--- parser3/src/targets/cgi/parser3.C 2026/09/27 19:40:38 1.371 +++ parser3/src/targets/cgi/parser3.C 2026/10/26 22:41:38 1.385 @@ -5,7 +5,7 @@ Authors: Konstantin Morshnev , Alexandr Petrosian */ -volatile const char * IDENT_PARSER3_C="$Id: parser3.C,v 1.371 2026/09/27 19:40:38 moko Exp $"; +volatile const char * IDENT_PARSER3_C="$Id: parser3.C,v 1.385 2026/10/26 22:41:38 moko Exp $"; #include "pa_config_includes.h" @@ -17,6 +17,8 @@ volatile const char * IDENT_PARSER3_C="$ #include "pa_threads.h" #include "pa_vconsole.h" #include "pa_sapi_info.h" +#include "pa_code_cache.h" +#include "pa_websocket.h" #ifdef _MSC_VER #include @@ -56,30 +58,26 @@ const char* parser3_mode = "cgi"; // $st static const char* filespec_to_process = 0; // [file] static const char* httpd_host_port = 0; // -p option static const char* config_filespec = 0; // -f option or from env or next to the executable if exists -static const char* log_filespec = 0; // -l option +static const char* log_filespec = 0; // -l option or from env, full disk path static bool mail_received = false; // -m option? [asked to parse incoming message to $mail:received] static const char* parser3_filespec = 0; // argv[0] static char** argv_extra = NULL; // for error logging static THREAD_LOCAL Request_info *request_info_4log = NULL; // global for correct log() reporting -static const char* filespec_4log = NULL; // null only if system-wide auto.p used +static const char* filespec_4log = NULL; // null only if system-wide auto.p used; full disk path, parser3.log goes next to it const char *parser3_log_filespec() { // $status:log-filename const char* slog=log_filespec; - if(!slog) - slog=getenv(PARSER_LOG_ENV_NAME); - if(!slog) - slog=getenv(REDIRECT_PREFIX PARSER_LOG_ENV_NAME); if(!slog) { - static char log_spec[MAX_STRING + 12 /* '/parser3.log' */]; + static THREAD_LOCAL char log_spec[MAX_STRING + 12 /* '/parser3.log' */]; pa_strncpy(log_spec, filespec_4log, MAX_STRING); - if(char* log_dir_pos=strrpbrk(log_spec, "/\\")){ + if(char* log_dir_pos=strrchr(log_spec, '/')){ strcpy(log_dir_pos, "/parser3.log"); } else { - // no path, just filename + // system-wide auto.p used: the current directory strcpy(log_spec, "./parser3.log"); } slog=log_spec; @@ -89,45 +87,57 @@ const char *parser3_log_filespec() { // // SAPI -static void pa_log(const char* fmt, va_list args) { - // use no memory [so that we could log out-of-memory error] - const char* slog=parser3_log_filespec(); - FILE *f=strcmp(slog,"-") ? fopen(slog, "at") : stderr; - - // fallback to stderr - if(!f) - f=stderr; +#ifndef O_CLOEXEC +#define O_CLOEXEC 0 +#endif +#ifndef STDERR_FILENO +#define STDERR_FILENO 2 +#endif - setbuf(f, 0); // stderr stream is unbuffered by default, but still... +static void pa_log(const char* stamp, const char* fmt, va_list args) { + // use no memory (to log out-of-memory error): the line is built on the stack and written at once + char line[64 + MAX_LOG_STRING + 1024 + 32 + 100]; // prefix + message + request info suffix: uri + method and the rest of it + size_t size=0; // prefix - time_t t=time(0); - if(const char* stamp=ctime(&t)) { // never saw that +#ifndef _MSC_VER + char stamp_buffer[32]; +#endif + if(!stamp){ + time_t t=time(0); +#ifdef _MSC_VER + stamp=ctime(&t); // the buffer is per thread +#else + stamp=ctime_r(&t, stamp_buffer); // the buffer of ctime() is shared by the threads +#endif + } + if(stamp) { // never saw that if(size_t len=strlen(stamp)) // saw once stamp being ="" - fprintf(f, "[%.