
void rewinddir(DIR* pdir)
{
seekdir(pdir, 0);
}
int closedir(DIR* pdir)
{
const vfs_entry_t* vfs = get_vfs_for_index(pdir->dd_vfs_idx);
struct _reent* r = __getreent();
if (vfs == NULL) {
errno = EBADF;
return -1;
}
int ret;
CHECK_AND_CALL(ret, r, vfs, closedir, pdir);
return ret;
}
int mkdir(const char* name, mode_t mode)
{
const vfs_entry_t* vfs = get_vfs_for_path(name);
struct _reent* r = __getreent();
if (vfs == NULL) {
__errno_r(r) = ENOENT;
return -1;
}
const char* path_within_vfs = translate_path(vfs, name);
int ret;
CHECK_AND_CALL(ret, r, vfs, mkdir, path_within_vfs, mode);
return ret;
}
int rmdir(const char* name)
{
const vfs_entry_t* vfs = get_vfs_for_path(name);
struct _reent* r = __getreent();
if (vfs == NULL) {
__errno_r(r) = ENOENT;
return -1;
}
const char* path_within_vfs = translate_path(vfs, name);
int ret;
CHECK_AND_CALL(ret, r, vfs, rmdir, path_within_vfs);
return ret;
}
int fcntl(int fd, int cmd, ...)
{
const vfs_entry_t* vfs = get_vfs_for_fd(fd);
const int local_fd = get_local_fd(vfs, fd);
struct _reent* r = __getreent();
if (vfs == NULL || local_fd < 0) {
__errno_r(r) = EBADF;
return -1;
}
int ret;
va_list args;
va_start(args, cmd);
CHECK_AND_CALL(ret, r, vfs, fcntl, local_fd, cmd, args);
va_end(args);
return ret;
}
int ioctl(int fd, int cmd, ...)
{
const vfs_entry_t* vfs = get_vfs_for_fd(fd);
const int local_fd = get_local_fd(vfs, fd);
struct _reent* r = __getreent();
if (vfs == NULL || local_fd < 0) {
__errno_r(r) = EBADF;
return -1;
}
int ret;
va_list args;
va_start(args, cmd);
CHECK_AND_CALL(ret, r, vfs, ioctl, local_fd, cmd, args);
va_end(args);
return ret;
}
int fsync(int fd)
{
const vfs_entry_t* vfs = get_vfs_for_fd(fd);
const int local_fd = get_local_fd(vfs, fd);
struct _reent* r = __getreent();
if (vfs == NULL || local_fd < 0) {
__errno_r(r) = EBADF;
return -1;
}
int ret;
CHECK_AND_CALL(ret, r, vfs, fsync, local_fd);
return ret;
}
int access(const char *path, int amode)
{
int ret;
const vfs_entry_t* vfs = get_vfs_for_path(path);
struct _reent* r = __getreent();
if (vfs == NULL) {
__errno_r(r) = ENOENT;
return -1;
}
const char* path_within_vfs = translate_path(vfs, path);
CHECK_AND_CALL(ret, r, vfs, access, path_within_vfs, amode);
return ret;
}
static void call_end_selects(int end_index, const fds_triple_t *vfs_fds_triple)
{
for (int i = 0; i < end_index; ++i) {
const vfs_entry_t *vfs = get_vfs_for_index(i);
const fds_triple_t *item = &vfs_fds_triple[i];
if (vfs && vfs->vfs.end_select && item->isset) {
vfs->vfs.end_select();
}
}
}
static inline bool esp_vfs_safe_fd_isset(int fd, const fd_set *fds)
{
return fds && FD_ISSET(fd, fds);
}
static int set_global_fd_sets(const fds_triple_t *vfs_fds_triple, int size, fd_set *readfds, fd_set *writefds, fd_set *errorfds)
{
int ret = 0;
for (int i = 0; i < size; ++i) {
const fds_triple_t *item = &vfs_fds_triple[i];
if (item->isset) {
for (int fd = 0; fd < MAX_FDS; ++fd) {
const int local_fd = s_fd_table[fd].local_fd; // single read -> no locking is required
if (readfds && esp_vfs_safe_fd_isset(local_fd, &item->readfds)) {
ESP_LOGD(TAG, "FD %d in readfds was set from VFS ID %d", fd, i);
FD_SET(fd, readfds);
++ret;
}
if (writefds && esp_vfs_safe_fd_isset(local_fd, &item->writefds)) {
