lighttpd1.4/src/mod_scgi.c

#include <sys/types.h>
#include <unistd.h>
#include <errno.h>
#include <fcntl.h>
#include <string.h>
#include <stdlib.h>
#include <ctype.h>
#include <assert.h>
#include <signal.h>

#include "buffer.h"
#include "server.h"
#include "keyvalue.h"
#include "log.h"

#include "http_chunk.h"
#include "fdevent.h"
#include "connections.h"
#include "response.h"
#include "joblist.h"

#include "plugin.h"

#include "inet_ntop_cache.h"

#include <stdio.h>

#ifdef HAVE_SYS_FILIO_H
# include <sys/filio.h>
#endif

#include "sys-socket.h"

#ifdef HAVE_SYS_UIO_H
#include <sys/uio.h>
#endif
#ifdef HAVE_SYS_WAIT_H
#include <sys/wait.h>
#endif

#include "version.h"

enum {EOL_UNSET, EOL_N, EOL_RN};

/*
 *
 * TODO:
 *
 * - add timeout for a connect to a non-scgi process
 *   (use state_timestamp + state)
 *
 */

typedef struct scgi_proc {
	size_t id; /* id will be between 1 and max_procs */
	buffer *socket; /* config.socket + "-" + id */
	unsigned port;  /* config.port + pno */

	pid_t pid;   /* PID of the spawned process (0 if not spawned locally) */


	size_t load; /* number of requests waiting on this process */

	time_t last_used; /* see idle_timeout */
	size_t requests;  /* see max_requests */
	struct scgi_proc *prev, *next; /* see first */

	time_t disable_ts; /* replace by host->something */

	int is_local;

	enum { PROC_STATE_UNSET,            /* init-phase */
			PROC_STATE_RUNNING, /* alive */
			PROC_STATE_DIED_WAIT_FOR_PID,
			PROC_STATE_KILLED,  /* was killed as we don't have the load anymore */
			PROC_STATE_DIED,    /* marked as dead, should be restarted */
			PROC_STATE_DISABLED /* proc disabled as it resulted in an error */
	} state;
} scgi_proc;

typedef struct {
	/* list of processes handling this extension
	 * sorted by lowest load
	 *
	 * whenever a job is done move it up in the list
	 * until it is sorted, move it down as soon as the
	 * job is started
	 */
	scgi_proc *first;
	scgi_proc *unused_procs;

	/*
	 * spawn at least min_procs, at max_procs.
	 *
	 * as soon as the load of the first entry
	 * is max_load_per_proc we spawn a new one
	 * and add it to the first entry and give it
	 * the load
	 *
	 */

	unsigned short min_procs;
	unsigned short max_procs;
	size_t num_procs;    /* how many procs are started */
	size_t active_procs; /* how many of them are really running */

	unsigned short max_load_per_proc;

	/*
	 * kick the process from the list if it was not
	 * used for idle_timeout until min_procs is
	 * reached. this helps to get the processlist
	 * small again we had a small peak load.
	 *
	 */

	unsigned short idle_timeout;

	/*
	 * time after a disabled remote connection is tried to be re-enabled
	 *
	 *
	 */

	unsigned short disable_time;

	/*
	 * same scgi processes get a little bit larger
	 * than wanted. max_requests_per_proc kills a
	 * process after a number of handled requests.
	 *
	 */
	size_t max_requests_per_proc;


	/* config */

	/*
	 * host:port
	 *
	 * if host is one of the local IP adresses the
	 * whole connection is local
	 *
	 * if tcp/ip should be used host AND port have
	 * to be specified
	 *
	 */
	buffer *host;
	unsigned short port;

	/*
	 * Unix Domain Socket
	 *
	 * instead of TCP/IP we can use Unix Domain Sockets
	 * - more secure (you have fileperms to play with)
	 * - more control (on locally)
	 * - more speed (no extra overhead)
	 */
	buffer *unixsocket;

	/* if socket is local we can start the scgi
	 * process ourself
	 *
	 * bin-path is the path to the binary
	 *
	 * check min_procs and max_procs for the number
	 * of process to start-up
	 */
	buffer *bin_path;

	/* bin-path is set bin-environment is taken to
	 * create the environement before starting the
	 * FastCGI process
	 *
	 */
	array *bin_env;

	array *bin_env_copy;

	/*
	 * docroot-translation between URL->phys and the
	 * remote host
	 *
	 * reasons:
	 * - different dir-layout if remote
	 * - chroot if local
	 *
	 */
	buffer *docroot;

