diff options
author | Vikas Gorur <vikas@zresearch.com> | 2009-02-18 17:36:07 +0530 |
---|---|---|
committer | Vikas Gorur <vikas@zresearch.com> | 2009-02-18 17:36:07 +0530 |
commit | 77adf4cd648dce41f89469dd185deec6b6b53a0b (patch) | |
tree | 02e155a5753b398ee572b45793f889b538efab6b /mod_glusterfs/lighttpd/1.4 | |
parent | f3b2e6580e5663292ee113c741343c8a43ee133f (diff) |
Added all files
Diffstat (limited to 'mod_glusterfs/lighttpd/1.4')
-rw-r--r-- | mod_glusterfs/lighttpd/1.4/Makefile.am | 3 | ||||
-rw-r--r-- | mod_glusterfs/lighttpd/1.4/Makefile.am.diff | 29 | ||||
-rw-r--r-- | mod_glusterfs/lighttpd/1.4/README.txt | 57 | ||||
-rw-r--r-- | mod_glusterfs/lighttpd/1.4/mod_glusterfs.c | 1709 | ||||
-rw-r--r-- | mod_glusterfs/lighttpd/1.4/mod_glusterfs.h | 29 |
5 files changed, 1827 insertions, 0 deletions
diff --git a/mod_glusterfs/lighttpd/1.4/Makefile.am b/mod_glusterfs/lighttpd/1.4/Makefile.am new file mode 100644 index 00000000000..eda329111dc --- /dev/null +++ b/mod_glusterfs/lighttpd/1.4/Makefile.am @@ -0,0 +1,3 @@ +EXTRA_DIST = Makefile.am.diff mod_glusterfs.c mod_glusterfs.h README.txt + +CLEANFILES = diff --git a/mod_glusterfs/lighttpd/1.4/Makefile.am.diff b/mod_glusterfs/lighttpd/1.4/Makefile.am.diff new file mode 100644 index 00000000000..375696b5d8f --- /dev/null +++ b/mod_glusterfs/lighttpd/1.4/Makefile.am.diff @@ -0,0 +1,29 @@ +--- lighttpd-1.4.19/src/Makefile.am 2008-04-16 18:42:18.000000000 +0400 ++++ lighttpd-1.4.19.mod/src/Makefile.am 2008-04-16 18:41:11.000000000 +0400 +@@ -1,4 +1,4 @@ +-AM_CFLAGS = $(FAM_CFLAGS) ++AM_CFLAGS = $(FAM_CFLAGS) -D_FILE_OFFSET_BITS=64 -D__USE_FILE_OFFSET64 + + noinst_PROGRAMS=proc_open lemon # simple-fcgi #graphic evalo bench ajp ssl error_test adserver gen-license + sbin_PROGRAMS=lighttpd lighttpd-angel +@@ -241,6 +241,11 @@ + mod_accesslog_la_LDFLAGS = -module -export-dynamic -avoid-version -no-undefined + mod_accesslog_la_LIBADD = $(common_libadd) + ++lib_LTLIBRARIES += mod_glusterfs.la ++mod_glusterfs_la_SOURCES = mod_glusterfs.c ++mod_glusterfs_la_CFLAGS = $(AM_CFLAGS) ++mod_glusterfs_la_LDFLAGS = -module -export-dynamic -avoid-version -no-undefined -lglusterfsclient -lpthread ++mod_glusterfs_la_LIBADD = $(common_libadd) + + hdr = server.h buffer.h network.h log.h keyvalue.h \ + response.h request.h fastcgi.h chunk.h \ +@@ -254,7 +259,7 @@ + configparser.h mod_ssi_exprparser.h \ + sys-mmap.h sys-socket.h mod_cml.h mod_cml_funcs.h \ + splaytree.h proc_open.h status_counter.h \ +- mod_magnet_cache.h ++ mod_magnet_cache.h mod_glusterfs.h + + DEFS= @DEFS@ -DLIBRARY_DIR="\"$(libdir)\"" -DSBIN_DIR="\"$(sbindir)\"" + diff --git a/mod_glusterfs/lighttpd/1.4/README.txt b/mod_glusterfs/lighttpd/1.4/README.txt new file mode 100644 index 00000000000..786a146e44d --- /dev/null +++ b/mod_glusterfs/lighttpd/1.4/README.txt @@ -0,0 +1,57 @@ +Introduction +============ +mod_glusterfs is a module written for lighttpd to speed up the access of files present on glusterfs. mod_glusterfs uses libglusterfsclient library provided for glusterfs and hence can be used without fuse (File System in User Space). + +Usage +===== +To use mod_glusterfs with lighttpd-1.4, copy mod_glusterfs.c and mod_glusterfs.h into src/ of lighttpd-1.4 source tree, and apply the Makefile.am.diff to src/Makefile.am. Re-run ./autogen.sh on the top level of the lighttpd-1.4 build tree and recompile. + +# cp mod_glusterfs.[ch] /home/glusterfs/lighttpd-1.4/src/ +# cp Makefile.am.diff /home/glusterfs/lighttpd-1.4/ +# cd /home/glusterfs/lighttpd-1.4 +# patch -p1 < Makefile.am.diff +# ./autogen.sh +# ./configure +# make +# make install + +Configuration +============= +* mod_glusterfs should be listed at the begining of the list server.modules in lighttpd.conf. + +Below is a snippet from lighttpd.conf concerning to mod_glusterfs. + +$HTTP["url"] =~ "^/glusterfs" { + glusterfs.prefix = "/glusterfs" + glusterfs.document-root = "/home/glusterfs/document-root" + glusterfs.logfile = "/var/log/glusterfs-logfile" + glusterfs.volume-specfile = "/etc/glusterfs/glusterfs.vol" + glusterfs.loglevel = "error" + glusterfs.cache-timeout = 300 + glusterfs.xattr-interface-size-limit = "65536" +} + +* $HTTP["url"] =~ "^/glusterfs" + A perl style regular expression used to match against the url. If regular expression matches the url, the url is handled by mod_glusterfs. Note that the pattern given here should match glusterfs.prefix. + +* glusterfs.prefix (COMPULSORY) + A string to be present at the starting of the file path in the url so that the file would be handled by glusterfs. + Eg., A GET request on the url http://www.example.com/glusterfs-prefix/some-dir/example-file will result in fetching of the file "/some-dir/example-file" from glusterfs mount if glusterfs.prefix is set to "/glusterfs-prefix". + +* glusterfs.volume-specfile (COMPULSORY) + Path to the the glusterfs volume specification file. + +* glusterfs.logfile (COMPULSORY) + Path to the glusterfs logfile. + +* glusterfs.loglevel (OPTIONAL, default = warning) + Allowed values are critical, error, warning, debug, none in the decreasing order of severity of error conditions. + +* glusterfs.cache-timeout (OPTIONAL, default = 0) + Timeout values for glusterfs stat and lookup cache. + +* glusterfs.document-root (COMPULSORY) + An absolute path, relative to which all the files are fetched from glusterfs. + +* glusterfs.xattr-interface-size-limit (OPTIONAL, default = 0) + Files with sizes upto and including this value are fetched through the extended attribute interface of glusterfs rather than the usual open-read-close set of operations. For files of small sizes, it is recommended to use extended attribute interface. diff --git a/mod_glusterfs/lighttpd/1.4/mod_glusterfs.c b/mod_glusterfs/lighttpd/1.4/mod_glusterfs.c new file mode 100644 index 00000000000..9de33b7c437 --- /dev/null +++ b/mod_glusterfs/lighttpd/1.4/mod_glusterfs.c @@ -0,0 +1,1709 @@ +/* + Copyright (c) 2008 Z RESEARCH, Inc. <http://www.zresearch.com> + This file is part of GlusterFS. + + GlusterFS