From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga11.intel.com (mga11.intel.com [192.55.52.93]) by dpdk.org (Postfix) with ESMTP id A2BC42A5E for ; Wed, 8 Jul 2015 12:35:20 +0200 (CEST) Received: from fmsmga001.fm.intel.com ([10.253.24.23]) by fmsmga102.fm.intel.com with ESMTP; 08 Jul 2015 03:35:19 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.15,431,1432623600"; d="scan'208";a="742876987" Received: from bricha3-mobl3.ger.corp.intel.com ([10.237.208.63]) by fmsmga001.fm.intel.com with SMTP; 08 Jul 2015 03:35:18 -0700 Received: by (sSMTP sendmail emulation); Wed, 08 Jul 2015 11:35:16 +0025 Date: Wed, 8 Jul 2015 11:35:16 +0100 From: Bruce Richardson To: "Mcnamara, John" Message-ID: <20150708103516.GA10744@bricha3-MOBL3> References: <1433492166-30758-1-git-send-email-jingjing.wu@intel.com> <1433917473-21508-2-git-send-email-jingjing.wu@intel.com> <9BB6961774997848B5B42BEC655768F8C59FF9@SHSMSX104.ccr.corp.intel.com> <2821398.avzWtSOKrm@xps13> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: Organization: Intel Shannon Ltd. User-Agent: Mutt/1.5.23 (2014-03-12) Cc: "dev@dpdk.org" Subject: Re: [dpdk-dev] [PATCH v4 1/4] ethdev: rename rte_eth_vmdq_mirror_conf X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 08 Jul 2015 10:35:21 -0000 On Wed, Jul 08, 2015 at 10:31:15AM +0000, Mcnamara, John wrote: > > -----Original Message----- > > From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Thomas Monjalon > > Sent: Tuesday, July 7, 2015 3:51 PM > > To: Wu, Jingjing > > Cc: dev@dpdk.org > > Subject: Re: [dpdk-dev] [PATCH v4 1/4] ethdev: rename > > rte_eth_vmdq_mirror_conf > > > > > Additional, about the validate-abi.sh, does it mean we need to fix all > > > the problems it reports? Or we can decide case by case. > > > Can a Low Severity problem be acceptable? > > > > We have to decide case by case. > > It makes ABI checking impossible to automate. > > That's why any help is welcome to check the git HEAD for ABI violation. > > Hi, > > Automated testing for ABI breaks is tricky since it requires some interpretation of the errors and the severity level. > > However, each report file contains a 2 line summary at the top that can be read and parsed. For example (with long lines wrapped): > > head -2 compat_reports/librte_mbuf.so/v2.0.0_to_ABI_CHECK/compat_report.html > > > > > This still requires interpretation but it can be used to search for categories of incompatibility. > > To test if a patchset breaks the API it is possible to do something like the following. Say, for example that you have committed 3 commits to your local master, you could tag and run the checker like this: > > git checkout master > > git tag -f ABI_CHECK_AFTER > git tag -f ABI_CHECK_BEFORE HEAD~3 > > ./scripts/validate-abi.sh ABI_CHECK_BEFORE ABI_CHECK_AFTER x86_64-native-linuxapp-gcc > > grep -lr "kind:binary;verdict:incompatible" compat_reports/ > > The grep could be extended to match other categories that you are interested in. However, it is still probably better to examine the reports manually. > > You can also use something this to run a git bisect to find a commit that introduced an ABI incompatibility: > > # Tag the head and some known good point in the history. > git checkout master > git tag -f ABI_CHECK_END > git tag -f ABI_CHECK_START v2.0.0 > > # Start git bisect with bad and good. > git bisect start ABI_CHECK_END ABI_CHECK_START > > # Run bisect with a script. > git bisect run ~/abi_bisect.sh > > Where the ~/abi_bisect.sh script that searches for ABI incompatibilities looks like this: > > #!/bin/sh > > git tag -f ABI_CHECK_END > > rm -rf compat_reports/ > > ./scripts/validate-abi.sh ABI_CHECK_START ABI_CHECK_END x86_64-native-linuxapp-gcc > > ! grep -lr "kind:binary;verdict:incompatible" compat_reports/ > > John. > -- I'd mod this post up as "+1 informative", except that this isn't slashdot. :-) Very helpful, John. Thanks. /Bruce