Tarantool development patches archive
 help / color / mirror / Atom feed
From: Cyrill Gorcunov via Tarantool-patches <tarantool-patches@dev.tarantool.org>
To: tml <tarantool-patches@dev.tarantool.org>
Cc: Mons Anderson <v.perepelitsa@corp.mail.ru>,
	Vladislav Shpilevoy <v.shpilevoy@tarantool.org>
Subject: [Tarantool-patches] [PATCH v15 02/11] module_cache: move module handling into own subsystem
Date: Fri,  5 Feb 2021 21:54:27 +0300	[thread overview]
Message-ID: <20210205185436.638894-3-gorcunov@gmail.com> (raw)
In-Reply-To: <20210205185436.638894-1-gorcunov@gmail.com>

The module handling should not be bound to particular
function implementation (we will have two users: already
existing functions for "_func" space, and a new upcoming
one which will be serving functions on read only instances).

For this sake all module related code is moved to
module_cache file where we do symbol resolving, calling
and tracking of symbol usage.

Part-of #4642

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
---
 src/box/CMakeLists.txt |   1 +
 src/box/func.c         | 473 +-----------------------------------
 src/box/func.h         |  83 +------
 src/box/func_def.h     |  14 --
 src/box/module_cache.c | 535 +++++++++++++++++++++++++++++++++++++++++
 src/box/module_cache.h | 139 +++++++++++
 6 files changed, 679 insertions(+), 566 deletions(-)
 create mode 100644 src/box/module_cache.c
 create mode 100644 src/box/module_cache.h

diff --git a/src/box/CMakeLists.txt b/src/box/CMakeLists.txt
index 19203f770..339e2c8a9 100644
--- a/src/box/CMakeLists.txt
+++ b/src/box/CMakeLists.txt
@@ -187,6 +187,7 @@ add_library(box STATIC
     sql_stmt_cache.c
     wal.c
     call.c
+    module_cache.c
     merger.c
     ibuf.c
     ${sql_sources}
diff --git a/src/box/func.c b/src/box/func.c
index 8030aa07c..4e7025fdb 100644
--- a/src/box/func.c
+++ b/src/box/func.c
@@ -31,30 +31,12 @@
 #include "func.h"
 #include "fiber.h"
 #include "trivia/config.h"
-#include "assoc.h"
 #include "lua/utils.h"
 #include "lua/call.h"
-#include "error.h"
-#include "errinj.h"
 #include "diag.h"
 #include "port.h"
 #include "schema.h"
 #include "session.h"
-#include "libeio/eio.h"
-#include <fcntl.h>
-#include <dlfcn.h>
-
-/**
- * Parsed symbol and package names.
- */
-struct func_name {
-	/** Null-terminated symbol name, e.g. "func" for "mod.submod.func" */
-	const char *sym;
-	/** Package name, e.g. "mod.submod" for "mod.submod.func" */
-	const char *package;
-	/** A pointer to the last character in ->package + 1 */
-	const char *package_end;
-};
 
 struct func_c {
 	/** Function object base class. */
@@ -63,427 +45,6 @@ struct func_c {
 	struct module_sym mod_sym;
 };
 
