Adding upstream version 2.4.56.upstream/2.4.56

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

1 files changed, 1169 insertions, 0 deletions

diff --git a/modules/md/md_store_fs.c b/modules/md/md_store_fs.c
new file mode 100644
index 0000000..35c24b4
--- /dev/null
+++ b/modules/md/md_store_fs.c
@@ -0,0 +1,1169 @@
+/* Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements.  See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License.  You may obtain a copy of the License at
+ *
+ *     http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+ 
+#include <assert.h>
+#include <stddef.h>
+#include <stdio.h>
+#include <stdlib.h>
+
+#include <apr_lib.h>
+#include <apr_file_info.h>
+#include <apr_file_io.h>
+#include <apr_fnmatch.h>
+#include <apr_hash.h>
+#include <apr_strings.h>
+
+#include "md.h"
+#include "md_crypt.h"
+#include "md_json.h"
+#include "md_log.h"
+#include "md_store.h"
+#include "md_store_fs.h"
+#include "md_util.h"
+#include "md_version.h"
+
+/**************************************************************************************************/
+/* file system based implementation of md_store_t */
+
+#define MD_STORE_VERSION        3
+#define MD_FS_LOCK_NAME         "store.lock"
+
+typedef struct {
+    apr_fileperms_t dir;
+    apr_fileperms_t file;
+} perms_t;
+
+typedef struct md_store_fs_t md_store_fs_t;
+struct md_store_fs_t {
+    md_store_t s;
+    
+    const char *base;       /* base directory of store */
+    perms_t def_perms;
+    perms_t group_perms[MD_SG_COUNT];
+    md_store_fs_cb *event_cb;
+    void *event_baton;
+    
+    md_data_t key;
+    int plain_pkey[MD_SG_COUNT];
+    
+    int port_80;
+    int port_443;
+
+    apr_file_t *global_lock;
+};
+
+#define FS_STORE(store)     (md_store_fs_t*)(((char*)store)-offsetof(md_store_fs_t, s))
+#define FS_STORE_JSON       "md_store.json"
+#define FS_STORE_KLEN       48
+
+static apr_status_t fs_load(md_store_t *store, md_store_group_t group, 
+                            const char *name, const char *aspect,  
+                            md_store_vtype_t vtype, void **pvalue, apr_pool_t *p);
+static apr_status_t fs_save(md_store_t *store, apr_pool_t *p, md_store_group_t group, 
+                            const char *name, const char *aspect,  
+                            md_store_vtype_t vtype, void *value, int create);
+static apr_status_t fs_remove(md_store_t *store, md_store_group_t group, 
+                              const char *name, const char *aspect, 
+                              apr_pool_t *p, int force);
+static apr_status_t fs_purge(md_store_t *store, apr_pool_t *p, 
+                             md_store_group_t group, const char *name);
+static apr_status_t fs_remove_nms(md_store_t *store, apr_pool_t *p, 
+                                  apr_time_t modified, md_store_group_t group, 
+                                  const char *name, const char *aspect);
+static apr_status_t fs_move(md_store_t *store, apr_pool_t *p, 
+                            md_store_group_t from, md_store_group_t to, 
+                            const char *name, int archive);
+static apr_status_t fs_rename(md_store_t *store, apr_pool_t *p, 
+                            md_store_group_t group, const char *from, const char *to);
+static apr_status_t fs_iterate(md_store_inspect *inspect, void *baton, md_store_t *store, 
+                               apr_pool_t *p, md_store_group_t group,  const char *pattern,
+                               const char *aspect, md_store_vtype_t vtype);
+static apr_status_t fs_iterate_names(md_store_inspect *inspect, void *baton, md_store_t *store, 
+                                     apr_pool_t *p, md_store_group_t group, const char *pattern);
+
+static apr_status_t fs_get_fname(const char **pfname, 
+                                 md_store_t *store, md_store_group_t group, 
+                                 const char *name, const char *aspect, 
+                                 apr_pool_t *p);
+static int fs_is_newer(md_store_t *store, md_store_group_t group1, md_store_group_t group2,  
+                       const char *name, const char *aspect, apr_pool_t *p);
+
+static apr_time_t fs_get_modified(md_store_t *store, md_store_group_t group,  
+                                  const char *name, const char *aspect, apr_pool_t *p);
+
+static apr_status_t fs_lock_global(md_store_t *store, apr_pool_t *p, apr_time_t max_wait);
+static void fs_unlock_global(md_store_t *store, apr_pool_t *p);
+
+static apr_status_t init_store_file(md_store_fs_t *s_fs, const char *fname, 
+                                    apr_pool_t *p, apr_pool_t *ptemp)
+{
+    md_json_t *json = md_json_create(p);
+    const char *key64;
+    apr_status_t rv;
+    
+    md_json_setn(MD_STORE_VERSION, json, MD_KEY_STORE, MD_KEY_VERSION, NULL);
+
+    md_data_pinit(&s_fs->key, FS_STORE_KLEN, p);
+    if (APR_SUCCESS != (rv = md_rand_bytes((unsigned char*)s_fs->key.data, s_fs->key.len, p))) {
+        return rv;
+    }
+        
+    key64 = md_util_base64url_encode(&s_fs->key, ptemp);
+    md_json_sets(key64, json, MD_KEY_KEY, NULL);
+    rv = md_json_fcreatex(json, ptemp, MD_JSON_FMT_INDENT, fname, MD_FPROT_F_UONLY);
+    memset((char*)key64, 0, strlen(key64));
+
+    return rv;
+}
+
+static apr_status_t rename_pkey(void *baton, apr_pool_t *p, apr_pool_t *ptemp, 
