Tarantool development patches archive
 help / color / mirror / Atom feed
From: Timur Safin <tsafin@tarantool.org>
To: v.shpilevoy@tarantool.org, alexander.turenko@tarantool.org
Cc: tarantool-patches@dev.tarantool.org
Subject: [Tarantool-patches] [PATCH 1.10 5/9] module api: luaT_temp_luastate & luaT_release_temp_luastate
Date: Fri, 25 Sep 2020 00:00:37 +0300	[thread overview]
Message-ID: <064abe8c26b65283e1e7e591796f779ea4b42b4b.1600955796.git.tsafin@tarantool.org> (raw)
In-Reply-To: <cover.1600955796.git.tsafin@tarantool.org>

Moved `luaT_temp_luastate` & `luaT_release_temp_luastate` from
merger to ustil.c and made them public.

They were necessary for external merger support

Part of #5273
---
 src/lua/utils.c | 49 +++++++++++++++++++++++++++++++++++++++++++++++++
 src/lua/utils.h | 35 +++++++++++++++++++++++++++++++++++
 2 files changed, 84 insertions(+)

diff --git a/src/lua/utils.c b/src/lua/utils.c
index e555607cf..58c54f8fb 100644
--- a/src/lua/utils.c
+++ b/src/lua/utils.c
@@ -1090,6 +1090,55 @@ luaT_newthread(struct lua_State *L)
 	return L1;
 }
 
+struct lua_State *
+luaT_temp_luastate(int *coro_ref, int *top)
+{
+	if (fiber()->storage.lua.stack != NULL) {
+		/*
+		 * Reuse existing stack. In the releasing function
+		 * we should drop a stack top to its initial
+		 * value to don't exhaust available slots by
+		 * many requests in row.
+		 */
+		struct lua_State *L = fiber()->storage.lua.stack;
+		*coro_ref = LUA_NOREF;
+		*top = lua_gettop(L);
+		return L;
+	}
+
+	/* Popped by luaL_ref(). */
+	struct lua_State *L = luaT_newthread(tarantool_L);
+	if (L == NULL)
+		return NULL;
+	/*
+	 * We should remove the reference to the newly created Lua
+	 * thread from tarantool_L, because of two reasons:
+	 *
+	 * First, if we'll push something to tarantool_L and
+	 * yield, then another fiber will not know that a stack
+	 * top is changed and may operate on a wrong slot.
+	 *
+	 * Second, many requests that push a value to tarantool_L
+	 * and yield may exhaust available slots on the stack. It
+	 * is limited by LUAI_MAXSTACK build time constant (~65K).
+	 *
+	 * We cannot just pop the value, but should keep the
+	 * reference in the registry while it is in use.
+	 * Otherwise it may be garbage collected.
+	 */
+	*coro_ref = luaL_ref(tarantool_L, LUA_REGISTRYINDEX);
+	*top = -1;
+	return L;
+}
+
+void
+luaT_release_temp_luastate(struct lua_State *L, int coro_ref, int top)
+{
+	if (top >= 0)
+		lua_settop(L, top);
+	luaL_unref(tarantool_L, LUA_REGISTRYINDEX, coro_ref);
+}
+
 int
 tarantool_lua_utils_init(struct lua_State *L)
 {
diff --git a/src/lua/utils.h b/src/lua/utils.h
index a818e5eee..4eb5ce566 100644
--- a/src/lua/utils.h
+++ b/src/lua/utils.h
@@ -530,6 +530,41 @@ luaL_iscallable(lua_State *L, int idx);
 struct lua_State *
 luaT_newthread(struct lua_State *L);
 
+/**
+ * Get a temporary Lua state.
+ *
+ * Use case: a function does not accept a Lua state as an argument
+ * to allow using from C code, but uses a Lua value, which is
+ * referenced in LUA_REGISTRYINDEX. A temporary Lua stack is needed
+ * to get and process the value.
+ *
+ * The resulting Lua state has a separate Lua stack, but the same
+ * globals and registry as `tarantool_L` (and all Lua states in
+ * tarantool at the moment of writing this).
+ *
+ * This Lua state should be used only from one fiber: otherwise
+ * one fiber may change the stack and another one will access a
+ * wrong stack slot when it will be scheduled for execution after
+ * yield.
+ *
+ * Return a Lua state on success and set @a coro_ref and @a top.
+ * These values should be passed to
+ * `luaT_release_temp_luastate()`, when the state is not needed
+ * anymore.
+ *
+ * Return NULL and set a diag at failure.
+ */
+struct lua_State *
+luaT_temp_luastate(int *coro_ref, int *top);
+
+/**
+ * Release a temporary Lua state.
+ *
+ * It complements `luaT_temp_luastate()`.
+ */
+void
+luaT_release_temp_luastate(struct lua_State *L, int coro_ref, int top);
+
 /** \endcond public */
 
 void
-- 
2.20.1

  parent reply	other threads:[~2020-09-24 21:01 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-09-24 21:00 [Tarantool-patches] [PATCH 1.10 0/9] RFC: module api: extend for external merger Lua module Timur Safin
2020-09-24 21:00 ` [Tarantool-patches] [PATCH 1.10 1/9] module api: export box_tuple_validate Timur Safin
2020-09-24 21:00 ` [Tarantool-patches] [PATCH 1.10 2/9] module api: export box_key_def_dup Timur Safin
2020-09-24 21:00 ` [Tarantool-patches] [PATCH 1.10 3/9] module api: luaT_newthread Timur Safin
2020-09-24 21:00 ` [Tarantool-patches] [PATCH 1.10 4/9] module api: luaL_register_module & luaL_register_type Timur Safin
2020-09-24 21:00 ` Timur Safin [this message]
2020-09-24 21:00 ` [Tarantool-patches] [PATCH 1.10 6/9] module api: luaL_checkibuf & luaL_checkconstchar Timur Safin
2020-09-24 21:00 ` [Tarantool-patches] [PATCH 1.10 7/9] RFC: module api: extend list of public symbols in 1.10 Timur Safin
2020-09-29  6:21   ` Alexander Turenko
2020-09-24 21:00 ` [Tarantool-patches] [PATCH 1.10 8/9] module api: add luaL_iscallable with support of cdata metatype Timur Safin
2020-09-24 21:00 ` [Tarantool-patches] [PATCH 1.10 9/9] module api: luaL_cdata_iscallable Timur Safin

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=064abe8c26b65283e1e7e591796f779ea4b42b4b.1600955796.git.tsafin@tarantool.org \
    --to=tsafin@tarantool.org \
    --cc=alexander.turenko@tarantool.org \
    --cc=tarantool-patches@dev.tarantool.org \
    --cc=v.shpilevoy@tarantool.org \
    --subject='Re: [Tarantool-patches] [PATCH 1.10 5/9] module api: luaT_temp_luastate & luaT_release_temp_luastate' \
    /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