From: Pablo Neira Ayuso Subject: Re: [PATCH] extensions: libxt_mark: Fix inversion code Date: Wed, 23 Dec 2015 11:09:26 +0100 Message-ID: <20151223100926.GB1952@salvia> References: <20151222212137.GA15975@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: netfilter-devel@vger.kernel.org To: Shivani Bhardwaj Return-path: Received: from mail.us.es ([193.147.175.20]:42706 "EHLO mail.us.es" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755293AbbLWKJi (ORCPT ); Wed, 23 Dec 2015 05:09:38 -0500 Received: from antivirus1-rhel7.int (unknown [192.168.2.11]) by mail.us.es (Postfix) with ESMTP id 3A4D810214B for ; Wed, 23 Dec 2015 11:09:37 +0100 (CET) Received: from antivirus1-rhel7.int (localhost [127.0.0.1]) by antivirus1-rhel7.int (Postfix) with ESMTP id 27C77EBAE7 for ; Wed, 23 Dec 2015 11:09:37 +0100 (CET) Received: from antivirus1-rhel7.int (localhost [127.0.0.1]) by antivirus1-rhel7.int (Postfix) with ESMTP id AA594DA843 for ; Wed, 23 Dec 2015 11:09:31 +0100 (CET) Content-Disposition: inline In-Reply-To: <20151222212137.GA15975@gmail.com> Sender: netfilter-devel-owner@vger.kernel.org List-ID: On Wed, Dec 23, 2015 at 02:51:37AM +0530, Shivani Bhardwaj wrote: > Fix the code associated with invert flag. > > Examples: > > $ sudo iptables-translate -I INPUT -p tcp -m mark ! --mark 0xa/0xa > nft insert rule ip filter INPUT ip protocol tcp mark and 0xa != 0xa counter > > $ sudo iptables-translate -I INPUT -p tcp -m mark ! --mark 0x1 > nft insert rule ip filter INPUT ip protocol tcp mark != 0x1 counter > > Signed-off-by: Shivani Bhardwaj > --- > extensions/libxt_mark.c | 26 ++++++++++++++++++-------- > 1 file changed, 18 insertions(+), 8 deletions(-) > > diff --git a/extensions/libxt_mark.c b/extensions/libxt_mark.c > index 3b96a34..3beccc8 100644 > --- a/extensions/libxt_mark.c > +++ b/extensions/libxt_mark.c > @@ -1,5 +1,6 @@ > #include > #include > +#include > #include > #include > > @@ -103,13 +104,14 @@ mark_save(const void *ip, const struct xt_entry_match *match) > } I suggest you add this to xtables.h: enum { XT_OP_EQ, XT_OP_NEQ, XT_OP_MAX }; > static void > -print_mark_xlate(struct xt_buf *buf, > - unsigned int mark, unsigned int mask) > +print_mark_xlate(struct xt_buf *buf, unsigned int mark, > + unsigned int mask, const char *str) > { > if (mask != 0xffffffffU) > - xt_buf_add(buf, " and 0x%x == 0x%x ", mark, mask); > + xt_buf_add(buf, " and 0x%x %s 0x%x ", mark, str, mask); > else > - xt_buf_add(buf, " 0x%x ", mark); > + xt_buf_add(buf, " %s0x%x ", > + !strcmp(str, "==") ? "" : "!= ", mark); So instead of passing 'str' as parameter, you can pass a uint32_t op (enums are always guaranteed to be uint32_t). Then xt_buf_add(buf, " %s0x%x ", op == XT_OP_EQ ? "" : "!=", mark); > static int > @@ -117,9 +119,13 @@ mark_mt_xlate(const struct xt_entry_match *match, > struct xt_buf *buf, int numeric) > { > const struct xt_mark_mtinfo1 *info = (const void *)match->data; > + const char *str = "=="; This code will be nicer using the approach I'm suggesting above. Please, apply this suggestion to all your pending patches and resubmit. Thanks Shivani.