+                                const char *dir, const char *name, 
+                                apr_filetype_e ftype)
+{
+    const char *from, *to;
+    apr_status_t rv = APR_SUCCESS;
+    
+    (void)baton;
+    (void)ftype;
+    if (   MD_OK(md_util_path_merge(&from, ptemp, dir, name, NULL))
+        && MD_OK(md_util_path_merge(&to, ptemp, dir, MD_FN_PRIVKEY, NULL))) {
+        md_log_perror(MD_LOG_MARK, MD_LOG_TRACE1, 0, p, "renaming %s/%s to %s", 
+                      dir, name, MD_FN_PRIVKEY);
+        return apr_file_rename(from, to, ptemp);
+    }
+    return rv;
+}
+
+static apr_status_t mk_pubcert(void *baton, apr_pool_t *p, apr_pool_t *ptemp, 
+                               const char *dir, const char *name, 
+                               apr_filetype_e ftype)
+{
+    md_cert_t *cert;
+    apr_array_header_t *chain, *pubcert;
+    const char *fname, *fpubcert;
+    apr_status_t rv = APR_SUCCESS;
+    
+    (void)baton;
+    (void)ftype;
+    (void)p;
+    if (   MD_OK(md_util_path_merge(&fpubcert, ptemp, dir, MD_FN_PUBCERT, NULL))
+        && APR_STATUS_IS_ENOENT(rv = md_chain_fload(&pubcert, ptemp, fpubcert))
+        && MD_OK(md_util_path_merge(&fname, ptemp, dir, name, NULL))
+        && MD_OK(md_cert_fload(&cert, ptemp, fname))
+        && MD_OK(md_util_path_merge(&fname, ptemp, dir, "chain.pem", NULL))) {
+        
+        rv = md_chain_fload(&chain, ptemp, fname);
+        if (APR_STATUS_IS_ENOENT(rv)) {
+            chain = apr_array_make(ptemp, 1, sizeof(md_cert_t*));
+            rv = APR_SUCCESS;
+        }
+        if (APR_SUCCESS == rv) {
+            pubcert = apr_array_make(ptemp, chain->nelts + 1, sizeof(md_cert_t*));
+            APR_ARRAY_PUSH(pubcert, md_cert_t *) = cert;
+            apr_array_cat(pubcert, chain);
+            rv = md_chain_fsave(pubcert, ptemp, fpubcert, MD_FPROT_F_UONLY);
+        }
+    }
+    return rv;
+}
+
+static apr_status_t upgrade_from_1_0(md_store_fs_t *s_fs, apr_pool_t *p, apr_pool_t *ptemp)
+{
+    md_store_group_t g;
+    apr_status_t rv = APR_SUCCESS;
+    
+    (void)ptemp;
+    /* Migrate pkey.pem -> privkey.pem */
+    for (g = MD_SG_NONE; g < MD_SG_COUNT && APR_SUCCESS == rv; ++g) {
+        rv = md_util_files_do(rename_pkey, s_fs, p, s_fs->base, 
+                              md_store_group_name(g), "*", "pkey.pem", NULL);
+    }
+    /* Generate fullcert.pem from cert.pem and chain.pem where missing */
+    rv = md_util_files_do(mk_pubcert, s_fs, p, s_fs->base, 
+                          md_store_group_name(MD_SG_DOMAINS), "*", MD_FN_CERT, NULL);
+    rv = md_util_files_do(mk_pubcert, s_fs, p, s_fs->base, 
+                          md_store_group_name(MD_SG_ARCHIVE), "*", MD_FN_CERT, NULL);
+    
+    return rv;
+}
+
+static apr_status_t read_store_file(md_store_fs_t *s_fs, const char *fname, 
+                                    apr_pool_t *p, apr_pool_t *ptemp)
+{
+    md_json_t *json;
+    const char *key64;
+    apr_status_t rv;
+    double store_version;
+    
+    if (MD_OK(md_json_readf(&json, p, fname))) {
+        store_version = md_json_getn(json, MD_KEY_STORE, MD_KEY_VERSION, NULL);
+        if (store_version <= 0.0) {
+            /* ok, an old one, compatible to 1.0 */
+            store_version = 1.0;
+        }
+        if (store_version > MD_STORE_VERSION) {
+            md_log_perror(MD_LOG_MARK, MD_LOG_ERR, 0, p, "version too new: %f", store_version);
+            return APR_EINVAL;
+        }
+
+        key64 = md_json_dups(p, json, MD_KEY_KEY, NULL);
+        if (!key64) {
+            md_log_perror(MD_LOG_MARK, MD_LOG_ERR, 0, p, "missing key: %s", MD_KEY_KEY);
+            return APR_EINVAL;
+        }
+        
+        md_util_base64url_decode(&s_fs->key, key64, p);
+        if (s_fs->key.len != FS_STORE_KLEN) {
+            md_log_perror(MD_LOG_MARK, MD_LOG_ERR, 0, p, "key length unexpected: %" APR_SIZE_T_FMT, 
+                          s_fs->key.len);
+            return APR_EINVAL;
+        }
+
+        /* Need to migrate format? */
+        if (store_version < MD_STORE_VERSION) {
+            if (store_version <= 1.0) {
+                md_log_perror(MD_LOG_MARK, MD_LOG_DEBUG, 0, p, "migrating store v1 -> v2");
+                rv = upgrade_from_1_0(s_fs, p, ptemp);
+            }
+            if (store_version <= 2.0) {
+                md_log_perror(MD_LOG_MARK, MD_LOG_DEBUG, 0, p, "migrating store v2 -> v3");
+                md_json_del(json, MD_KEY_VERSION, NULL);
+            }
+            
+            if (APR_SUCCESS == rv) {
+                md_json_setn(MD_STORE_VERSION, json, MD_KEY_STORE, MD_KEY_VERSION, NULL);
+                rv = md_json_freplace(json, ptemp, MD_JSON_FMT_INDENT, fname, MD_FPROT_F_UONLY);
+            }
+            md_log_perror(MD_LOG_MARK, MD_LOG_INFO, rv, p, "migrated store");
+        } 
+    }
+    return rv;
+}
+
+static apr_status_t setup_store_file(void *baton, apr_pool_t *p, apr_pool_t *ptemp, va_list ap)
+{
+    md_store_fs_t *s_fs = baton;
+    const char *fname;
+    apr_status_t rv;
+
+    (void)ap;
+    s_fs->plain_pkey[MD_SG_DOMAINS] = 1;
+    /* Added: the encryption of tls-alpn-01 certificate keys is not a security issue
+     * for these self-signed, short-lived certificates. Having them unencrypted let's
+     * use pass around the files insteak of an *SSL implementation dependent PKEY_something.