-/***
- * Split function name to symbol and package names.
- * For example, str = foo.bar.baz => sym = baz, package = foo.bar
- * @param str function name, e.g. "module.submodule.function".
- * @param[out] name parsed symbol and package names.
- */
-static void
-func_split_name(const char *str, struct func_name *name)
-{
-	name->package = str;
-	name->package_end = strrchr(str, '.');
-	if (name->package_end != NULL) {
-		/* module.submodule.function => module.submodule, function */
-		name->sym = name->package_end + 1; /* skip '.' */
-	} else {
-		/* package == function => function, function */
-		name->sym = name->package;
-		name->package_end = str + strlen(str);
-	}
-}
-
-/**
- * Arguments for luaT_module_find used by lua_cpcall()
- */
-struct module_find_ctx {
-	const char *package;
-	const char *package_end;
-	char *path;
-	size_t path_len;
-};
-
-/**
- * A cpcall() helper for module_find()
- */
-static int
-luaT_module_find(lua_State *L)
-{
-	struct module_find_ctx *ctx = (struct module_find_ctx *)
-		lua_topointer(L, 1);
-
-	/*
-	 * Call package.searchpath(name, package.cpath) and use
-	 * the path to the function in dlopen().
-	 */
-	lua_getglobal(L, "package");
-
-	lua_getfield(L, -1, "search");
-
-	/* Argument of search: name */
-	lua_pushlstring(L, ctx->package, ctx->package_end - ctx->package);
-
-	lua_call(L, 1, 1);
-	if (lua_isnil(L, -1))
-		return luaL_error(L, "module not found");
-	/* Convert path to absolute */
-	char resolved[PATH_MAX];
-	if (realpath(lua_tostring(L, -1), resolved) == NULL) {
-		diag_set(SystemError, "realpath");
-		return luaT_error(L);
-	}
-
-	snprintf(ctx->path, ctx->path_len, "%s", resolved);
-	return 0;
-}
-
-/**
- * Find path to module using Lua's package.cpath
- * @param package package name
- * @param package_end a pointer to the last byte in @a package + 1
- * @param[out] path path to shared library
- * @param path_len size of @a path buffer
- * @retval 0 on success
- * @retval -1 on error, diag is set
- */
-static int
-module_find(const char *package, const char *package_end, char *path,
-	    size_t path_len)
-{
-	struct module_find_ctx ctx = { package, package_end, path, path_len };
-	lua_State *L = tarantool_L;
-	int top = lua_gettop(L);
-	if (luaT_cpcall(L, luaT_module_find, &ctx) != 0) {
-		int package_len = (int) (package_end - package);
-		diag_set(ClientError, ER_LOAD_MODULE, package_len, package,
-			 lua_tostring(L, -1));
-		lua_settop(L, top);
-		return -1;
-	}
-	assert(top == lua_gettop(L)); /* cpcall discard results */
-	return 0;
-}
-
-static struct mh_strnptr_t *modules = NULL;
-
-static void
-module_gc(struct module *module);
-
-int
-module_init(void)
-{
-	modules = mh_strnptr_new();
-	if (modules == NULL) {
-		diag_set(OutOfMemory, sizeof(*modules), "malloc",
-			  "modules hash table");
-		return -1;
-	}
-	return 0;
-}
-
-void
-module_free(void)
-{
-	while (mh_size(modules) > 0) {
-		mh_int_t i = mh_first(modules);
-		struct module *module =
-			(struct module *) mh_strnptr_node(modules, i)->val;
-		/* Can't delete modules if they have active calls */
-		module_gc(module);
-	}
-	mh_strnptr_delete(modules);
-}
-
-/**
- * Look up a module in the modules cache.
- */
-static struct module *
-module_cache_find(const char *name, const char *name_end)
-{
-	mh_int_t i = mh_strnptr_find_inp(modules, name, name_end - name);
-	if (i == mh_end(modules))
-		return NULL;
-	return (struct module *)mh_strnptr_node(modules, i)->val;
-}
-
-/**
- * Save module to the module cache.
- */
-static inline int
-module_cache_put(struct module *module)
-{
-	size_t package_len = strlen(module->package);
-	uint32_t name_hash = mh_strn_hash(module->package, package_len);
-	const struct mh_strnptr_node_t strnode = {
-		module->package, package_len, name_hash, module};
-
-	if (mh_strnptr_put(modules, &strnode, NULL, NULL) == mh_end(modules)) {
-		diag_set(OutOfMemory, sizeof(strnode), "malloc", "modules");
-		return -1;
-	}
-	return 0;
-}
-
-/**
- * Delete a module from the module cache
- */
-static void
-module_cache_del(const char *name, const char *name_end)
-{
-	mh_int_t i = mh_strnptr_find_inp(modules, name, name_end - name);
-	if (i == mh_end(modules))
-		return;
-	mh_strnptr_del(modules, i, NULL);
-}
-
-/*
- * Load a dso.
- * Create a new symlink based on temporary directory and try to
- * load via this symink to load a dso twice for cases of a function
- * reload.
- */
-static struct module *
-module_load(const char *package, const char *package_end)
-{
-	char path[PATH_MAX];
-	if (module_find(package, package_end, path, sizeof(path)) != 0)
-		return NULL;
-
-	int package_len = package_end - package;
-	struct module *module = (struct module *)
-		malloc(sizeof(*module) + package_len + 1);
-	if (module == NULL) {
-		diag_set(OutOfMemory, sizeof(struct module) + package_len + 1,
-			 "malloc", "struct module");
-		return NULL;
-	}
-	memcpy(module->package, package, package_len);
-	module->package[package_len] = 0;
-	rlist_create(&module->funcs);
-	module->calls = 0;
-
-	const char *tmpdir = getenv("TMPDIR");
-	if (tmpdir == NULL)
-		tmpdir = "/tmp";
-	char dir_name[PATH_MAX];
-	int rc = snprintf(dir_name, sizeof(dir_name), "%s/tntXXXXXX", tmpdir);
-	if (rc < 0 || (size_t) rc >= sizeof(dir_name)) {
-		diag_set(SystemError, "failed to generate path to tmp dir");
-		goto error;
-	}
-
-	if (mkdtemp(dir_name) == NULL) {
-		diag_set(SystemError, "failed to create unique dir name: %s",
-			 dir_name);
-		goto error;
-	}
-	char load_name[PATH_MAX];
-	rc = snprintf(load_name, sizeof(load_name), "%s/%.*s." TARANTOOL_LIBEXT,