ESP_LOGD(TAG, "FD %d in writefds was set from VFS ID %d", fd, i);
FD_SET(fd, writefds);
++ret;
}
if (errorfds && esp_vfs_safe_fd_isset(local_fd, &item->errorfds)) {
ESP_LOGD(TAG, "FD %d in errorfds was set from VFS ID %d", fd, i);
FD_SET(fd, errorfds);
++ret;
}
}
}
}
return ret;
}
static void esp_vfs_log_fd_set(const char *fds_name, const fd_set *fds)
{
if (fds_name && fds) {
ESP_LOGD(TAG, "FDs in %s =", fds_name);
for (int i = 0; i < MAX_FDS; ++i) {
if (esp_vfs_safe_fd_isset(i, fds)) {
ESP_LOGD(TAG, "%d", i);
}
}
}
}
int esp_vfs_select(int nfds, fd_set *readfds, fd_set *writefds, fd_set *errorfds, struct timeval *timeout)
{
int ret = 0;
struct _reent* r = __getreent();
ESP_LOGD(TAG, "esp_vfs_select starts with nfds = %d", nfds);
if (timeout) {
ESP_LOGD(TAG, "timeout is %lds + %ldus", timeout->tv_sec, timeout->tv_usec);
}
esp_vfs_log_fd_set("readfds", readfds);
esp_vfs_log_fd_set("writefds", writefds);
esp_vfs_log_fd_set("errorfds", errorfds);
if (nfds > MAX_FDS || nfds < 0) {
ESP_LOGD(TAG, "incorrect nfds");
__errno_r(r) = EINVAL;
return -1;
}
fds_triple_t *vfs_fds_triple;
if ((vfs_fds_triple = calloc(s_vfs_count, sizeof(fds_triple_t))) == NULL) {
__errno_r(r) = ENOMEM;
ESP_LOGD(TAG, "calloc is unsuccessful");
return -1;
}
int (*socket_select)(int, fd_set *, fd_set *, fd_set *, struct timeval *) = NULL;
for (int fd = 0; fd < nfds; ++fd) {
_lock_acquire(&s_fd_table_lock);
const bool is_socket_fd = s_fd_table[fd].permanent;
const int vfs_index = s_fd_table[fd].vfs_index;
const int local_fd = s_fd_table[fd].local_fd;
_lock_release(&s_fd_table_lock);
if (vfs_index < 0) {
continue;
}
if (is_socket_fd) {
if (!socket_select) {
// no socket_select found yet so take a look
if (esp_vfs_safe_fd_isset(fd, readfds) ||
esp_vfs_safe_fd_isset(fd, writefds) ||
esp_vfs_safe_fd_isset(fd, errorfds)) {
const vfs_entry_t *vfs = s_vfs[vfs_index];
socket_select = vfs->vfs.socket_select;
}
}
continue;
}
fds_triple_t *item = &vfs_fds_triple[vfs_index]; // FD sets for VFS which belongs to fd
if (esp_vfs_safe_fd_isset(fd, readfds)) {
item->isset = true;
FD_SET(local_fd, &item->readfds);
FD_CLR(fd, readfds);
ESP_LOGD(TAG, "removing %d from readfds and adding as local FD %d to fd_set of VFS ID %d", fd, local_fd, vfs_index);
}
if (esp_vfs_safe_fd_isset(fd, writefds)) {
item->isset = true;
FD_SET(local_fd, &item->writefds);
FD_CLR(fd, writefds);
ESP_LOGD(TAG, "removing %d from writefds and adding as local FD %d to fd_set of VFS ID %d", fd, local_fd, vfs_index);
}
if (esp_vfs_safe_fd_isset(fd, errorfds)) {
item->isset = true;
FD_SET(local_fd, &item->errorfds);
FD_CLR(fd, errorfds);
ESP_LOGD(TAG, "removing %d from errorfds and adding as local FD %d to fd_set of VFS ID %d", fd, local_fd, vfs_index);
}
}
// all non-socket VFSs have their FD sets in vfs_fds_triple
// the global readfds, writefds and errorfds contain only socket FDs (if
// there any)
/* Semaphore used for waiting select events from other VFS drivers when socket
* select is not used (not registered or socket FDs are not observed by the
* given call of select)
*/
SemaphoreHandle_t select_sem = NULL;
if (!socket_select) {
// There is no socket VFS registered or select() wasn't called for
// any socket. Therefore, we will use our own signalization.