is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published + by the Free Software Foundation; either version 3 of the License, + or (at your option) any later version. + + GlusterFS is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see + <http://www.gnu.org/licenses/>. +*/ + +#include <ctype.h> +#include <stdlib.h> +#include <stdio.h> +#include <string.h> +#include <pthread.h> +#include <sys/types.h> +#include <fcntl.h> + +#include <sys/types.h> +#include <sys/stat.h> + +#include <errno.h> +#include <unistd.h> +#include <assert.h> + +#include "base.h" +#include "log.h" +#include "buffer.h" + +#include "plugin.h" + +#include "stat_cache.h" +#include "mod_glusterfs.h" +#include "etag.h" +#include "http_chunk.h" +#include "response.h" + +#include "fdevent.h" +#include <libglusterfsclient.h> + +#ifdef HAVE_ATTR_ATTRIBUTES_H +#include <attr/attributes.h> +#endif + +#ifdef HAVE_FAM_H +# include <fam.h> +#endif + +#include "sys-mmap.h" + +/* NetBSD 1.3.x needs it */ +#ifndef MAP_FAILED +# define MAP_FAILED -1 +#endif + +#ifndef O_LARGEFILE +# define O_LARGEFILE 0 +#endif + +#ifndef HAVE_LSTAT +#define lstat stat +#endif + +#if 0 +/* enables debug code for testing if all nodes in the stat-cache as accessable */ +#define DEBUG_STAT_CACHE +#endif + +#ifdef HAVE_LSTAT +#undef HAVE_LSTAT +#endif + +#define GLUSTERFS_FILE_CHUNK (FILE_CHUNK + 1) + +/* Keep this value large. Each glusterfs_async_read of GLUSTERFS_CHUNK_SIZE results in a network_backend_write of the read data*/ + +#define GLUSTERFS_CHUNK_SIZE 8192 + +/** + * this is a staticfile for a lighttpd plugin + * + */ + +typedef struct glusterfs_async_local { + int op_ret; + int op_errno; + char async_read_complete; + off_t length; + size_t read_bytes; + glusterfs_read_buf_t *buf; + pthread_mutex_t lock; + pthread_cond_t cond; +} glusterfs_async_local_t; + + +typedef struct { + unsigned long fd; + void *buf; + buffer *glusterfs_path; + /* off_t response_content_length; */ + int prefix; +}mod_glusterfs_ctx_t; + +/* plugin config for all request/connections */ +typedef struct { + buffer *logfile; + buffer *loglevel; + buffer *specfile; + buffer *prefix; + buffer *xattr_file_size; + buffer *document_root; + array *exclude_exts; + unsigned short cache_timeout; + + /* FIXME: its a pointer, hence cant be short */ + unsigned long handle; +} plugin_config; + +static int (*network_backend_write)(struct server *srv, connection *con, int fd, chunkqueue *cq); + +typedef struct { + PLUGIN_DATA; + buffer *range_buf; + plugin_config **config_storage; + + plugin_config conf; +} plugin_data; + +typedef struct { + chunkqueue *cq; + glusterfs_read_buf_t *buf; + size_t length; +}mod_glusterfs_chunkqueue; + +#ifdef HAVE_FAM_H +typedef struct { + FAMRequest *req; + FAMConnection *fc; + + buffer *name; + + int version; +} fam_dir_entry; +#endif + +/* the directory name is too long to always compare on it + * - we need a hash + * - the hash-key is used as sorting criteria for a tree + * - a splay-tree is used as we can use the caching effect of it + */ + +/* we want to cleanup the stat-cache every few seconds, let's say 10 + * + * - remove entries which are outdated since 30s + * - remove entries which are fresh but havn't been used since 60s + * - if we don't have a stat-cache entry for a directory, release it from the monitor + */ + +#ifdef DEBUG_STAT_CACHE +typedef struct { + int *ptr; + + size_t used; + size_t size; +} fake_keys; + +static fake_keys ctrl; +#endif + +int +mod_glusterfs_readv_async_cbk (glusterfs_read_buf_t *buf, + void *cbk_data) +{ + glusterfs_async_local_t *local = cbk_data; + pthread_mutex_lock (&local->lock); + { + local->async_read_complete = 1; + local->buf = buf; + + pthread_cond_signal (&local->cond); + } + pthread_mutex_unlock (&local->lock); + + return 0; +} + +static int +mod_glusterfs_read_async (server *srv, connection *con, chunk *glusterfs_chunk) +{ + glusterfs_async_local_t local; + off_t end = 0; + int nbytes; + int complete; + chunkqueue *cq = NULL; + chunk *c = NULL; + off_t offset = glusterfs_chunk->file.start; + size_t length = glusterfs_chunk->file.length; + long fd = (long)glusterfs_chunk->file.name; + + pthread_cond_init (&local.cond, NULL); + pthread_mutex_init (&local.lock, NULL); + + //local.fd = fd; + memset (&local, 0, sizeof (local)); + + if (length > 0) + end = offset + length; + + cq = chunkqueue_init (); + if (!cq) { + con->http_status = 500; + return HANDLER_FINISHED; + } + + do { + glusterfs_read_buf_t *buf; + int i; + if (length > 0) { + nbytes = end - offset; + if (nbytes > GLUSTERFS_CHUNK_SIZE) + nbytes = GLUSTERFS_CHUNK_SIZE; + } else + nbytes = GLUSTERFS_CHUNK_SIZE; + + glusterfs_read_async(fd, + nbytes, + offset, + mod_glusterfs_readv_async_cbk, + (void *)&local); + + pthread_mutex_lock (&local.lock); + { + while (!local.async_read_complete) { + pthread_cond_wait (&local.cond, &local.lock); + } + + local.op_ret = local.buf->op_ret; + local.op_errno = local.buf->op_errno; + + local.async_read_complete = 0; + buf = local.buf; + + if ((int)length < 0) + complete = (local.buf->op_ret <= 0); + else { + local.read_bytes += local.buf->op_ret; + complete = ((local.read_bytes == length) || (local.buf->op_ret <= 0)); + } + } + pthread_mutex_unlock (&local.lock); + + if (local.op_ret > 0) { + unsigned long check = 0; + for (i = 0; i < buf->count; i++) { + buffer *nw_write_buf = buffer_init (); + + check += buf->vector[i].iov_len; + + nw_write_buf->used = nw_write_buf->size = buf->vector[i].iov_len + 1; + nw_write_buf->ptr = buf->vector[i].iov_base; + + // buffer_copy_memory (nw_write_buf, buf->vector[i].iov_base, buf->vector[i].iov_len + 1); + offset += local.op_ret; + chunkqueue_append_buffer_weak (cq, nw_write_buf); + } + + network_backend_write (srv, con, con->fd, cq); + + if (chunkqueue_written (cq) != local.op_ret) { + mod_glusterfs_chunkqueue *gf_cq; + glusterfs_chunk->file.start = offset; + if ((int)glusterfs_chunk->file.length > 0) + glusterfs_chunk->file.length -= local.read_bytes; + + gf_cq = calloc (1, sizeof (*gf_cq)); + /* ERR_ABORT (gf_cq); */ + gf_cq->cq = cq; + gf_cq->buf = buf; + gf_cq->length = local.op_ret; + glusterfs_chunk->file.mmap.start = (char *)gf_cq; + return local.read_bytes; + } + + for (c = cq->first ; c; c = c->next) + c->mem->ptr = NULL; + + chunkqueue_reset (cq); + } + + glusterfs_free (buf); + } while (!complete); + + chunkqueue_free (cq); + glusterfs_close (fd); + + if (local.op_ret < 0) + con->http_status = 500; + + return (local.op_ret < 0 ? HANDLER_FINISHED : HANDLER_GO_ON); +} + +int mod_glusterfs_network_backend_write(struct server *srv, connection *con, int fd, chunkqueue *cq) +{ + chunk *c, *prev, *first; + int chunks_written = 0; + int error = 0; + + for (first = prev = c = cq->first; c; c = c->next, chunks_written++) { + + if (c->type == MEM_CHUNK && c->mem->used && !c->mem->ptr) { + if (cq->first != c) { + prev->next = NULL; + + /* call stored network_backend_write */ + network_backend_write (srv, con, fd, cq); + + prev->next = c; + } + cq->first = c->next; + + if (c->file.fd < 0) { + error = HANDLER_ERROR; + break; + } + + if (c->file.mmap.start) { + chunk *tmp; + mod_glusterfs_chunkqueue *gf_cq = (mod_glusterfs_chunkqueue *)c->file.mmap.start; + + network_backend_write (srv, con, fd, gf_cq->cq); + + if ((size_t)chunkqueue_written (gf_cq->cq) != gf_cq->length) { + cq->first = first; + return chunks_written; + } + for (tmp = gf_cq->cq->first ; tmp; tmp = tmp->next) + tmp->mem->ptr = NULL; + + chunkqueue_free (gf_cq->cq); + glusterfs_free (gf_cq->buf); + free (gf_cq); + c->file.mmap.start = NULL; + } + + mod_glusterfs_read_async (srv, con, c); //c->file.fd, c->file.start, -1);//c->file.length); + if (c->file.mmap.start) { + /* pending chunkqueue from mod_glusterfs_read_async to be written to network */ + cq->first = first; + return chunks_written; + } + + buffer_free (c->mem); + c->mem = NULL; + + c->type = FILE_CHUNK; + c->offset = c->file.length = 0; + c->file.name = NULL; + + if (first == c) + first = c->next; + + if (cq->last == c) + cq->last = NULL; + + prev->next = c->next; + + free(c); + } + prev = c; + } + + network_backend_write (srv, con, fd, cq); + + cq->first = first; + + return chunks_written; +} + +int chunkqueue_append_glusterfs_file (connection *con, long fd, off_t offset, off_t len) +{ + chunk *c = NULL; + c = chunkqueue_get_append_tempfile (con->write_queue); + + if (c->file.is_temp) { + close (c->file.fd); + unlink (c->file.name->ptr); + } + + c->type = MEM_CHUNK; + + c->mem = buffer_init (); + c->mem->used = len + 1; + c->mem->ptr = NULL; + c->offset = 0; + + /* buffer_copy_string_buffer (c->file.name, fn); */ + buffer_free (c->file.name); + + /* fd returned by libglusterfsclient is a pointer */ + c->file.name = (buffer *)fd; + c->file.start = offset; + c->file.length = len; + + //c->file.fd = fd; + c->file.mmap.start = NULL; + return 0; +} + +/* init the plugin data */ +INIT_FUNC(mod_glusterfs_init) { + plugin_data *p; + + p = calloc(1, sizeof(*p)); + /* ERR_ABORT (p); */ + network_backend_write = NULL; + + return p; +} + +/* detroy the plugin data */ +FREE_FUNC(mod_glusterfs_free) { + plugin_data *p = p_d; + + UNUSED (srv); + + if (!p) return HANDLER_GO_ON; + + if (p->config_storage) { + size_t i; + for (i = 0; i < srv->config_context->used; i++) { + plugin_config *s = p->config_storage[i]; + + buffer_free (s->logfile); + buffer_free (s->loglevel); + buffer_free (s->specfile); + buffer_free (s->prefix); + buffer_free (s->xattr_file_size); + buffer_free (s->document_root); + array_free (s->exclude_exts); + + free (s); + } + free (p->config_storage); + } + buffer_free (p->range_buf); + + free (p); + + return HANDLER_GO_ON; +} + +SETDEFAULTS_FUNC(mod_glusterfs_set_defaults) { + plugin_data *p = p_d; + size_t i = 0; + + config_values_t cv[] = { + { "glusterfs.logfile", NULL, T_CONFIG_STRING, T_CONFIG_SCOPE_CONNECTION }, + + { "glusterfs.loglevel", NULL, T_CONFIG_STRING, T_CONFIG_SCOPE_CONNECTION }, + + { "glusterfs.volume-specfile", NULL, T_CONFIG_STRING, T_CONFIG_SCOPE_CONNECTION }, + + { "glusterfs.cache-timeout", NULL, T_CONFIG_SHORT, T_CONFIG_SCOPE_CONNECTION }, + + { "glusterfs.exclude-extensions", NULL, T_CONFIG_ARRAY, T_CONFIG_SCOPE_CONNECTION }, + + /*TODO: get the prefix from config_conext and remove glusterfs.prefix from conf file */ + { "glusterfs.prefix", NULL, T_CONFIG_STRING, T_CONFIG_SCOPE_CONNECTION }, + + { "glusterfs.xattr-interface-size-limit", NULL, T_CONFIG_STRING, T_CONFIG_SCOPE_CONNECTION }, + + { "glusterfs.document-root", NULL, T_CONFIG_STRING, T_CONFIG_SCOPE_CONNECTION }, + + { NULL, NULL, T_CONFIG_UNSET, T_CONFIG_SCOPE_UNSET } + }; + + p->config_storage = calloc(1, srv->config_context->used * sizeof(specific_config *)); + /* ERR_ABORT (p->config_storage);*/ + p->range_buf = buffer_init (); + + for (i = 0; i < srv->config_context->used; i++) { + plugin_config *s; + + s = calloc(1, sizeof(plugin_config)); + /* ERR_ABORT (s); */ + s->logfile = buffer_init (); + s->loglevel = buffer_init (); + s->specfile = buffer_init (); + s->document_root = buffer_init (); + s->exclude_exts = array_init (); + s->prefix = buffer_init (); + s->xattr_file_size = buffer_init (); + + cv[0].destination = s->logfile; + cv[1].destination = s->loglevel; + cv[2].destination = s->specfile; + cv[3].destination = &s->cache_timeout; + cv[4].destination = s->exclude_exts; + cv[5].destination = s->prefix; + cv[6].destination = s->xattr_file_size; + cv[7].destination = s->document_root; + p->config_storage[i] = s; + + if (0 != config_insert_values_global(srv, ((data_config *)srv->config_context->data[i])->value, cv)) { + return HANDLER_FINISHED; + } + } + + return HANDLER_GO_ON; +} + +#define PATCH(x) \ + p->conf.x = s->x; + +static int mod_glusterfs_patch_connection(server *srv, connection *con, plugin_data *p) { + size_t i, j; + plugin_config *s; + + /* skip the first, the global context */ + /* glusterfs related config can only occur inside $HTTP["url"] == "<glusterfs-prefix>" */ + p->conf.logfile = NULL; + p->conf.loglevel = NULL; + p->conf.specfile = NULL; + p->conf.cache_timeout = 0; + p->conf.exclude_exts = NULL; + p->conf.prefix = NULL; + p->conf.xattr_file_size = NULL; + p->conf.document_root = NULL; + + for (i = 1; i < srv->config_context->used; i++) { + data_config *dc = (data_config *)srv->config_context->data[i]; + s = p->config_storage[i]; + + /* condition didn't match */ + if (!config_check_cond(srv, con, dc)) continue; + + /* merge config */ + for (j = 0; j < dc->value->used; j++) { + data_unset *du = dc->value->data[j]; + + if (buffer_is_equal_string (du->key, CONST_STR_LEN("glusterfs.logfile"))) { + PATCH (logfile); + } else if (buffer_is_equal_string (du->key, CONST_STR_LEN("glusterfs.loglevel"))) { + PATCH (loglevel); + } else if (buffer_is_equal_string (du->key, CONST_STR_LEN ("glusterfs.volume-specfile"))) { + PATCH (specfile); + } else if (buffer_is_equal_string (du->key, CONST_STR_LEN("glusterfs.cache-timeout"))) { + PATCH (cache_timeout); + } else if (buffer_is_equal_string (du->key, CONST_STR_LEN ("glusterfs.exclude-extensions"))) { + PATCH (exclude_exts); + } else if (buffer_is_equal_string (du->key, CONST_STR_LEN ("glusterfs.prefix"))) { + PATCH (prefix); + } else if (buffer_is_equal_string (du->key, CONST_STR_LEN ("glusterfs.xattr-interface-size-limit"))) { + PATCH (xattr_file_size); + } else if (buffer_is_equal_string (du->key, CONST_STR_LEN ("glusterfs.document-root"))) { + PATCH (document_root); + } + } + } + return 0; +} + +#undef PATCH + +static int http_response_parse_range(server *srv, connection *con, plugin_data *p) { + int multipart = 0; + int error; + off_t start, end; + const char *s, *minus; + char *boundary = "fkj49sn38dcn3"; + data_string *ds; + stat_cache_entry *sce = NULL; + buffer *content_type = NULL; + size_t size = 0; + mod_glusterfs_ctx_t *ctx = con->plugin_ctx[p->id]; + + if (p->conf.xattr_file_size && p->conf.xattr_file_size->ptr) { + size = atoi (p->conf.xattr_file_size->ptr); + } + + if (HANDLER_ERROR == stat_cache_get_entry(srv, con, con->physical.path, &sce)) { + SEGFAULT(); + } + + start = 0; + end = sce->st.st_size - 1; + + con->response.content_length = 0; + + if (NULL != (ds = (data_string *)array_get_element(con->response.headers, "Content-Type"))) { + content_type = ds->value; + } + + for (s = con->request.http_range, error = 0; + !error && *s && NULL != (minus = strchr(s, '-')); ) { + char *err; + off_t la, le; + + if (s == minus) { + /* -<stop> */ + + le = strtoll(s, &err, 10); + + if (le == 0) { + /* RFC 2616 - 14.35.1 */ + + con->http_status = 416; + error = 1; + } else if (*err == '\0') { + /* end */ + s = err; + + end = sce->st.st_size - 1; + start = sce->st.st_size + le; + } else if (*err == ',') { + multipart = 1; + s = err + 1; + + end = sce->st.st_size - 1; + start = sce->st.st_size + le; + } else { + error = 1; + } + + } else if (*(minus+1) == '\0' || *(minus+1) == ',') { + /* <start>- */ + + la = strtoll(s, &err, 10); + + if (err == minus) { + /* ok */ + + if (*(err + 1) == '\0') { + s = err + 1; + + end = sce->st.st_size - 1; + start = la; + + } else if (*(err + 1) == ',') { + multipart = 1; + s = err + 2; + + end = sce->st.st_size - 1; + start = la; + } else { + error = 1; + } + } else { + /* error */ + error = 1; + } + } else { + /* <start>-<stop> */ + + la = strtoll(s, &err, 10); + + if (err == minus) { + le = strtoll(minus+1, &err, 10); + + /* RFC 2616 - 14.35.1 */ + if (la > le) { + error = 1; + } + + if (*err == '\0') { + /* ok, end*/ + s = err; + + end = le; + start = la; + } else if (*err == ',') { + multipart = 1; + s = err + 1; + + end = le; + start = la; + } else { + /* error */ + + error = 1; + } + } else { + /* error */ + + error = 1; + } + } + + if (!error) { + if (start < 0) start = 0; + + /* RFC 2616 - 14.35.1 */ + if (end > sce->st.st_size - 1) end = sce->st.st_size - 1; + + if (start > sce->st.st_size - 1) { + error = 1; + + con->http_status = 416; + } + } + + if (!error) { + if (multipart) { + /* write boundary-header */ + buffer *b; + + b = chunkqueue_get_append_buffer(con->write_queue); + + buffer_copy_string(b, "\r\n--"); + buffer_append_string(b, boundary); + + /* write Content-Range */ + buffer_append_string(b, "\r\nContent-Range: bytes "); + buffer_append_off_t(b, start); + buffer_append_string(b, "-"); + buffer_append_off_t(b, end); + buffer_append_string(b, "/"); + buffer_append_off_t(b, sce->st.st_size); + + buffer_append_string(b, "\r\nContent-Type: "); + buffer_append_string_buffer(b, content_type); + + /* write END-OF-HEADER */ + buffer_append_string(b, "\r\n\r\n"); + + con->response.content_length += b->used - 1; + + } + + /* path = con->physical.path->ptr + p->conf.prefix->used - 1 + con->physical.doc_root->used - 1; */ + /* + fd = glusterfs_open (p->conf.handle, path, O_RDONLY); + if (fd < 0) + return HANDLER_ERROR; + */ + /* fn = buffer_init_string (path); */ + if ((size_t)sce->st.st_size > size) { + chunkqueue_append_glusterfs_file(con, ctx->fd, start, end - start + 1); + } else { + if (!start) { + buffer *mem = buffer_init (); + mem->ptr = ctx->buf; + mem->used = mem->size = sce->st.st_size + 1; + http_chunk_append_buffer (srv, con, mem); + ctx->buf = NULL; + } else { + chunkqueue_append_mem (con->write_queue, ((char *)ctx->buf) + start, end - start + 1); + } + } + + con->response.content_length += end - start + 1; + } + } + + if (ctx->buf) { + free (ctx->buf); + ctx->buf = NULL; + } + + /* something went wrong */ + if (error) return -1; + + if (multipart) { + /* add boundary end */ + buffer *b; + + b = chunkqueue_get_append_buffer(con->write_queue); + + buffer_copy_string_len(b, "\r\n--", 4); + buffer_append_string(b, boundary); + buffer_append_string_len(b, "--\r\n", 4); + + con->response.content_length += b->used - 1; + + /* set header-fields */ + + buffer_copy_string(p->range_buf, "multipart/byteranges; boundary="); + buffer_append_string(p->range_buf, boundary); + + /* overwrite content-type */ + response_header_overwrite(srv, con, CONST_STR_LEN("Content-Type"), CONST_BUF_LEN(p->range_buf)); + } else { + /* add Content-Range-header */ + + buffer_copy_string(p->range_buf, "bytes "); + buffer_append_off_t(p->range_buf, start); + buffer_append_string(p->range_buf, "-"); + buffer_append_off_t(p->range_buf, end); + buffer_append_string(p->range_buf, "/"); + buffer_append_off_t(p->range_buf, sce->st.st_size); + + response_header_insert(srv, con, CONST_STR_LEN("Content-Range"), CONST_BUF_LEN(p->range_buf)); + } + + /* ok, the file is set-up */ + return 0; +} + +PHYSICALPATH_FUNC(mod_glusterfs_handle_physical) { + plugin_data *p = p_d; + stat_cache_entry *sce; + mod_glusterfs_ctx_t *plugin_ctx = NULL; + size_t size = 0; + + if (con->http_status != 0) return HANDLER_GO_ON; + if (con->uri.path->used == 0) return HANDLER_GO_ON; + if (con->physical.path->used == 0) return HANDLER_GO_ON; + + if (con->mode != DIRECT) return HANDLER_GO_ON; + + /* + network_backend_write = srv->network_backend_write; + srv->network_backend_write = mod_glusterfs_network_backend_write; + */ + + switch (con->request.http_method) { + case HTTP_METHOD_GET: + case HTTP_METHOD_POST: + case HTTP_METHOD_HEAD: + break; + + default: + return HANDLER_GO_ON; + } + + mod_glusterfs_patch_connection(srv, con, p); + if (!p->conf.prefix || p->conf.prefix->used == 0) { + return HANDLER_GO_ON; + } + + if (!p->conf.document_root || p->conf.document_root->used == 0) { + log_error_write(srv, __FILE__, __LINE__, "s", "glusterfs.document-root is not specified"); + con->http_status = 500; + return HANDLER_FINISHED; + } + + if (p->conf.handle <= 0) { + glusterfs_init_ctx_t ctx; + + if (!p->conf.specfile || p->conf.specfile->used == 0) { + return HANDLER_GO_ON; + } + memset (&ctx, 0, sizeof (ctx)); + + ctx.specfile = p->conf.specfile->ptr; + ctx.logfile = p->conf.logfile->ptr; + ctx.loglevel = p->conf.loglevel->ptr; + ctx.lookup_timeout = ctx.stat_timeout = p->conf.cache_timeout; + + p->conf.handle = (long)glusterfs_init (&ctx); + + if (p->conf.handle <= 0) { + con->http_status = 500; + log_error_write(srv, __FILE__, __LINE__, "sbs", "glusterfs initialization failed, please check your configuration. Glusterfs logfile ", p->conf.logfile, "might contain details"); + return HANDLER_FINISHED; + } + } + + size = 0; + if (p->conf.xattr_file_size && p->conf.xattr_file_size->ptr) + size = atoi (p->conf.xattr_file_size->ptr); + + if (!con->plugin_ctx[p->id]) { +/* FIXME: what if multiple files are requested from a single connection? */ +/* TODO: check whether this works fine for HTTP protocol 1.1 */ + + buffer *tmp_buf = buffer_init_buffer (con->physical.basedir); + + plugin_ctx = calloc (1, sizeof (*plugin_ctx)); + /* ERR_ABORT (plugin_ctx); */ + con->plugin_ctx[p->id] = plugin_ctx; + + buffer_append_string_buffer (tmp_buf, p->conf.prefix); + buffer_path_simplify (tmp_buf, tmp_buf); + + plugin_ctx->prefix = tmp_buf->used - 1; + if (tmp_buf->ptr[plugin_ctx->prefix - 1] == '/') + plugin_ctx->prefix--; + + buffer_free (tmp_buf); + } else + /*FIXME: error!! error!! */ + plugin_ctx = con->plugin_ctx[p->id]; + + + if (size) + { + plugin_ctx->buf = MALLOC (size); + /* ERR_ABORT (plugin_ctx->buf); */ + } + + plugin_ctx->glusterfs_path = buffer_init (); + buffer_copy_string_buffer (plugin_ctx->glusterfs_path, p->conf.document_root); + buffer_append_string (plugin_ctx->glusterfs_path, "/"); + buffer_append_string (plugin_ctx->glusterfs_path, con->physical.path->ptr + plugin_ctx->prefix); + buffer_path_simplify (plugin_ctx->glusterfs_path, plugin_ctx->glusterfs_path); + + if (glusterfs_stat_cache_get_entry (srv, con, (libglusterfs_handle_t )p->conf.handle, plugin_ctx->glusterfs_path, con->physical.path, plugin_ctx->buf, size, &sce) == HANDLER_ERROR) { + if (errno == ENOENT) + con->http_status = 404; + else + con->http_status = 403; + + free (plugin_ctx->buf); + buffer_free (plugin_ctx->glusterfs_path); + plugin_ctx->glusterfs_path = NULL; + plugin_ctx->buf = NULL; + + free (plugin_ctx); + con->plugin_ctx[p->id] = NULL; + + return HANDLER_FINISHED; + } + + if (!(S_ISREG (sce->st.st_mode) && (size_t)sce->st.st_size <= size)) { + free (plugin_ctx->buf); + plugin_ctx->buf = NULL; + } + + return HANDLER_GO_ON; +} + +static int http_chunk_append_len(server *srv, connection *con, size_t len) { + size_t i, olen = len, j; + buffer *b; + + b = srv->tmp_chunk_len; + + if (len == 0) { + buffer_copy_string(b, "0"); + } else { + for (i = 0; i < 8 && len; i++) { + len >>= 4; + } + + /* i is the number of hex digits we have */ + buffer_prepare_copy(b, i + 1); + + for (j = i-1, len = olen; j+1 > 0; j--) { + b->ptr[j] = (len & 0xf) + (((len & 0xf) <= 9) ? '0' : 'a' - 10); + len >>= 4; + } + b->used = i; + b->ptr[b->used++] = '\0'; + } + + buffer_append_string(b, "\r\n"); + chunkqueue_append_buffer(con->write_queue, b); + + return 0; +} + +int http_chunk_append_glusterfs_file_chunk(server *srv, connection *con, long fd, off_t offset, off_t len) { + chunkqueue *cq; + + if (!con) return -1; + + cq = con->write_queue; + + if (con->response.transfer_encoding & HTTP_TRANSFER_ENCODING_CHUNKED) { + http_chunk_append_len(srv, con, len); + } + + chunkqueue_append_glusterfs_file (con, fd, offset, len); + + if (con->response.transfer_encoding & HTTP_TRANSFER_ENCODING_CHUNKED && len > 0) { + chunkqueue_append_mem(cq, "\r\n", 2 + 1); + } + + return 0; +} + +int http_chunk_append_glusterfs_mem(server *srv, connection *con, const char * mem, size_t len) { + chunkqueue *cq = NULL; + buffer *buf = NULL; + + if (!con) return -1; + + cq = con->write_queue; + + if (len == 0) { + if (con->response.transfer_encoding & HTTP_TRANSFER_ENCODING_CHUNKED) { + chunkqueue_append_mem(cq, "0\r\n\r\n", 5 + 1); + } else { + chunkqueue_append_mem(cq, "", 1); + } + return 0; + } + + if (con->response.transfer_encoding & HTTP_TRANSFER_ENCODING_CHUNKED) { + http_chunk_append_len(srv, con, len - 1); + } + + buf = buffer_init (); + + buf->used = len + 1; + buf->ptr = (char *)mem; + chunkqueue_append_buffer_weak (cq, buf); + + if (con->response.transfer_encoding & HTTP_TRANSFER_ENCODING_CHUNKED) { + chunkqueue_append_mem(cq, "\r\n", 2 + 1); + } + + return 