From: Nick Kew Date: Sun, 26 Aug 2007 14:21:28 +0000 (+0000) Subject: Backport Chris's mod_dbd fixes X-Git-Tag: 2.2.6~114 X-Git-Url: http://git.ipfire.org/cgi-bin/gitweb.cgi?a=commitdiff_plain;h=f9669e2e2aafc97e9c3e9e1f14d49b17da6a11d0;p=thirdparty%2Fapache%2Fhttpd.git Backport Chris's mod_dbd fixes git-svn-id: https://svn.apache.org/repos/asf/httpd/httpd/branches/2.2.x@569818 13f79535-47bb-0310-9956-ffa450edef68 --- diff --git a/CHANGES b/CHANGES index 62f5a566736..bed6238a914 100644 --- a/CHANGES +++ b/CHANGES @@ -1,6 +1,31 @@  -*- coding: utf-8 -*- Changes with Apache 2.2.6 + *) mod_dbd: Introduce configuration groups to allow inheritance by virtual + hosts of database configurations from the main server. Determine the + minimal set of distinct configurations and share connection pools + whenever possible. Allow virtual hosts to override inherited SQL + statements. PR 41302. [Chris Darroch] + + *) mod_dbd: Create memory sub-pools for each DB connection and close + DB connections in a pool cleanup function. Ensure prepared statements + are destroyed before DB connection is closed. When using reslists, + prevent segfaults when child processes exit, and stop memory leakage + of ap_dbd_t structures. Avoid use of global s->process->pool, which + isn't destroyed by exiting child processes in most multi-process MPMs. + PR 39985. [Chris Darroch, Nick Kew] + + *) mod_dbd: Handle error conditions in dbd_construct() properly. + Simplify ap_dbd_open() and use correct arguments to apr_dbd_error() + when non-threaded. Register correct cleanup data in non-threaded + ap_dbd_acquire() and ap_dbd_cacquire(). Clean up configuration data + and merge function. Use ap_log_error() wherever possible. + [Chris Darroch, Nick Kew] + + *) mod_dbd: Stash DBD connections in request_config of initial request + only, or else sub-requests and internal redirections may cause + entire DBD pool to be stashed in a single HTTP request. [Chris Darroch] + *) main core: Emit errors during the initial apr_app_initialize() or apr_pool_create() (when apr-based error reporting is not ready). [William Rowe] diff --git a/STATUS b/STATUS index 045e52313a6..076555fd918 100644 --- a/STATUS +++ b/STATUS @@ -122,51 +122,6 @@ PATCHES PROPOSED TO BACKPORT FROM TRUNK: I'll +1 it when that's done. wrowe: ditto's - good concept. - * mod_dbd: initialise resources and prepared statements in all vhosts - PRs: 42327, 41302 - Trunk: N/A due to other changes (r503931 et al) - Patch by bojan: http://issues.apache.org/bugzilla/attachment.cgi?id=20161 - +1: niq - -1: chrisd: I think r503931 is actually the right thing to backport here, - which admittedly probably requires backporting r491729, r491884, - r492394 (trivial), and r496831 first. While more complex than a - quick patch, I think the explanation in the fourth section of this - post explains why there's no simple fix for the virtual-host - inheritance issues: - - http://marc.info/?l=apache-httpd-dev&m=116742014418304&w=2 - - Doing this work would also let us close at least three PRs. - PR 42327 seems like a duplicate of 41302 (both would be closed - by r503931), and r496831 would close PR 39985 as well. - - * mod_dbd: Rename functions and variables for consistency, and move - some functions around for readability, prior to making - functional changes for PRs 39985, 41302, and other issues. - Trunk version of patches: - http://svn.apache.org/viewvc?view=rev&revision=491729 - http://svn.apache.org/viewvc?view=rev&revision=492394 - (relevant portions only) - http://svn.apache.org/viewvc?view=rev&revision=521230 - 2.2.x version of patch: - http://people.apache.org/~chrisd/patches/mod_dbd_pools_groups/mod_dbd-1tidy-2.2.x.patch - +1: chrisd, fuankg - +1: niq (to chris's mod_dbd updates collectively) - - * mod_dbd: Handle error conditions in dbd_construct() properly. - Simplify ap_dbd_open() and use correct arguments to apr_dbd_error() - when non-threaded. Register correct cleanup data in non-threaded - ap_dbd_acquire() and ap_dbd_cacquire(). Clean up configuration data - and merge function. Use ap_log_error() wherever possible. - Trunk version of patches: - http://svn.apache.org/viewvc?view=rev&revision=491884 - http://svn.apache.org/viewvc?view=rev&revision=492394 - (relevant portions only) - 2.2.x version of patch: - http://people.apache.org/~chrisd/patches/mod_dbd_pools_groups/mod_dbd-2misc-2.2.x.patch - +1: chrisd, fuankg - +1: niq (to chris's mod_dbd updates collectively) - * beos MPM: Create pmain pool and run modules' child_init hooks when entering ap_mpm_run(), then destroy pmain when exiting ap_mpm_run(). Otherwise modules' child_init hooks appear to never be executed. @@ -178,42 +133,6 @@ PATCHES PROPOSED TO BACKPORT FROM TRUNK: http://people.apache.org/~chrisd/patches/mod_dbd_pools_groups/mpm_child_init-beos-2.2.x.patch +0: chrisd (abstaining; unable to test) - * mod_dbd: Create memory sub-pools for each DB connection and close - DB connections in a pool cleanup function. Ensure prepared statements - are destroyed before DB connection is closed. When using reslists, - prevent segfaults when child processes exit, and stop memory leakage - of ap_dbd_t structures. Avoid use of global s->process->pool, which - isn't destroyed by exiting child processes in most multi-process MPMs. - See also the third section of: - http://marc.info/?l=apache-httpd-dev&m=116742014418304&w=2 - and the eventual solution as described in the comments in the - dbd_setup() function. - PR: 39985 - Trunk version of patch: - http://svn.apache.org/viewvc?view=rev&revision=496831 - 2.2.x version of patch: - http://people.apache.org/~chrisd/patches/mod_dbd_pools_groups/mod_dbd-3pools-2.2.x.patch - +1: chrisd, fuankg - +1: niq (to chris's mod_dbd updates collectively) - - * mod_dbd: Introduce configuration groups to allow inheritance by virtual - hosts of database configurations from the main server. Determine the - minimal set of distinct configurations and share connection pools - whenever possible. Allow virtual hosts to override inherited SQL - statements. See also the fourth section of: - http://marc.info/?l=apache-httpd-dev&m=116742014418304&w=2 - and the eventual solution as implemented in the dbd_post_config() - function. Obviates r424798, r432560, r432562, and r466641. - Could be improved by the addition of configuration - sections. (Any takers? :-) - PR: 41302 - Trunk version of patch: - http://svn.apache.org/viewvc?view=rev&revision=503931 - 2.2.x version of patch: - http://people.apache.org/~chrisd/patches/mod_dbd_pools_groups/mod_dbd-4groups-2.2.x.patch - +1: chrisd, fuankg - +1: niq (to chris's mod_dbd updates collectively) - * mod_proxy: Improve traces in ap_proxy_http_process_response() to investigate PR37770. Trunk version of patch: diff --git a/modules/database/mod_dbd.c b/modules/database/mod_dbd.c index 4df4f4b65b4..9914875caf1 100644 --- a/modules/database/mod_dbd.c +++ b/modules/database/mod_dbd.c @@ -20,15 +20,21 @@ * http://apache.webthing.com/database/ */ -#include +#include "apr_reslist.h" +#include "apr_strings.h" +#include "apr_hash.h" +#include "apr_tables.h" +#include "apr_lib.h" +#include "apr_dbd.h" + +#define APR_WANT_MEMFUNC +#define APR_WANT_STRFUNC +#include "apr_want.h" #include "http_protocol.h" #include "http_config.h" #include "http_log.h" #include "http_request.h" -#include "apr_reslist.h" -#include "apr_strings.h" -#include "apr_dbd.h" #include "mod_dbd.h" extern module AP_MODULE_DECLARE_DATA dbd_module; @@ -40,126 +46,199 @@ extern module AP_MODULE_DECLARE_DATA dbd_module; #define NMAX_SET 0x4 #define EXPTIME_SET 0x8 -typedef struct dbd_prepared { - const char *label; - const char *query; - struct dbd_prepared *next; -} dbd_prepared; -typedef struct svr_cfg { +typedef struct { + server_rec *server; const char *name; const char *params; int persist; - dbd_prepared *prepared; #if APR_HAS_THREADS - apr_thread_mutex_t *mutex; - apr_pool_t *pool; - apr_reslist_t *dbpool; int nmin; int nkeep; int nmax; int exptime; + int set; +#endif + apr_hash_t *queries; +} dbd_cfg_t; + +typedef struct dbd_group_t dbd_group_t; + +struct dbd_group_t { + dbd_cfg_t *cfg; + dbd_group_t *next; + apr_pool_t *pool; +#if APR_HAS_THREADS + apr_thread_mutex_t *mutex; + apr_reslist_t *reslist; + int destroyed; #else - ap_dbd_t *conn; + ap_dbd_t *rec; #endif - unsigned int set; +}; + +typedef struct { + dbd_cfg_t *cfg; + dbd_group_t *group; } svr_cfg; typedef enum { cmd_name, cmd_params, cmd_persist, cmd_min, cmd_keep, cmd_max, cmd_exp } cmd_parts; -static apr_hash_t *dbd_prepared_defns; +static apr_pool_t *config_pool; +static dbd_group_t *group_list; /* a default DBDriver value that'll generate meaningful error messages */ static const char *const no_dbdriver = "[DBDriver unset]"; -#define ISINT(val) \ - for (p = val; *p; ++p) \ - if (!isdigit(*p)) \ - return "Argument must be numeric!" -static const char *dbd_param(cmd_parms *cmd, void *cfg, const char *val) +/* A default nmin of >0 will help with generating meaningful + * startup error messages if the database is down. + */ +#define DEFAULT_NMIN 1 +#define DEFAULT_NKEEP 2 +#define DEFAULT_NMAX 10 +#define DEFAULT_EXPTIME 300 + +static void *create_dbd_config(apr_pool_t *pool, server_rec *s) +{ + svr_cfg *svr = apr_pcalloc(pool, sizeof(svr_cfg)); + dbd_cfg_t *cfg = svr->cfg = apr_pcalloc(pool, sizeof(dbd_cfg_t)); + + cfg->server = s; + cfg->name = no_dbdriver; /* to generate meaningful error messages */ + cfg->params = ""; /* don't risk segfault on misconfiguration */ + cfg->persist = -1; +#if APR_HAS_THREADS + cfg->nmin = DEFAULT_NMIN; + cfg->nkeep = DEFAULT_NKEEP; + cfg->nmax = DEFAULT_NMAX; + cfg->exptime = DEFAULT_EXPTIME; +#endif + cfg->queries = apr_hash_make(pool); + + return svr; +} + +static void *merge_dbd_config(apr_pool_t *pool, void *basev, void *addv) +{ + dbd_cfg_t *base = ((svr_cfg*) basev)->cfg; + dbd_cfg_t *add = ((svr_cfg*) addv)->cfg; + svr_cfg *svr = apr_pcalloc(pool, sizeof(svr_cfg)); + dbd_cfg_t *new = svr->cfg = apr_pcalloc(pool, sizeof(dbd_cfg_t)); + + new->server = add->server; + new->name = (add->name != no_dbdriver) ? add->name : base->name; + new->params = strcmp(add->params, "") ? add->params : base->params; + new->persist = (add->persist != -1) ? add->persist : base->persist; +#if APR_HAS_THREADS + new->nmin = (add->set&NMIN_SET) ? add->nmin : base->nmin; + new->nkeep = (add->set&NKEEP_SET) ? add->nkeep : base->nkeep; + new->nmax = (add->set&NMAX_SET) ? add->nmax : base->nmax; + new->exptime = (add->set&EXPTIME_SET) ? add->exptime : base->exptime; +#endif + new->queries = apr_hash_overlay(pool, add->queries, base->queries); + + return svr; +} + +#define ISINT(val) do { \ + const char *p; \ + \ + for (p = val; *p; ++p) { \ + if (!apr_isdigit(*p)) { \ + return "Argument must be numeric!"; \ + } \ + } \ + } while (0) + +static const char *dbd_param(cmd_parms *cmd, void *dconf, const char *val) { - const char *p; const apr_dbd_driver_t *driver = NULL; - svr_cfg *svr = (svr_cfg*) ap_get_module_config - (cmd->server->module_config, &dbd_module); + svr_cfg *svr = ap_get_module_config(cmd->server->module_config, + &dbd_module); + dbd_cfg_t *cfg = svr->cfg; switch ((long) cmd->info) { case cmd_name: - svr->name = val; + cfg->name = val; /* loading the driver involves once-only dlloading that is * best done at server startup. This also guarantees that * we won't return an error later. */ - switch (apr_dbd_get_driver(cmd->pool, svr->name, &driver)) { + switch (apr_dbd_get_driver(cmd->pool, cfg->name, &driver)) { case APR_ENOTIMPL: - return apr_psprintf(cmd->pool, "DBD: No driver for %s", svr->name); + return apr_psprintf(cmd->pool, "DBD: No driver for %s", cfg->name); case APR_EDSOOPEN: return apr_psprintf(cmd->pool, +#ifdef NETWARE + "DBD: Can't load driver file dbd%s.nlm", +#else "DBD: Can't load driver file apr_dbd_%s.so", - svr->name); +#endif + cfg->name); case APR_ESYMNOTFOUND: return apr_psprintf(cmd->pool, "DBD: Failed to load driver apr_dbd_%s_driver", - svr->name); + cfg->name); } break; case cmd_params: - svr->params = val; + cfg->params = val; break; #if APR_HAS_THREADS case cmd_min: ISINT(val); - svr->nmin = atoi(val); - svr->set |= NMIN_SET; + cfg->nmin = atoi(val); + cfg->set |= NMIN_SET; break; case cmd_keep: ISINT(val); - svr->nkeep = atoi(val); - svr->set |= NKEEP_SET; + cfg->nkeep = atoi(val); + cfg->set |= NKEEP_SET; break; case cmd_max: ISINT(val); - svr->nmax = atoi(val); - svr->set |= NMAX_SET; + cfg->nmax = atoi(val); + cfg->set |= NMAX_SET; break; case cmd_exp: ISINT(val); - svr->exptime = atoi(val); - svr->set |= EXPTIME_SET; + cfg->exptime = atoi(val); + cfg->set |= EXPTIME_SET; break; #endif } + return NULL; } -static const char *dbd_param_flag(cmd_parms *cmd, void *cfg, int flag) + +static const char *dbd_param_flag(cmd_parms *cmd, void *dconf, int flag) { - svr_cfg *svr = (svr_cfg*) ap_get_module_config - (cmd->server->module_config, &dbd_module); + svr_cfg *svr = ap_get_module_config(cmd->server->module_config, + &dbd_module); switch ((long) cmd->info) { case cmd_persist: - svr->persist = flag; + svr->cfg->persist = flag; break; } + return NULL; } -DBD_DECLARE_NONSTD(void) ap_dbd_prepare(server_rec *s, const char *query, - const char *label) -{ - dbd_prepared *prepared = apr_pcalloc(s->process->pool, sizeof(dbd_prepared)); - const char *key = apr_psprintf(s->process->pool, "%pp", s); - prepared->label = label; - prepared->query = query; - prepared->next = apr_hash_get(dbd_prepared_defns, key, APR_HASH_KEY_STRING); - apr_hash_set(dbd_prepared_defns, key, APR_HASH_KEY_STRING, prepared); -} -static const char *dbd_prepare(cmd_parms *cmd, void *cfg, const char *query, + +static const char *dbd_prepare(cmd_parms *cmd, void *dconf, const char *query, const char *label) { + if (!label) { + label = query; + query = ""; + } + ap_dbd_prepare(cmd->server, query, label); + return NULL; } + static const command_rec dbd_cmds[] = { AP_INIT_TAKE1("DBDriver", dbd_param, (void*)cmd_name, RSRC_CONF, "SQL Driver"), @@ -167,8 +246,9 @@ static const command_rec dbd_cmds[] = { "SQL Driver Params"), AP_INIT_FLAG("DBDPersist", dbd_param_flag, (void*)cmd_persist, RSRC_CONF, "Use persistent connection/pool"), - AP_INIT_TAKE2("DBDPrepareSQL", dbd_prepare, NULL, RSRC_CONF, - "Prepared SQL statement, label"), + AP_INIT_TAKE12("DBDPrepareSQL", dbd_prepare, NULL, RSRC_CONF, + "SQL statement to prepare (or nothing, to override " + "statement inherited from main server) and label"), #if APR_HAS_THREADS AP_INIT_TAKE1("DBDMin", dbd_param, (void*)cmd_min, RSRC_CONF, "Minimum number of connections"), @@ -183,370 +263,542 @@ static const command_rec dbd_cmds[] = { #endif {NULL} }; -static void *dbd_merge(apr_pool_t *pool, void *BASE, void *ADD) { - svr_cfg *base = (svr_cfg*) BASE; - svr_cfg *add = (svr_cfg*) ADD; - svr_cfg *cfg = apr_pcalloc(pool, sizeof(svr_cfg)); - cfg->name = (add->name != no_dbdriver) ? add->name : base->name; - cfg->params = strcmp(add->params, "") ? add->params : base->params; - cfg->persist = (add->persist == -1) ? base->persist : add->persist; -#if APR_HAS_THREADS - cfg->nmin = (add->set&NMIN_SET) ? add->nmin : base->nmin; - cfg->nkeep = (add->set&NKEEP_SET) ? add->nkeep : base->nkeep; - cfg->nmax = (add->set&NMAX_SET) ? add->nmax : base->nmax; - cfg->exptime = (add->set&EXPTIME_SET) ? add->exptime : base->exptime; -#endif - cfg->set = add->set | base->set; - cfg->prepared = (add->prepared != NULL) ? add->prepared : base->prepared; - return (void*) cfg; + +static int dbd_pre_config(apr_pool_t *pconf, apr_pool_t *plog, + apr_pool_t *ptemp) +{ + config_pool = pconf; + group_list = NULL; + return OK; } -/* A default nmin of >0 will help with generating meaningful - * startup error messages if the database is down. - */ -#define DEFAULT_NMIN 1 -#define DEFAULT_NKEEP 2 -#define DEFAULT_NMAX 10 -#define DEFAULT_EXPTIME 300 -static void *dbd_cfg(apr_pool_t *p, server_rec *x) + +DBD_DECLARE_NONSTD(void) ap_dbd_prepare(server_rec *s, const char *query, + const char *label) +{ + svr_cfg *svr; + + svr = ap_get_module_config(s->module_config, &dbd_module); + if (!svr) { + /* some modules may call from within