Received: from eggs.gnu.org ([2001:4830:134:3::10]:44881) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1XkDrv-0006Zb-BZ for qemu-devel@nongnu.org; Fri, 31 Oct 2014 11:12:48 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1Xk6qJ-0007FQ-1N for qemu-devel@nongnu.org; Fri, 31 Oct 2014 03:41:59 -0400 Message-ID: <54533D41.6010704@msgid.tls.msk.ru> Date: Fri, 31 Oct 2014 10:41:53 +0300 From: Michael Tokarev MIME-Version: 1.0 References: <5453178B.2040504@huawei.com> <54533742.1000109@msgid.tls.msk.ru> <54533B60.2080008@huawei.com> In-Reply-To: <54533B60.2080008@huawei.com> Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Subject: Re: [Qemu-devel] [PATCH 3/4] pidfile: stop making pidfile error a special case List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Gonglei Cc: "qemu-trivial@nongnu.org" , "qemu-devel@nongnu.org" , Markus Armbruster 31.10.2014 10:33, Gonglei wrote: [] >>>> While at it, fix wrong identation in os_daemonize(). >>> >>> s/identation/identification/ >> >> No, the original word was the right one. > > Sorry, I can't find 'identation' both dictionary and Google. > Your meaning 'indentation'? Damn. I missed that one. It is "indentation" inedeed :) (but not identification). [] >>>> do { >>>> len = read(fds[0], &status, 1); >>>> } while (len < 0 && errno == EINTR); >>>> - if (len != 1) { >>>> - exit(1); >>>> - } >>> >>> Does this check need to be removed? >> >> Yes, because... >>> >>>> - else if (status == 1) { >>>> - fprintf(stderr, "Could not acquire pidfile\n"); >>>> - exit(1); >>>> - } else { >>>> - exit(0); >>>> - } >>>> - } else if (pid < 0) { >>>> - exit(1); >>>> - } >>>> + >>>> + exit(len == 1 && status == 0 ? 0 : 1); >> >> ...it is checked here, note the 'len == 1' part of the condition. > > If len != 1, the original code exit with 1, after your changes, > it will exit with 0. Right? with len != 1, the condition 'len == 1 && status == 0 ? 0 : 1' evaluates to 1. Maybe I can add a comment here: + /* only exit successfully if our child actually + * wrote a one-byte zero to our pipe */ + exit(len == 1 && status == 0 ? 0 : 1); See the result at http://git.corpit.ru/?p=qemu.git;a=shortlog;h=refs/heads/trivial-patches-next Thanks, /mjt