*s] [%u] ", (int)len-1, stamp, (unsigned int)pa_get_thread_id() ); + size=pa_snprintf(line, 64, "[%.*s] [%u] ", (int)len-1 /* exclude \n */, stamp, (unsigned int)pa_get_thread_id() ); } // message - char buf[MAX_LOG_STRING]; - size_t size=vsnprintf(buf, MAX_LOG_STRING, fmt, args); - size=remove_crlf(buf, buf+size); - fwrite(buf, size, 1, f); - - if(request_info_4log && request_info_4log->method) { - fprintf(f, " [uri=%s, method=%s, cl=%lu]\n", request_info_4log->uri ? request_info_4log->uri : "", request_info_4log->method, (unsigned long)request_info_4log->content_length); - } else - fputs(" [no request info]\n", f); + size+=remove_crlf(line+size, line+size+pa_vsnprintf(line+size, MAX_LOG_STRING, fmt, args)); - if(f!=stderr) - fclose(f); + // suffix + if(request_info_4log && request_info_4log->method) + size+=pa_snprintf(line+size, sizeof(line)-size, " [uri=%.1024s, method=%.32s, cl=%lu]\n", request_info_4log->uri ? request_info_4log->uri : "", request_info_4log->method, (unsigned long)request_info_4log->content_length); else - fflush(f); + size+=pa_snprintf(line+size, sizeof(line)-size, " [no request info]\n"); + + const char* slog=parser3_log_filespec(); + int fd=strcmp(slog,"-") ? pa_open(slog, O_WRONLY | O_APPEND | O_CREAT | O_CLOEXEC, 0666) : -1; + if(fd<0) + fd=STDERR_FILENO; // fallback to stderr + PA_UNUSED int ignore_result=write(fd, line, (unsigned int)size); + if(fd!=STDERR_FILENO) + close(fd); } void pa_log(const char* fmt, ...) { va_list args; va_start(args,fmt); - pa_log(fmt, args); + pa_log(0, fmt, args); va_end(args); } @@ -135,7 +145,7 @@ void pa_log(const char* fmt, ...) { void SAPI::log(SAPI_Info&, const char* fmt, ...) { va_list args; va_start(args,fmt); - pa_log(fmt, args); + pa_log(0, fmt, args); va_end(args); } @@ -144,13 +154,13 @@ void SAPI::die(const char* fmt, ...) { // logging first, first vsnprintf va_start(args,fmt); - pa_log(fmt, args); + pa_log(0, fmt, args); va_end(args); // inform user, second vsnprintf va_start(args, fmt); char message[MAX_STRING]; - vsnprintf(message, MAX_STRING, fmt, args); + pa_vsnprintf(message, MAX_STRING, fmt, args); SAPI::send_error(sapi_info_4log ? *sapi_info_4log : *sapi_info, message); exit(1); @@ -194,17 +204,11 @@ size_t SAPI::send_body(SAPI_Info& info, } static const char* full_disk_path(const char* file_name = "") { - char* result; - if(file_name[0]=='/' || is_os_absolute_path(file_name)){ - result=pa_strdup(file_name); - } else { - char cwd[MAX_STRING]; - result=pa_strcat(getcwd(cwd, MAX_STRING) ? cwd : "", "/", file_name); - } -#ifdef WIN32 - back_slashes_to_slashes(result); -#endif - return result; + const char* path=backslashes_to_slashes(file_name); + if(path[0]=='/' || is_os_absolute_path(path)) // '/' includes the UNC path //server, as there are no web paths here; C: is left for is_os_absolute_path + return path; + char cwd[MAX_STRING]; + return pa_strcat(backslashes_to_slashes(getcwd(cwd, MAX_STRING) ? cwd : ""), "/", path); } static void log_signal(const char* signal_name) { @@ -234,17 +238,14 @@ static const char* locate_config(const c if(!filespec_4log) filespec_4log=getenv(REDIRECT_PREFIX PARSER_CONFIG_ENV_NAME); if(!filespec_4log){ - const char* exec_dir_pos = strrpbrk(executable_path, "/\\"); + const char* exec_dir_pos = strrchr(executable_path, '/'); #ifdef SYSTEM_CONFIG_FILE if(exec_dir_pos){ #endif - // next to the executable - if(!exec_dir_pos || (exec_dir_pos==executable_path+1 && *executable_path=='.')){ - // when just parser3 or ./parser3 full path should be used to avoid "parser already configured" - filespec_4log=full_disk_path(AUTO_FILE_NAME); - } else { - filespec_4log=pa_strcat(pa_strdup(executable_path, exec_dir_pos - executable_path), "/" AUTO_FILE_NAME); - } + // next to the executable; full path should be used to avoid "parser already configured" + filespec_4log=full_disk_path(!exec_dir_pos || (exec_dir_pos==executable_path+1 && *executable_path=='.') + ? AUTO_FILE_NAME // parser3 or ./parser3 + : pa_strcat(pa_strdup(executable_path, exec_dir_pos - executable_path), "/" AUTO_FILE_NAME)); if(entry_exists(filespec_4log)) return filespec_4log; #ifdef SYSTEM_CONFIG_FILE @@ -256,7 +257,8 @@ static const char* locate_config(const c #endif return NULL; } - return filespec_4log; + // when -f auto.p full path should be used to avoid "parser already configured" + return filespec_4log=full_disk_path(filespec_4log); } #ifdef WIN32 @@ -292,19 +294,11 @@ static const char* maybe_reconstruct_IIS return original; } -static const char* maybe_back_slashes_to_slashes(const char* original){ - char *result=pa_strdup(original); - back_slashes_to_slashes(result); - return result; -} - #define MAYBE_RECONSTRUCT_IIS_STATUS_IN_QS(s) maybe_reconstruct_IIS_status_in_qs(s) -#define MAYBE_BACK_SLASHES_TO_SLASHES(s) maybe_back_slashes_to_slashes(s) #else #define MAYBE_RECONSTRUCT_IIS_STATUS_IN_QS(s) s -#define MAYBE_BACK_SLASHES_TO_SLASHES(s) s #endif @@ -337,7 +331,7 @@ static void config_handler(SAPI_Info &in Request_info request_info; RequestInfoController ric(&request_info, &info); - request_info.document_root = full_disk_path(); + request_info.set_document_root(full_disk_path()); request_info.uri = ""; request_info.argv = argv_extra; @@ -346,7 +340,13 @@ static void config_handler(SAPI_Info &in // only once config_filespec = locate_config(config_filespec, parser3_filespec); // process main auto.p only - r.core(config_filespec, false, String::Empty); +#ifdef CODE_CACHE + Code_cache_manager::start(); +#endif + r.core(config_filespec, String::Empty); +#ifdef CODE_CACHE + Code_cache_manager::finish(r); +#endif } static void connection_handler(SAPI_Info_HTTPD &info, HTTPD_Connection &connection) { @@ -358,7 +358,7 @@ static void connection_handler(SAPI_Info return; // ignore "void" connections info.populate_env(); - request_info.document_root = full_disk_path(); + request_info.set_document_root(full_disk_path()); request_info.path_translated = filespec_to_process; request_info.method = connection.method(); request_info.query_string = connection.query(); @@ -371,8 +371,10 @@ static void connection_handler(SAPI_Info // prepare to process request Request r(info, request_info, String::Language(String::L_HTML|String::L_OPTIMIZE_BIT)); + WebSocket websocket(connection); + r.websocket=&websocket; // process the request - r.core(config_filespec, strcasecmp(request_info.method, "HEAD")==0, main_method_name, &httpd_class_name); + r.core(config_filespec, main_method_name, &httpd_class_name); } catch(const Exception& e) { // exception in connection handling SAPI::log(info, "%s", e.comment()); const char* status = info.exception_http_status(e.type()); @@ -386,6 +388,8 @@ DWORD WINAPI connection_thread(void *arg #else static void *connection_thread(void *arg){ #endif + Temp_gc_references gc_references; // for pointers GC does not see, the very start of the thread + HTTPD_Connection &connection=*(HTTPD_Connection*)arg; SAPI_Info_HTTPD info(connection); @@ -399,6 +403,22 @@ static void *connection_thread(void *arg return 0; } +static void reject_connection(HTTPD_Connection &connection) { + // main thread only + static unsigned long total=0; + static time_t logged=0; + + total++; + time_t now=time(0); + if(now!