config directive handlers, and + * if these are called in a server context that contains no mod_dbd + * config directives, then we have to create our own server config + */ + svr = create_dbd_config(config_pool, s); + ap_set_module_config(s->module_config, &dbd_module, svr); + } + + if (apr_hash_get(svr->cfg->queries, label, APR_HASH_KEY_STRING) + && strcmp(query, "")) { + ap_log_error(APLOG_MARK, APLOG_WARNING, 0, s, + "conflicting SQL statements with label %s", label); + } + + apr_hash_set(svr->cfg->queries, label, APR_HASH_KEY_STRING, query); +} + +typedef struct { + const char *label, *query; +} dbd_query_t; + +static int dbd_post_config(apr_pool_t *pconf, apr_pool_t *plog, + apr_pool_t *ptemp, server_rec *s) { - svr_cfg *svr = (svr_cfg*) apr_pcalloc(p, sizeof(svr_cfg)); - svr->params = ""; /* don't risk segfault on misconfiguration */ - svr->name = no_dbdriver; /* to generate meaningful error messages */ - svr->persist = -1; + server_rec *sp; + apr_array_header_t *add_queries = apr_array_make(ptemp, 10, + sizeof(dbd_query_t)); + + for (sp = s; sp; sp = sp->next) { + svr_cfg *svr = ap_get_module_config(sp->module_config, &dbd_module); + dbd_cfg_t *cfg = svr->cfg; + apr_hash_index_t *hi_first = apr_hash_first(ptemp, cfg->queries); + dbd_group_t *group; + + /* dbd_setup in 2.2.3 and under was causing spurious error messages + * when dbd isn't configured. We can stop that with a quick check here + * together with a similar check in ap_dbd_open (where being + * unconfigured is a genuine error that must be reported). + */ + if (cfg->name == no_dbdriver || !cfg->persist) { + continue; + } + + for (group = group_list; group; group = group->next) { + dbd_cfg_t *group_cfg = group->cfg; + apr_hash_index_t *hi; + int group_ok = 1; + + if (strcmp(cfg->name, group_cfg->name) + || strcmp(cfg->params, group_cfg->params)) { + continue; + } + #if APR_HAS_THREADS - svr->nmin = DEFAULT_NMIN; - svr->nkeep = DEFAULT_NKEEP; - svr->nmax = DEFAULT_NMAX; - svr->exptime = DEFAULT_EXPTIME; + if (cfg->nmin != group_cfg->nmin + || cfg->nkeep != group_cfg->nkeep + || cfg->nmax != group_cfg->nmax + || cfg->exptime != group_cfg->exptime) { + continue; + } #endif - return svr; + + add_queries->nelts = 0; + + for (hi = hi_first; hi; hi = apr_hash_next(hi)) { + const char *label, *query; + const char *group_query; + + apr_hash_this(hi, (void*) &label, NULL, (void*) &query); + + group_query = apr_hash_get(group_cfg->queries, label, + APR_HASH_KEY_STRING); + + if (!group_query) { + dbd_query_t *add_query = apr_array_push(add_queries); + + add_query->label = label; + add_query->query = query; + } + else if (strcmp(query, group_query)) { + group_ok = 0; + break; + } + } + + if (group_ok) { + int i; + + for (i = 0; i < add_queries->nelts; ++i) { + dbd_query_t *add_query = ((dbd_query_t*) add_queries->elts) + + i; + + apr_hash_set(group_cfg->queries, add_query->label, + APR_HASH_KEY_STRING, add_query->query); + } + + svr->group = group; + break; + } + } + + if (!svr->group) { + svr->group = group = apr_pcalloc(pconf, sizeof(dbd_group_t)); + + group->cfg = cfg; + + group->next = group_list; + group_list = group; + } + } + + return OK; } -static apr_status_t dbd_prepared_init(apr_pool_t *pool, svr_cfg *svr, - ap_dbd_t *dbd) + +static apr_status_t dbd_prepared_init(apr_pool_t *pool, dbd_cfg_t *cfg, + ap_dbd_t *rec) { - dbd_prepared *p; - apr_status_t ret = APR_SUCCESS; - apr_dbd_prepared_t *stmt; - dbd->prepared = apr_hash_make(pool); + apr_hash_index_t *hi; + apr_status_t rv = APR_SUCCESS; + + rec->prepared = apr_hash_make(pool); + + for (hi = apr_hash_first(pool, cfg->queries); hi; + hi = apr_hash_next(hi)) { + const char *label, *query; + apr_dbd_prepared_t *stmt; + + apr_hash_this(hi, (void*) &label, NULL, (void*) &query); + + if (!strcmp(query, "")) { + continue; + } - for (p = svr->prepared; p; p = p->next) { stmt = NULL; - if (apr_dbd_prepare(dbd->driver, pool, dbd->handle, p->query, - p->label, &stmt) == 0) { - apr_hash_set(dbd->prepared, p->label, APR_HASH_KEY_STRING, stmt); + if (apr_dbd_prepare(rec->driver, pool, rec->handle, query, + label, &stmt)) { + rv = APR_EGENERAL; } else { - ret = APR_EGENERAL; + apr_hash_set(rec->prepared, label, APR_HASH_KEY_STRING, stmt); } } - return ret; + + return rv; } -/************ svr cfg: manage db connection pool ****************/ + +static apr_status_t dbd_close(void *data) +{ + ap_dbd_t *rec = data; + + return apr_dbd_close(rec->driver, rec->handle); +} + +#if APR_HAS_THREADS +static apr_status_t dbd_destruct(void *data, void *params, apr_pool_t *pool) +{ + dbd_group_t *group = params; + + if (!group->destroyed) { + ap_dbd_t *rec = data; + + apr_pool_destroy(rec->pool); + } + + return APR_SUCCESS; +} +#endif + /* an apr_reslist_constructor for SQL connections * Also use this for opening in non-reslist modes, since it gives * us all the error-handling in one place. */ -static apr_status_t dbd_construct(void **db, void *params, apr_pool_t *pool) +static apr_status_t dbd_construct(void **data_ptr, + void *params, apr_pool_t *pool) { - svr_cfg *svr = (svr_cfg*) params; - ap_dbd_t *rec = apr_pcalloc(pool, sizeof(ap_dbd_t)); + dbd_group_t *group = params; + dbd_cfg_t *cfg = group->cfg; + apr_pool_t *rec_pool, *prepared_pool; + ap_dbd_t *rec; apr_status_t rv; - /* this pool is mostly so dbd_close can destroy the prepared stmts */ - rv = apr_pool_create(&rec->pool, pool); + rv = apr_pool_create(&rec_pool, pool); if (rv != APR_SUCCESS) { - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, pool, - "DBD: Failed to create memory pool"); + ap_log_error(APLOG_MARK, APLOG_CRIT, rv, cfg->server, + "DBD: Failed to create memory pool"); + return rv; } -/* The driver is loaded at config time now, so this just checks a hash. - * If that changes, the driver DSO could be registered to unload against - * our pool, which is probably not what we want. Error checking isn't - * necessary now, but in case that changes in the future ... - */ - rv = apr_dbd_get_driver(rec->pool, svr->name, &rec->driver); - switch (rv) { - case APR_ENOTIMPL: - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, rec->pool, - "DBD: driver for %s not available", svr->name); - return rv; - case APR_EDSOOPEN: - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, rec->pool, - "DBD: can't find driver for %s", svr->name); - return rv; - case APR_ESYMNOTFOUND: - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, rec->pool, - "DBD: driver for %s is invalid or corrupted", svr->name); - return rv; - default: - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, rec->pool, - "DBD: mod_dbd not compatible with apr in get_driver"); + rec = apr_pcalloc(rec_pool, sizeof(ap_dbd_t)); + + rec->pool = rec_pool; + + /* The driver is loaded at config time now, so this just checks a hash. + * If that changes, the driver DSO could be registered to unload against + * our pool, which is probably not what we want. Error checking isn't + * necessary now, but in case that changes in the future ... + */ + rv = apr_dbd_get_driver(rec->pool, cfg->name, &rec->driver); + if (rv != APR_SUCCESS) { + switch (rv) { + case APR_ENOTIMPL: + ap_log_error(APLOG_MARK, APLOG_ERR, rv, cfg->server, + "DBD: driver for %s not available", cfg->name); + break; + case APR_EDSOOPEN: + ap_log_error(APLOG_MARK, APLOG_ERR, rv, cfg->server, + "DBD: can't find driver for %s", cfg->name); + break; + case APR_ESYMNOTFOUND: + ap_log_error(APLOG_MARK, APLOG_ERR, rv, cfg->server, + "DBD: driver for %s is invalid or corrupted", + cfg->name); + break; + default: + ap_log_error(APLOG_MARK, APLOG_ERR, rv, cfg->server, + "DBD: mod_dbd not compatible with APR in get_driver"); + break; + } + + apr_pool_destroy(rec->pool); return rv; - case APR_SUCCESS: - break; } - rv = apr_dbd_open(rec->driver, rec->pool, svr->params, &rec->handle); - switch (rv) { - case APR_EGENERAL: - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, rec->pool, - "DBD: Can't connect to %s", svr->name); + rv = apr_dbd_open(rec->driver, rec->pool, cfg->params, &rec->handle); + if (rv != APR_SUCCESS) { + switch (rv) { + case APR_EGENERAL: + ap_log_error(APLOG_MARK, APLOG_ERR, rv, cfg->server, + "DBD: Can't connect to %s", cfg->name); + break; + default: + ap_log_error(APLOG_MARK, APLOG_ERR, rv, cfg->server, + "DBD: mod_dbd not compatible with APR in open"); + break; + } + + apr_pool_destroy(rec->pool); return rv; - default: - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, rec->pool, - "DBD: mod_dbd not compatible with apr in open"); + } + + apr_pool_cleanup_register(rec->pool, rec, dbd_close, + apr_pool_cleanup_null); + + /* we use a sub-pool for the prepared statements for each connection so + * that they will be cleaned up first, before the connection is closed + */ + rv = apr_pool_create(&prepared_pool, rec->pool); + if (rv != APR_SUCCESS) { + ap_log_error(APLOG_MARK, APLOG_CRIT, rv, cfg->server, + "DBD: Failed to create memory pool"); + + apr_pool_destroy(rec->pool); return rv; - case APR_SUCCESS: - break; } - *db = rec; - rv = dbd_prepared_init(rec->pool, svr, rec); + + rv = dbd_prepared_init(prepared_pool, cfg, rec); if (rv != APR_SUCCESS) { const char *errmsg = apr_dbd_error(rec->driver, rec->handle, rv); - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, rec->pool, - "DBD: failed to initialise prepared SQL statements: %s", - (errmsg ? errmsg : "[???]")); + ap_log_error(APLOG_MARK, APLOG_ERR, rv, cfg->server, + "DBD: failed to prepare SQL statements: %s", + (errmsg ? errmsg : "[???]")); + + apr_pool_destroy(rec->pool); + return rv; } - return rv; -} -static apr_status_t dbd_close(void *CONN) -{ - ap_dbd_t *conn = CONN; - apr_status_t rv = apr_dbd_close(conn->driver, conn->handle); - apr_pool_destroy(conn->pool); - return rv; + + *data_ptr = rec; + + return APR_SUCCESS; } + #if APR_HAS_THREADS -static apr_status_t dbd_destruct(void *sql, void *params, apr_pool_t *pool) +static apr_status_t dbd_destroy(void *data) { - return dbd_close(sql); + dbd_group_t *group = data; + + group->destroyed = 1; + + return APR_SUCCESS; } -static apr_status_t dbd_setup(apr_pool_t *pool, svr_cfg *svr) +static apr_status_t dbd_setup(server_rec *s, dbd_group_t *group) { + dbd_cfg_t *cfg = group->cfg; apr_status_t rv; - /* create a pool just for the reslist from a process-lifetime pool; - * that pool (s->process->pool in the dbd_setup_lock case, - * whatever was passed to ap_run_child_init in the dbd_setup_init case) - * will be shared with other threads doing other non-mod_dbd things - * so we can't use it for the reslist directly + /* We create the reslist using a sub-pool of the pool passed to our + * child_init hook. No other threads can be here because we're + * either in the child_init phase or dbd_setup_lock() acquired our mutex. + * No