----- Original Message ----- > From: "Ulrich Weigand" <[email protected]> > To: "Hal Finkel" <[email protected]> > Cc: "llvm cfe" <[email protected]>, "Chandler Carruth" > <[email protected]>, "Commit Messages and Patches for > LLVM" <[email protected]>, "Rafael Espíndola" > <[email protected]>, "Bill Schmidt" > <[email protected]> > Sent: Thursday, January 29, 2015 1:12:38 PM > Subject: Re: [PATCH] Work around ppc64 TLS linker bug (Re: [LPM] Try again to > appease powerpc64 in its self host.) > > Hal Finkel <[email protected]> wrote on 29.01.2015 19:37:31: > > > > I'm not completely sure if this is the correct place to add this, > > > but the patch seems to work for me. With this clang patch, I'm > > > able to successfully bootstrap clang again (with the initial-exec > > > TLS attribute in Chandler's latest change removed again). > > > > > > OK to check in? > > > > > > (See attached file: clang-ppc64-notlsopt.patch) > > > > Yes, LGTM. We do need a test case, however (I think that something > > like test/Driver/Xlinker-args.c is essentially what is needed). > > I've added a check to test/Driver/ppc-features.cpp which did already > check for linker options on PowerPC. Checked in as r227480. > > Chandler, it would probably be best to remove the initial-exec > TLS model attribute again now that we use --no-tls-optimize ...
Feel free to do this yourself too ;) -Hal > > Bye, > Ulrich > > -- Hal Finkel Assistant Computational Scientist Leadership Computing Facility Argonne National Laboratory _______________________________________________ cfe-commits mailing list [email protected] http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