+     */
+    s_fs->plain_pkey[MD_SG_CHALLENGES] = 1;
+    s_fs->plain_pkey[MD_SG_TMP] = 1;
+    
+    if (!MD_OK(md_util_path_merge(&fname, ptemp, s_fs->base, FS_STORE_JSON, NULL))) {
+        return rv;
+    }
+    
+read:
+    if (MD_OK(md_util_is_file(fname, ptemp))) {
+        rv = read_store_file(s_fs, fname, p, ptemp);
+    }
+    else if (APR_STATUS_IS_ENOENT(rv)
+        && APR_STATUS_IS_EEXIST(rv = init_store_file(s_fs, fname, p, ptemp))) {
+        goto read;
+    }
+    return rv;
+}
+
+apr_status_t md_store_fs_init(md_store_t **pstore, apr_pool_t *p, const char *path)
+{
+    md_store_fs_t *s_fs;
+    apr_status_t rv = APR_SUCCESS;
+    
+    s_fs = apr_pcalloc(p, sizeof(*s_fs));
+
+    s_fs->s.load = fs_load;
+    s_fs->s.save = fs_save;
+    s_fs->s.remove = fs_remove;
+    s_fs->s.move = fs_move;
+    s_fs->s.rename = fs_rename;
+    s_fs->s.purge = fs_purge;
+    s_fs->s.iterate = fs_iterate;
+    s_fs->s.iterate_names = fs_iterate_names;
+    s_fs->s.get_fname = fs_get_fname;
+    s_fs->s.is_newer = fs_is_newer;
+    s_fs->s.get_modified = fs_get_modified;
+    s_fs->s.remove_nms = fs_remove_nms;
+    s_fs->s.lock_global = fs_lock_global;
+    s_fs->s.unlock_global = fs_unlock_global;
+
+    /* by default, everything is only readable by the current user */ 
+    s_fs->def_perms.dir = MD_FPROT_D_UONLY;
+    s_fs->def_perms.file = MD_FPROT_F_UONLY;
+
+    /* Account information needs to be accessible to httpd child processes.
+     * private keys are, similar to staging, encrypted. */
+    s_fs->group_perms[MD_SG_ACCOUNTS].dir = MD_FPROT_D_UALL_WREAD;
+    s_fs->group_perms[MD_SG_ACCOUNTS].file = MD_FPROT_F_UALL_WREAD;
+    s_fs->group_perms[MD_SG_STAGING].dir = MD_FPROT_D_UALL_WREAD;
+    s_fs->group_perms[MD_SG_STAGING].file = MD_FPROT_F_UALL_WREAD;
+    /* challenges dir and files are readable by all, no secrets involved */ 
+    s_fs->group_perms[MD_SG_CHALLENGES].dir = MD_FPROT_D_UALL_WREAD;
+    s_fs->group_perms[MD_SG_CHALLENGES].file = MD_FPROT_F_UALL_WREAD;
+    /* OCSP data is readable by all, no secrets involved */ 
+    s_fs->group_perms[MD_SG_OCSP].dir = MD_FPROT_D_UALL_WREAD;
+    s_fs->group_perms[MD_SG_OCSP].file = MD_FPROT_F_UALL_WREAD;
+
+    s_fs->base = apr_pstrdup(p, path);
+
+    rv = md_util_is_dir(s_fs->base, p);
+    if (APR_STATUS_IS_ENOENT(rv)) {
+        md_log_perror(MD_LOG_MARK, MD_LOG_INFO, rv, p,
+            "store directory does not exist, creating %s", s_fs->base);
+        rv = apr_dir_make_recursive(s_fs->base, s_fs->def_perms.dir, p);
+        if (APR_SUCCESS != rv) goto cleanup;
+        rv = apr_file_perms_set(s_fs->base, MD_FPROT_D_UALL_WREAD);
+        if (APR_STATUS_IS_ENOTIMPL(rv)) {
+            rv = APR_SUCCESS;
+        }
+        if (APR_SUCCESS != rv) goto cleanup;
+    }
+    else if (APR_SUCCESS != rv) {
+        md_log_perror(MD_LOG_MARK, MD_LOG_DEBUG, rv, p,
+            "not a plain directory, maybe a symlink? %s", s_fs->base);
+    }
+
+    rv = md_util_pool_vdo(setup_store_file, s_fs, p, NULL);
+    if (APR_SUCCESS != rv) {
+        md_log_perror(MD_LOG_MARK, MD_LOG_ERR, rv, p, "init fs store at %s", s_fs->base);
+    }
+cleanup:
+    *pstore = (rv == APR_SUCCESS)? &(s_fs->s) : NULL;
+    return rv;
+}
+
+apr_status_t md_store_fs_default_perms_set(md_store_t *store, 
+                                           apr_fileperms_t file_perms,
+                                           apr_fileperms_t dir_perms)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    
+    s_fs->def_perms.file = file_perms;
+    s_fs->def_perms.dir = dir_perms;
+    return APR_SUCCESS;
+}
+
+apr_status_t md_store_fs_group_perms_set(md_store_t *store, md_store_group_t group, 
+                                         apr_fileperms_t file_perms,
+                                         apr_fileperms_t dir_perms)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    
+    if (group >= (sizeof(s_fs->group_perms)/sizeof(s_fs->group_perms[0]))) {
+        return APR_ENOTIMPL;
+    }
+    s_fs->group_perms[group].file = file_perms;
+    s_fs->group_perms[group].dir = dir_perms;
+    return APR_SUCCESS;
+}
+
+apr_status_t md_store_fs_set_event_cb(struct md_store_t *store, md_store_fs_cb *cb, void *baton)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    
+    s_fs->event_cb = cb;
+    s_fs->event_baton = baton;
+    return APR_SUCCESS;
+}
+
+static const perms_t *gperms(md_store_fs_t *s_fs, md_store_group_t group)
+{
+    if (group >= (sizeof(s_fs->group_perms)/sizeof(s_fs->group_perms[0]))
+        || !s_fs->group_perms[group].dir) {
+        return &s_fs->def_perms;
+    }
+    return &s_fs->group_perms[group];
+}
+
+static apr_status_t fs_get_fname(const char **pfname, 
+                                 md_store_t *store, md_store_group_t group, 
+                                 const char *name, const char *aspect, 
+                                 apr_pool_t *p)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    if (group == MD_SG_NONE) {
+        return md_util_path_merge(pfname, p, s_fs->base, aspect, NULL);
+    }
+    return md_util_path_merge(pfname, p, 
+                              s_fs->base, md_store_group_name(group), name, aspect, NULL);
+}
+
+static apr_status_t fs_get_dname(const char **pdname, 