-		      dir_name, package_len, package);
-	if (rc < 0 || (size_t) rc >= sizeof(dir_name)) {
-		diag_set(SystemError, "failed to generate path to DSO");
-		goto error;
-	}
-
-	struct stat st;
-	if (stat(path, &st) < 0) {
-		diag_set(SystemError, "failed to stat() module %s", path);
-		goto error;
-	}
-
-	int source_fd = open(path, O_RDONLY);
-	if (source_fd < 0) {
-		diag_set(SystemError, "failed to open module %s file for" \
-			 " reading", path);
-		goto error;
-	}
-	int dest_fd = open(load_name, O_WRONLY|O_CREAT|O_TRUNC,
-			   st.st_mode & 0777);
-	if (dest_fd < 0) {
-		diag_set(SystemError, "failed to open file %s for writing ",
-			 load_name);
-		close(source_fd);
-		goto error;
-	}
-
-	off_t ret = eio_sendfile_sync(dest_fd, source_fd, 0, st.st_size);
-	close(source_fd);
-	close(dest_fd);
-	if (ret != st.st_size) {
-		diag_set(SystemError, "failed to copy DSO %s to %s",
-			 path, load_name);
-		goto error;
-	}
-
-	module->handle = dlopen(load_name, RTLD_NOW | RTLD_LOCAL);
-	if (unlink(load_name) != 0)
-		say_warn("failed to unlink dso link %s", load_name);
-	if (rmdir(dir_name) != 0)
-		say_warn("failed to delete temporary dir %s", dir_name);
-	if (module->handle == NULL) {
-		diag_set(ClientError, ER_LOAD_MODULE, package_len,
-			  package, dlerror());
-		goto error;
-	}
-	struct errinj *e = errinj(ERRINJ_DYN_MODULE_COUNT, ERRINJ_INT);
-	if (e != NULL)
-		++e->iparam;
-	return module;
-error:
-	free(module);
-	return NULL;
-}
-
-static void
-module_delete(struct module *module)
-{
-	struct errinj *e = errinj(ERRINJ_DYN_MODULE_COUNT, ERRINJ_INT);
-	if (e != NULL)
-		--e->iparam;
-	dlclose(module->handle);
-	TRASH(module);
-	free(module);
-}
-
-/*
- * Check if a dso is unused and can be closed.
- */
-static void
-module_gc(struct module *module)
-{
-	if (rlist_empty(&module->funcs) && module->calls == 0)
-		module_delete(module);
-}
-
-/*
- * Import a function from the module.
- */
-static box_function_f
-module_sym(struct module *module, const char *name)
-{
-	box_function_f f = (box_function_f)dlsym(module->handle, name);
-	if (f == NULL) {
-		diag_set(ClientError, ER_LOAD_FUNCTION, name, dlerror());
-		return NULL;
-	}
-	return f;
-}
-
-int
-module_sym_load(struct module_sym *mod_sym)
-{
-	assert(mod_sym->addr == NULL);
-
-	struct func_name name;
-	func_split_name(mod_sym->name, &name);
-
-	/*
-	 * In case if module has been loaded already by
-	 * some previous call we can eliminate redundant
-	 * loading and take it from the cache.
-	 */
-	struct module *cached, *module;
-	cached = module_cache_find(name.package, name.package_end);
-	if (cached == NULL) {
-		module = module_load(name.package, name.package_end);
-		if (module == NULL)
-			return -1;
-		if (module_cache_put(module) != 0) {
-			module_delete(module);
-			return -1;
-		}
-	} else {
-		module = cached;
-	}
-
-	mod_sym->addr = module_sym(module, name.sym);
-	if (mod_sym->addr == NULL) {
-		if (cached == NULL) {
-			/*
-			 * In case if it was a first load we should
-			 * clean the cache immediately otherwise
-			 * the module continue being referenced even
-			 * if there will be no use of it.
-			 *
-			 * Note the module_sym set an error thus be
-			 * careful to not wipe it.
-			 */
-			module_cache_del(name.package, name.package_end);
-			module_delete(module);
-		}
-		return -1;
-	}
-	mod_sym->module = module;
-	rlist_add(&module->funcs, &mod_sym->item);
-	return 0;
-}
-
-void
-module_sym_unload(struct module_sym *mod_sym)
-{
-	if (mod_sym->module == NULL)
-		return;
-
-	rlist_del(&mod_sym->item);
-	if (rlist_empty(&mod_sym->module->funcs)) {
-		struct func_name name;
-		func_split_name(mod_sym->name, &name);
-		module_cache_del(name.package, name.package_end);
-	}
-	module_gc(mod_sym->module);
-
-	mod_sym->module = NULL;
-	mod_sym->addr = NULL;
-}
-
-int
-module_reload(const char *package, const char *package_end, struct module **module)
-{
-	struct module *old_module = module_cache_find(package, package_end);
-	if (old_module == NULL) {
-		/* Module wasn't loaded - do nothing. */
-		*module = NULL;
-		return 0;
-	}
-
-	struct module *new_module = module_load(package, package_end);
-	if (new_module == NULL)
-		return -1;
-
-	struct module_sym *mod_sym, *tmp;
-	rlist_foreach_entry_safe(mod_sym, &old_module->funcs, item, tmp) {
-		struct func_name name;
-		func_split_name(mod_sym->name, &name);
-		mod_sym->addr = module_sym(new_module, name.sym);
-		if (mod_sym->addr == NULL)
-			goto restore;
-		mod_sym->module = new_module;
-		rlist_move(&new_module->funcs, &mod_sym->item);
-	}
-	module_cache_del(package, package_end);
-	if (module_cache_put(new_module) != 0)
-		goto restore;
-	module_gc(old_module);
-	*module = new_module;
-	return 0;
-restore:
-	/*
-	 * Some old-dso func can't be load from new module, restore old
-	 * functions.
-	 */
-	do {
-		struct func_name name;
-		func_split_name(mod_sym->name, &name);
-		mod_sym->addr = module_sym(old_module, name.sym);
-		if (mod_sym->addr == NULL) {
-			/*
-			 * Something strange was happen, an early loaden
-			 * function was not found in an old dso.
-			 */
-			panic("Can't restore module function, "
-			      "server state is inconsistent");
-		}
-		mod_sym->module = old_module;
-		rlist_move(&old_module->funcs, &mod_sym->item);
-	} while (mod_sym != rlist_first_entry(&old_module->funcs,
-					      struct module_sym,
-					      item));
-	assert(rlist_empty(&new_module->funcs));
-	module_delete(new_module);
-	return -1;
-}
-
 static struct func *
 func_c_new(struct func_def *def);
 
