On Tue, Jan 26, 2016 at 03:06:43PM +0100, Richard Biener wrote: > > Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk? > > Can you split out the non -fsanitize part? It is ok.
Ok, I've committed following patch: 2016-01-26 Jakub Jelinek <ja...@redhat.com> PR lto/69254 * lto-wrapper.c (merge_and_complain): Handle -fcilkplus. (append_compiler_options): Handle -fcilkplus. (append_linker_options): Ignore -fno-{openmp,openacc,cilkplus}. --- gcc/lto-wrapper.c.jj 2016-01-23 00:13:00.632019027 +0100 +++ gcc/lto-wrapper.c 2016-01-25 15:59:49.778877313 +0100 @@ -277,6 +293,7 @@ merge_and_complain (struct cl_decoded_op case OPT_fwrapv: case OPT_fopenmp: case OPT_fopenacc: + case OPT_fcilkplus: case OPT_fcheck_pointer_bounds: /* For selected options we can merge conservatively. */ for (j = 0; j < *decoded_options_count; ++j) @@ -505,6 +546,7 @@ append_compiler_options (obstack *argv_o case OPT_fwrapv: case OPT_fopenmp: case OPT_fopenacc: + case OPT_fcilkplus: case OPT_ftrapv: case OPT_fstrict_overflow: case OPT_foffload_abi_: @@ -558,6 +601,15 @@ append_linker_options (obstack *argv_obs ??? We fail to diagnose a possible mismatch here. */ continue; + case OPT_fopenmp: + case OPT_fopenacc: + case OPT_fcilkplus: + /* Ignore -fno-XXX form of these options, as otherwise + corresponding builtins will not be enabled. */ + if (option->value == 0) + continue; + break; + default: break; } Jakub