	/*
	 * check_local tell you if the phys file is stat()ed
	 * or not. FastCGI doesn't care if the service is
	 * remote. If the web-server side doesn't contain
	 * the scgi-files we should not stat() for them
	 * and say '404 not found'.
	 */
	unsigned short check_local;

	/*
	 * append PATH_INFO to SCRIPT_FILENAME
	 *
	 * php needs this if cgi.fix_pathinfo is provied
	 *
	 */

	/*
	 * workaround for program when prefix="/"
	 *
	 * rule to build PATH_INFO is hardcoded for when check_local is disabled
	 * enable this option to use the workaround
	 *
	 */

	unsigned short fix_root_path_name;
	ssize_t load; /* replace by host->load */

	size_t max_id; /* corresponds most of the time to
	num_procs.

	only if a process is killed max_id waits for the process itself
	to die and decrements its afterwards */
} scgi_extension_host;

/*
 * one extension can have multiple hosts assigned
 * one host can spawn additional processes on the same
 *   socket (if we control it)
 *
 * ext -> host -> procs
 *    1:n     1:n
 *
 * if the scgi process is remote that whole goes down
 * to
 *
 * ext -> host -> procs
 *    1:n     1:1
 *
 * in case of PHP and FCGI_CHILDREN we have again a procs
 * but we don't control it directly.
 *
 */

typedef struct {
	buffer *key; /* like .php */

	int note_is_sent;
	scgi_extension_host **hosts;

	size_t used;
	size_t size;
} scgi_extension;

typedef struct {
	scgi_extension **exts;

	size_t used;
	size_t size;
} scgi_exts;


typedef struct {
	scgi_exts *exts;

	int debug;
} plugin_config;

typedef struct {
	char **ptr;

	size_t size;
	size_t used;
} char_array;

/* generic plugin data, shared between all connections */
typedef struct {
	PLUGIN_DATA;

	buffer *scgi_env;

	buffer *path;
	buffer *parse_response;

	plugin_config **config_storage;

	plugin_config conf; /* this is only used as long as no handler_ctx is setup */
} plugin_data;

/* connection specific data */
typedef enum { FCGI_STATE_INIT, FCGI_STATE_CONNECT, FCGI_STATE_PREPARE_WRITE,
		FCGI_STATE_WRITE, FCGI_STATE_READ
} scgi_connection_state_t;

typedef struct {
	buffer  *response;
	size_t   response_len;
	int      response_type;
	int      response_padding;

	scgi_proc *proc;
	scgi_extension_host *host;

	scgi_connection_state_t state;
	time_t   state_timestamp;

	int      reconnects; /* number of reconnect attempts */

	read_buffer *rb;
	chunkqueue *wb;

	buffer   *response_header;

	int       delayed;   /* flag to mark that the connect() is delayed */

	size_t    request_id;
	int       fd;        /* fd to the scgi process */
	int       fde_ndx;   /* index into the fd-event buffer */

	pid_t     pid;
	int       got_proc;

	plugin_config conf;

	connection *remote_conn;  /* dumb pointer */
	plugin_data *plugin_data; /* dumb pointer */
} handler_ctx;


/* ok, we need a prototype */
static handler_t scgi_handle_fdevent(void *s, void *ctx, int revents);

int scgi_proclist_sort_down(server *srv, scgi_extension_host *host, scgi_proc *proc);



static handler_ctx * handler_ctx_init() {
	handler_ctx * hctx;

	hctx = calloc(1, sizeof(*hctx));
	assert(hctx);

	hctx->fde_ndx = -1;

	hctx->response = buffer_init();
	hctx->response_header = buffer_init();

	hctx->request_id = 0;
	hctx->state = FCGI_STATE_INIT;
	hctx->proc = NULL;

	hctx->response_len = 0;
	hctx->response_type = 0;
	hctx->response_padding = 0;
	hctx->fd = -1;

	hctx->reconnects = 0;

	hctx->wb = chunkqueue_init();

	return hctx;
}

static void handler_ctx_free(handler_ctx *hctx) {
	buffer_free(hctx->response);
	buffer_free(hctx->response_header);

	chunkqueue_free(hctx->wb);

	if (hctx->rb) {
		if (hctx->rb->ptr) free(hctx->rb->ptr);
		free(hctx->rb);
	}

	free(hctx);
}

static scgi_proc *scgi_process_init() {
	scgi_proc *f;

	f = calloc(1, sizeof(*f));
	f->socket = buffer_init();

	f->prev = NULL;
	f->next = NULL;

	return f;
}

static void scgi_process_free(scgi_proc *f) {
	if (!f) return;

	scgi_process_free(f->next);

	buffer_free(f->socket);