0; +} + + + +URIHANDLER_FUNC(mod_glusterfs_subrequest) { + plugin_data *p = p_d; + stat_cache_entry *sce = NULL; + int s_len; + char allow_caching = 1; + size_t size = 0; + mod_glusterfs_ctx_t *ctx = con->plugin_ctx[p->id]; + + /* someone else has done a decision for us */ + if (con->http_status != 0) return HANDLER_GO_ON; + if (con->uri.path->used == 0) return HANDLER_GO_ON; + if (con->physical.path->used == 0) return HANDLER_GO_ON; + + /* someone else has handled this request */ + if (con->mode != DIRECT) return HANDLER_GO_ON; + + /* we only handle GET, POST and HEAD */ + switch(con->request.http_method) { + case HTTP_METHOD_GET: + case HTTP_METHOD_POST: + case HTTP_METHOD_HEAD: + break; + default: + return HANDLER_GO_ON; + } + + mod_glusterfs_patch_connection(srv, con, p); + + if (!p->conf.prefix || !p->conf.prefix->used) + return HANDLER_GO_ON; + + s_len = con->uri.path->used - 1; + /* ignore certain extensions */ + /* + for (k = 0; k < p->conf.exclude_exts->used; k++) { + data_string *ds; + ds = (data_string *)p->conf.exclude_exts->data[k]; + + if (ds->value->used == 0) continue; + + if (!strncmp (ds->value->ptr, con->uri.path->ptr, strlen (ds->value->ptr))) + break; + } + + if (k == p->conf.exclude_exts->used) { + return HANDLER_GO_ON; + } + */ + + if (con->conf.log_request_handling) { + log_error_write(srv, __FILE__, __LINE__, "s", "-- serving file from glusterfs"); + } + + if (HANDLER_ERROR == stat_cache_get_entry(srv, con, con->physical.path, &sce)) { + con->http_status = 403; + + log_error_write(srv, __FILE__, __LINE__, "sbsb", + "not a regular file:", con->uri.path, + "->", con->physical.path); + + free (ctx); + con->plugin_ctx[p->id] = NULL; + + return HANDLER_FINISHED; + } + + if (con->uri.path->ptr[s_len] == '/' || !S_ISREG(sce->st.st_mode)) { + free (ctx); + con->plugin_ctx[p->id] = NULL; + return HANDLER_FINISHED; + } + + if (p->conf.xattr_file_size && p->conf.xattr_file_size->ptr) + size = atoi (p->conf.xattr_file_size->ptr); + + if ((size_t)sce->st.st_size > size) { + ctx->fd = glusterfs_open ((libglusterfs_handle_t ) ((unsigned long)p->conf.handle), ctx->glusterfs_path->ptr, O_RDONLY, 0); + + if (((long)ctx->fd) == 0) { + con->http_status = 403; + free (ctx); + con->plugin_ctx[p->id] = NULL; + return HANDLER_FINISHED; + } + } + + buffer_free (ctx->glusterfs_path); + ctx->glusterfs_path = NULL; + + /* we only handline regular files */ +#ifdef HAVE_LSTAT + if ((sce->is_symlink == 1) && !con->conf.follow_symlink) { + con->http_status = 403; + + if (con->conf.log_request_handling) { + log_error_write(srv, __FILE__, __LINE__, "s", "-- access denied due symlink restriction"); + log_error_write(srv, __FILE__, __LINE__, "sb", "Path :", con->physical.path); + } + + buffer_reset(con->physical.path); + free (ctx); + con->plugin_ctx[p->id] = NULL; + return HANDLER_FINISHED; + } +#endif + if (!S_ISREG(sce->st.st_mode)) { + con->http_status = 404; + + if (con->conf.log_file_not_found) { + log_error_write(srv, __FILE__, __LINE__, "sbsb", + "not a regular file:", con->uri.path, + "->", sce->name); + } + + free (ctx); + con->plugin_ctx[p->id] = NULL; + + return HANDLER_FINISHED; + } + + /* mod_compress might set several data directly, don't overwrite them */ + + /* set response content-type, if not set already */ + + if (NULL == array_get_element(con->response.headers, "Content-Type")) { + if (buffer_is_empty(sce->content_type)) { + /* we are setting application/octet-stream, but also announce that + * this header field might change in the seconds few requests + * + * This should fix the aggressive caching of FF and the script download + * seen by the first installations + */ + response_header_overwrite(srv, con, CONST_STR_LEN("Content-Type"), CONST_STR_LEN("application/octet-stream")); + + allow_caching = 0; + } else { + response_header_overwrite(srv, con, CONST_STR_LEN("Content-Type"), CONST_BUF_LEN(sce->content_type)); + } + } + + if (con->conf.range_requests) { + response_header_overwrite(srv, con, CONST_STR_LEN("Accept-Ranges"), CONST_STR_LEN("bytes")); + } + + /* TODO: Allow Cachable requests */ +#if 0 + if (allow_caching) { + if (p->conf.etags_used && con->etag_flags != 0 && !buffer_is_empty(sce->etag)) { + if (NULL == array_get_element(con->response.headers, "ETag")) { + /* generate e-tag */ + etag_mutate(con->physical.etag, sce->etag); + + response_header_overwrite(srv, con, CONST_STR_LEN("ETag"), CONST_BUF_LEN(con->physical.etag)); + } + } + + /* prepare header */ + if (NULL == (ds = (data_string *)array_get_element(con->response.headers, "Last-Modified"))) { + mtime = strftime_cache_get(srv, sce->st.st_mtime); + response_header_overwrite(srv, con, CONST_STR_LEN("Last-Modified"), CONST_BUF_LEN(mtime)); + } else { + mtime = ds->value; + } + + if (HANDLER_FINISHED == http_response_handle_cachable(srv, con, mtime)) { + free (ctx); + con->plugin_ctx[p->id] = NULL; + return HANDLER_FINISHED; + } + } +#endif + + /*TODO: Read about etags */ + if (con->request.http_range && con->conf.range_requests) { + int do_range_request = 1; + data_string *ds = NULL; + buffer *mtime = NULL; + /* check if we have a conditional GET */ + + /* prepare header */ + if (NULL == (ds = (data_string *)array_get_element(con->response.headers, "Last-Modified"))) { + mtime = strftime_cache_get(srv, sce->st.st_mtime); + response_header_overwrite(srv, con, CONST_STR_LEN("Last-Modified"), CONST_BUF_LEN(mtime)); + } else { + mtime = ds->value; + } + + if (NULL != (ds = (data_string *)array_get_element(con->request.headers, "If-Range"))) { + /* if the value is the same as our ETag, we do a Range-request, + * otherwise a full 200 */ + + if (ds->value->ptr[0] == '"') { + /** + * client wants a ETag + */ + if (!con->physical.etag) { + do_range_request = 0; + } else if (!buffer_is_equal(ds->value, con->physical.etag)) { + do_range_request = 0; + } + } else if (!mtime) { + /** + * we don't have a Last-Modified and can match the If-Range: + * + * sending all + */ + do_range_request = 0; + } else if (!buffer_is_equal(ds->value, mtime)) { + do_range_request = 0; + } + } + + if (do_range_request) { + /* content prepared, I'm done */ + con->file_finished = 1; + + if (0 == http_response_parse_range(srv, con, p)) { + con->http_status = 206; + } + + free (ctx); + con->plugin_ctx[p->id] = NULL; + return HANDLER_FINISHED; + } + } + + /* if we are still here, prepare body */ + + /* we add it here for all requests + * the HEAD request will drop it afterwards again + */ + /*TODO check whether 1 should be subtracted */ + + if (p->conf.xattr_file_size && p->conf.xattr_file_size->ptr) + size = atoi (p->conf.xattr_file_size->ptr); + + if (size < (size_t)sce->st.st_size) { + http_chunk_append_glusterfs_file_chunk (srv, con, ctx->fd, 0, sce->st.st_size); + } else { + http_chunk_append_glusterfs_mem (srv, con, ctx->buf, sce->st.st_size); + } + + con->http_status = 200; + con->file_finished = 1; + + free (ctx); + con->plugin_ctx[p->id] = NULL; + + return HANDLER_FINISHED; +} + +#if 0 +URIHANDLER_FUNC(mod_glusterfs_request_done) +{ + mod_glusterfs_read_buf_t *cur = first, *prev; + while (cur) { + prev = cur; + glusterfs_free (cur->buf); + cur = cur->next; + free (prev); + } + first = NULL + } +#endif + +/* this function is called at dlopen() time and inits the callbacks */ +CONNECTION_FUNC(mod_glusterfs_connection_reset) +{ + (void) p_d; + (void) con; + if (!network_backend_write) + network_backend_write = srv->network_backend_write; + + srv->network_backend_write = mod_glusterfs_network_backend_write; + + return HANDLER_GO_ON; +} + +int mod_glusterfs_plugin_init(plugin *p) { + p->version = LIGHTTPD_VERSION_ID; + p->name = buffer_init_string("glusterfs"); + p->init = mod_glusterfs_init; + p->handle_physical = mod_glusterfs_handle_physical; + p->handle_subrequest_start = mod_glusterfs_subrequest; + // p->handle_request_done = mod_glusterfs_request_done; + p->set_defaults = mod_glusterfs_set_defaults; + p->connection_reset = mod_glusterfs_connection_reset; + p->cleanup = mod_glusterfs_free; + + p->data = NULL; + + return 0; +} + + +/* mod_glusterfs_stat_cache */ +static stat_cache_entry * stat_cache_entry_init(void) { + stat_cache_entry *sce = NULL; + + sce = calloc(1, sizeof(*sce)); + /* ERR_ABORT (sce); */ + + sce->name = buffer_init(); + sce->etag = buffer_init(); + sce->content_type = buffer_init(); + + return sce; +} + +#ifdef HAVE_FAM_H +static fam_dir_entry * fam_dir_entry_init(void) { + fam_dir_entry *fam_dir = NULL; + + fam_dir = calloc(1, sizeof(*fam_dir)); + /* ERR_ABORT (fam_dir); */ + + fam_dir->name = buffer_init(); + + return fam_dir; +} + +static void fam_dir_entry_free(void *data) { + fam_dir_entry *fam_dir = data; + + if (!fam_dir) return; + + FAMCancelMonitor(fam_dir->fc, fam_dir->req); + + buffer_free(fam_dir->name); + free(fam_dir->req); + + free(fam_dir); +} +#endif + +#ifdef HAVE_XATTR +static int stat_cache_attr_get(buffer *buf, char *name) { + int attrlen; + int ret; + + attrlen = 1024; + buffer_prepare_copy(buf, attrlen); + attrlen--; + if(0 == (ret = attr_get(name, "Content-Type", buf->ptr, &attrlen, 0))) { + buf->used = attrlen + 1; + buf->ptr[attrlen] = '\0'; + } + return ret; +} +#endif + +/* the famous DJB hash function for strings */ +static uint32_t hashme(buffer *str) { + uint32_t hash = 5381; + const char *s; + for (s = str->ptr; *s; s++) { + hash = ((hash << 5) + hash) + *s; + } + + hash &= ~(1 << 31); /* strip the highest bit */ + + return hash; +} + + +#ifdef HAVE_LSTAT +static int stat_cache_lstat(server *srv, buffer *dname, struct stat *lst) { + if (lstat(dname->ptr, lst) == 0) { + return S_ISLNK(lst->st_mode) ? 0 : 1; + } + else { + log_error_write(srv, __FILE__, __LINE__, "sbs", + "lstat failed for:", + dname, strerror(errno)); + }; + return -1; +} +#endif + +/*** + * + * + * + * returns: + * - HANDLER_FINISHED on cache-miss (don't forget to reopen the file) + * - HANDLER_ERROR on stat() failed -> see errno for problem + */ + +handler_t glusterfs_stat_cache_get_entry(server *srv, + connection *con, + libglusterfs_handle_t handle, + buffer *glusterfs_path, + buffer *name, + void *buf, + size_t size, + stat_cache_entry **ret_sce) +{ +#ifdef HAVE_FAM_H + fam_dir_entry *fam_dir = NULL; + int dir_ndx = -1; + splay_tree *dir_node = NULL; +#endif + stat_cache_entry *sce = NULL; + stat_cache *sc; + struct stat st; + size_t k; +#ifdef DEBUG_STAT_CACHE + size_t i; +#endif + int file_ndx; + splay_tree *file_node = NULL; + + *ret_sce = NULL; + memset (&st, 0, sizeof (st)); + + /* + * check if the directory for this file has changed + */ + + sc = srv->stat_cache; + + buffer_copy_string_buffer(sc->hash_key, name); + buffer_append_long(sc->hash_key, con->conf.follow_symlink); + + file_ndx = hashme(sc->hash_key); + sc->files = splaytree_splay(sc->files, file_ndx); + +#ifdef DEBUG_STAT_CACHE + for (i = 0; i < ctrl.used; i++) { + if (ctrl.ptr[i] == file_ndx) break; + } +#endif + + if (sc->files && (sc->files->key == file_ndx)) { +#ifdef DEBUG_STAT_CACHE + /* it was in the cache */ + assert(i < ctrl.used); +#endif + + /* we have seen this file already and + * don't stat() it again in the same second */ + + file_node = sc->files; + + sce = file_node->data; + + /* check if the name is the same, we might have a collision */ + + if (buffer_is_equal(name, sce->name)) { + if (srv->srvconf.stat_cache_engine == STAT_CACHE_ENGINE_SIMPLE) { + if (sce->stat_ts == srv->cur_ts && !buf) { + *ret_sce = sce; + return HANDLER_GO_ON; + } + } + } else { + /* oops, a collision, + * + * file_node is used by the FAM check below to see if we know this file + * and if we can save a stat(). + * + * BUT, the sce is not reset here as the entry into the cache is ok, we + * it is just not pointing to our requested file. + * + * */ + + file_node = NULL; + } + } else { +#ifdef DEBUG_STAT_CACHE + if (i != ctrl.used) { + fprintf(stderr, "%s.