other threads will use this sub-pool after this, except via + * reslist calls, which have an internal mutex. + * + * We need to short-circuit the cleanup registered internally by + * apr_reslist_create(). We do this by registering dbd_destroy() + * as a cleanup afterwards, so that it will run before the reslist's + * internal cleanup. + * + * If we didn't do this, then we could free memory twice when the pool + * was destroyed. When apr_pool_destroy() runs, it first destroys all + * all the per-connection sub-pools created in dbd_construct(), and + * then it runs the reslist's cleanup. The cleanup calls dbd_destruct() + * on each resource, which would then attempt to destroy the sub-pools + * a second time. */ - rv = apr_pool_create(&svr->pool, pool); + rv = apr_reslist_create(&group->reslist, + cfg->nmin, cfg->nkeep, cfg->nmax, + apr_time_from_sec(cfg->exptime), + dbd_construct, dbd_destruct, group, + group->pool); if (rv != APR_SUCCESS) { - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, pool, - "DBD: Failed to create reslist memory pool"); + ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, + "DBD: failed to initialise"); return rv; } - rv = apr_reslist_create(&svr->dbpool, svr->nmin, svr->nkeep, svr->nmax, - apr_time_from_sec(svr->exptime), - dbd_construct, dbd_destruct, svr, svr->pool); - if (rv == APR_SUCCESS) { - apr_pool_cleanup_register(svr->pool, svr->dbpool, - (void*)apr_reslist_destroy, - apr_pool_cleanup_null); - } - else { - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, svr->pool, - "DBD: failed to initialise"); - apr_pool_destroy(svr->pool); - svr->pool = NULL; - } + apr_pool_cleanup_register(group->pool, group, dbd_destroy, + apr_pool_cleanup_null); - return rv; + return APR_SUCCESS; } +#endif + static apr_status_t dbd_setup_init(apr_pool_t *pool, server_rec *s) { - svr_cfg *svr = ap_get_module_config(s->module_config, &dbd_module); - apr_status_t rv; + dbd_group_t *group; + apr_status_t rv = APR_SUCCESS; - /* dbd_setup in 2.2.3 and under was causing spurious error messages - * when dbd isn't configured. We can stop that with a quick check here - * together with a similar check in ap_dbd_open (where being - * unconfigured is a genuine error that must be reported). - */ - if (svr->name == no_dbdriver) { - return APR_SUCCESS; - } + for (group = group_list; group; group = group->next) { + apr_status_t rv2; - if (!svr->persist) { - return APR_SUCCESS; - } + rv2 = apr_pool_create(&group->pool, pool); + if (rv2 != APR_SUCCESS) { + ap_log_error(APLOG_MARK, APLOG_CRIT, rv2, s, + "DBD: Failed to create reslist cleanup memory pool"); + return rv2; + } - rv = dbd_setup(pool, svr); - if (rv == APR_SUCCESS) { - return rv; - } +#if APR_HAS_THREADS + rv2 = dbd_setup(s, group); + if (rv2 == APR_SUCCESS) { + continue; + } + else if (rv == APR_SUCCESS) { + rv = rv2; + } - /* we failed, so create a mutex so that subsequent competing callers - * to ap_dbd_open can serialize themselves while they retry - */ - rv = apr_thread_mutex_create(&svr->mutex, APR_THREAD_MUTEX_DEFAULT, pool); - if (rv != APR_SUCCESS) { - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, pool, - "DBD: Failed to create thread mutex"); + /* we failed, so create a mutex so that subsequent competing callers + * to ap_dbd_open can serialize themselves while they retry + */ + rv2 = apr_thread_mutex_create(&group->mutex, + APR_THREAD_MUTEX_DEFAULT, pool); + if (rv2 != APR_SUCCESS) { + ap_log_error(APLOG_MARK, APLOG_CRIT, rv2, s, + "DBD: Failed to create thread mutex"); + return rv2; + } +#endif } + return rv; } -static apr_status_t dbd_setup_lock(apr_pool_t *pool, server_rec *s) + +#if APR_HAS_THREADS +static apr_status_t dbd_setup_lock(server_rec *s, dbd_group_t *group) { - svr_cfg *svr = ap_get_module_config(s->module_config, &dbd_module); - apr_status_t rv, rv2 = APR_SUCCESS; + apr_status_t rv = APR_SUCCESS, rv2; /* several threads could be here at the same time, all trying to * initialize the reslist because dbd_setup_init failed to do so */ - if (!svr->mutex) { + if (!group->mutex) { /* we already logged an error when the mutex couldn't be created */ return APR_EGENERAL; } - rv = apr_thread_mutex_lock(svr->mutex); - if (rv != APR_SUCCESS) { - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, pool, - "DBD: Failed to acquire thread mutex"); - return rv; + rv2 = apr_thread_mutex_lock(group->mutex); + if (rv2 != APR_SUCCESS) { + ap_log_error(APLOG_MARK, APLOG_ERR, rv2, s, + "DBD: Failed to acquire thread mutex"); + return rv2; } - if (!svr->dbpool) { - rv2 = dbd_setup(s->process->pool, svr); + if (!group->reslist) { + rv = dbd_setup(s, group); } - rv = apr_thread_mutex_unlock(svr->mutex); - if (rv != APR_SUCCESS) { - ap_log_perror(APLOG_MARK, APLOG_CRIT, rv, pool, - "DBD: Failed to release thread mutex"); - if (rv2 == APR_SUCCESS) { - rv2 = rv; + rv2 = apr_thread_mutex_unlock(group->mutex); + if (rv2 != APR_SUCCESS) { + ap_log_error(APLOG_MARK, APLOG_ERR, rv2, s, + "DBD: Failed to release thread mutex"); + if (rv == APR_SUCCESS) { + rv = rv2; } } - return rv2; + + return rv; } #endif - /* Functions we export for modules to use: - open acquires a connection from the pool (opens one if necessary) - close releases it back in to the pool */ -DBD_DECLARE_NONSTD(void) ap_dbd_close(server_rec *s, ap_dbd_t *sql) +DBD_DECLARE_NONSTD(void) ap_dbd_close(server_rec *s, ap_dbd_t *rec) { svr_cfg *svr = ap_get_module_config(s->module_config, &dbd_module); - if (!svr->persist) { - dbd_close((void*) sql); + + if (!svr->cfg->persist) { + apr_pool_destroy(rec->pool); } #if APR_HAS_THREADS else { - apr_reslist_release(svr->dbpool, sql); + apr_reslist_release(svr->group->reslist, rec); } #endif } -#define arec ((ap_dbd_t*)rec) -#if APR_HAS_THREADS + +static apr_status_t dbd_check(apr_pool_t *pool, server_rec *s, ap_dbd_t *rec) +{ + svr_cfg *svr; + apr_status_t rv = apr_dbd_check_conn(rec->driver, pool, rec->handle); + const char *errmsg; + + if ((rv == APR_SUCCESS) || (rv == APR_ENOTIMPL)) { + return APR_SUCCESS; + } + + errmsg = apr_dbd_error(rec->driver, rec->handle, rv); + if (!errmsg) { + errmsg = "(unknown)"; + } + + svr = ap_get_module_config(s->module_config, &dbd_module); + ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, + "DBD [%s] Error: %s", svr->cfg->name, errmsg); + return rv; +} + DBD_DECLARE_NONSTD(ap_dbd_t*) ap_dbd_open(apr_pool_t *pool, server_rec *s) { - void *rec = NULL; svr_cfg *svr = ap_get_module_config(s->module_config, &dbd_module); - apr_status_t rv = APR_SUCCESS; - const char *errmsg; + dbd_group_t *group = svr->group; + dbd_cfg_t *cfg = svr->cfg; + ap_dbd_t *rec = NULL; +#if APR_HAS_THREADS + apr_status_t rv; +#endif /* If nothing is configured, we shouldn't be here */ - if (svr->name == no_dbdriver) { - ap_log_perror(APLOG_MARK, APLOG_ERR, 0, pool, "DBD: not configured"); + if (cfg->name == no_dbdriver) { + ap_log_error(APLOG_MARK, APLOG_ERR, 0, s, "DBD: not configured"); return NULL; } - if (!svr->persist) { + if (!cfg->persist) { /* Return a once-only connection */ - rv = dbd_construct(&rec, svr, s->process->pool); - return (rv == APR_SUCCESS) ? arec : NULL; + group = apr_pcalloc(pool, sizeof(dbd_group_t)); + + group->cfg = cfg; + + dbd_construct((void*) &rec, group, pool); + return rec; } - if (!svr->dbpool) { - if (dbd_setup_lock(pool, s) != APR_SUCCESS) { +#if APR_HAS_THREADS + if (!group->reslist) { + if (dbd_setup_lock(s, group) != APR_SUCCESS) { return NULL; } } - rv = apr_reslist_acquire(svr->dbpool, &rec); + + rv = apr_reslist_acquire(group->reslist, (void*) &rec); if (rv != APR_SUCCESS) { - ap_log_perror(APLOG_MARK, APLOG_ERR, rv, pool, - "Failed to acquire DBD connection from pool!"); + ap_log_error(APLOG_MARK, APLOG_ERR, rv, s, + "Failed to acquire DBD connection from pool!"); return NULL; } - rv = apr_dbd_check_conn(arec->driver, pool, arec->handle); - if ((rv != APR_SUCCESS) && (rv != APR_ENOTIMPL)) { - errmsg = apr_dbd_error(arec->driver, arec->handle, rv); - if (!errmsg) { - errmsg = "(unknown)"; - } - ap_log_perror(APLOG_MARK, APLOG_ERR, rv, pool, - "DBD[%s] Error: %s", svr->name, errmsg ); - apr_reslist_invalidate(svr->dbpool, rec); - return NULL; - } - return arec; -} -#else -DBD_DECLARE_NONSTD(ap_dbd_t*) ap_dbd_open(apr_pool_t *pool, server_rec *s) -{ - apr_status_t rv = APR_SUCCESS; - const char *errmsg; - void *rec = NULL; - svr_cfg *svr = ap_get_module_config(s->module_config, &dbd_module); - /* If nothing is configured, we shouldn't be here */ - if (svr->name == no_dbdriver) { - ap_log_perror(APLOG_MARK, APLOG_ERR, 0, pool, "DBD: not configured"); + if (dbd_check(pool, s, rec) != APR_SUCCESS) { + apr_reslist_invalidate(group->reslist, rec); return NULL; } - - if (!svr->persist) { - /* Return a once-only connection */ - rv = dbd_construct(&rec, svr, s->process->pool); - return (rv == APR_SUCCESS) ? arec : NULL; - } - -/* since we're in nothread-land, we can mess with svr->conn with impunity */ -/* If we have a persistent connection and it's good, we'll use it */ - if (svr->conn) { - rv = apr_dbd_check_conn(svr->conn->driver, pool, svr->conn->handle); - if ((rv != APR_SUCCESS) && (rv != APR_ENOTIMPL)) { - errmsg = apr_dbd_error(arec->driver, arec->handle, rv); - if (!errmsg) { - errmsg = "(unknown)"; - } - ap_log_perror(APLOG_MARK, APLOG_ERR, rv, pool, - "DBD[%s] Error: %s", svr->name, errmsg); - svr->conn = NULL; +#else + /* If we have a persistent connection and it's good, we'll use it; + * since this is non-threaded, we can update without a mutex + */ + rec = group->rec; + if (rec) { + if (dbd_check(pool, s, rec) != APR_SUCCESS) { + apr_pool_destroy(rec->pool); + rec = NULL; } } -/* We don't have a connection right now, so we'll open one */ - if (!svr->conn) { - if (dbd_construct(&rec, svr, s->process->pool) == APR_SUCCESS) { - svr->conn = arec ; - apr_pool_cleanup_register(s->process->pool, svr->conn, - dbd_close, apr_pool_cleanup_null); - } + + /* We don't have a connection right now, so we'll open one */ + if (!rec) { + dbd_construct((void*) &rec, group, group->pool); + group->rec = rec; } - return svr->conn; -} #endif + + return rec; +} + #if APR_HAS_THREADS typedef struct { - ap_dbd_t *conn; - apr_reslist_t *dbpool; -} dbd_pool_rec; -static apr_status_t dbd_release(void *REQ) + ap_dbd_t *rec; + apr_reslist_t *reslist; +} dbd_acquire_t; + +static apr_status_t dbd_release(void *data) { - dbd_pool_rec *req = REQ; - apr_reslist_release(req->dbpool, req->conn); + dbd_acquire_t *acq = data; + apr_reslist_release(acq->reslist, acq->rec); return APR_SUCCESS; } + DBD_DECLARE_NONSTD(ap_dbd_t *) ap_dbd_acquire(request_rec *r) { - svr_cfg *svr; - dbd_pool_rec *req; + dbd_acquire_t *acq; while (!ap_is_initial_req(r)) { if (r->prev) { @@ -557,54 +809,52 @@ DBD_DECLARE_NONSTD(ap_dbd_t *) ap_dbd_acquire(request_rec *r) } } - req = ap_get_module_config(r->request_config, &dbd_module); - if (!req) { - req = apr_palloc(r->pool, sizeof(dbd_pool_rec)); - req->conn = ap_dbd_open(r->pool, r->server); - if (req->conn) { - svr = ap_get_module_config(r->server->module_config, &dbd_module); - ap_set_module_config(r->request_config, &dbd_module, req); - if (svr->persist) { - req->dbpool = svr->dbpool; - apr_pool_cleanup_register(r->pool, req, dbd_release, - apr_pool_cleanup_null); - } - else { - apr_pool_cleanup_register(r->pool, req->conn, dbd_close, + acq = ap_get_module_config(r->request_config, &dbd_module); + if (!acq) { + acq = apr_palloc(r->pool, sizeof(dbd_acquire_t)); + acq->rec = ap_dbd_open(r->pool, r->server); + if (acq->rec) { + svr_cfg *svr = ap_get_module_config(r->server->module_config, + &dbd_module); + + ap_set_module_config(r->request_config, &dbd_module, acq); + if (svr->cfg->persist) { + acq->reslist = svr->group->reslist; + apr_pool_cleanup_register(r->pool, acq, dbd_release, apr_pool_cleanup_null); } } } - return req->conn; + + return acq->rec; } + DBD_DECLARE_NONSTD(ap_dbd_t *) ap_dbd_cacquire(conn_rec *c) { - svr_cfg *svr; - dbd_pool_rec *req = ap_get_module_config(c->conn_config, &dbd_module); - if (!req) { - req = apr_palloc(c->pool, sizeof(dbd_pool_rec)); - req->conn = ap_dbd_open(c->pool, c->base_server); - if (req->conn) { - svr = ap_get_module_config(c->base_server->module_config, &dbd_module); - ap_set_module_config(c->conn_config, &dbd_module, req); - if (svr->persist) { - req->dbpool = svr->dbpool; - apr_pool_cleanup_register(c->pool, req, dbd_release, - apr_pool_cleanup_null); - } - else { - apr_pool_cleanup_register(c->pool, req->conn, dbd_close, + dbd_acquire_t *acq = ap_get_module_config(c->conn_config, &dbd_module); + + if (!acq) { + acq = apr_palloc(c->pool, sizeof(dbd_acquire_t)); + acq->rec = ap_dbd_open(c->pool, c->base_server); + if (acq->rec) { + svr_cfg *svr = ap_get_module_config(c->base_server->module_config, + &dbd_module); + + ap_set_module_config(c->conn_config, &dbd_module, acq); + if (svr->cfg->persist) { + acq->reslist = svr->group->reslist; + apr_pool_cleanup_register(c->pool, acq, dbd_release, apr_pool_cleanup_null); } } } - return req->conn; + + return acq->rec; } #else DBD_DECLARE_NONSTD(ap_dbd_t *) ap_dbd_acquire(request_rec *r) { - svr_cfg *svr; - ap_dbd_t *ret; + ap_dbd_t *rec; while (!ap_is_initial_req(r)) { if (r->prev) { @@ -615,80 +865,54 @@ DBD_DECLARE_NONSTD(ap_dbd_t *) ap_dbd_acquire(request_rec *r) } } - ret = ap_get_module_config(r->request_config, &dbd_module); - if (!ret) { - svr = ap_get_module_config(r->server->module_config, &dbd_module); - ret = ap_dbd_open(r->pool, r->server); - if (ret) { - ap_set_module_config(r->request_config, &dbd_module, ret); - if (!svr->persist) { - apr_pool_cleanup_register(r->pool, svr->conn, dbd_close, - apr_pool_cleanup_null); - } - /* if persist then dbd_open registered cleanup on proc pool */ + rec = ap_get_module_config(r->request_config, &dbd_module); + if (!rec) { + rec = ap_dbd_open(r->pool, r->server); + if (rec) { + ap_set_module_config(r->request_config, &dbd_module, rec); } } - return ret; + + return rec; } + DBD_DECLARE_NONSTD(ap_dbd_t *) ap_dbd_cacquire(conn_rec *c) { - svr_cfg *svr; - ap_dbd_t *ret = ap_get_module_config(c->conn_config, &dbd_module); - if (!ret) { - svr = ap_get_module_config(c->base_server->module_config, &dbd_module); - ret = ap_dbd_open(c->pool, c->base_server); - if (ret) { - ap_set_module_config(c->conn_config, &dbd_module, ret); - if (!svr->persist) { - apr_pool_cleanup_register(c->pool, svr->conn, dbd_close, - apr_pool_cleanup_null); - } - /* if persist then dbd_open registered cleanup on proc pool */ + ap_dbd_t *rec = ap_get_module_config(c->conn_config, &dbd_module); + + if (!rec) { + rec = ap_dbd_open(c->pool, c->base_server); + if (rec) { + ap_set_module_config(c->conn_config, &dbd_module, rec); } } - return ret; + + return rec; } #endif -static int dbd_pre_config(apr_pool_t *p, apr_pool_t *plog, apr_pool_t *ptemp) -{ - dbd_prepared_defns = apr_hash_make(ptemp); - return OK; -} -static int dbd_post_config(apr_pool_t *pconf, apr_pool_t *plog, - apr_pool_t *ptemp, server_rec *s) -{ - svr_cfg *svr; - server_rec *sp; - for (sp = s; sp; sp = sp->next) { - const char *key = apr_psprintf(s->process->pool, "%pp", s); - svr = ap_get_module_config(sp->module_config, &dbd_module); - svr->prepared = apr_hash_get(dbd_prepared_defns, key, - APR_HASH_KEY_STRING); - } - return OK; -} static void dbd_hooks(apr_pool_t *pool) { -#if APR_HAS_THREADS + ap_hook_pre_config(dbd_pre_config, NULL, NULL, APR_HOOK_MIDDLE); + ap_hook_post_config(dbd_post_config, NULL, NULL, APR_HOOK_MIDDLE); ap_hook_child_init((void*)dbd_setup_init, NULL, NULL, APR_HOOK_MIDDLE); -#endif + + APR_REGISTER_OPTIONAL_FN(ap_dbd_prepare); APR_REGISTER_OPTIONAL_FN(ap_dbd_open); APR_REGISTER_OPTIONAL_FN(ap_dbd_close); APR_REGISTER_OPTIONAL_FN(ap_dbd_acquire); APR_REGISTER_OPTIONAL_FN(ap_dbd_cacquire); - APR_REGISTER_OPTIONAL_FN(ap_dbd_prepare); + apr_dbd_init(pool); - ap_hook_pre_config(dbd_pre_config, NULL, NULL, APR_HOOK_MIDDLE); - ap_hook_post_config(dbd_post_config, NULL, NULL, APR_HOOK_MIDDLE); } module AP_MODULE_DECLARE_DATA dbd_module = { STANDARD20_MODULE_STUFF, NULL, NULL, - dbd_cfg, - dbd_merge, + create_dbd_config, + merge_dbd_config, dbd_cmds, dbd_hooks }; +