	free(f);
}

static scgi_extension_host *scgi_host_init() {
	scgi_extension_host *f;

	f = calloc(1, sizeof(*f));

	f->host = buffer_init();
	f->unixsocket = buffer_init();
	f->docroot = buffer_init();
	f->bin_path = buffer_init();
	f->bin_env = array_init();
	f->bin_env_copy = array_init();

	return f;
}

static void scgi_host_free(scgi_extension_host *h) {
	if (!h) return;

	buffer_free(h->host);
	buffer_free(h->unixsocket);
	buffer_free(h->docroot);
	buffer_free(h->bin_path);
	array_free(h->bin_env);
	array_free(h->bin_env_copy);

	scgi_process_free(h->first);
	scgi_process_free(h->unused_procs);

	free(h);

}

static scgi_exts *scgi_extensions_init() {
	scgi_exts *f;

	f = calloc(1, sizeof(*f));

	return f;
}

static void scgi_extensions_free(scgi_exts *f) {
	size_t i;

	if (!f) return;

	for (i = 0; i < f->used; i++) {
		scgi_extension *fe;
		size_t j;

		fe = f->exts[i];

		for (j = 0; j < fe->used; j++) {
			scgi_extension_host *h;

			h = fe->hosts[j];

			scgi_host_free(h);
		}

		buffer_free(fe->key);
		free(fe->hosts);

		free(fe);
	}

	free(f->exts);

	free(f);
}

static int scgi_extension_insert(scgi_exts *ext, buffer *key, scgi_extension_host *fh) {
	scgi_extension *fe;
	size_t i;

	/* there is something */

	for (i = 0; i < ext->used; i++) {
		if (buffer_is_equal(key, ext->exts[i]->key)) {
			break;
		}
	}

	if (i == ext->used) {
		/* filextension is new */
		fe = calloc(1, sizeof(*fe));
		assert(fe);
		fe->key = buffer_init();
		buffer_copy_string_buffer(fe->key, key);

		/* */

		if (ext->size == 0) {
			ext->size = 8;
			ext->exts = malloc(ext->size * sizeof(*(ext->exts)));
			assert(ext->exts);
		} else if (ext->used == ext->size) {
			ext->size += 8;
			ext->exts = realloc(ext->exts, ext->size * sizeof(*(ext->exts)));
			assert(ext->exts);
		}
		ext->exts[ext->used++] = fe;
	} else {
		fe = ext->exts[i];
	}

	if (fe->size == 0) {
		fe->size = 4;
		fe->hosts = malloc(fe->size * sizeof(*(fe->hosts)));
		assert(fe->hosts);
	} else if (fe->size == fe->used) {
		fe->size += 4;
		fe->hosts = realloc(fe->hosts, fe->size * sizeof(*(fe->hosts)));
		assert(fe->hosts);
	}

	fe->hosts[fe->used++] = fh;

	return 0;

}

INIT_FUNC(mod_scgi_init) {
	plugin_data *p;

	p = calloc(1, sizeof(*p));

	p->scgi_env = buffer_init();

	p->path = buffer_init();
	p->parse_response = buffer_init();

	return p;
}


FREE_FUNC(mod_scgi_free) {
	plugin_data *p = p_d;

	UNUSED(srv);

	buffer_free(p->scgi_env);
	buffer_free(p->path);
	buffer_free(p->parse_response);

	if (p->config_storage) {
		size_t i, j, n;
		for (i = 0; i < srv->config_context->used; i++) {
			plugin_config *s = p->config_storage[i];
			scgi_exts *exts;

			if (!s) continue;

			exts = s->exts;

			for (j = 0; j < exts->used; j++) {
				scgi_extension *ex;

				ex = exts->exts[j];

				for (n = 0; n < ex->used; n++) {
					scgi_proc *proc;
					scgi_extension_host *host;

					host = ex->hosts[n];

					for (proc = host->first; proc; proc = proc->next) {
						if (proc->pid != 0) kill(proc->pid, SIGTERM);

						if (proc->is_local &&
						    !buffer_is_empty(proc->socket)) {
							unlink(proc->socket->ptr);
						}
					}

					for (proc = host->unused_procs; proc; proc = proc->next) {
						if (proc->pid != 0) kill(proc->pid, SIGTERM);

						if (proc->is_local &&
						    !buffer_is_empty(proc->socket)) {
							unlink(proc->socket->ptr);
						}
					}
				}
			}

			scgi_extensions_free(s->exts);

			free(s);
		}
		free(p->config_storage);
	}

	free(p);