%d: %08x was already inserted but not found in cache, %s\n", __FILE__, __LINE__, file_ndx, name->ptr); + } + assert(i == ctrl.used); +#endif + } + /* + * *lol* + * - open() + fstat() on a named-pipe results in a (intended) hang. + * - stat() if regular file + open() to see if we can read from it is better + * + * */ + if (-1 == glusterfs_lookup(handle, glusterfs_path->ptr, buf, size, &st)) { + return HANDLER_ERROR; + } + + if (NULL == sce) { + int osize = 0; + + if (sc->files) { + osize = sc->files->size; + } + + sce = stat_cache_entry_init(); + buffer_copy_string_buffer(sce->name, name); + + sc->files = splaytree_insert(sc->files, file_ndx, sce); +#ifdef DEBUG_STAT_CACHE + if (ctrl.size == 0) { + ctrl.size = 16; + ctrl.used = 0; + ctrl.ptr = MALLOC(ctrl.size * sizeof(*ctrl.ptr)); + /* ERR_ABORT (ctrl.ptr); */ + } else if (ctrl.size == ctrl.used) { + ctrl.size += 16; + ctrl.ptr = realloc(ctrl.ptr, ctrl.size * sizeof(*ctrl.ptr)); + /* ERR_ABORT (ctrl.ptr); */ + } + + ctrl.ptr[ctrl.used++] = file_ndx; + + assert(sc->files); + assert(sc->files->data == sce); + assert(osize + 1 == splaytree_size(sc->files)); +#endif + } + + sce->st = st; + sce->stat_ts = srv->cur_ts; + + /* catch the obvious symlinks + * + * this is not a secure check as we still have a race-condition between + * the stat() and the open. We can only solve this by + * 1. open() the file + * 2. fstat() the fd + * + * and keeping the file open for the rest of the time. But this can + * only be done at network level. + * + * per default it is not a symlink + * */ +#ifdef HAVE_LSTAT + sce->is_symlink = 0; + + /* we want to only check for symlinks if we should block symlinks. + */ + if (!con->conf.follow_symlink) { + if (stat_cache_lstat(srv, name, &lst) == 0) { +#ifdef DEBUG_STAT_CACHE + log_error_write(srv, __FILE__, __LINE__, "sb", + "found symlink", name); +#endif + sce->is_symlink = 1; + } + + /* + * we assume "/" can not be symlink, so + * skip the symlink stuff if our path is / + **/ + else if ((name->used > 2)) { + buffer *dname; + char *s_cur; + + dname = buffer_init(); + buffer_copy_string_buffer(dname, name); + + while ((s_cur = strrchr(dname->ptr,'/'))) { + *s_cur = '\0'; + dname->used = s_cur - dname->ptr + 1; + if (dname->ptr == s_cur) { +#ifdef DEBUG_STAT_CACHE + log_error_write(srv, __FILE__, __LINE__, "s", "reached /"); +#endif + break; + } +#ifdef DEBUG_STAT_CACHE + log_error_write(srv, __FILE__, __LINE__, "sbs", + "checking if", dname, "is a symlink"); +#endif + if (stat_cache_lstat(srv, dname, &lst) == 0) { + sce->is_symlink = 1; +#ifdef DEBUG_STAT_CACHE + log_error_write(srv, __FILE__, __LINE__, "sb", + "found symlink", dname); +#endif + break; + }; + }; + buffer_free(dname); + }; + }; +#endif + + if (S_ISREG(st.st_mode)) { + /* determine mimetype */ + buffer_reset(sce->content_type); + + for (k = 0; k < con->conf.mimetypes->used; k++) { + data_string *ds = (data_string *)con->conf.mimetypes->data[k]; + buffer *type = ds->key; + + if (type->used == 0) continue; + + /* check if the right side is the same */ + if (type->used > name->used) continue; + + if (0 == strncasecmp(name->ptr + name->used - type->used, type->ptr, type->used - 1)) { + buffer_copy_string_buffer(sce->content_type, ds->value); + break; + } + } + etag_create(sce->etag, &(sce->st), con->etag_flags); +#ifdef HAVE_XATTR + if (con->conf.use_xattr && buffer_is_empty(sce->content_type)) { + stat_cache_attr_get(sce->content_type, name->ptr); + } +#endif + } else if (S_ISDIR(st.st_mode)) { + etag_create(sce->etag, &(sce->st), con->etag_flags); + } + +#ifdef HAVE_FAM_H + if (sc->fam && + (srv->srvconf.stat_cache_engine == STAT_CACHE_ENGINE_FAM)) { + /* is this directory already registered ? */ + if (!dir_node) { + fam_dir = fam_dir_entry_init(); + fam_dir->fc = sc->fam; + + buffer_copy_string_buffer(fam_dir->name, sc->dir_name); + + fam_dir->version = 1; + + fam_dir->req = calloc(1, sizeof(FAMRequest)); + /* ERR_ABORT (fam_dir->req); */ + + if (0 != FAMMonitorDirectory(sc->fam, fam_dir->name->ptr, + fam_dir->req, fam_dir)) { + + log_error_write(srv, __FILE__, __LINE__, "sbsbs", + "monitoring dir failed:", + fam_dir->name, + "file:", name, + FamErrlist[FAMErrno]); + + fam_dir_entry_free(fam_dir); + } else { + int osize = 0; + + if (sc->dirs) { + osize = sc->dirs->size; + } + + sc->dirs = splaytree_insert(sc->dirs, dir_ndx, fam_dir); + assert(sc->dirs); + assert(sc->dirs->data == fam_dir); + assert(osize == (sc->dirs->size - 1)); + } + } else { + fam_dir = dir_node->data; + } + + /* bind the fam_fc to the stat() cache entry */ + + if (fam_dir) { + sce->dir_version = fam_dir->version; + sce->dir_ndx = dir_ndx; + } + } +#endif + + *ret_sce = sce; + + return HANDLER_GO_ON; +} + +/** + * remove stat() from cache which havn't been stat()ed for + * more than 10 seconds + * + * + * walk though the stat-cache, collect the ids which are too old + * and remove them in a second loop + */ + +static int stat_cache_tag_old_entries(server *srv, splay_tree *t, int *keys, size_t *ndx) { + stat_cache_entry *sce; + + if (!t) return 0; + + stat_cache_tag_old_entries(srv, t->left, keys, ndx); + stat_cache_tag_old_entries(srv, t->right, keys, ndx); + + sce = t->data; + + if (srv->cur_ts - sce->stat_ts > 2) { + keys[(*ndx)++] = t->key; + } + + return 0; +} diff --git a/mod_glusterfs/lighttpd/1.4/mod_glusterfs.h b/mod_glusterfs/lighttpd/1.4/mod_glusterfs.h new file mode 100644 index 00000000000..5dc65cbcf4e --- /dev/null +++ b/mod_glusterfs/lighttpd/1.4/mod_glusterfs.h @@ -0,0 +1,29 @@ +/* + Copyright (c) 2008 Z RESEARCH, Inc. <http://www.zresearch.com> + This file is part of GlusterFS. + + GlusterFS is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published + by the Free Software Foundation; either version 3 of the License, + or (at your option) any later version. + + GlusterFS is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see + <http://www.gnu.org/licenses/>. +*/ + +#ifndef _MOD_GLUSTERFS_FILE_CACHE_H_ +#define _MOD_GLUSTERFS_FILE_CACHE_H_ + +#include "stat_cache.h" +#include <libglusterfsclient.h> +#include "base.h" + +handler_t glusterfs_stat_cache_get_entry(server *srv, connection *con, libglusterfs_handle_t handle, buffer *glusterfs_path, buffer *name, void *buf, size_t size, stat_cache_entry **fce); + +#endif |