From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:48354) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1XX9So-000259-9g for qemu-devel@nongnu.org; Thu, 25 Sep 2014 09:52:15 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1XX9Sf-0002V9-2i for qemu-devel@nongnu.org; Thu, 25 Sep 2014 09:52:06 -0400 Received: from mail-we0-x22c.google.com ([2a00:1450:400c:c03::22c]:47626) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1XX9Se-0002Tj-Rw for qemu-devel@nongnu.org; Thu, 25 Sep 2014 09:51:56 -0400 Received: by mail-we0-f172.google.com with SMTP id p10so7565775wes.3 for ; Thu, 25 Sep 2014 06:51:50 -0700 (PDT) Sender: Paolo Bonzini Message-ID: <54241DF0.8050001@redhat.com> Date: Thu, 25 Sep 2014 15:51:44 +0200 From: Paolo Bonzini MIME-Version: 1.0 References: <1411632395-10516-1-git-send-email-olaf@aepfle.de> <20140925104954.GE19882@stefanha-thinkpad.redhat.com> In-Reply-To: <20140925104954.GE19882@stefanha-thinkpad.redhat.com> Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 8bit Subject: Re: [Qemu-devel] [PATCH] build: preserve debug symbols with --enable-debug-info List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Stefan Hajnoczi , Olaf Hering Cc: Peter Maydell , Cole Robinson , Michael Tokarev , qemu-devel@nongnu.org, Stefan Weil Il 25/09/2014 12:49, Stefan Hajnoczi ha scritto: > On Thu, Sep 25, 2014 at 10:06:35AM +0200, Olaf Hering wrote: >> During code review for xen I noticed that --enable-debug-info >> would still strip the binaries because strip_opt= defaults to >> yes. If --enable-debug-info is passed to configure it has to be >> assumed that not only the compiled binaries have debugsymbols, >> also the installed binaries should keep the symbols. The >> requirement to pass also --disable-strip looks odd. > > Perhaps package maintainers rely on installed binaries not having > debug symbols? If so, that should be taken care of by the distribution. Of course, a distribution is free to separate the debug info and ship it as a separate package; in that case, it makes sense to distribute stripped binaries. But I think discarding symbols on "make install" is in general a bad idea, especially for long-lived processes such as QEMU where you often have non-reproducible bugs. If symbols are gone, even the simplest bug becomes basically impossible to diagnose from a core dump. The GNU Makefile standards have "make install" and "make install-strip" targets. It would be nice to add "make install-strip" and at the same time flip the default from --enable-strip to --disable-strip. Paolo