+                                 md_store_t *store, md_store_group_t group, 
+                                 const char *name, apr_pool_t *p)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    if (group == MD_SG_NONE) {
+        *pdname = s_fs->base;
+        return APR_SUCCESS;
+    }
+    return md_util_path_merge(pdname, p, s_fs->base, md_store_group_name(group), name, NULL);
+}
+
+static void get_pass(const char **ppass, apr_size_t *plen, 
+                     md_store_fs_t *s_fs, md_store_group_t group)
+{
+    if (s_fs->plain_pkey[group]) {
+        *ppass = NULL;
+        *plen = 0;
+    }
+    else {
+        *ppass = (const char *)s_fs->key.data;
+        *plen = s_fs->key.len;
+    }
+}
+ 
+static apr_status_t fs_fload(void **pvalue, md_store_fs_t *s_fs, const char *fpath, 
+                             md_store_group_t group, md_store_vtype_t vtype, 
+                             apr_pool_t *p, apr_pool_t *ptemp)
+{
+    apr_status_t rv;
+    const char *pass;
+    apr_size_t pass_len;
+    
+    if (pvalue != NULL) {
+        switch (vtype) {
+            case MD_SV_TEXT:
+                rv = md_text_fread8k((const char **)pvalue, p, fpath);
+                break;
+            case MD_SV_JSON:
+                rv = md_json_readf((md_json_t **)pvalue, p, fpath);
+                break;
+            case MD_SV_CERT:
+                rv = md_cert_fload((md_cert_t **)pvalue, p, fpath);
+                break;
+            case MD_SV_PKEY:
+                get_pass(&pass, &pass_len, s_fs, group);
+                rv = md_pkey_fload((md_pkey_t **)pvalue, p, pass, pass_len, fpath);
+                break;
+            case MD_SV_CHAIN:
+                rv = md_chain_fload((apr_array_header_t **)pvalue, p, fpath);
+                break;
+            default:
+                rv = APR_ENOTIMPL;
+                break;
+        }
+        md_log_perror(MD_LOG_MARK, MD_LOG_TRACE3, rv, ptemp, 
+                      "loading type %d from %s", vtype, fpath);
+    }
+    else { /* check for existence only */
+        rv = md_util_is_file(fpath, p);
+    }
+    return rv;
+}
+
+static apr_status_t pfs_load(void *baton, apr_pool_t *p, apr_pool_t *ptemp, va_list ap)
+{
+    md_store_fs_t *s_fs = baton;
+    const char *fpath, *name, *aspect;
+    md_store_vtype_t vtype;
+    md_store_group_t group;
+    void **pvalue;
+    apr_status_t rv;
+    
+    group = (md_store_group_t)va_arg(ap, int);
+    name = va_arg(ap, const char *);
+    aspect = va_arg(ap, const char *);
+    vtype = (md_store_vtype_t)va_arg(ap, int);
+    pvalue= va_arg(ap, void **);
+        
+    if (MD_OK(fs_get_fname(&fpath, &s_fs->s, group, name, aspect, ptemp))) {
+        rv = fs_fload(pvalue, s_fs, fpath, group, vtype, p, ptemp);
+    }
+    return rv;
+}
+
+static apr_status_t dispatch(md_store_fs_t *s_fs, md_store_fs_ev_t ev, unsigned int group, 
+                             const char *fname, apr_filetype_e ftype, apr_pool_t *p)
+{
+    (void)ev;
+    if (s_fs->event_cb) {
+        return s_fs->event_cb(s_fs->event_baton, &s_fs->s, MD_S_FS_EV_CREATED, 
+                              group, fname, ftype, p);
+    }
+    return APR_SUCCESS;
+}
+
+static apr_status_t mk_group_dir(const char **pdir, md_store_fs_t *s_fs, 
+                                 md_store_group_t group, const char *name,
+                                 apr_pool_t *p)
+{
+    const perms_t *perms;
+    apr_status_t rv;
+    
+    perms = gperms(s_fs, group);
+
+    *pdir = NULL;
+    rv = fs_get_dname(pdir, &s_fs->s, group, name, p);
+    if ((APR_SUCCESS != rv) || (MD_SG_NONE == group)) goto cleanup;
+
+    rv = md_util_is_dir(*pdir, p);
+    if (APR_STATUS_IS_ENOENT(rv)) {
+        md_log_perror(MD_LOG_MARK, MD_LOG_TRACE3, rv, p, "not a directory, creating %s", *pdir);
+        rv = apr_dir_make_recursive(*pdir, perms->dir, p);
+        if (APR_SUCCESS != rv) goto cleanup;
+        dispatch(s_fs, MD_S_FS_EV_CREATED, group, *pdir, APR_DIR, p);
+    }
+
+    rv = apr_file_perms_set(*pdir, perms->dir);
+    md_log_perror(MD_LOG_MARK, MD_LOG_TRACE3, rv, p, "mk_group_dir %s perm set", *pdir);
+    if (APR_STATUS_IS_ENOTIMPL(rv)) {
+        rv = APR_SUCCESS;
+    }
+cleanup:
+    if (APR_SUCCESS != rv) {
+        md_log_perror(MD_LOG_MARK, MD_LOG_ERR, rv, p, "mk_group_dir %d %s",
+            group, (*pdir? *pdir : (name? name : "(null)")));
+    }
+    return rv;
+}
+
+static apr_status_t pfs_is_newer(void *baton, apr_pool_t *p, apr_pool_t *ptemp, va_list ap)
+{
+    md_store_fs_t *s_fs = baton;
+    const char *fname1, *fname2, *name, *aspect;
+    md_store_group_t group1, group2;
+    apr_finfo_t inf1, inf2;
+    int *pnewer;
+    apr_status_t rv;
+    
+    (void)p;
+    group1 = (md_store_group_t)va_arg(ap, int);
+    group2 = (md_store_group_t)va_arg(ap, int);
+    name = va_arg(ap, const char*);
+    aspect = va_arg(ap, const char*);
+    pnewer = va_arg(ap, int*);
+    
+    *pnewer = 0;
+    if (   MD_OK(fs_get_fname(&fname1, &s_fs->s, group1, name, aspect, ptemp))
+        && MD_OK(fs_get_fname(&fname2, &s_fs->s, group2, name, aspect, ptemp))
+        && MD_OK(apr_stat(&inf1, fname1, APR_FINFO_MTIME, ptemp))
+        && MD_OK(apr_stat(&inf2, fname2, APR_FINFO_MTIME, ptemp))) {
+        *pnewer = inf1.mtime > inf2.mtime;
+    }
+
+    return rv;
+}
+