=logged){ // not more than once a second + logged=now; + pa_log("max-connections limit %d reached, rejecting connection from %s (%lu rejected in total)", HTTPD_Server::max_connections, connection.remote_addr, total); + } + + SAPI_Info_HTTPD info(connection); + SAPI::send_error(info, "too many connections", "503"); +} + static void httpd_mode() { config_handler(*sapi_info); @@ -411,20 +431,29 @@ static void httpd_mode() { while(1){ #ifndef _MSC_VER pid_t pid=1; - if(HTTPD_Server::mode == HTTPD_Server::PARALLEL) - while (waitpid((pid_t)(-1), 0, WNOHANG) > 0); #endif try { HTTPD_Connection connection; - if(!connection.accept(sock, 500)) + bool accepted=connection.accept(sock, 500); +#ifndef _MSC_VER + if(HTTPD_Server::mode == HTTPD_Server::PARALLEL) + while (waitpid((pid_t)(-1), 0, WNOHANG) > 0) + HTTPD_Server::track_connection(-1); // the child has exited, some of them while we were waiting +#endif + if(!accepted) continue; + if(HTTPD_Server::too_many_connections()){ + reject_connection(connection); // no thread or process is started + continue; + } + switch (HTTPD_Server::mode) { case HTTPD_Server::MULTITHREADED: #ifdef _MSC_VER if (!GC_CreateThread(0, 0, connection_thread, new HTTPD_Connection(connection), 0, 0)) throw Exception("httpd.fork", 0, "thread creation failed"); - connection.sock=INVALID_SOCKET; + connection.sock=INVALID_SOCKET; // thread owns the connection now break; #else #ifdef HAVE_TLS @@ -435,22 +464,25 @@ static void httpd_mode() { if(int result=GC_pthread_create(&thread, &attr, connection_thread, new HTTPD_Connection(connection))) throw Exception("httpd.fork", 0, "thread creation failed (%d)", result); - connection.sock=INVALID_SOCKET; + connection.sock=INVALID_SOCKET; // thread owns the connection now break; #endif case HTTPD_Server::PARALLEL: pid=fork(); if(pid<0) throw Exception("httpd.fork", 0, "fork failed: %s (%d)", strerror(errno), errno); - if(pid>0) - continue; // parent should close connection.sock as well + if(pid>0){ + closesocket(connection.sock); // parent should close connection.sock as well + connection.sock=INVALID_SOCKET; // child owns the connection now + continue; + } #endif case HTTPD_Server::SEQUENTIAL: // and fork child SAPI_Info_HTTPD info(connection); connection_handler(info, connection); } - // closing connection socket in HTTPD_Connection destructor + // SEQUENTIAL or error: closing connection socket in HTTPD_Connection destructor } catch(const Exception& e) { // exception in accept or send_error pa_log("%s", e.comment()); } @@ -475,7 +507,7 @@ static void real_parser_handler(bool cgi const char* request_method=getenv("REQUEST_METHOD"); if(!filespec_to_process){ - SAPI::send_error(*sapi_info, "Parser/" PARSER_VERSION); + SAPI::send_error(*sapi_info, "Parser/" PARSER_VERSION_WITH_FEATURES); exit(1); } @@ -493,17 +525,16 @@ static void real_parser_handler(bool cgi if(!path_info) SAPI::die("parser3: illegal CGI call (missing PATH_INFO)"); - request_info.document_root = getenv("DOCUMENT_ROOT"); - if(!request_info.document_root) { + const