if ((select_sem = xSemaphoreCreateBinary()) == NULL) {
free(vfs_fds_triple);
__errno_r(r) = ENOMEM;
ESP_LOGD(TAG, "cannot create select_sem");
return -1;
}
}
for (int i = 0; i < s_vfs_count; ++i) {
const vfs_entry_t *vfs = get_vfs_for_index(i);
fds_triple_t *item = &vfs_fds_triple[i];
if (vfs && vfs->vfs.start_select && item->isset) {
// call start_select for all non-socket VFSs with has at least one FD set in readfds, writefds, or errorfds
// note: it can point to socket VFS but item->isset will be false for that
ESP_LOGD(TAG, "calling start_select for VFS ID %d with the following local FDs", i);
esp_vfs_log_fd_set("readfds", &item->readfds);
esp_vfs_log_fd_set("writefds", &item->writefds);
esp_vfs_log_fd_set("errorfds", &item->errorfds);
esp_err_t err = vfs->vfs.start_select(nfds, &item->readfds, &item->writefds, &item->errorfds, &select_sem);
if (err != ESP_OK) {
call_end_selects(i, vfs_fds_triple);
(void) set_global_fd_sets(vfs_fds_triple, s_vfs_count, readfds, writefds, errorfds);
if (select_sem) {
vSemaphoreDelete(select_sem);
select_sem = NULL;
}
free(vfs_fds_triple);
__errno_r(r) = EINTR;
ESP_LOGD(TAG, "start_select failed");
return -1;
}
}
}
if (socket_select) {
ESP_LOGD(TAG, "calling socket_select with the following FDs");
esp_vfs_log_fd_set("readfds", readfds);
esp_vfs_log_fd_set("writefds", writefds);
esp_vfs_log_fd_set("errorfds", errorfds);
ret = socket_select(nfds, readfds, writefds, errorfds, timeout);
ESP_LOGD(TAG, "socket_select returned %d and the FDs are the following", ret);
esp_vfs_log_fd_set("readfds", readfds);
esp_vfs_log_fd_set("writefds", writefds);
esp_vfs_log_fd_set("errorfds", errorfds);
} else {
if (readfds) {
FD_ZERO(readfds);
}
if (writefds) {
FD_ZERO(writefds);
}
if (errorfds) {
FD_ZERO(errorfds);
}
TickType_t ticks_to_wait = portMAX_DELAY;
if (timeout) {
uint32_t timeout_ms = timeout->tv_sec * 1000 + timeout->tv_usec / 1000;
ticks_to_wait = timeout_ms / portTICK_PERIOD_MS;
ESP_LOGD(TAG, "timeout is %dms", timeout_ms);
}
ESP_LOGD(TAG, "waiting without calling socket_select");
xSemaphoreTake(select_sem, ticks_to_wait);
}
call_end_selects(s_vfs_count, vfs_fds_triple); // for VFSs for start_select was called before
if (ret >= 0) {
ret += set_global_fd_sets(vfs_fds_triple, s_vfs_count, readfds, writefds, errorfds);
}
if (select_sem) {
vSemaphoreDelete(select_sem);
select_sem = NULL;
}
free(vfs_fds_triple);
ESP_LOGD(TAG, "esp_vfs_select returns %d", ret);
esp_vfs_log_fd_set("readfds", readfds);
esp_vfs_log_fd_set("writefds", writefds);
esp_vfs_log_fd_set("errorfds", errorfds);
return ret;
}
void esp_vfs_select_triggered(SemaphoreHandle_t *signal_sem)
{
if (signal_sem && (*signal_sem)) {
xSemaphoreGive(*signal_sem);
} else {
// Another way would be to go through s_fd_table and find the VFS
// which has a permanent FD. But in order to avoid to lock
// s_fd_table_lock we go through the VFS table.
for (int i = 0; i < s_vfs_count; ++i) {
const vfs_entry_t *vfs = s_vfs[i];
if (vfs != NULL && vfs->vfs.stop_socket_select != NULL) {
vfs->vfs.stop_socket_select();
break;
}
}
}
}
void esp_vfs_select_triggered_isr(SemaphoreHandle_t *signal_sem, BaseType_t *woken)
{
if (signal_sem && (*signal_sem)) {
xSemaphoreGiveFromISR(*signal_sem, woken);
} else {
// Another way would be to go through s_fd_table and find the VFS
// which has a permanent FD. But in order to avoid to lock
// s_fd_table_lock we go through the VFS table.
for (int i = 0; i < s_vfs_count; ++i) {
const vfs_entry_t *vfs = s_vfs[i];
if (vfs != NULL && vfs->vfs.stop_socket_select_isr != NULL) {
vfs->vfs.stop_socket_select_isr(woken);
break;
}
}
}
}