@@ -563,39 +124,9 @@ func_c_call(struct func *base, struct port *args, struct port *ret)
 {
 	assert(base->vtab == &func_c_vtab);
 	assert(base != NULL && base->def->language == FUNC_LANGUAGE_C);
-	struct func_c *func = (struct func_c *) base;
-	if (func->mod_sym.addr == NULL) {
-		if (module_sym_load(&func->mod_sym) != 0)
-			return -1;
-	}
-
-	struct region *region = &fiber()->gc;
-	size_t region_svp = region_used(region);
-	uint32_t data_sz;
-	const char *data = port_get_msgpack(args, &data_sz);
-	if (data == NULL)
-		return -1;
-
-	port_c_create(ret);
-	box_function_ctx_t ctx = { ret };
 
-	/* Module can be changed after function reload. */
-	struct module *module = func->mod_sym.module;
-	assert(module != NULL);
-	++module->calls;
-	int rc = func->mod_sym.addr(&ctx, data, data + data_sz);
-	--module->calls;
-	module_gc(module);
-	region_truncate(region, region_svp);
-	if (rc != 0) {
-		if (diag_last_error(&fiber()->diag) == NULL) {
-			/* Stored procedure forget to set diag  */
-			diag_set(ClientError, ER_PROC_C, "unknown error");
-		}
-		port_destroy(ret);
-		return -1;
-	}
-	return rc;
+	struct func_c *func = (struct func_c *)base;
+	return module_sym_call(&func->mod_sym, args, ret);
 }
 
 static struct func_vtab func_c_vtab = {
diff --git a/src/box/func.h b/src/box/func.h
index 9a7f17446..11a466b28 100644
--- a/src/box/func.h
+++ b/src/box/func.h
@@ -34,7 +34,8 @@
 #include <stddef.h>
 #include <stdint.h>
 #include <stdbool.h>
-#include "small/rlist.h"
+
+#include "module_cache.h"
 #include "func_def.h"
 #include "user_def.h"
 
@@ -44,42 +45,6 @@ extern "C" {
 
 struct func;
 
-/**
- * Dynamic shared module.
- */
-struct module {
-	/** Module dlhandle. */
-	void *handle;
-	/** List of imported functions. */
-	struct rlist funcs;
-	/** Count of active calls. */
-	size_t calls;
-	/** Module's package name. */
-	char package[0];
-};
-
-/**
- * Callable symbol bound to a module.
- */
-struct module_sym {
-	/**
-	 * Anchor for module membership.
-	 */
-	struct rlist item;
-	/**
-	 * For C functions, address of the function.
-	 */
-	box_function_f addr;
-	/**
-	 * A module the symbol belongs to.
-	 */
-	struct module *module;
-	/**
-	 * Function name definition.
-	 */
-	char *name;
-};
-
 /** Virtual method table for func object. */
 struct func_vtab {
 	/** Call function with given arguments. */
@@ -106,18 +71,6 @@ struct func {
 	struct access access[BOX_USER_MAX];
 };
 
-/**
- * Initialize modules subsystem.
- */
-int
-module_init(void);
-
-/**
- * Cleanup modules subsystem.
- */
-void
-module_free(void);
-
 struct func *
 func_new(struct func_def *def);
 
@@ -130,38 +83,6 @@ func_delete(struct func *func);
 int
 func_call(struct func *func, struct port *args, struct port *ret);
 
-/**
- * Resolve C entry (find the respective DLL and fetch the
- * symbol from it).
- *
- * @param mod_sym module symbol pointer.
- * @retval -1 on error.
- * @retval 0 on success.
- */
-int
-module_sym_load(struct module_sym *mod_sym);
-
-/**
- * Unload module symbol and drop it from the package
- * cache if there is no users left.
- *
- * @param mod_sym module symbol pointer.
- */
-void
-module_sym_unload(struct module_sym *mod_sym);
-
-/**
- * Reload dynamically loadable module.
- *
- * @param package name begin pointer.
- * @param package_end package_end name end pointer.
- * @param[out] module a pointer to store module object on success.
- * @retval -1 on error.
- * @retval 0 on success.
- */
-int
-module_reload(const char *package, const char *package_end, struct module **module);
-
 #if defined(__cplusplus)
 } /* extern "C" */
 #endif /* defined(__cplusplus) */
diff --git a/src/box/func_def.h b/src/box/func_def.h
index d99d89190..75cd6a0d3 100644
--- a/src/box/func_def.h
+++ b/src/box/func_def.h
@@ -168,20 +168,6 @@ func_def_dup(struct func_def *def);
 int
 func_def_check(struct func_def *def);
 