char* document_root = getenv("DOCUMENT_ROOT"); + if(!document_root) { // IIS or fcgiwrap minimalistic setup ssize_t prefix_len = strlen(filespec_to_process) - strlen(path_info); if(prefix_len < 0 || strcmp(filespec_to_process + prefix_len, path_info) != 0) SAPI::die("parser3: illegal CGI call (invalid PATH_INFO in reinventing DOCUMENT_ROOT)"); - char* document_root = new(PointerFreeGC) char[prefix_len + 1/*0*/]; - memcpy(document_root, filespec_to_process, prefix_len); document_root[prefix_len] = 0; - request_info.document_root = document_root; + document_root = pa_strdup(filespec_to_process, prefix_len); } + request_info.set_document_root(document_root); request_info.uri = request_info.strip_absolute_uri(getenv("REQUEST_URI")); if(request_info.uri) { // apache & others stuck to standards @@ -532,7 +563,7 @@ static void real_parser_handler(bool cgi request_info.uri = request_info.query_string && *request_info.query_string ? pa_strcat(path_info, "?", request_info.query_string) : path_info; } } else{ - request_info.document_root = full_disk_path(); + request_info.set_document_root(full_disk_path()); request_info.uri = ""; } @@ -547,14 +578,15 @@ static void real_parser_handler(bool cgi pa_log("request_info: method=%s, uri=%s, q=%s, dr=%s, pt=%s", request_info.method, request_info.uri, request_info.query_string, request_info.document_root, request_info.path_translated); #endif - // prepare to process request - Request r(*sapi_info, request_info, cgi ? String::Language(String::L_HTML|String::L_OPTIMIZE_BIT) : String::L_AS_IS); { + // prepare to process request + Request r(*sapi_info, request_info, cgi ? String::Language(String::L_HTML|String::L_OPTIMIZE_BIT) : String::L_AS_IS); // initing ::request ptr for signal handlers RequestController rc(&r); // process the request - r.core(locate_config(config_filespec, parser3_filespec), strcasecmp(request_info.method, "HEAD")==0); + r.core(locate_config(config_filespec, parser3_filespec)); // clearing ::request in RequestController destructor to prevent signal handlers from accessing invalid memory + // then the request is destroyed, before the libraries are unloaded: its pool cleanups call them (memcached) } // finalize libraries @@ -616,7 +648,7 @@ static void usage(const char* message=NU " -l log_file Use this log file (/path/to/parser3.log)\n" " -p [host:]port Start web server on this port\n" " -h Display usage information (this message)\n", - PARSER_VERSION, + PARSER_VERSION_WITH_FEATURES, parser3_filespec ? parser3_filespec : "parser3" ); exit(EINVAL); } @@ -628,13 +660,15 @@ static void usage(const char* message=NU } int main(int argc, char *argv[]) { + Temp_gc_references gc_references; // for pointers GC does not see + #ifdef PA_DEBUG_CGI_ENTRY_EXIT pa_log("main: entry"); #endif if(!argc || !argv[0]) usage(); - parser3_filespec = MAYBE_BACK_SLASHES_TO_SLASHES(argv[0]); + parser3_filespec = backslashes_to_slashes(argv[0]); umask(2); @@ -703,6 +737,14 @@ int main(int argc, char *argv[]) { argv_extra=carg; } + // prepare full disk path log_filespec here, as logging must not allocate + if(!log_filespec) + log_filespec=getenv(PARSER_LOG_ENV_NAME); + if(!log_filespec) + log_filespec=getenv(REDIRECT_PREFIX PARSER_LOG_ENV_NAME); + if(log_filespec && strcmp(log_filespec, "-")) // "-" is stderr + log_filespec=full_disk_path(log_filespec); + #ifdef _MSC_VER setmode(fileno(stdin), _O_BINARY); setmode(fileno(stdout), _O_BINARY);