[PATCH] D39831: [Driver] Make the use of relax relocations a per target option

2017-11-21 Thread Petr Hosek via Phabricator via cfe-commits
This revision was automatically updated to reflect the committed changes. Closed by commit rL318816: [Driver] Make the use of relax relocations a per target option (authored by phosek). Changed prior to commit: https://reviews.llvm.org/D39831?vs=123838&id=123870#toc Repository: rL LLVM http

Re: [PATCH] D39831: [Driver] Make the use of relax relocations a per target option

2017-11-21 Thread Rafael Avila de Espindola via cfe-commits
Petr Hosek writes: > That's the Fuchsia CMake cache file which is used to build Fuchsia > toolchain, it's not needed there anymore because Fuchsia Clang driver now > handles this. I haven't touched Clang's CMakeLists.txt which defines > the ENABLE_X86_RELAX_RELOCATIONS option. Oops, I guess I sh

Re: [PATCH] D39831: [Driver] Make the use of relax relocations a per target option

2017-11-21 Thread Petr Hosek via cfe-commits
That's the Fuchsia CMake cache file which is used to build Fuchsia toolchain, it's not needed there anymore because Fuchsia Clang driver now handles this. I haven't touched Clang's CMakeLists.txt which defines the ENABLE_X86_RELAX_RELOCATIONS option. On Tue, Nov 21, 2017 at 3:14 PM Rafael Avila de

Re: [PATCH] D39831: [Driver] Make the use of relax relocations a per target option

2017-11-21 Thread Rafael Avila de Espindola via cfe-commits
I am probably missing something, but the patch has -# This is a "Does your linker support it?" option that only applies -# to x86-64 ELF targets. All Fuchsia target linkers do support it. -# For x86-64 Linux, it's supported by LLD and by GNU linkers since -# binutils 2.27, so one can hope that al

Re: [PATCH] D39831: [Driver] Make the use of relax relocations a per target option

2017-11-21 Thread Petr Hosek via cfe-commits
I kept the CMake option, so by default the driver will use the value set through CMake as before but individual targets can now set their platform default. On Tue, Nov 21, 2017 at 1:30 PM Rafael Avila de Espindola < rafael.espind...@gmail.com> wrote: > Petr Hosek via Phabricator writes: > > > >

Re: [PATCH] D39831: [Driver] Make the use of relax relocations a per target option

2017-11-21 Thread Rafael Avila de Espindola via cfe-commits
Petr Hosek via Phabricator writes: > -# This is a "Does your linker support it?" option that only applies > -# to x86-64 ELF targets. All Fuchsia target linkers do support it. > -# For x86-64 Linux, it's supported by LLD and by GNU linkers since > -# binutils 2.27, so one can hope that all Linu

[PATCH] D39831: [Driver] Make the use of relax relocations a per target option

2017-11-21 Thread Roland McGrath via Phabricator via cfe-commits
mcgrathr accepted this revision. mcgrathr added a comment. This revision is now accepted and ready to land. LGTM Repository: rL LLVM https://reviews.llvm.org/D39831 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi

[PATCH] D39831: [Driver] Make the use of relax relocations a per target option

2017-11-21 Thread Petr Hosek via Phabricator via cfe-commits
phosek updated this revision to Diff 123838. phosek added a comment. Added a test case. Repository: rL LLVM https://reviews.llvm.org/D39831 Files: cmake/caches/Fuchsia-stage2.cmake include/clang/Driver/ToolChain.h lib/Driver/ToolChain.cpp lib/Driver/ToolChains/Clang.cpp lib/Driver/

[PATCH] D39831: [Driver] Make the use of relax relocations a per target option

2017-11-21 Thread Roland McGrath via Phabricator via cfe-commits
mcgrathr added a comment. test case? Repository: rL LLVM https://reviews.llvm.org/D39831 ___ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

[PATCH] D39831: [Driver] Make the use of relax relocations a per target option

2017-11-09 Thread Petr Hosek via Phabricator via cfe-commits
phosek created this revision. Herald added a subscriber: mgorny. The support for relax relocations is dependent on the linker and different toolchains within the same compiler can be using different linkers some of which may or may not support relax relocations. Give toolchains the option