-/**
- * API of C stored function.
- */
-
-struct port;
-
-struct box_function_ctx {
-	struct port *port;
-};
-
-typedef struct box_function_ctx box_function_ctx_t;
-typedef int (*box_function_f)(box_function_ctx_t *ctx,
-	     const char *args, const char *args_end);
-
 #ifdef __cplusplus
 }
 #endif
diff --git a/src/box/module_cache.c b/src/box/module_cache.c
new file mode 100644
index 000000000..4d89baf9b
--- /dev/null
+++ b/src/box/module_cache.c
@@ -0,0 +1,535 @@
+/*
+ * SPDX-License-Identifier: BSD-2-Clause
+ *
+ * Copyright 2010-2021, Tarantool AUTHORS, please see AUTHORS file.
+ */
+
+#include <dlfcn.h>
+#include <fcntl.h>
+#include <stdlib.h>
+#include <string.h>
+
+#include "assoc.h"
+#include "diag.h"
+#include "error.h"
+#include "errinj.h"
+#include "fiber.h"
+#include "port.h"
+
+#include "error.h"
+#include "lua/utils.h"
+#include "libeio/eio.h"
+
+#include "module_cache.h"
+
+/** Modules name to descriptor hash. */
+static struct mh_strnptr_t *mod_hash = NULL;
+
+/**
+ * Parsed symbol and package names.
+ */
+struct func_name {
+	/**
+	 * Null-terminated symbol name, e.g.
+	 * "func" for "mod.submod.func".
+	 */
+	const char *sym;
+	/**
+	 * Package name, e.g. "mod.submod" for
+	 * "mod.submod.func".
+	 */
+	const char *package;
+	/**
+	 * A pointer to the last character in ->package + 1.
+	 */
+	const char *package_end;
+};
+
+/***
+ * Split function name to symbol and package names.
+ *
+ * For example, str = foo.bar.baz => sym = baz, package = foo.bar
+ *
+ * @param str function name, e.g. "module.submodule.function".
+ * @param[out] name parsed symbol and a package name.
+ */
+static void
+func_split_name(const char *str, struct func_name *name)
+{
+	name->package = str;
+	name->package_end = strrchr(str, '.');
+	if (name->package_end != NULL) {
+		/* module.submodule.function => module.submodule, function */
+		name->sym = name->package_end + 1; /* skip '.' */
+	} else {
+		/* package == function => function, function */
+		name->sym = name->package;
+		name->package_end = str + strlen(str);
+	}
+}
+
+/**
+ * Look up a module in the modules cache.
+ */
+static struct module *
+module_cache_find(const char *name, const char *name_end)
+{
+	mh_int_t e = mh_strnptr_find_inp(mod_hash, name, name_end - name);
+	if (e == mh_end(mod_hash))
+		return NULL;
+	return mh_strnptr_node(mod_hash, e)->val;
+}
+
+/**
+ * Save a module to the modules cache.
+ */
+static int
+module_cache_add(struct module *module)
+{
+	size_t package_len = strlen(module->package);
+	const struct mh_strnptr_node_t nd = {
+		.str	= module->package,
+		.len	= package_len,
+		.hash	= mh_strn_hash(module->package, package_len),
+		.val	= module,
+	};
+
+	if (mh_strnptr_put(mod_hash, &nd, NULL, NULL) == mh_end(mod_hash)) {
+		diag_set(OutOfMemory, sizeof(nd), "malloc",
+			 "module cache node");
+		return -1;
+	}
+	return 0;
+}
+
+/**
+ * Delete a module from the modules cache.
+ */
+static void
+module_cache_del(const char *name, const char *name_end)
+{
+	mh_int_t e = mh_strnptr_find_inp(mod_hash, name, name_end - name);
+	if (e != mh_end(mod_hash))
+		mh_strnptr_del(mod_hash, e, NULL);
+}
+
+/**
+ * Arguments for luaT_module_find used by lua_cpcall().
+ */
+struct module_find_ctx {
+	const char *package;
+	const char *package_end;
+	char *path;
+	size_t path_len;
+};
+
+/**
+ * A cpcall() helper for module_find().
+ */
+static int
+luaT_module_find(lua_State *L)
+{
+	struct module_find_ctx *ctx = (void *)lua_topointer(L, 1);
+
+	/*
+	 * Call package.searchpath(name, package.cpath) and use
+	 * the path to the function in dlopen().
+	 */
+	lua_getglobal(L, "package");
+	lua_getfield(L, -1, "search");
+
+	/* Argument of search: name */
+	lua_pushlstring(L, ctx->package, ctx->package_end - ctx->package);
+
+	lua_call(L, 1, 1);
+	if (lua_isnil(L, -1))
+		return luaL_error(L, "module not found");
+
+	/* Convert path to absolute */
+	char resolved[PATH_MAX];
+	if (realpath(lua_tostring(L, -1), resolved) == NULL) {
+		diag_set(SystemError, "realpath");
+		return luaT_error(L);
+	}
+
+	snprintf(ctx->path, ctx->path_len, "%s", resolved);
+	return 0;
+}
+
+/**
+ * Find a path to a module using Lua's package.cpath.
+ *
+ * @param package package name
+ * @param package_end a pointer to the last byte in @a package + 1
+ * @param[out] path path to shared library
+ * @param path_len size of @a path buffer
+ *
+ * @retval 0 on success
+ * @retval -1 on error, diag is set
+ */
+static int
+module_find(const char *package, const char *package_end,
+	    char *path, size_t path_len)
+{
+	struct module_find_ctx ctx = {
+		.package	= package,
+		.package_end	= package_end,
+		.path		= path,
+		.path_len	= path_len,
+	};
+	lua_State *L = tarantool_L;
+	int top = lua_gettop(L);
+	if (luaT_cpcall(L, luaT_module_find, &ctx) != 0) {
+		diag_set(ClientError, ER_LOAD_MODULE,
+			 (int)(package_end - package),
+			 package, lua_tostring(L, -1));
+		lua_settop(L, top);
+		return -1;
+	}
+	assert(top == lua_gettop(L)); /* cpcall discard results */
+	return 0;
+}
+
+/**
+ * Load dynamic shared object, ie module library.
+ *
+ * Create a new symlink based on temporary directory
+ * and try to load via this symink to load a dso twice
+ * for cases of a function reload.
+ */
+static struct module *
+module_load(const char *package, const char *package_end)
+{
+	char path[PATH_MAX];
+	if (module_find(package, package_end, path, sizeof(path)) != 0)
