On Fri, Apr 06, 2012 at 11:18:20AM +0200, René Ladan wrote: > 2012/4/6 Lars Engels <[email protected]>: > > On Mon, Apr 02, 2012 at 10:46:23AM +0200, René Ladan wrote: > >> 2012/4/2 Xin Li <[email protected]>: > >> > On 03/18/12 13:11, Lars Engels wrote: > >> >> Build of Chromium 17.0.963.79 fails to build on r233121, amd64: > >> >> > >> >> CXX(target) > >> >> out/Release/obj.target/webkit_gpu/webkit/gpu/webgraphicscontext3d_in_process_impl.o > >> >> > >> >> > >> > LINK(host) out/Release/protoc > >> >> AR(host) out/Release/obj.host/v8/tools/gyp/libv8_nosnapshot.a RULE > >> >> sync_proto_genproto_0 > >> >> out/Release/pyproto/chrome/browser/sync/protocol/sync_pb2.py gmake: > >> >> *** [out/Release/pyproto/chrome/browser/sync/protocol/sync_pb2.py] > >> >> Floating point exception: 8 (core dumped) gmake: *** Waiting for > >> >> unfinished jobs.... *** [do-build] Error code 1 > >> >> > >> >> Stop in /usr/ports/www/chromium. *** [build] Error code 1 > >> > > >> > Hit the same thing and compiling with GCC46 either works around it, or > >> > fixed the problem. Haven't investigated the underlying issue though. > >> > > >> Fixed probably, chromium also builds with top-of-tree clang (from > >> upstream) with > >> http://trillian.chruetertee.ch/chromium/changeset/318 applied. > > > > Unfortunately it's still not fixed for me. Building with CLANG: > > > > LINK(host) out/Release/protoc > > LINK(host) out/Release/protoc: Finished > > AR(host) out/Release/obj.host/v8/tools/gyp/libv8_nosnapshot.a > > RULE sync_proto_genproto_0 > > out/Release/pyproto/chrome/browser/sync/protocol/app_notification_specifics_pb2.py > > gmake: *** > > [out/Release/pyproto/chrome/browser/sync/protocol/app_notification_specifics_pb2.py] > > Floating point exception: 8 (core dumped) > > > > # clang -v > > FreeBSD clang version 3.0 (tags/RELEASE_30/final 145349) 20111210 > > Target: x86_64-unknown-freebsd10.0 > > Thread model: posix > > > Your clang is probably too old, clang/llvm from upstream trunk worked for me. > The clang in base contains a bug which miscompiles protoc resulting in > the SIGFPE even > with the patch added in the last commit.
Yes, that maybe. FWIW I succeeded building chromium with gcc-4.6 now. :)
pgptSP9TnV34L.pgp
Description: PGP signature
