No OneTemporary
Actions

Size

35 KB

Referenced Files

None

Subscribers

None

View Options

	diff --git a/c_src/exile.c b/c_src/exile.c
	index a9c5ea6..e706bc8 100644
	--- a/c_src/exile.c
	+++ b/c_src/exile.c
	@@ -1,589 +1,591 @@
	#ifndef _POSIX_C_SOURCE
	#define _POSIX_C_SOURCE 200809L
	#endif

	#include "erl_nif.h"
	#include <errno.h>
	#include <fcntl.h>
	#include <signal.h>
	#include <stdbool.h>
	#include <stdio.h>
	#include <string.h>
	#include <sys/types.h>
	#include <sys/wait.h>
	#include <unistd.h>

	//#define DEBUG

	#ifdef DEBUG
	#define debug(...) \
	do { \
	enif_fprintf(stderr, __VA_ARGS__); \
	enif_fprintf(stderr, "\n"); \
	} while (0)
	#define start_timing() ErlNifTime __start = enif_monotonic_time(ERL_NIF_USEC)
	#define elapsed_microseconds() (enif_monotonic_time(ERL_NIF_USEC) - __start)
	#else
	#define debug(...)
	#define start_timing()
	#define elapsed_microseconds() 0
	#endif

	#define error(...) \
	do { \
	enif_fprintf(stderr, __VA_ARGS__); \
	enif_fprintf(stderr, "\n"); \
	} while (0)

	#define MAKE_OK(term) enif_make_tuple2(env, ATOM_OK, term)
	#define MAKE_ERROR(term) enif_make_tuple2(env, ATOM_ERROR, term)

	#define GET_CTX(env, arg, ctx) \
	do { \
	ExilePriv *data = enif_priv_data(env); \
	if (enif_get_resource(env, arg, data->rt, (void **)&ctx) == false) { \
	return MAKE_ERROR(ATOM_INVALID_CTX); \
	} \
	} while (0);

	static const int PIPE_READ = 0;
	static const int PIPE_WRITE = 1;
	static const int PIPE_CLOSED = -1;
	static const int CMD_EXIT = -1;
	static const int MAX_ARGUMENTS = 20;
	static const int MAX_ARGUMENT_LEN = 1024;

	/* We are choosing an exit code which is not reserved see:
	* https://www.tldp.org/LDP/abs/html/exitcodes.html. */
	static const int FORK_EXEC_FAILURE = 125;

	static ERL_NIF_TERM ATOM_TRUE;
	static ERL_NIF_TERM ATOM_FALSE;
	static ERL_NIF_TERM ATOM_OK;
	static ERL_NIF_TERM ATOM_ERROR;
	static ERL_NIF_TERM ATOM_UNDEFINED;
	static ERL_NIF_TERM ATOM_INVALID_CTX;
	static ERL_NIF_TERM ATOM_PIPE_CLOSED;
	+static ERL_NIF_TERM ATOM_EAGAIN;

	/* command exit types */
	static ERL_NIF_TERM ATOM_EXIT;
	static ERL_NIF_TERM ATOM_SIGNALED;
	static ERL_NIF_TERM ATOM_STOPPED;

	enum exec_status {
	SUCCESS,
	PIPE_CREATE_ERROR,
	PIPE_FLAG_ERROR,
	FORK_ERROR,
	PIPE_DUP_ERROR,
	NULL_DEV_OPEN_ERROR,
	};

	enum exit_type { NORMAL_EXIT, SIGNALED, STOPPED };

	typedef struct ExilePriv {
	ErlNifResourceType *rt;
	} ExilePriv;

	typedef struct ExecContext {
	int cmd_input_fd;
	int cmd_output_fd;
	int exit_status; // can be exit status or signal number depending on exit_type
	enum exit_type exit_type;
	pid_t pid;
	} ExecContext;

	typedef struct StartProcessResult {
	bool success;
	int err;
	ExecContext context;
	} StartProcessResult;

	/* TODO: assert if the external process is exit (?) */
	static void rt_dtor(ErlNifEnv env, void obj) {
	debug("Exile rt_dtor called\n");
	}

	static void rt_stop(ErlNifEnv env, void obj, int fd, int is_direct_call) {
	debug("Exile rt_stop called\n");
	}

	static void rt_down(ErlNifEnv env, void obj, ErlNifPid *pid,
	ErlNifMonitor *monitor) {
	debug("Exile rt_down called\n");
	}

	static ErlNifResourceTypeInit rt_init = {rt_dtor, rt_stop, rt_down};

	static int set_flag(int fd, int flags) {
	return fcntl(fd, F_SETFL, fcntl(fd, F_GETFL) \| flags);
	}

	static void close_all(int pipes[2][2]) {
	for (int i = 0; i < 2; i++) {
	if (pipes[i][PIPE_READ] > 0)
	close(pipes[i][PIPE_READ]);
	if (pipes[i][PIPE_WRITE] > 0)
	close(pipes[i][PIPE_WRITE]);
	}
	}