+		return NULL;
+
+	int package_len = package_end - package;
+	struct module *module = malloc(sizeof(*module) + package_len + 1);
+	if (module == NULL) {
+		diag_set(OutOfMemory, sizeof(*module) + package_len + 1,
+			 "malloc", "struct module");
+		return NULL;
+	}
+	memcpy(module->package, package, package_len);
+	module->package[package_len] = 0;
+	rlist_create(&module->funcs_list);
+	module->calls = 0;
+
+	const char *tmpdir = getenv("TMPDIR");
+	if (tmpdir == NULL)
+		tmpdir = "/tmp";
+
+	char dir_name[PATH_MAX];
+	int rc = snprintf(dir_name, sizeof(dir_name), "%s/tntXXXXXX", tmpdir);
+	if (rc < 0 || (size_t)rc >= sizeof(dir_name)) {
+		diag_set(SystemError, "failed to generate path to tmp dir");
+		goto error;
+	}
+
+	if (mkdtemp(dir_name) == NULL) {
+		diag_set(SystemError, "failed to create unique dir name: %s",
+			 dir_name);
+		goto error;
+	}
+
+	char load_name[PATH_MAX];
+	rc = snprintf(load_name, sizeof(load_name), "%s/%.*s." TARANTOOL_LIBEXT,
+		      dir_name, package_len, package);
+	if (rc < 0 || (size_t)rc >= sizeof(dir_name)) {
+		diag_set(SystemError, "failed to generate path to DSO");
+		goto error;
+	}
+
+	struct stat st;
+	if (stat(path, &st) < 0) {
+		diag_set(SystemError, "failed to stat() module %s", path);
+		goto error;
+	}
+
+	int source_fd = open(path, O_RDONLY);
+	if (source_fd < 0) {
+		diag_set(SystemError, "failed to open module %s", path);
+		goto error;
+	}
+
+	int dest_fd = open(load_name, O_WRONLY | O_CREAT | O_TRUNC,
+			   st.st_mode & (S_IRWXU | S_IRWXG | S_IRWXO));
+	if (dest_fd < 0) {
+		diag_set(SystemError, "failed to open file %s for writing ",
+			 load_name);
+		close(source_fd);
+		goto error;
+	}
+
+	off_t ret = eio_sendfile_sync(dest_fd, source_fd, 0, st.st_size);
+	close(source_fd);
+	close(dest_fd);
+	if (ret != st.st_size) {
+		diag_set(SystemError, "failed to copy DSO %s to %s",
+			 path, load_name);
+		goto error;
+	}
+
+	module->handle = dlopen(load_name, RTLD_NOW | RTLD_LOCAL);
+	if (unlink(load_name) != 0)
+		say_warn("failed to unlink dso link %s", load_name);
+	if (rmdir(dir_name) != 0)
+		say_warn("failed to delete temporary dir %s", dir_name);
+	if (module->handle == NULL) {
+		diag_set(ClientError, ER_LOAD_MODULE, package_len,
+			  package, dlerror());
+		goto error;
+	}
+
+	struct errinj *e = errinj(ERRINJ_DYN_MODULE_COUNT, ERRINJ_INT);
+	if (e != NULL)
+		++e->iparam;
+	return module;
+
+error:
+	free(module);
+	return NULL;
+}
+
+/**
+ * Delete a module.
+ */
+static void
+module_delete(struct module *module)
+{
+	struct errinj *e = errinj(ERRINJ_DYN_MODULE_COUNT, ERRINJ_INT);
+	if (e != NULL)
+		--e->iparam;
+	dlclose(module->handle);
+	TRASH(module);
+	free(module);
+}
+
+/**
+ * Check if a module is unused and delete it then.
+ */
+static void
+module_gc(struct module *module)
+{
+	if (rlist_empty(&module->funcs_list) && module->calls == 0)
+		module_delete(module);
+}
+
+/**
+ * Import a function from a module.
+ */
+static box_function_f
+module_sym(struct module *module, const char *name)
+{
+	box_function_f f = dlsym(module->handle, name);
+	if (f == NULL) {
+		diag_set(ClientError, ER_LOAD_FUNCTION, name, dlerror());
+		return NULL;
+	}
+	return f;
+}
+
+int
+module_sym_load(struct module_sym *mod_sym)
+{
+	assert(mod_sym->addr == NULL);
+
+	struct func_name name;
+	func_split_name(mod_sym->name, &name);
+
+	/*
+	 * In case if module has been loaded already by
+	 * some previous call we can eliminate redundant
+	 * loading and take it from the cache.
+	 */
+	struct module *cached, *module;
+	cached = module_cache_find(name.package, name.package_end);
+	if (cached == NULL) {
+		module = module_load(name.package, name.package_end);
+		if (module == NULL)
+			return -1;
+		if (module_cache_add(module) != 0) {
+			module_delete(module);
+			return -1;
+		}
+	} else {
+		module = cached;
+	}
+
+	mod_sym->addr = module_sym(module, name.sym);
+	if (mod_sym->addr == NULL) {
+		if (cached == NULL) {
+			/*
+			 * In case if it was a first load we should
+			 * clean the cache immediately otherwise
+			 * the module continue being referenced even
+			 * if there will be no use of it.
+			 *
+			 * Note the module_sym set an error thus be
+			 * careful to not wipe it.
+			 */
+			module_cache_del(name.package, name.package_end);
+			module_delete(module);
+		}
+		return -1;
+	}
+
+	mod_sym->module = module;
+	rlist_add(&module->funcs_list, &mod_sym->item);
+	return 0;
+}
+
+void
+module_sym_unload(struct module_sym *mod_sym)
+{
+	if (mod_sym->addr == NULL)
+		return;
+
+	rlist_del(&mod_sym->item);
+	if (rlist_empty(&mod_sym->module->funcs_list)) {
+		struct func_name name;
+		func_split_name(mod_sym->name, &name);
+		module_cache_del(name.package, name.package_end);
+	}
+	module_gc(mod_sym->module);
+
+	mod_sym->module = NULL;
+	mod_sym->addr = NULL;
+}
+
+int
+module_sym_call(struct module_sym *mod_sym, struct port *args,
+		struct port *ret)
+{
+	if (mod_sym->addr == NULL) {
+		if (module_sym_load(mod_sym) != 0)
+			return -1;
+	}
+
+	struct region *region = &fiber()->gc;
+	size_t region_svp = region_used(region);
+
+	uint32_t data_sz;
+	const char *data = port_get_msgpack(args, &data_sz);
+	if (data == NULL)
+		return -1;
+
+	port_c_create(ret);
+	box_function_ctx_t ctx = {
+		.port = ret,
+	};
+
+	/*
+	 * Module can be changed after function reload. Also
+	 * keep in mind that stored C procedure may yield inside.
