Standard I/O and Console Output

[中文]

ESP-IDF provides C standard I/O facilities, such as stdin, stdout, and stderr streams, as well as C standard library functions such as printf() which operate on these streams.

As common in POSIX systems, these streams are buffering wrappers around file descriptors:

  • stdin is a buffered stream for reading input from the user, wrapping file descriptor STDIN_FILENO (0).

  • stdout is a buffered stream for writing output to the user, wrapping STDOUT_FILENO (1).

  • stderr is a buffered stream for writing error messages to the user, wrapping STDERR_FILENO (2).

In ESP-IDF, there is no practical distinction between stdout and stderr, as both streams are sent to the same physical interface. Most applications will use only stdout. For example, ESP-IDF logging functions always write to stdout regardless of the log level.

The underlying stdin, stdout, and stderr file descriptors are implemented based on VFS drivers.

On ESP32-C2, ESP-IDF provides implementations of VFS drivers for I/O over:

  • UART

  • "Null" (no output)

Standard I/O is not limited to these options, though. See below on enabling custom destinations for standard I/O.

Note

The console path is separate from an application UART that you open with the UART driver on other pins. If a UART peripheral provides console output, a USB-to-UART bridge chip is required. Many development boards already include this bridge chip onboard for console communication between the PC and the default UART0. If a USB or USB Serial/JTAG peripheral provides console output, no adapter chip is needed. The USB connector on the development board connects directly to the chip's USB D+/D- pins. Traffic on other UART ports does not appear in idf.py monitor unless a given port is configured as the console. For details, see Establish Serial Connection with ESP32-C2.

Configuration

Built-in implementations of standard I/O can be selected using several Kconfig options:

  • CONFIG_ESP_CONSOLE_UART_DEFAULT — Enables UART with default options (pin numbers, baud rate) for standard I/O.

  • CONFIG_ESP_CONSOLE_UART_CUSTOM — Enables UART for standard I/O, with TX/RX pin numbers and baud rate configurable via Kconfig.

  • CONFIG_ESP_CONSOLE_NONE — Disables standard I/O. If this option is selected, stdin, stdout, and stderr will be mapped to /dev/null and won't produce any output or generate any input.

Enabling one of these options will cause the corresponding VFS driver to be built into the application and used to open stdin, stdout, and stderr streams. Data written to stdout and stderr will be sent over the selected interface, and input from the selected interface will be available on stdin.

Standard Streams and FreeRTOS Tasks

ESP-IDF provides two different implementations of standard I/O streams based on the selected LibC implementation defined by CONFIG_LIBC. The behavior of stdin, stdout, and stderr streams differs between these implementations, particularly regarding how they are shared across FreeRTOS tasks.

Common to both implementations, each stream (stdin, stdout, stderr) has a mutex associated with it to protect the stream from concurrent access by multiple tasks. For example, if two tasks are writing to stdout at the same time, the mutex ensures that the outputs from each task are not mixed together.

Newlib

In ESP-IDF, to save RAM, FILE objects for stdin, stdout, and stderr are shared between all FreeRTOS tasks, but the pointers to these objects are unique for every task. This means that:

  • It is possible to change stdin, stdout, and stderr for any given task without affecting other tasks, e.g., by doing stdin = fopen("/dev/uart/1", "r").

  • To change the default stdin, stdout, stderr streams for new tasks, modify _GLOBAL_REENT->_stdin (_stdout, _stderr) before creating the task.

  • Closing default stdin, stdout, or stderr using fclose closes the FILE stream object, which will affect all other tasks.

Picolibc

According to the POSIX standard, all default stdin, stdout, and stderr streams are global and shared between all FreeRTOS tasks. This means that:

  • Modifying stdin, stdout, or stderr will affect all other tasks. It is not possible to change standard I/O streams for specific tasks.

  • If a thread-local stream is needed, it should be implemented in the application code by opening a file stream and using it within tasks, e.g., fscanf(), fprintf(), etc.

Blocking and non-blocking I/O

UART

By default, UART VFS uses simplified functions for reading from and writing to UART. Writes busy-wait until all data is put into UART FIFO, and reads are non-blocking, returning only the data present in the FIFO. Due to this non-blocking read behavior, higher level C library calls, such as fscanf("%d\n", &var);, might not have desired results.

Applications which use the UART driver can instruct VFS to use the driver's interrupt driven, blocking read and write functions instead. This can be done using a call to the uart_vfs_dev_use_driver() function. It is also possible to revert to the basic non-blocking functions using a call to uart_vfs_dev_use_nonblocking().

When the interrupt-driven driver is installed, it is also possible to enable/disable non-blocking behavior using fcntl function with O_NONBLOCK flag.

Newline conversion

VFS drivers provide an optional newline conversion feature for input and output. Internally, most applications send and receive lines terminated by the LF (\n) character. Different terminal programs may require different line termination, such as CR or CRLF.