	/* This is not ideal, but as of now there is no portable way to do this */
	static void close_all_fds() {
	int fd_limit = (int)sysconf(_SC_OPEN_MAX);
	for (int i = STDERR_FILENO + 1; i < fd_limit; i++)
	close(i);
	}

	static StartProcessResult start_proccess(char *args[], bool stderr_to_console) {
	StartProcessResult result = {.success = false};
	pid_t pid;
	int pipes[2][2] = {{0, 0}, {0, 0}};

	if (pipe(pipes[STDIN_FILENO]) == -1 \|\| pipe(pipes[STDOUT_FILENO]) == -1) {
	result.err = errno;
	perror("[exile] failed to create pipes");
	close_all(pipes);
	return result;
	}

	const int r_cmdin = pipes[STDIN_FILENO][PIPE_READ];
	const int w_cmdin = pipes[STDIN_FILENO][PIPE_WRITE];

	const int r_cmdout = pipes[STDOUT_FILENO][PIPE_READ];
	const int w_cmdout = pipes[STDOUT_FILENO][PIPE_WRITE];

	if (set_flag(r_cmdin, O_CLOEXEC) < 0 \|\| set_flag(w_cmdout, O_CLOEXEC) < 0 \|\|
	set_flag(w_cmdin, O_CLOEXEC \| O_NONBLOCK) < 0 \|\|
	set_flag(r_cmdout, O_CLOEXEC \| O_NONBLOCK) < 0) {
	result.err = errno;
	perror("[exile] failed to set flags for pipes");
	close_all(pipes);
	return result;
	}

	/* TODO: fork() can be expensive. especially in mac os. we should report
	* correct reduction cost for this to avoid potential scheduler collapse */
	switch (pid = fork()) {

	case -1:
	result.err = errno;
	perror("[exile] failed to fork");
	close_all(pipes);
	return result;

	case 0: // child

	close(STDIN_FILENO);
	close(STDOUT_FILENO);

	if (dup2(r_cmdin, STDIN_FILENO) < 0) {
	perror("[exile] failed to dup to stdin");

	/* We are assuming FORK_EXEC_FAILURE exit code wont be used by the command
	* we are running. Technically we can not assume any exit code here. The
	* parent can not differentiate between exit before `exec` and the normal
	* command exit.
	* One correct way to solve this might be to have a separate
	* pipe shared between child and parent and signaling the parent by
	* closing it or writing to it. */
	_exit(FORK_EXEC_FAILURE);
	}
	if (dup2(w_cmdout, STDOUT_FILENO) < 0) {
	perror("[exile] failed to dup to stdout");
	_exit(FORK_EXEC_FAILURE);
	}

	if (stderr_to_console != true) {
	close(STDERR_FILENO);
	int dev_null = open("/dev/null", O_WRONLY);

	if (dev_null == -1) {
	perror("[exile] failed to open /dev/null");
	_exit(FORK_EXEC_FAILURE);
	}

	if (dup2(dev_null, STDERR_FILENO) < 0) {
	perror("[exile] failed to dup stderr");
	_exit(FORK_EXEC_FAILURE);
	}

	close(dev_null);
	}

	close_all_fds();

	execvp(args[0], args);
	perror("[exile] execvp(): failed");

	_exit(FORK_EXEC_FAILURE);

	default: // parent
	/* close file descriptors used by child */
	close(r_cmdin);
	close(w_cmdout);

	result.success = true;
	result.context.pid = pid;
	result.context.cmd_input_fd = w_cmdin;
	result.context.cmd_output_fd = r_cmdout;

	return result;
	}
	}

	/* TODO: return appropriate error instead returning generic "badarg" error */
	static ERL_NIF_TERM exec_proc(ErlNifEnv *env, int argc,
	const ERL_NIF_TERM argv[]) {
	char tmp[MAX_ARGUMENTS][MAX_ARGUMENT_LEN + 1];
	char *exec_args[MAX_ARGUMENTS + 1];

	unsigned int args_len;
	if (enif_get_list_length(env, argv[0], &args_len) != true)
	return enif_make_badarg(env);

	if (args_len > MAX_ARGUMENTS)
	return enif_make_badarg(env);

	ERL_NIF_TERM head, tail, list = argv[0];
	for (unsigned int i = 0; i < args_len; i++) {
	if (enif_get_list_cell(env, list, &head, &tail) != true)
	return enif_make_badarg(env);

	if (enif_get_string(env, head, tmp[i], MAX_ARGUMENT_LEN, ERL_NIF_LATIN1) <
	1)
	return enif_make_badarg(env);
	exec_args[i] = tmp[i];
	list = tail;
	}
	exec_args[args_len] = NULL;

	bool stderr_to_console = true;
	int tmp_int;
	if (enif_get_int(env, argv[1], &tmp_int) != true)
	return enif_make_badarg(env);
	stderr_to_console = tmp_int == 1 ? true : false;

	struct ExilePriv *data = enif_priv_data(env);
	StartProcessResult result = start_proccess(exec_args, stderr_to_console);
	ExecContext *ctx = NULL;
	ERL_NIF_TERM term;

	if (result.success) {
	ctx = enif_alloc_resource(data->rt, sizeof(ExecContext));
	ctx->cmd_input_fd = result.context.cmd_input_fd;
	ctx->cmd_output_fd = result.context.cmd_output_fd;
	ctx->pid = result.context.pid;

	debug("pid: %d cmd_in_fd: %d cmd_out_fd: %d", ctx->pid, ctx->cmd_input_fd,
	ctx->cmd_output_fd);

	term = enif_make_resource(env, ctx);