+	 */
+	struct module *module = mod_sym->module;
+	assert(module != NULL);
+	++module->calls;
+	int rc = mod_sym->addr(&ctx, data, data + data_sz);
+	--module->calls;
+	module_gc(module);
+	region_truncate(region, region_svp);
+
+	if (rc != 0) {
+		if (diag_last_error(&fiber()->diag) == NULL) {
+			/* Stored procedure forget to set diag  */
+			diag_set(ClientError, ER_PROC_C, "unknown error");
+		}
+		port_destroy(ret);
+		return -1;
+	}
+
+	return rc;
+}
+
+int
+module_reload(const char *package, const char *package_end,
+	      struct module **module)
+{
+	struct module *old = module_cache_find(package, package_end);
+	if (old == NULL) {
+		/* Module wasn't loaded - do nothing. */
+		*module = NULL;
+		return 0;
+	}
+
+	struct module *new = module_load(package, package_end);
+	if (new == NULL)
+		return -1;
+
+	struct module_sym *mod_sym, *tmp;
+	rlist_foreach_entry_safe(mod_sym, &old->funcs_list, item, tmp) {
+		struct func_name name;
+		func_split_name(mod_sym->name, &name);
+
+		mod_sym->addr = module_sym(new, name.sym);
+		if (mod_sym->addr == NULL) {
+			say_error("module: reload %s, symbol %s not found",
+				  package, name.sym);
+			goto restore;
+		}
+
+		mod_sym->module = new;
+		rlist_move(&new->funcs_list, &mod_sym->item);
+	}
+
+	module_cache_del(package, package_end);
+	if (module_cache_add(new) != 0)
+		goto restore;
+
+	module_gc(old);
+	*module = new;
+	return 0;
+
+restore:
+	/*
+	 * Some old-dso func can't be load from new module,
+	 * restore old functions.
+	 */
+	do {
+		struct func_name name;
+		func_split_name(mod_sym->name, &name);
+		mod_sym->addr = module_sym(old, name.sym);
+		if (mod_sym->addr == NULL) {
+			/*
+			 * Something strange was happen, an early loaden
+			 * function was not found in an old dso.
+			 */
+			panic("Can't restore module function, "
+			      "server state is inconsistent");
+		}
+		mod_sym->module = old;
+		rlist_move(&old->funcs_list, &mod_sym->item);
+	} while (mod_sym != rlist_first_entry(&old->funcs_list,
+					      struct module_sym,
+					      item));
+	assert(rlist_empty(&new->funcs_list));
+	module_delete(new);
+	return -1;
+}
+
+int
+module_init(void)
+{
+	mod_hash = mh_strnptr_new();
+	if (mod_hash == NULL) {
+		diag_set(OutOfMemory, sizeof(*mod_hash),
+			 "malloc", "modules hash table");
+		return -1;
+	}
+	return 0;
+}
+
+void
+module_free(void)
+{
+	while (mh_size(mod_hash) > 0) {
+		mh_int_t i = mh_first(mod_hash);
+		struct module *m = mh_strnptr_node(mod_hash, i)->val;
+		module_gc(m);
+	}
+	mh_strnptr_delete(mod_hash);
+	mod_hash = NULL;
+}
diff --git a/src/box/module_cache.h b/src/box/module_cache.h
new file mode 100644
index 000000000..0f5d2b64a
--- /dev/null
+++ b/src/box/module_cache.h
@@ -0,0 +1,139 @@
+/*
+ * SPDX-License-Identifier: BSD-2-Clause
+ *
+ * Copyright 2010-2021, Tarantool AUTHORS, please see AUTHORS file.
+ */
+
+#pragma once
+
+#include "small/rlist.h"
+
+#if defined(__cplusplus)
+extern "C" {
+#endif /* defined(__cplusplus) */
+
+
+/**
+ * API of C stored function.
+ */
+
+struct port;
+
+struct box_function_ctx {
+	struct port *port;
+};
+
+typedef struct box_function_ctx box_function_ctx_t;
+typedef int (*box_function_f)(box_function_ctx_t *ctx,
+			      const char *args,
+			      const char *args_end);
+
+/**
+ * Dynamic shared module.
+ */
+struct module {
+	/**
+	 * Module dlhandle.
+	 */
+	void *handle;
+	/**
+	 * List of associated symbols (functions).
+	 */
+	struct rlist funcs_list;
+	/**
+	 * Count of active calls.
+	 */
+	size_t calls;
+	/**
+	 * Module's package name.
+	 */
+	char package[0];
+};
+
+/**
+ * Callable symbol bound to a module.
+ */
+struct module_sym {
+	/**
+	 * Anchor for module::funcs_list.
+	 */
+	struct rlist item;
+	/**
+	 * For C functions, address of the function.
+	 */
+	box_function_f addr;
+	/**
+	 * A module the symbol belongs to.
+	 */
+	struct module *module;
+	/**
+	 * Symbol (function) name definition.
+	 */
+	char *name;
+};
+
+/**
+ * Load a new module symbol.
+ *
+ * @param mod_sym symbol to load.
+ *
+ * @returns 0 on succse, -1 otherwise, diag is set.
+ */
+int
+module_sym_load(struct module_sym *mod_sym);
+
+/**
+ * Unload a module's symbol.
+ *
+ * @param mod_sym symbol to unload.
+ */
+void
+module_sym_unload(struct module_sym *mod_sym);
+
+/**
+ * Execute a module symbol (run a function).
+ *
+ * The function packs function arguments into a message pack
+ * and send it as a function argument. Function may return
+ * results via @a ret stream.
+ *
+ * @param mod_sym module symbol to run.
+ * @param args function arguments.
+ * @param[out] ret execution results.
+ *
+ * @returns 0 on success, -1 otherwise, diag is set.
+ */
+int
+module_sym_call(struct module_sym *mod_sym, struct port *args,
+		struct port *ret);
+
+/**
+ * Reload a module and all associated symbols.
+ *
+ * @param package shared library path start.
+ * @param package_end shared library path end.
+ * @param[out] module pointer to the reloaded module.
+ *
+ * @return 0 on succes, -1 otherwise, diag is set.
+ */
+int
+module_reload(const char *package, const char *package_end,
+	      struct module **module);
+
+/**
+ * Initialize modules subsystem.
+ *
+ * @return 0 on succes, -1 otherwise, diag is set.
+ */
+int
+module_init(void);
+
+/**
+ * Free modules subsystem.
+ */
+void
+module_free(void);
+
+#if defined(__cplusplus)
+}
+#endif /* defined(__plusplus) */
-- 
2.29.2