Applications can configure this behavior globally using the following Kconfig options:

It is also possible to configure line ending conversion for the specific VFS driver:

Buffering

By default, standard I/O streams are line buffered. This means that data written to the stream is not sent to the underlying device until a newline character is written, or the buffer is full. This means, for example, that if you call printf("Hello"), the text will not be sent to the UART until you call printf("\n") or the stream buffer fills up due to other prints.

This behavior can be changed using the setvbuf() function. For example, to disable buffering for stdout:

setvbuf(stdout, NULL, _IONBF, 0);

You can also use setvbuf() to increase the buffer size, or switch to fully buffered mode.

Custom channels for standard I/O

To send application output to a custom channel (for example, a WebSocket connection), it is possible to create a custom VFS driver. See the VFS documentation for details. The VFS driver has to implement at least the following functions:

  • open() and close()

  • write()

  • read() — only if the custom channel is also used for input

  • fstat() — recommended, to provide correct buffering behavior for the I/O streams

  • fcntl() — only if non-blocking I/O has to be supported

Once you have created a custom VFS driver, use esp_vfs_register_fs() to register it with VFS. Then, use fopen() to redirect stdout and stderr to the custom channel. For example:

FILE *f = fopen("/dev/mychannel", "w");
if (f == NULL) {
    // handle the error here
}
stdout = f;
stderr = f;

Note that logging functions (ESP_LOGE(), etc.) write their output to stdout. Keep this in mind when using logging within the implementation of your custom VFS (or any components which it calls). For example, if the custom VFS driver's write() operation fails and uses ESP_LOGE() to log the error, this will cause the output to be sent to stdout, which would again call the custom VFS driver's write() operation. This would result in an infinite loop. It is recommended to keep track of this re-entry condition in the VFS driver's write() implementation, and return immediately if the write operation is still in progress.

Console I/O multiplexer

In addition to redirecting stdout and stderr manually (as shown above), ESP-IDF provides a console I/O multiplexer that lets an application register one or more custom VFS backends and switch between them at runtime, without reassigning the stdin/stdout/stderr streams itself.

The multiplexer is exposed on /dev/console and manages backends in two roles:

  • Primary — the active read and write backend. Both application input (stdin) and output (stdout/stderr) go through it. Primaries are kept on a stack: pushing a new primary suspends the previous one, and popping restores it. The default console selected in Kconfig sits at the base of the stack and is always kept as the ultimate fallback.

  • Auxiliary — a write-only sink. Every byte written to stdout and stderr is fanned out to all registered auxiliaries in addition to the primary. This is useful, for example, to mirror console output to a log file or a network connection while keeping the interactive console on the physical interface.

The maximum number of backends that can be registered at once (primaries plus auxiliaries, including the Kconfig default) is set by CONFIG_ESP_STDIO_MAX_VFS_ENTRIES.

API overview

The multiplexer API is declared in esp_stdio.h and is available when CONFIG_VFS_SUPPORT_IO is enabled:

  • esp_stdio_register_io() — register a VFS backend as a write-only auxiliary and return an opaque handle. The backend starts receiving the write fan-out immediately.

  • esp_stdio_push_primary() — promote a previously registered handle to the active read and write primary. The current primary is suspended on the stack.

  • esp_stdio_pop_primary() — remove a primary from the stack, returning it to auxiliary status. Passing NULL removes the current (top) primary; passing a specific handle removes it from wherever it sits in the stack.

  • esp_stdio_unregister_io() — remove a backend from the multiplexer entirely. If it is currently the active primary, it is popped first.

The backend is described by an esp_stdio_io_config_t structure:

  • vfs_ops — pointer to the VFS operations table (must not be NULL).

  • vfs_ctx — context pointer forwarded to every VFS callback.

  • path — path passed to open() inside the driver, for example "/".

Note

The ops table, context, and path string passed in the configuration must remain valid until the backend is removed with esp_stdio_unregister_io().

Example

The following example registers a custom backend, makes it the active console, and later restores the previous one:

#include "esp_stdio.h"

// A VFS operations table implemented by the application.
extern const esp_vfs_fs_ops_t my_console_vfs;

esp_stdio_handle_t handle;
esp_stdio_io_config_t config = {
    .vfs_ops = &my_console_vfs,
    .vfs_ctx = NULL,
    .path = "/",
};

// Register as a write-only auxiliary: output is now mirrored to it.
ESP_ERROR_CHECK(esp_stdio_register_io(&config, &handle));

// Promote it to the active read + write primary.
ESP_ERROR_CHECK(esp_stdio_push_primary(handle));

// ... application uses the custom console for stdin/stdout/stderr ...

// Restore the previous primary (the custom backend reverts to auxiliary).
ESP_ERROR_CHECK(esp_stdio_pop_primary(handle));

// Remove the backend from the multiplexer entirely.
ESP_ERROR_CHECK(esp_stdio_unregister_io(handle));

The same re-entrancy caveat described above for logging applies here: the backend's write() implementation must not trigger logging that would recursively write to the console.

Was this page helpful?