	/* resource should be collected beam GC when there are no more references */
	enif_release_resource(ctx);

	return MAKE_OK(term);
	} else {
	return MAKE_ERROR(enif_make_int(env, result.err));
	}
	}

	static int select_write(ErlNifEnv env, ExecContext ctx) {
	int retval = enif_select(env, ctx->cmd_input_fd, ERL_NIF_SELECT_WRITE, ctx,
	NULL, ATOM_UNDEFINED);
	if (retval != 0)
	perror("select_write()");

	return retval;
	}

	static ERL_NIF_TERM write_proc(ErlNifEnv *env, int argc,
	const ERL_NIF_TERM argv[]) {
	if (argc != 2)
	enif_make_badarg(env);

	ExecContext *ctx = NULL;
	GET_CTX(env, argv[0], ctx);

	if (ctx->cmd_input_fd == PIPE_CLOSED)
	return MAKE_ERROR(ATOM_PIPE_CLOSED);

	ErlNifBinary bin;
	/* TODO: should not use enif_inspect_binary */
	if (enif_inspect_binary(env, argv[1], &bin) != true)
	return enif_make_badarg(env);

	unsigned int result = write(ctx->cmd_input_fd, bin.data, bin.size);

	/* TODO: branching is ugly, cleanup required */
	if (result >= bin.size) { // request completely satisfied
	return MAKE_OK(enif_make_int(env, result));
	} else if (result >= 0) { // request partially satisfied
	int retval = select_write(env, ctx);
	if (retval != 0)
	return MAKE_ERROR(enif_make_int(env, retval));
	return MAKE_OK(enif_make_int(env, result));
	} else if (errno == EAGAIN) { // busy
	int retval = select_write(env, ctx);
	if (retval != 0)
	return MAKE_ERROR(enif_make_int(env, retval));
	- return MAKE_ERROR(enif_make_int(env, EAGAIN));
	+ return MAKE_ERROR(ATOM_EAGAIN);
	} else { // Error
	perror("write()");
	return MAKE_ERROR(enif_make_int(env, errno));
	}
	}

	static ERL_NIF_TERM close_pipe(ErlNifEnv *env, int argc,
	const ERL_NIF_TERM argv[]) {
	ExecContext *ctx = NULL;
	GET_CTX(env, argv[0], ctx);

	int kind;
	enif_get_int(env, argv[1], &kind);

	int result;
	switch (kind) {
	case 0:
	if (ctx->cmd_input_fd == PIPE_CLOSED) {
	return ATOM_OK;
	} else {
	result = close(ctx->cmd_input_fd);
	if (result == 0) {
	ctx->cmd_input_fd = PIPE_CLOSED;
	return ATOM_OK;
	} else {
	perror("cmd_input_fd close()");
	return MAKE_ERROR(enif_make_int(env, errno));
	}
	}
	case 1:
	if (ctx->cmd_output_fd == PIPE_CLOSED) {
	return ATOM_OK;
	} else {
	result = close(ctx->cmd_output_fd);
	if (result == 0) {
	ctx->cmd_output_fd = PIPE_CLOSED;
	return ATOM_OK;
	} else {
	perror("cmd_output_fd close()");
	return MAKE_ERROR(enif_make_int(env, errno));
	}
	}
	default:
	debug("invalid file descriptor type");
	return enif_make_badarg(env);
	}
	}

	static int select_read(ErlNifEnv env, ExecContext ctx) {
	int retval = enif_select(env, ctx->cmd_output_fd, ERL_NIF_SELECT_READ, ctx,
	NULL, ATOM_UNDEFINED);
	if (retval != 0)
	perror("select_read()");
	return retval;
	}

	static ERL_NIF_TERM read_proc(ErlNifEnv *env, int argc,
	const ERL_NIF_TERM argv[]) {
	if (argc != 2)
	enif_make_badarg(env);

	ExecContext *ctx = NULL;
	GET_CTX(env, argv[0], ctx);

	if (ctx->cmd_output_fd == PIPE_CLOSED)
	return MAKE_ERROR(ATOM_PIPE_CLOSED);

	bool is_buffered = true;
	int size;
	enif_get_int(env, argv[1], &size);

	if (size == -1) {
	size = 65535;
	is_buffered = false;
	} else if (size > 65535 \|\| size < 1) {
	enif_make_badarg(env);
	}

	unsigned char buf[size];
	int result = read(ctx->cmd_output_fd, buf, sizeof(buf));

	ERL_NIF_TERM bin_term = 0;
	if (result >= 0) {
	ErlNifBinary bin;
	enif_alloc_binary(result, &bin);
	/* TODO: we should use the erl binary for `read` itself instead of
	* allocating again */
	memcpy(bin.data, buf, result);
	bin_term = enif_make_binary(env, &bin);
	}

