From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mailman by lists.gnu.org with tmda-scanned (Exim 4.43) id 1MTrj4-0002vC-7N for qemu-devel@nongnu.org; Thu, 23 Jul 2009 02:24:22 -0400 Received: from exim by lists.gnu.org with spam-scanned (Exim 4.43) id 1MTriy-0002v0-Hv for qemu-devel@nongnu.org; Thu, 23 Jul 2009 02:24:20 -0400 Received: from [199.232.76.173] (port=47440 helo=monty-python.gnu.org) by lists.gnu.org with esmtp (Exim 4.43) id 1MTriy-0002ux-C6 for qemu-devel@nongnu.org; Thu, 23 Jul 2009 02:24:16 -0400 In-Reply-To: (malc's message of "Thu\, 23 Jul 2009 07\:28\:16 +0400 \(MSD\)") References: <4A67594C.7080908@codemonkey.ws> <200907222333.21770.paul@codesourcery.com> From: Juan Quintela Date: Thu, 23 Jul 2009 08:10:00 +0200 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Subject: [Qemu-devel] Re: [PATCH 00/37] Generate a proper LIBS variable List-Id: qemu-devel.nongnu.org List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: malc Cc: Paolo Bonzini , Paul Brook , qemu-devel@nongnu.org malc wrote: > On Thu, 23 Jul 2009, Juan Quintela wrote: > >> Paul Brook wrote: >> >> >> I'd rather see the us generate a config-host.ld that contained the >> >> >> output of ld --verbose -v | grep SEARCH_DIR and then do: >> >> > >> >> > Only for x86_64 or all architectures? Once here. >> >> >> >> For all I would think. My only concern is that we'll break non-GNU ld's >> >> but we'll cross that bridge when we come to it. >> > >> > We should only be using linker scripts for the userspace emulation. >> >> State of afairs: >> a- we use linker scripts for linux-user and bsd-user >> b- we don't use linker script on darwin-user >> c- ia64 uses linker script for target-softmmu (only arch at all) >> >> The reason for changing the SEARCH_PATH is because: >> a- We changed to use a LIBS variable for everything >> b- In x86_64, it is common to have locally compiled libraries in >> /usr/local/lib not /usr/local/lib64 (case of libvdeplug.so >> for Anthony. >> >> What to do? >> a- revert back to use diferent LIBS variable for each binary > > a Ok, we have to questions here: a- should we support other SEARCH_PATHS in linker scripts b- should we use a diffrente LIBS variable for target-softmmu and *-user I think that we definitely want "a". I tend to think that using the same libraries for all binaries is a good idea, but I think we shouldn't confuse the isues. Please comment in my series to use the default SEARCH_PATH of the linker on the linker script. Otherwise, is there any reason to have that different SEARCH_PATH? Later, Juan. >> b- change the linker scripts search path, and don't allow ever >> user compiled libraries when using a linker-script. >> >> Thanks for the input about the liker scripts use. >> >> Later, Juan. >> >>