From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp56.i.mail.ru (smtp56.i.mail.ru [217.69.128.36]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by dev.tarantool.org (Postfix) with ESMTPS id 1B131445321 for ; Thu, 16 Jul 2020 23:11:51 +0300 (MSK) Date: Thu, 16 Jul 2020 23:11:04 +0300 From: Alexander Turenko Message-ID: <20200716201104.xkexzk2tpgtrgth3@tkn_work_nb> References: <20200701203650.GB5559@tarantool.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <20200701203650.GB5559@tarantool.org> Subject: Re: [Tarantool-patches] [PATCH v2 2/3] merger: clean fiber-local Lua stack after next() List-Id: Tarantool development patches List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Igor Munkin Cc: tarantool-patches@dev.tarantool.org, Vladislav Shpilevoy > > Now those functions may be called only from Lua and if the fiber-local > > Lua state is present it is the same as one that is passed to a Lua/C > > Typo: s/present/presented/. Cited from [1]: | "As presented" (verb) connotes deliberate placement. "As present" | (adjective) just means it's there. 'It is there' meanging fits better here, IMHO. Isn't I miss something about grammar here? [1]: https://www.instructionalsolutions.com/blog/bid/102954/Tables-in-Report-Writing-Presented-or-Present > > The merge_source_next() implementations do not leave any garbage on a > > Lua stack at success path, but may left something when an error occurs > > (say, when a Lua iterator generator returns more then two values). I > > Typo: s/then/than/. Thanks! Fixed. > > @@ -206,14 +216,10 @@ luaT_temp_luastate(int *coro_ref) > > * It is the other half of `luaT_temp_luastate()`. > > */ > > static void > > -luaT_release_temp_luastate(int coro_ref) > > +luaT_release_temp_luastate(struct lua_State *L, int coro_ref, int top) > > { > > - /* > > - * FIXME: The reusable fiber-local Lua state is not > > - * unreferenced here (coro_ref == LUA_REFNIL), but > > - * it must be truncated to its past top to prevent > > - * stack overflow. > > - */ > > + if (top >= 0) > > + lua_settop(L, top); > > luaL_unref(tarantool_L, LUA_REGISTRYINDEX, coro_ref); > > Minor: You can just either restore top value for fiber-local Lua state > or unreference Lua coroutine without restoring a pointer to its stack > top slot. As a result you need to preserve the top value only for the > first case (i.e. when the coro_ref is LUA_NOREF) and ignore the value > for all other cases. Are you propose the following? | if (top >= 0) | lua_settop(L, top); | else | luaL_unref(tarantool_L, LUA_REGISTRYINDEX, coro_ref); When I look into the code locally, it does not seem to be logical: why decision whether we should unreference a state should be made based on `top` value? Now the code seem to be logical for me: if `top` was saved, then we drop it. If reference was saved, we unreference it. Vlad even proposed to drop `top >= 0`, which works in fact, but Lua Reference Manual does not guarantee it (it was discussed within this mailing thread). > > +static int > > +lbox_check_merge_source_call_next(struct lua_State *L) > > +{ > > + assert(lua_gettop(L) == 1); > > + > > + /* > > + * Ensure that there is reusable temporary Lua stack. > > + * > > + * Note: It may be the same as L (and usually do). > > Minor: It would be nice to mention (at least for inquisitive persons) a > case when differs from the given for Lua-born fibers. No-no, it is always so for a Lua born fiber. It seems I should reword the comment in a more strict way and explain why I use explicitly despite the fact that it is always the same as . The new comment: | /* | * Ensure that there is a reusable temporary Lua stack. | * | * Note: It is the same as for a Lua born fiber (at | * least at the moment of writing), but it is the | * implementation detail and the test looks more clean | * when we don't lean on this fact. | */