	/* TODO: branching is ugly, cleanup required */
	if (result >= size \|\|
	(is_buffered == false && result >= 0)) { // request completely satisfied
	return MAKE_OK(bin_term);
	} else if (result > 0) { // request partially satisfied
	int retval = select_read(env, ctx);
	if (retval != 0)
	return MAKE_ERROR(enif_make_int(env, retval));
	return MAKE_OK(bin_term);
	} else if (result == 0) { // EOF
	return MAKE_OK(bin_term);
	} else if (errno == EAGAIN) { // busy
	int retval = select_read(env, ctx);
	if (retval != 0)
	return MAKE_ERROR(enif_make_int(env, retval));
	- return MAKE_ERROR(enif_make_int(env, EAGAIN));
	+ return MAKE_ERROR(ATOM_EAGAIN);
	} else { // Error
	perror("read()");
	return MAKE_ERROR(enif_make_int(env, errno));
	}
	}

	static ERL_NIF_TERM is_alive(ErlNifEnv *env, int argc,
	const ERL_NIF_TERM argv[]) {
	ExecContext *ctx = NULL;
	GET_CTX(env, argv[0], ctx);

	if (ctx->pid == CMD_EXIT)
	return ATOM_FALSE;

	int result = kill(ctx->pid, 0);

	if (result == 0) {
	return ATOM_TRUE;
	} else {
	return ATOM_FALSE;
	}
	}

	static ERL_NIF_TERM terminate_proc(ErlNifEnv *env, int argc,
	const ERL_NIF_TERM argv[]) {
	ExecContext *ctx = NULL;
	GET_CTX(env, argv[0], ctx);
	if (ctx->pid == CMD_EXIT)
	return MAKE_OK(enif_make_int(env, 0));

	return MAKE_OK(enif_make_int(env, kill(ctx->pid, SIGTERM)));
	}

	static ERL_NIF_TERM kill_proc(ErlNifEnv *env, int argc,
	const ERL_NIF_TERM argv[]) {
	ExecContext *ctx = NULL;
	GET_CTX(env, argv[0], ctx);
	if (ctx->pid == CMD_EXIT)
	return MAKE_OK(enif_make_int(env, 0));

	return MAKE_OK(enif_make_int(env, kill(ctx->pid, SIGKILL)));
	}

	static ERL_NIF_TERM make_exit_term(ErlNifEnv env, ExecContext ctx) {
	switch (ctx->exit_type) {
	case NORMAL_EXIT:
	return MAKE_OK(
	enif_make_tuple2(env, ATOM_EXIT, enif_make_int(env, ctx->exit_status)));
	case SIGNALED:
	/* exit_status here points to signal number */
	return MAKE_OK(enif_make_tuple2(env, ATOM_SIGNALED,
	enif_make_int(env, ctx->exit_status)));
	case STOPPED:
	return MAKE_OK(enif_make_tuple2(env, ATOM_STOPPED,
	enif_make_int(env, ctx->exit_status)));
	default:
	error("Invalid wait status");
	return MAKE_ERROR(ATOM_UNDEFINED);
	}
	}

	static ERL_NIF_TERM wait_proc(ErlNifEnv *env, int argc,
	const ERL_NIF_TERM argv[]) {
	ExecContext *ctx = NULL;
	GET_CTX(env, argv[0], ctx);

	if (ctx->pid == CMD_EXIT)
	return make_exit_term(env, ctx);

	int status;
	int wpid = waitpid(ctx->pid, &status, WNOHANG);

	if (wpid == ctx->pid) {
	ctx->pid = CMD_EXIT;

	if (WIFEXITED(status)) {
	ctx->exit_type = NORMAL_EXIT;
	ctx->exit_status = WEXITSTATUS(status);
	} else if (WIFSIGNALED(status)) {
	ctx->exit_type = SIGNALED;
	ctx->exit_status = WTERMSIG(status);
	} else if (WIFSTOPPED(status)) {
	ctx->exit_type = STOPPED;
	ctx->exit_status = 0;
	}

	return make_exit_term(env, ctx);
	} else if (wpid != 0) {
	perror("waitpid()");
	}
	ERL_NIF_TERM term = enif_make_tuple2(env, enif_make_int(env, wpid),
	enif_make_int(env, status));
	return MAKE_ERROR(term);
	}

	static ERL_NIF_TERM os_pid(ErlNifEnv *env, int argc,
	const ERL_NIF_TERM argv[]) {
	ExecContext *ctx = NULL;
	GET_CTX(env, argv[0], ctx);
	if (ctx->pid == CMD_EXIT)
	return MAKE_OK(enif_make_int(env, 0));

	return MAKE_OK(enif_make_int(env, ctx->pid));
	}

	static int on_load(ErlNifEnv env, void *priv, ERL_NIF_TERM load_info) {
	struct ExilePriv *data = enif_alloc(sizeof(struct ExilePriv));
	if (!data)
	return 1;

	data->rt =
	enif_open_resource_type_x(env, "exile_resource", &rt_init,
	ERL_NIF_RT_CREATE \| ERL_NIF_RT_TAKEOVER, NULL);

