Hi Peter, On Wed, May 22, 2013 at 6:26 PM, Peter Maydell <peter.mayd...@linaro.org> wrote: > On 22 May 2013 08:47, Edgar E. Iglesias <edgar.igles...@gmail.com> wrote: >> On Wed, May 22, 2013 at 11:50:22AM +1000, peter.crosthwa...@xilinx.com wrote: >>> From: Peter Crosthwaite <peter.crosthwa...@xilinx.com> >>> Currently QEMU provides a local clone of the file libfdt_env.h in >>> /include. This file is supposed to come with the libfdt package and is >>> only needed for broken installs of libfdt. Unfortunately, libfdt 1.3 >>> stable is tagged with this bug so we cant realistically remove the >>> clone. So instead, use a configure time probe - check to see if the >>> host is already providing this file, and if so, don't use QEMUs >>> local clone of libfdt.h. >>> >>> Moved include/libfdt_env.h to include/libfdt/libfdt_env.h so it can be >>> treated as a special case. >>> >>> For newer (development) versions of libfdt this probe will succeed and >>> include/libfdt/libfdt_env.h will simply be ignored. > >> Should we consider removing our builtin version and stick to either >> using the hosts or when that fails suggest the submoduled dtc pkg? > > I agree that that sounds like a better idea: just add the #include > to our existing fdt probe, and then we'll simply ignore system libfdt > if it's not new enough. The relevant changes got committed to libfdt > git in 2008, after all.... >
I'm happy with this respin assuming Peters dtc-is-compulsory series goes through before. Otherwise people with dtc 1.3.0 installed will get cop a silent failure when qemu borks on their previously-working system 1.3.0 and falls backs to just building with no dtc at all. I'll remake in the meantime. Regards, Peter > thanks > -- PMM >