+static int fs_is_newer(md_store_t *store, md_store_group_t group1, md_store_group_t group2,  
+                       const char *name, const char *aspect, apr_pool_t *p)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    int newer = 0;
+    apr_status_t rv;
+    
+    rv = md_util_pool_vdo(pfs_is_newer, s_fs, p, group1, group2, name, aspect, &newer, NULL);
+    if (APR_SUCCESS == rv) {
+        return newer;
+    }
+    return 0;
+}
+
+static apr_status_t pfs_get_modified(void *baton, apr_pool_t *p, apr_pool_t *ptemp, va_list ap)
+{
+    md_store_fs_t *s_fs = baton;
+    const char *fname, *name, *aspect;
+    md_store_group_t group;
+    apr_finfo_t inf;
+    apr_time_t *pmtime;
+    apr_status_t rv;
+    
+    (void)p;
+    group = (md_store_group_t)va_arg(ap, int);
+    name = va_arg(ap, const char*);
+    aspect = va_arg(ap, const char*);
+    pmtime = va_arg(ap, apr_time_t*);
+    
+    *pmtime = 0;
+    if (   MD_OK(fs_get_fname(&fname, &s_fs->s, group, name, aspect, ptemp))
+        && MD_OK(apr_stat(&inf, fname, APR_FINFO_MTIME, ptemp))) {
+        *pmtime = inf.mtime;
+    }
+
+    return rv;
+}
+
+static apr_time_t fs_get_modified(md_store_t *store, md_store_group_t group,  
+                                  const char *name, const char *aspect, apr_pool_t *p)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    apr_time_t mtime;
+    apr_status_t rv;
+    
+    rv = md_util_pool_vdo(pfs_get_modified, s_fs, p, group, name, aspect, &mtime, NULL);
+    if (APR_SUCCESS == rv) {
+        return mtime;
+    }
+    return 0;
+}
+ 
+static apr_status_t pfs_save(void *baton, apr_pool_t *p, apr_pool_t *ptemp, va_list ap)
+{
+    md_store_fs_t *s_fs = baton;
+    const char *gdir, *dir, *fpath, *name, *aspect;
+    md_store_vtype_t vtype;
+    md_store_group_t group;
+    void *value;
+    int create;
+    apr_status_t rv;
+    const perms_t *perms;
+    const char *pass;
+    apr_size_t pass_len;
+    
+    group = (md_store_group_t)va_arg(ap, int);
+    name = va_arg(ap, const char*);
+    aspect = va_arg(ap, const char*);
+    vtype = (md_store_vtype_t)va_arg(ap, int);
+    value = va_arg(ap, void *);
+    create = va_arg(ap, int);
+    
+    perms = gperms(s_fs, group);
+    
+    if (   MD_OK(mk_group_dir(&gdir, s_fs, group, NULL, p)) 
+        && MD_OK(mk_group_dir(&dir, s_fs, group, name, p))
+        && MD_OK(md_util_path_merge(&fpath, ptemp, dir, aspect, NULL))) {
+        
+        md_log_perror(MD_LOG_MARK, MD_LOG_TRACE2, 0, ptemp, "storing in %s", fpath);
+        switch (vtype) {
+            case MD_SV_TEXT:
+                rv = (create? md_text_fcreatex(fpath, perms->file, p, value)
+                      : md_text_freplace(fpath, perms->file, p, value));
+                break;
+            case MD_SV_JSON:
+                rv = (create? md_json_fcreatex((md_json_t *)value, p, MD_JSON_FMT_INDENT, 
+                                               fpath, perms->file)
+                      : md_json_freplace((md_json_t *)value, p, MD_JSON_FMT_INDENT, 
+                                         fpath, perms->file));
+                break;
+            case MD_SV_CERT:
+                rv = md_cert_fsave((md_cert_t *)value, ptemp, fpath, perms->file);
+                break;
+            case MD_SV_PKEY:
+                /* Take care that we write private key with access only to the user,
+                 * unless we write the key encrypted */
+                get_pass(&pass, &pass_len, s_fs, group);
+                rv = md_pkey_fsave((md_pkey_t *)value, ptemp, pass, pass_len, 
+                                   fpath, (pass && pass_len)? perms->file : MD_FPROT_F_UONLY);
+                break;
+            case MD_SV_CHAIN:
+                rv = md_chain_fsave((apr_array_header_t*)value, ptemp, fpath, perms->file);
+                break;
+            default:
+                return APR_ENOTIMPL;
+        }
+        if (APR_SUCCESS == rv) {
+            rv = dispatch(s_fs, MD_S_FS_EV_CREATED, group, fpath, APR_REG, p);
+        }
+    }
+    return rv;
+}
+
+static apr_status_t pfs_remove(void *baton, apr_pool_t *p, apr_pool_t *ptemp, va_list ap)
+{
+    md_store_fs_t *s_fs = baton;
+    const char *dir, *name, *fpath, *groupname, *aspect;
+    apr_status_t rv;
+    int force;
+    apr_finfo_t info;
+    md_store_group_t group;
+    
+    (void)p;
+    group = (md_store_group_t)va_arg(ap, int);
+    name = va_arg(ap, const char*);
+    aspect = va_arg(ap, const char *);
+    force = va_arg(ap, int);
+    
+    groupname = md_store_group_name(group);
+    
+    if (   MD_OK(md_util_path_merge(&dir, ptemp, s_fs->base, groupname, name, NULL))
+        && MD_OK(md_util_path_merge(&fpath, ptemp, dir, aspect, NULL))) {
+        md_log_perror(MD_LOG_MARK, MD_LOG_TRACE1, 0, ptemp, "start remove of md %s/%s/%s", 
+                      groupname, name, aspect);
+
+        if (!MD_OK(apr_stat(&info, dir, APR_FINFO_TYPE, ptemp))) {
+            if (APR_ENOENT == rv && force) {
+                return APR_SUCCESS;
+            }
+            return rv;
+        }
+    
+        rv = apr_file_remove(fpath, ptemp);
+        if (APR_ENOENT == rv && force) {
+            rv = APR_SUCCESS;
+        }
+    }
+    return rv;
+}
+
+static apr_status_t fs_load(md_store_t *store, md_store_group_t group, 