	ATOM_TRUE = enif_make_atom(env, "true");
	ATOM_FALSE = enif_make_atom(env, "false");
	ATOM_OK = enif_make_atom(env, "ok");
	ATOM_ERROR = enif_make_atom(env, "error");
	ATOM_UNDEFINED = enif_make_atom(env, "undefined");
	ATOM_INVALID_CTX = enif_make_atom(env, "invalid_exile_exec_ctx");
	ATOM_PIPE_CLOSED = enif_make_atom(env, "closed_pipe");
	ATOM_EXIT = enif_make_atom(env, "exit");
	ATOM_SIGNALED = enif_make_atom(env, "signaled");
	ATOM_STOPPED = enif_make_atom(env, "stopped");
	+ ATOM_EAGAIN = enif_make_atom(env, "eagain");

	priv = (void )data;

	return 0;
	}

	static void on_unload(ErlNifEnv env, void priv) {
	debug("exile unload");
	enif_free(priv);
	}

	/* TODO: we can use dirty schedulers conditionally at compile time by checking
	* if they are available or not (?) */
	static ErlNifFunc nif_funcs[] = {
	{"exec_proc", 2, exec_proc, 0},
	{"write_proc", 2, write_proc, 0},
	{"read_proc", 2, read_proc, 0},
	{"close_pipe", 2, close_pipe, 0},
	{"terminate_proc", 1, terminate_proc, 0},
	{"wait_proc", 1, wait_proc, 0},
	{"kill_proc", 1, kill_proc, 0},
	{"is_alive", 1, is_alive, 0},
	{"os_pid", 1, os_pid, 0},
	};

	ERL_NIF_INIT(Elixir.Exile.ProcessNif, nif_funcs, &on_load, NULL, NULL,
	&on_unload)
	diff --git a/lib/exile/process.ex b/lib/exile/process.ex
	index 8e2cfc7..6e1eaa2 100644
	--- a/lib/exile/process.ex
	+++ b/lib/exile/process.ex
	@@ -1,378 +1,377 @@
	defmodule Exile.Process do
	@moduledoc """
	GenServer which wraps spawned external command.

	One should use `Exile.stream!` over `Exile.Process`. stream internally manages this server for you. Use this only if you need more control over the life-cycle OS process.

	## Overview
	`Exile.Process` is an alternative primitive for Port. It has different interface and approach to running external programs to solve the issues associated with the ports.

	### When compared to Port
	* it is demand driven. User explicitly has to `read` output of the command and the progress of the external command is controlled using OS pipes. so unlike Port, this never cause memory issues in beam by loading more than we can consume
	* it can close stdin of the program explicitly
	* does not create zombie process. It always tries to cleanup resources

	At high level it makes non-blocking asynchronous system calls to execute and interact with the external program. It completely bypasses beam implementation for the same using NIF. It uses `select()` system call for asynchronous IO. Most of the system calls are non-blocking, so it does not has adverse effect on scheduler. Issues such as "scheduler collapse".

	### Obligatory NIF warning
	As with any NIF based solution, bugs or issues in Exile implementation can bring down the beam VM. But NIF implementation is comparatively small and mostly uses POSIX system calls, spawned external processes are still completely isolated at OS level and the port issues it tries to solve are critical.
	"""

	alias Exile.ProcessNif
	require Logger
	use GenServer

	- defmacro eagain(), do: 35
	defmacro fork_exec_failure(), do: 125

	# delay between retries when io is busy (in milliseconds)
	@default_opts %{io_busy_wait: 1, stderr_to_console: false}

	def start_link(cmd, args, opts \\ %{}) do
	opts = Map.merge(@default_opts, opts)
	GenServer.start(__MODULE__, %{cmd: cmd, args: args, opts: opts})
	end

	def close_stdin(process) do
	GenServer.call(process, :close_stdin, :infinity)
	end

	def write(process, binary) do
	GenServer.call(process, {:write, binary}, :infinity)
	end

	def read(process, size) when is_integer(size) do
	GenServer.call(process, {:read, size}, :infinity)
	end

	def read(process) do
	GenServer.call(process, {:read, nil}, :infinity)
	end

	def kill(process, signal) when signal in [:sigkill, :sigterm] do
	GenServer.call(process, {:kill, signal}, :infinity)
	end

	def await_exit(process, timeout \\ :infinity) do
	GenServer.call(process, {:await_exit, timeout}, :infinity)
	end

	def os_pid(process) do
	GenServer.call(process, :os_pid, :infinity)
	end

	def stop(process), do: GenServer.call(process, :stop, :infinity)

	## Server

	defmodule Pending do
	defstruct bin: [], remaining: 0, client_pid: nil
	end

	defstruct [
	:cmd,
	:cmd_args,
	:opts,
	:errno,
	:context,
	:status,
	await: %{},
	pending_read: nil,
	pending_write: nil
	]

	alias __MODULE__

	def init(%{cmd: cmd, args: args, opts: opts}) do
	path = :os.find_executable(to_charlist(cmd))

	unless path do
	raise "Command not found: #{cmd}"
	end

	state = %__MODULE__{
	cmd: path,
	cmd_args: args,
	opts: opts,
	errno: nil,
	status: :init,
	await: %{},
	pending_read: %Pending{},
	pending_write: %Pending{}
	}

