From mboxrd@z Thu Jan 1 00:00:00 1970 From: Piyush Pangtey Subject: Re: [RFC] nft: Remove memory-leak Date: Mon, 14 Mar 2016 14:18:57 +0530 Message-ID: <56E67AF9.1030105@gmail.com> References: <20160313120843.GA6919@fate> <20160313202256.GD1548@breakpoint.cc> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Cc: netfilter-devel@vger.kernel.org To: Florian Westphal Return-path: Received: from mail-pf0-f175.google.com ([209.85.192.175]:34896 "EHLO mail-pf0-f175.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755315AbcCNIsn (ORCPT ); Mon, 14 Mar 2016 04:48:43 -0400 Received: by mail-pf0-f175.google.com with SMTP id n5so93462775pfn.2 for ; Mon, 14 Mar 2016 01:48:42 -0700 (PDT) In-Reply-To: <20160313202256.GD1548@breakpoint.cc> Sender: netfilter-devel-owner@vger.kernel.org List-ID: On Monday 14 March 2016 01:52 AM, Florian Westphal wrote: > Piyush Pangtey wrote: >> Memory leak due to the absence of matching free : >> ==4797== 189 bytes in 27 blocks are definitely lost in loss record 22 of 51 >> ==4797== at 0x4C2AB80: malloc (in >> /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so) >> ==4797== by 0x57A3839: strdup (strdup.c:42) >> ==4797== by 0x41C00D: xstrdup (utils.c:64) >> ==4797== by 0x411E4B: netlink_delinearize_chain.isra.3 (netlink.c:717) >> ==4797== by 0x411F20: list_chain_cb (netlink.c:748) >> ==4797== by 0x504A943: nft_chain_list_foreach (chain.c:1015) >> ==4797== by 0x41455E: netlink_list_chains (netlink.c:771) >> ==4797== by 0x4078EF: cache_init_objects (rule.c:90) >> ==4797== by 0x4078EF: cache_init (rule.c:130) >> ==4797== by 0x4078EF: cache_update (rule.c:147) >> ==4797== by 0x40FB09: cmd_evaluate (evaluate.c:2475) >> ==4797== by 0x4296BC: nft_parse (parser_bison.y:655) >> ==4797== by 0x4064CC: nft_run (main.c:231) >> ==4797== by 0x42CE57: cli_complete (cli.c:124) > .. >> --- a/src/netlink.c >> +++ b/src/netlink.c >> @@ -714,12 +714,12 @@ static struct chain *netlink_delinearize_chain(struct netlink_ctx *ctx, >> chain->priority = >> nftnl_chain_get_s32(nlc, NFTNL_CHAIN_PRIO); >> chain->type = >> - xstrdup(nftnl_chain_get_str(nlc, NFTNL_CHAIN_TYPE)); >> + nftnl_chain_get_str(nlc, NFTNL_CHAIN_TYPE); > > This looks wrong, nlc lifetime is not tied to chain object. > Seems chain_free() needs xfree calls for type and ->dev ? > Yes.Thanks for commenting. I tried this as a workaround, because I was not sure about nlc :). I sent a patch for adding matching free calls to chain_free().