+                            const char *name, const char *aspect,  
+                            md_store_vtype_t vtype, void **pvalue, apr_pool_t *p)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    return md_util_pool_vdo(pfs_load, s_fs, p, group, name, aspect, vtype, pvalue, NULL);
+}
+
+static apr_status_t fs_save(md_store_t *store, apr_pool_t *p, md_store_group_t group, 
+                            const char *name, const char *aspect,  
+                            md_store_vtype_t vtype, void *value, int create)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    return md_util_pool_vdo(pfs_save, s_fs, p, group, name, aspect, 
+                            vtype, value, create, NULL);
+}
+
+static apr_status_t fs_remove(md_store_t *store, md_store_group_t group, 
+                              const char *name, const char *aspect, 
+                              apr_pool_t *p, int force)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    return md_util_pool_vdo(pfs_remove, s_fs, p, group, name, aspect, force, NULL);
+}
+
+static apr_status_t pfs_purge(void *baton, apr_pool_t *p, apr_pool_t *ptemp, va_list ap)
+{
+    md_store_fs_t *s_fs = baton;
+    const char *dir, *name, *groupname;
+    md_store_group_t group;
+    apr_status_t rv;
+    
+    (void)p;
+    group = (md_store_group_t)va_arg(ap, int);
+    name = va_arg(ap, const char*);
+    
+    groupname = md_store_group_name(group);
+
+    if (MD_OK(md_util_path_merge(&dir, ptemp, s_fs->base, groupname, name, NULL))) {
+        /* Remove all files in dir, there should be no sub-dirs */
+        rv = md_util_rm_recursive(dir, ptemp, 1);
+    }
+    if (!APR_STATUS_IS_ENOENT(rv)) {
+        md_log_perror(MD_LOG_MARK, MD_LOG_TRACE2, rv, ptemp, "purge %s/%s (%s)", groupname, name, dir);
+    }
+    return APR_SUCCESS;
+}
+
+static apr_status_t fs_purge(md_store_t *store, apr_pool_t *p, 
+                             md_store_group_t group, const char *name)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    return md_util_pool_vdo(pfs_purge, s_fs, p, group, name, NULL);
+}
+
+/**************************************************************************************************/
+/* iteration */
+
+typedef struct {
+    md_store_fs_t *s_fs;
+    md_store_group_t group;
+    const char *pattern;
+    const char *aspect;
+    md_store_vtype_t vtype;
+    md_store_inspect *inspect;
+    const char *dirname;
+    void *baton;
+    apr_time_t ts;
+} inspect_ctx;
+
+static apr_status_t insp(void *baton, apr_pool_t *p, apr_pool_t *ptemp, 
+                         const char *dir, const char *name, apr_filetype_e ftype)
+{
+    inspect_ctx *ctx = baton;
+    apr_status_t rv;
+    void *value;
+    const char *fpath;
+ 
+    (void)ftype;   
+    md_log_perror(MD_LOG_MARK, MD_LOG_TRACE3, 0, ptemp, "inspecting value at: %s/%s", dir, name);
+    if (APR_SUCCESS == (rv = md_util_path_merge(&fpath, ptemp, dir, name, NULL))) {
+        rv = fs_fload(&value, ctx->s_fs, fpath, ctx->group, ctx->vtype, p, ptemp);
+        if (APR_SUCCESS == rv 
+            && !ctx->inspect(ctx->baton, ctx->dirname, name, ctx->vtype, value, p)) {
+            return APR_EOF;
+        }
+        else if (APR_STATUS_IS_ENOENT(rv)) {
+            rv = APR_SUCCESS;
+        }
+    } 
+    return rv;
+}
+
+static apr_status_t insp_dir(void *baton, apr_pool_t *p, apr_pool_t *ptemp, 
+                             const char *dir, const char *name, apr_filetype_e ftype)
+{
+    inspect_ctx *ctx = baton;
+    apr_status_t rv;
+    const char *fpath;
+ 
+    (void)ftype;
+    md_log_perror(MD_LOG_MARK, MD_LOG_TRACE3, 0, ptemp, "inspecting dir at: %s/%s", dir, name);
+    if (MD_OK(md_util_path_merge(&fpath, p, dir, name, NULL))) {
+        ctx->dirname = name;
+        rv = md_util_files_do(insp, ctx, p, fpath, ctx->aspect, NULL);
+        if (APR_STATUS_IS_ENOENT(rv)) {
+            rv = APR_SUCCESS;
+        }
+    } 
+    return rv;
+}
+
+static apr_status_t fs_iterate(md_store_inspect *inspect, void *baton, md_store_t *store, 
+                               apr_pool_t *p, md_store_group_t group, const char *pattern, 
+                               const char *aspect, md_store_vtype_t vtype)
+{
+    const char *groupname;
+    apr_status_t rv;
+    inspect_ctx ctx;
+    
+    ctx.s_fs = FS_STORE(store);
+    ctx.group = group;
+    ctx.pattern = pattern;
+    ctx.aspect = aspect;
+    ctx.vtype = vtype;
+    ctx.inspect = inspect;
+    ctx.baton = baton;
+    groupname = md_store_group_name(group);
+
+    rv = md_util_files_do(insp_dir, &ctx, p, ctx.s_fs->base, groupname, pattern, NULL);
+    
+    return rv;
+}
+
+static apr_status_t insp_name(void *baton, apr_pool_t *p, apr_pool_t *ptemp, 
+                              const char *dir, const char *name, apr_filetype_e ftype)
+{
+    inspect_ctx *ctx = baton;
+    
+    (void)ftype;
+    (void)p;
+    md_log_perror(MD_LOG_MARK, MD_LOG_TRACE3, 0, ptemp, "inspecting name at: %s/%s", dir, name);
+    return ctx->inspect(ctx->baton, dir, name, 0, NULL, ptemp);
+}
+
+static apr_status_t fs_iterate_names(md_store_inspect *inspect, void *baton, md_store_t *store, 
+                                     apr_pool_t *p, md_store_group_t group, const char *pattern)
+{
+    const char *groupname;
+    apr_status_t rv;