	return HANDLER_GO_ON;
}

static int env_add(char_array *env, const char *key, size_t key_len, const char *val, size_t val_len) {
	char *dst;
	size_t i;

	if (!key || !val) return -1;

	dst = malloc(key_len + val_len + 3);
	memcpy(dst, key, key_len);
	dst[key_len] = '=';
	/* add the \0 from the value */
	memcpy(dst + key_len + 1, val, val_len + 1);

	for (i = 0; i < env->used; i++) {
		if (0 == strncmp(dst, env->ptr[i], key_len + 1)) {
			/* don't care about free as we are in a forked child which is going to exec(...) */
			/* free(env->ptr[i]); */
			env->ptr[i] = dst;
			return 0;
		}
	}

	if (env->size == 0) {
		env->size = 16;
		env->ptr = malloc(env->size * sizeof(*env->ptr));
	} else if (env->size == env->used) {
		env->size += 16;
		env->ptr = realloc(env->ptr, env->size * sizeof(*env->ptr));
	}

	env->ptr[env->used++] = dst;

	return 0;
}

static int scgi_spawn_connection(server *srv,
				 plugin_data *p,
				 scgi_extension_host *host,
				 scgi_proc *proc) {
	int scgi_fd;
	int socket_type, status;
	struct timeval tv = { 0, 100 * 1000 };
#ifdef HAVE_SYS_UN_H
	struct sockaddr_un scgi_addr_un;
#endif
	struct sockaddr_in scgi_addr_in;
	struct sockaddr *scgi_addr;

	socklen_t servlen;

#ifndef HAVE_FORK
	return -1;
#endif

	if (p->conf.debug) {
		log_error_write(srv, __FILE__, __LINE__, "sdb",
				"new proc, socket:", proc->port, proc->socket);
	}

	if (!buffer_is_empty(proc->socket)) {
		memset(&scgi_addr, 0, sizeof(scgi_addr));

#ifdef HAVE_SYS_UN_H
		scgi_addr_un.sun_family = AF_UNIX;
		strcpy(scgi_addr_un.sun_path, proc->socket->ptr);

#ifdef SUN_LEN
		servlen = SUN_LEN(&scgi_addr_un);
#else
		/* stevens says: */
		servlen = proc->socket->used + sizeof(scgi_addr_un.sun_family);
#endif
		socket_type = AF_UNIX;
		scgi_addr = (struct sockaddr *) &scgi_addr_un;
#else
		log_error_write(srv, __FILE__, __LINE__, "s",
				"ERROR: Unix Domain sockets are not supported.");
		return -1;
#endif
	} else {
		scgi_addr_in.sin_family = AF_INET;

		if (buffer_is_empty(host->host)) {
			scgi_addr_in.sin_addr.s_addr = htonl(INADDR_ANY);
		} else {
			struct hostent *he;

			/* set a usefull default */
			scgi_addr_in.sin_addr.s_addr = htonl(INADDR_ANY);


			if (NULL == (he = gethostbyname(host->host->ptr))) {
				log_error_write(srv, __FILE__, __LINE__,
						"sdb", "gethostbyname failed: ",
						h_errno, host->host);
				return -1;
			}

			if (he->h_addrtype != AF_INET) {
				log_error_write(srv, __FILE__, __LINE__, "sd", "addr-type != AF_INET: ", he->h_addrtype);
				return -1;
			}

			if (he->h_length != sizeof(struct in_addr)) {
				log_error_write(srv, __FILE__, __LINE__, "sd", "addr-length != sizeof(in_addr): ", he->h_length);
				return -1;
			}

			memcpy(&(scgi_addr_in.sin_addr.s_addr), he->h_addr_list[0], he->h_length);

		}
		scgi_addr_in.sin_port = htons(proc->port);
		servlen = sizeof(scgi_addr_in);

		socket_type = AF_INET;
		scgi_addr = (struct sockaddr *) &scgi_addr_in;
	}

	if (-1 == (scgi_fd = socket(socket_type, SOCK_STREAM, 0))) {
		log_error_write(srv, __FILE__, __LINE__, "ss",
				"failed:", strerror(errno));
		return -1;
	}

	if (-1 == connect(scgi_fd, scgi_addr, servlen)) {
		/* server is not up, spawn in  */
		pid_t child;
		int val;

		if (!buffer_is_empty(proc->socket)) {
			unlink(proc->socket->ptr);
		}

		close(scgi_fd);

		/* reopen socket */
		if (-1 == (scgi_fd = socket(socket_type, SOCK_STREAM, 0))) {
			log_error_write(srv, __FILE__, __LINE__, "ss",
				"socket failed:", strerror(errno));
			return -1;
		}