	{:ok, state, {:continue, nil}}
	end

	def handle_continue(nil, state) do
	exec_args = Enum.map(state.cmd_args, &to_charlist/1)
	stderr_to_console = if state.opts.stderr_to_console, do: 1, else: 0

	case ProcessNif.exec_proc([state.cmd \| exec_args], stderr_to_console) do
	{:ok, context} ->
	start_watcher(context)
	{:noreply, %Process{state \| context: context, status: :start}}

	{:error, errno} ->
	raise "Failed to start command: #{state.cmd}, errno: #{errno}"
	end
	end

	def handle_call(:stop, _from, state) do
	# watcher will take care of termination of external process

	# TODO: pending write and read should receive "stopped" return
	# value instead of exit signal
	{:stop, :normal, :ok, state}
	end

	def handle_call(_, _from, %{status: {:exit, status}}), do: {:reply, {:error, {:exit, status}}}

	def handle_call({:await_exit, timeout}, from, state) do
	tref =
	if timeout != :infinity do
	Elixir.Process.send_after(self(), {:await_exit_timeout, from}, timeout)
	else
	nil
	end

	state = put_timer(state, from, :timeout, tref)
	check_exit(state, from)
	end

	def handle_call({:write, binary}, from, state) when is_binary(binary) do
	pending = %Pending{bin: binary, client_pid: from}
	do_write(%Process{state \| pending_write: pending})
	end

	def handle_call({:read, bytes}, from, state) do
	pending = %Pending{remaining: bytes, client_pid: from}
	do_read(%Process{state \| pending_read: pending})
	end

	def handle_call(:close_stdin, _from, state), do: do_close(state, :stdin)

	def handle_call(:os_pid, _from, state), do: {:reply, ProcessNif.os_pid(state.context), state}

	def handle_call({:kill, signal}, _from, state) do
	do_kill(state.context, signal)
	{:reply, :ok, %{state \| status: {:exit, :killed}}}
	end

	def handle_info({:check_exit, from}, state), do: check_exit(state, from)

	def handle_info({:await_exit_timeout, from}, state) do
	cancel_timer(state, from, :check)

	receive do
	{:check_exit, ^from} -> :ok
	after
	0 -> :ok
	end

	GenServer.reply(from, :timeout)
	{:noreply, clear_await(state, from)}
	end

	def handle_info({:select, context, _ref, :ready_output}, state) do
	do_write(%Process{state \| context: context})
	end

	def handle_info({:select, context, _ref, :ready_input}, state) do
	do_read(%Process{state \| context: context})
	end

	def handle_info(msg, _state), do: raise(msg)

	defp do_write(%Process{pending_write: pending} = state) do
	case ProcessNif.write_proc(state.context, pending.bin) do
	{:ok, size} ->
	if size < byte_size(pending.bin) do
	binary = binary_part(pending.bin, size, byte_size(pending.bin) - size)
	{:noreply, %{state \| pending_write: %Pending{bin: binary}}}
	else
	GenServer.reply(pending.client_pid, :ok)
	{:noreply, %{state \| pending_write: %Pending{}}}
	end

	- {:error, eagain()} ->
	+ {:error, :eagain} ->
	{:noreply, state}

	{:error, errno} ->
	GenServer.reply(pending.client_pid, {:error, errno})
	{:noreply, %{state \| errno: errno}}
	end
	end

	defp do_read(%Process{pending_read: %Pending{remaining: nil} = pending} = state) do
	case ProcessNif.read_proc(state.context, -1) do
	{:ok, <<>>} ->
	GenServer.reply(pending.client_pid, {:eof, []})
	{:noreply, state}

	{:ok, binary} ->
	GenServer.reply(pending.client_pid, {:ok, binary})
	{:noreply, state}

	- {:error, eagain()} ->
	+ {:error, :eagain} ->
	{:noreply, state}

	{:error, errno} ->
	GenServer.reply(pending.client_pid, {:error, errno})
	{:noreply, %{state \| errno: errno}}
	end
	end

	defp do_read(%Process{pending_read: pending} = state) do
	case ProcessNif.read_proc(state.context, pending.remaining) do
	{:ok, <<>>} ->
	GenServer.reply(pending.client_pid, {:eof, pending.bin})
	{:noreply, %Process{state \| pending_read: %Pending{}}}

	{:ok, binary} ->
	if byte_size(binary) < pending.remaining do
	pending = %Pending{
	pending
	\| bin: [pending.bin \| binary],
	remaining: pending.remaining - byte_size(binary)
	}

	{:noreply, %Process{state \| pending_read: pending}}
	else
	GenServer.reply(pending.client_pid, {:ok, [state.pending_read.bin \| binary]})
	{:noreply, %Process{state \| pending_read: %Pending{}}}
	end

	- {:error, eagain()} ->
	+ {:error, :eagain} ->
	{:noreply, state}

	{:error, errno} ->
	GenServer.reply(pending.client_pid, {:error, errno})
	{:noreply, %{state \| pending_read: %Pending{}, errno: errno}}
	end
	end

	defp check_exit(state, from) do
	case ProcessNif.wait_proc(state.context) do
	{:ok, {:exit, fork_exec_failure()}} ->
	GenServer.reply(from, {:error, :failed_to_execute})
	cancel_timer(state, from, :timeout)
	{:noreply, clear_await(state, from)}

	{:ok, status} ->
	GenServer.reply(from, {:ok, status})
	cancel_timer(state, from, :timeout)
	{:noreply, clear_await(state, from)}

	{:error, {0, _}} ->
	# Ideally we should not poll and we should handle this with SIGCHLD signal
	tref = Elixir.Process.send_after(self(), {:check_exit, from}, state.opts.io_busy_wait)
	{:noreply, put_timer(state, from, :check, tref)}

	{:error, {-1, status}} ->
	GenServer.reply(from, {:error, status})
	cancel_timer(state, from, :timeout)
	{:noreply, clear_await(state, from)}
	end
	end

	defp do_kill(context, :sigkill), do: ProcessNif.kill_proc(context)

	defp do_kill(context, :sigterm), do: ProcessNif.terminate_proc(context)

	defp do_close(state, type) do
	case ProcessNif.close_pipe(state.context, stream_type(type)) do
	:ok ->
	{:reply, :ok, state}

	{:error, errno} ->
	raise errno
	{:reply, {:error, errno}, %Process{state \| errno: errno}}
	end
	end

	defp clear_await(state, from) do
	%Process{state \| await: Map.delete(state.await, from)}
	end

	defp cancel_timer(state, from, key) do
	case get_timer(state, from, key) do
	nil -> :ok
	tref -> Elixir.Process.cancel_timer(tref)
	end
	end

	defp put_timer(state, from, key, timer) do
	if Map.has_key?(state.await, from) do
	await = put_in(state.await, [from, key], timer)
	%Process{state \| await: await}
	else
	%Process{state \| await: %{from => %{key => timer}}}
	end
	end

	defp get_timer(state, from, key), do: get_in(state.await, [from, key])

	# Try to gracefully terminate external proccess if the genserver associated with the process is killed
	defp start_watcher(context) do
	process_server = self()
	watcher_pid = spawn(fn -> watcher(process_server, context) end)

	receive do
	{^watcher_pid, :done} -> :ok
	end
	end

	defp stream_type(:stdin), do: 0
	defp stream_type(:stdout), do: 1

	defp process_exit?(context) do
	match?({:ok, _}, ProcessNif.wait_proc(context))
	end

	defp process_exit?(context, timeout) do
	if process_exit?(context) do
	true
	else
	:timer.sleep(timeout)
	process_exit?(context)
	end
	end

	# for proper process exit parent of the child must wait() for
	# child processes termination exit and "pickup" after the exit
	# (receive child exit_status). Resources acquired by child such as
	# file descriptors won't be released even if the child process
	# itself is terminated.
	defp watcher(process_server, context) do
	ref = Elixir.Process.monitor(process_server)
	send(process_server, {self(), :done})

	receive do
	{:DOWN, ^ref, :process, ^process_server, _reason} ->
	try do
	process_exit?(context) && throw(:done)

	Logger.debug(fn -> "Stopping external program" end)
	ProcessNif.close_pipe(context, stream_type(:stdin))
	ProcessNif.close_pipe(context, stream_type(:stdout))

	# at max we wait for 100ms for program to exit
	process_exit?(context, 100) && throw(:done)

	Logger.debug("Failed to stop external program gracefully. attempting SIGTERM")
	ProcessNif.terminate_proc(context)
	process_exit?(context, 100) && throw(:done)

	Logger.debug("Failed to stop external program with SIGTERM. attempting SIGKILL")
	ProcessNif.kill_proc(context)
	process_exit?(context, 1000) && throw(:done)

	Logger.error("[exile] failed to kill external process")
	raise "Failed to kill external process"
	catch
	:done -> Logger.debug(fn -> "Exited external program successfully" end)
	end
	end
	end
	end
	diff --git a/test/exile/process_test.exs b/test/exile/process_test.exs
	index c554e84..30d14d4 100644
	--- a/test/exile/process_test.exs
	+++ b/test/exile/process_test.exs
	@@ -1,172 +1,176 @@
	defmodule Exile.ProcessTest do
	use ExUnit.Case, async: true
	alias Exile.Process

	test "read" do
	{:ok, s} = Process.start_link("echo", ["test"])
	- assert {:ok, "test\n"} == Process.read(s)
	- assert {:eof, []} == Process.read(s)
	+ assert {:eof, iodata} = Process.read(s, 100)
	+ assert IO.iodata_to_binary(iodata) == "test\n"
	assert :ok == Process.close_stdin(s)
	assert {:ok, {:exit, 0}} == Process.await_exit(s, 500)
	end

	test "write" do
	{:ok, s} = Process.start_link("cat", [])
	assert :ok == Process.write(s, "hello")
	- assert {:ok, "hello"} == Process.read(s)
	+ assert {:ok, iodata} = Process.read(s, 5)
	+ assert IO.iodata_to_binary(iodata) == "hello"
	+
	assert :ok == Process.write(s, "world")
	- assert {:ok, "world"} == Process.read(s)
	+ assert {:ok, iodata} = Process.read(s, 5)
	+ assert IO.iodata_to_binary(iodata) == "world"
	+
	assert :ok == Process.close_stdin(s)
	assert {:eof, []} == Process.read(s)
	assert {:ok, {:exit, 0}} == Process.await_exit(s, 100)
	end

	test "stdin close" do
	logger = start_events_collector()

	# base64 produces output only after getting EOF from stdin. we
	# collect events in order and assert that we can still read from
	# stdout even after closing stdin
	{:ok, s} = Process.start_link("base64", [])

	# parallel reader should be blocked till we close stdin
	start_parallel_reader(s, logger)
	:timer.sleep(50)

	assert :ok == Process.write(s, "hello")
	add_event(logger, {:write, "hello"})
	assert :ok == Process.write(s, "world")
	add_event(logger, {:write, "world"})
	:timer.sleep(50)

	assert :ok == Process.close_stdin(s)
	add_event(logger, :input_close)
	assert {:ok, {:exit, 0}} == Process.await_exit(s, 50)

	assert [
	{:write, "hello"},
	{:write, "world"},
	:input_close,
	{:read, "aGVsbG93b3JsZA==\n"},
	:eof
	] == get_events(logger)
	end

	test "external command termination on stop" do
	{:ok, s} = Process.start_link("cat", [])
	{:ok, os_pid} = Process.os_pid(s)
	assert os_process_alive?(os_pid)

	Process.stop(s)
	:timer.sleep(100)

	refute os_process_alive?(os_pid)
	end

	test "external command kill on stop" do
	# cat command hangs waiting for EOF
	{:ok, s} = Process.start_link(fixture("ignore_sigterm.sh"), [])

	{:ok, os_pid} = Process.os_pid(s)
	assert os_process_alive?(os_pid)
	Process.stop(s)

	if os_process_alive?(os_pid) do
	:timer.sleep(3000)
	refute os_process_alive?(os_pid)
	else
	:ok
	end
	end

	test "exit status" do
	{:ok, s} = Process.start_link("sh", ~w(-c "exit 2"))
	assert {:ok, {:exit, 2}} == Process.await_exit(s, 500)
	end

	test "if we are leaking file descriptor" do
	# we are only printing FD, TYPE, NAME with respective prefix
	{:ok, s} = Process.start_link(fixture("opened_fds.sh"), [])
	:timer.seconds(100)
	- {:eof, iodata} = Process.read(s, 1000)
	+ {:eof, iodata} = Process.read(s, 10000)
	assert {:ok, {:exit, 0}} == Process.await_exit(s, 500)

	open_files = parse_lsof(iodata)
	assert [%{fd: "0", name: _, type: "PIPE"}, %{type: "PIPE", fd: "1", name: _}] = open_files
	end

	test "process kill with pending write" do
	{:ok, s} = Process.start_link("cat", [])
	{:ok, os_pid} = Process.os_pid(s)

	large_data =
	Stream.cycle(["test"]) \|> Stream.take(500_000) \|> Enum.to_list() \|> IO.iodata_to_binary()

	task =
	Task.async(fn ->
	try do
	Process.write(s, large_data)
	catch
	:exit, reason -> reason
	end
	end)

	:timer.sleep(200)
	Process.stop(s)
	:timer.sleep(3000)

	refute os_process_alive?(os_pid)
	assert {:normal, _} = Task.await(task)
	end

	def start_parallel_reader(proc_server, logger) do
	spawn_link(fn -> reader_loop(proc_server, logger) end)
	end

	def reader_loop(proc_server, logger) do
	case Process.read(proc_server) do
	{:ok, data} ->
	add_event(logger, {:read, data})
	reader_loop(proc_server, logger)

	{:eof, []} ->
	add_event(logger, :eof)
	end
	end

	def start_events_collector do
	{:ok, ordered_events} = Agent.start(fn -> [] end)
	ordered_events
	end

	def add_event(agent, event) do
	:ok = Agent.update(agent, fn events -> events ++ [event] end)
	end

	def get_events(agent) do
	Agent.get(agent, & &1)
	end

	defp os_process_alive?(pid) do
	match?({_, 0}, System.cmd("ps", ["-p", to_string(pid)]))
	end

	defp fixture(script) do
	Path.join([__DIR__, "../scripts", script])
	end

	defp parse_lsof(iodata) do
	String.split(IO.iodata_to_binary(iodata), "\n", trim: true)
	\|> Enum.reduce([], fn
	"f" <> fd, acc -> [%{fd: fd} \| acc]
	"t" <> type, [h \| acc] -> [Map.put(h, :type, type) \| acc]
	"n" <> name, [h \| acc] -> [Map.put(h, :name, name) \| acc]
	_, acc -> acc
	end)
	\|> Enum.reverse()
	\|> Enum.reject(fn
	%{fd: fd} when fd in ["255", "cwd", "txt"] -> true
	_ -> false
	end)
	end
	end

File Metadata

Mime Type: text/x-diff
Expires: Sat, Nov 30, 10:38 PM (1 d, 9 h)
Storage Engine: blob
Storage Format: Raw Data
Storage Handle: 41556
Default Alt Text: (35 KB)

No OneTemporaryActions

View Options

File Metadata

Event Timeline

No OneTemporary
Actions