+    inspect_ctx ctx;
+    
+    ctx.s_fs = FS_STORE(store);
+    ctx.group = group;
+    ctx.pattern = pattern;
+    ctx.inspect = inspect;
+    ctx.baton = baton;
+    groupname = md_store_group_name(group);
+
+    rv = md_util_files_do(insp_name, &ctx, p, ctx.s_fs->base, groupname, pattern, NULL);
+    
+    return rv;
+}
+
+static apr_status_t remove_nms_file(void *baton, apr_pool_t *p, apr_pool_t *ptemp, 
+                                    const char *dir, const char *name, apr_filetype_e ftype)
+{
+    inspect_ctx *ctx = baton;
+    const char *fname;
+    apr_finfo_t inf;
+    apr_status_t rv = APR_SUCCESS;
+
+    (void)p;
+    if (APR_DIR == ftype) goto leave;
+    if (APR_SUCCESS != (rv = md_util_path_merge(&fname, ptemp, dir, name, NULL))) goto leave;
+    if (APR_SUCCESS != (rv = apr_stat(&inf, fname, APR_FINFO_MTIME, ptemp))) goto leave;
+    if (inf.mtime >= ctx->ts) goto leave;
+
+    md_log_perror(MD_LOG_MARK, MD_LOG_TRACE3, 0, ptemp, "remove_nms file: %s/%s", dir, name);
+    rv = apr_file_remove(fname, ptemp);
+
+leave:
+    return rv;
+}
+
+static apr_status_t remove_nms_dir(void *baton, apr_pool_t *p, apr_pool_t *ptemp, 
+                                   const char *dir, const char *name, apr_filetype_e ftype)
+{
+    inspect_ctx *ctx = baton;
+    apr_status_t rv;
+    const char *fpath;
+ 
+    (void)ftype;
+    md_log_perror(MD_LOG_MARK, MD_LOG_TRACE3, 0, ptemp, "remove_nms dir at: %s/%s", dir, name);
+    if (MD_OK(md_util_path_merge(&fpath, p, dir, name, NULL))) {
+        ctx->dirname = name;
+        rv = md_util_files_do(remove_nms_file, ctx, p, fpath, ctx->aspect, NULL);
+        if (APR_STATUS_IS_ENOENT(rv)) {
+            rv = APR_SUCCESS;
+        }
+    } 
+    return rv;
+}
+
+static apr_status_t fs_remove_nms(md_store_t *store, apr_pool_t *p, 
+                                  apr_time_t modified, md_store_group_t group, 
+                                  const char *name, const char *aspect)
+{
+    const char *groupname;
+    apr_status_t rv;
+    inspect_ctx ctx;
+    
+    ctx.s_fs = FS_STORE(store);
+    ctx.group = group;
+    ctx.pattern = name;
+    ctx.aspect = aspect;
+    ctx.ts = modified;
+    groupname = md_store_group_name(group);
+
+    rv = md_util_files_do(remove_nms_dir, &ctx, p, ctx.s_fs->base, groupname, name, NULL);
+    
+    return rv;
+}
+
+/**************************************************************************************************/
+/* moving */
+
+static apr_status_t pfs_move(void *baton, apr_pool_t *p, apr_pool_t *ptemp, va_list ap)
+{
+    md_store_fs_t *s_fs = baton;
+    const char *name, *from_group, *to_group, *from_dir, *to_dir, *arch_dir, *dir;
+    md_store_group_t from, to;
+    int archive;
+    apr_status_t rv;
+    
+    (void)p;
+    from = (md_store_group_t)va_arg(ap, int);
+    to = (md_store_group_t)va_arg(ap, int);
+    name = va_arg(ap, const char*);
+    archive = va_arg(ap, int);
+    
+    from_group = md_store_group_name(from);
+    to_group = md_store_group_name(to);
+    if (!strcmp(from_group, to_group)) {
+        return APR_EINVAL;
+    }
+
+    if (   !MD_OK(md_util_path_merge(&from_dir, ptemp, s_fs->base, from_group, name, NULL))
+        || !MD_OK(md_util_path_merge(&to_dir, ptemp, s_fs->base, to_group, name, NULL))) {
+        goto out;
+    }
+    
+    if (!MD_OK(md_util_is_dir(from_dir, ptemp))) {
+        md_log_perror(MD_LOG_MARK, MD_LOG_DEBUG, rv, ptemp, "source is no dir: %s", from_dir);
+        goto out;
+    }
+    
+    if (MD_OK(archive? md_util_is_dir(to_dir, ptemp) : APR_ENOENT)) {
+        int n = 1;
+        const char *narch_dir;
+
+        if (    !MD_OK(md_util_path_merge(&dir, ptemp, s_fs->base, 
+                                          md_store_group_name(MD_SG_ARCHIVE), NULL))
+            || !MD_OK(apr_dir_make_recursive(dir, MD_FPROT_D_UONLY, ptemp))
+            || !MD_OK(md_util_path_merge(&arch_dir, ptemp, dir, name, NULL))) {
+            goto out;
+        }
+        
+#ifdef WIN32
+        /* WIN32 and handling of files/dirs. What can one say? */
+        
+        while (n < 1000) {
+            narch_dir = apr_psprintf(ptemp, "%s.%d", arch_dir, n);
+            rv = md_util_is_dir(narch_dir, ptemp);
+            if (APR_STATUS_IS_ENOENT(rv)) {
+                md_log_perror(MD_LOG_MARK, MD_LOG_TRACE1, rv, ptemp, "using archive dir: %s", 
+                              narch_dir);
+                break;
+            }
+            else {
+                ++n;
+                narch_dir = NULL;
+            }
+        }
+
+#else   /* ifdef WIN32 */
+
+        while (n < 1000) {
+            narch_dir = apr_psprintf(ptemp, "%s.%d", arch_dir, n);
+            if (MD_OK(apr_dir_make(narch_dir, MD_FPROT_D_UONLY, ptemp))) {
+                md_log_perror(MD_LOG_MARK, MD_LOG_TRACE1, rv, ptemp, "using archive dir: %s", 
+                              narch_dir);
+                break;
+            }
+            else if (APR_EEXIST == rv) {
+                ++n;
+                narch_dir = NULL;
+            }
+            else {
+                md_log_perror(MD_LOG_MARK, MD_LOG_ERR, rv, ptemp, "creating archive dir: %s", 
+                              narch_dir);
+                goto out;
+            }
+        }
+         
+#endif   /* ifdef WIN32 (else part) */
+        
+        if (!narch_dir) {
+            md_log_perror(MD_LOG_MARK, MD_LOG_ERR, rv, ptemp, "ran out of numbers less than 1000 "
+                          "while looking for an available one in %s to archive the data "
+                          "from %s. Either something is generally wrong or you need to "
+                          "clean up some of those directories.", arch_dir, from_dir);
+            rv = APR_EGENERAL;
+            goto out;
+        }
+        
+        if (!MD_OK(apr_file_rename(to_dir, narch_dir, ptemp))) {
+                md_log_perror(MD_LOG_MARK, MD_LOG_ERR, rv, ptemp, "rename from %s to %s",
+                              to_dir, narch_dir);
+                goto out;
+        }
+        if (!MD_OK(apr_file_rename(from_dir, to_dir, ptemp))) {
+            md_log_perror(MD_LOG_MARK, MD_LOG_ERR, rv, ptemp, "rename from %s to %s",
+                          from_dir, to_dir);
+            apr_file_rename(narch_dir, to_dir, ptemp);
+            goto out;
+        }
+        if (MD_OK(dispatch(s_fs, MD_S_FS_EV_MOVED, to, to_dir, APR_DIR, ptemp))) {
+            rv = dispatch(s_fs, MD_S_FS_EV_MOVED, MD_SG_ARCHIVE, narch_dir, APR_DIR, ptemp);
+        }
+    }
+    else if (APR_STATUS_IS_ENOENT(rv)) {
+        if (APR_SUCCESS != (rv = apr_file_rename(from_dir, to_dir, ptemp))) {
+            md_log_perror(MD_LOG_MARK, MD_LOG_ERR, rv, ptemp, "rename from %s to %s",
+                          from_dir, to_dir);
+            goto out;
+        }
+    }
+    else {
+        md_log_perror(MD_LOG_MARK, MD_LOG_DEBUG, rv, ptemp, "target is no dir: %s", to_dir);
+        goto out;
+    }
+    
+out:
+    return rv;
+}
+
+static apr_status_t fs_move(md_store_t *store, apr_pool_t *p, 
+                            md_store_group_t from, md_store_group_t to, 
+                            const char *name, int archive)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    return md_util_pool_vdo(pfs_move, s_fs, p, from, to, name, archive, NULL);
+}
+
+static apr_status_t pfs_rename(void *baton, apr_pool_t *p, apr_pool_t *ptemp, va_list ap)
+{
+    md_store_fs_t *s_fs = baton;
+    const char *group_name, *from_dir, *to_dir;
+    md_store_group_t group;
+    const char *from, *to;
+    apr_status_t rv;
+    
+    (void)p;
+    group = (md_store_group_t)va_arg(ap, int);
+    from = va_arg(ap, const char*);
+    to = va_arg(ap, const char*);
+    
+    group_name = md_store_group_name(group);
+    if (   !MD_OK(md_util_path_merge(&from_dir, ptemp, s_fs->base, group_name, from, NULL))
+        || !MD_OK(md_util_path_merge(&to_dir, ptemp, s_fs->base, group_name, to, NULL))) {
+        goto out;
+    }
+    
+    if (APR_SUCCESS != (rv = apr_file_rename(from_dir, to_dir, ptemp))
+        && !APR_STATUS_IS_ENOENT(rv)) {
+        md_log_perror(MD_LOG_MARK, MD_LOG_ERR, rv, ptemp, "rename from %s to %s",
+                      from_dir, to_dir);
+        goto out;
+    }
+out:
+    return rv;
+}
+
+static apr_status_t fs_rename(md_store_t *store, apr_pool_t *p, 
+                            md_store_group_t group, const char *from, const char *to)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    return md_util_pool_vdo(pfs_rename, s_fs, p, group, from, to, NULL);
+}
+
+static apr_status_t fs_lock_global(md_store_t *store, apr_pool_t *p, apr_time_t max_wait)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+    apr_status_t rv;
+    const char *lpath;
+    apr_time_t end;
+
+    if (s_fs->global_lock) {
+        rv = APR_EEXIST;
+        md_log_perror(MD_LOG_MARK, MD_LOG_ERR, rv, p, "already locked globally");
+        goto cleanup;
+    }
+
+    rv = md_util_path_merge(&lpath, p, s_fs->base, MD_FS_LOCK_NAME, NULL);
+    if (APR_SUCCESS != rv) goto cleanup;
+    end = apr_time_now() + max_wait;
+
+    md_log_perror(MD_LOG_MARK, MD_LOG_TRACE1, 0, p,
+                  "acquire global lock: %s", lpath);
+    while (apr_time_now() < end) {
+        rv = apr_file_open(&s_fs->global_lock, lpath,
+                           (APR_FOPEN_WRITE|APR_FOPEN_CREATE),
+                           MD_FPROT_F_UALL_GREAD, p);
+        if (APR_SUCCESS != rv) {
+            md_log_perror(MD_LOG_MARK, MD_LOG_TRACE1, rv, p,
+                          "unable to create/open lock file: %s",
+                          lpath);
+            goto next_try;
+        }
+        rv = apr_file_lock(s_fs->global_lock,
+                           APR_FLOCK_EXCLUSIVE|APR_FLOCK_NONBLOCK);
+        if (APR_SUCCESS == rv) {
+            goto cleanup;
+        }
+        md_log_perror(MD_LOG_MARK, MD_LOG_TRACE1, rv, p,
+                      "unable to obtain lock on: %s",
+                      lpath);
+
+    next_try:
+        if (s_fs->global_lock) {
+            apr_file_close(s_fs->global_lock);
+            s_fs->global_lock = NULL;
+        }
+        apr_sleep(apr_time_from_msec(100));
+    }
+    rv = APR_EGENERAL;
+    md_log_perror(MD_LOG_MARK, MD_LOG_TRACE1, rv, p,
+                  "acquire global lock: %s", lpath);
+
+cleanup:
+    return rv;
+}
+
+static void fs_unlock_global(md_store_t *store, apr_pool_t *p)
+{
+    md_store_fs_t *s_fs = FS_STORE(store);
+
+    (void)p;
+    if (s_fs->global_lock) {
+        apr_file_close(s_fs->global_lock);
+        s_fs->global_lock = NULL;
+    }
+}