  parent reply	other threads:[~2021-02-05 18:55 UTC|newest]

Thread overview: 21+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-02-05 18:54 [Tarantool-patches] [PATCH v15 00/11] box: implement cmod Lua module Cyrill Gorcunov via Tarantool-patches
2021-02-05 18:54 ` [Tarantool-patches] [PATCH v15 01/11] box/func: factor out c function entry structure Cyrill Gorcunov via Tarantool-patches
2021-02-05 18:54 ` Cyrill Gorcunov via Tarantool-patches [this message]
2021-02-07 17:20   ` [Tarantool-patches] [PATCH v15 02/11] module_cache: move module handling into own subsystem Vladislav Shpilevoy via Tarantool-patches
2021-02-05 18:54 ` [Tarantool-patches] [PATCH v15 03/11] module_cache: direct update a cache value on reload Cyrill Gorcunov via Tarantool-patches
2021-02-05 18:54 ` [Tarantool-patches] [PATCH v15 04/11] module_cache: rename calls to ref in module structure Cyrill Gorcunov via Tarantool-patches
2021-02-05 18:54 ` [Tarantool-patches] [PATCH v15 05/11] module_cache: add comment about weird resolving Cyrill Gorcunov via Tarantool-patches
2021-02-05 18:54 ` [Tarantool-patches] [PATCH v15 06/11] module_cache: module_reload - drop redundant parameter Cyrill Gorcunov via Tarantool-patches
2021-02-05 18:54 ` [Tarantool-patches] [PATCH v15 07/11] module_cache: use references as a main usage counter Cyrill Gorcunov via Tarantool-patches
2021-02-07 17:20   ` Vladislav Shpilevoy via Tarantool-patches
2021-02-08 11:54     ` Cyrill Gorcunov via Tarantool-patches
2021-02-05 18:54 ` [Tarantool-patches] [PATCH v15 08/11] module_cache: make module to carry hash it belongs to Cyrill Gorcunov via Tarantool-patches
2021-02-07 17:20   ` Vladislav Shpilevoy via Tarantool-patches
2021-02-05 18:54 ` [Tarantool-patches] [PATCH v15 09/11] module_cache: use own hash for box.schema.func requests Cyrill Gorcunov via Tarantool-patches
2021-02-07 17:20   ` Vladislav Shpilevoy via Tarantool-patches
2021-02-05 18:54 ` [Tarantool-patches] [PATCH v15 10/11] box/cmod: implement cmod Lua module Cyrill Gorcunov via Tarantool-patches
2021-02-07 17:20   ` Vladislav Shpilevoy via Tarantool-patches
2021-02-05 18:54 ` [Tarantool-patches] [PATCH v15 11/11] test: box/cfunc -- add cmod test Cyrill Gorcunov via Tarantool-patches
2021-02-07 17:20   ` Vladislav Shpilevoy via Tarantool-patches
2021-02-07 17:21     ` Vladislav Shpilevoy via Tarantool-patches
2021-02-06 17:55 ` [Tarantool-patches] [PATCH v15 00/11] box: implement cmod Lua module Vladislav Shpilevoy via Tarantool-patches

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20210205185436.638894-3-gorcunov@gmail.com \
    --to=tarantool-patches@dev.tarantool.org \
    --cc=gorcunov@gmail.com \
    --cc=v.perepelitsa@corp.mail.ru \
    --cc=v.shpilevoy@tarantool.org \
    --subject='Re: [Tarantool-patches] [PATCH v15 02/11] module_cache: move module handling into own subsystem' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox