[Libclc-dev] [PATCH 1/1] configure: drop -fno-builtin

[Matt Arsenault via Libclc-dev]

> On Jun 24, 2016, at 12:15, Jan Vesely via Libclc-dev <libclc-dev at lists.llvm.org> wrote:
> 
> We are using builtins.
> Some builtins (copysign, fmax, fmin, ...) are translated to lirbary calls.
> Using -fno-builtin makes those calls tagged nobuiltin and starting with 273039,
> they are no longer translated to llvm opcodes, resulting in failed function call;
> "<unknown>:0:0: in function test_1_copysign_float void (float addrspace(1)*, float addrspace(1)*, float addrspace(1)*): unsupported call to function copysignf"
> Fixes: r273039
> 
> Signed-off-by: Jan Vesely <jan.vesely at rutgers.edu>
> ---
> The alternative is to keep -fno-builtin and avoid using some __builtins.
> 
> configure.py | 1 -
> 1 file changed, 1 deletion(-)
> 
> diff --git a/configure.py b/configure.py
> index b96faee..d3a3977 100755
> --- a/configure.py
> +++ b/configure.py
> @@ -181,7 +181,6 @@ for target in targets:
>   for device in available_targets[target]['devices']:
>     # The rule for building a .bc file for the specified architecture using clang.
>     clang_bc_flags = "-target %s -I`dirname $in` %s " \
> -                     "-fno-builtin " \
>                      "-D__CLC_INTERNAL " \
>                      "-emit-llvm" % (target, clang_cl_includes)
>     if device['gpu'] != '':
> -- 
> 2.7.4
> 
> _______________________________________________
> Libclc-dev mailing list
> Libclc-dev at lists.llvm.org
> http://lists.llvm.org/cgi-bin/mailman/listinfo/libclc-dev

I think we should continue using it, but fix the builtins we expect to really be emitting intrinsics: http://reviews.llvm.org/D5896 <http://reviews.llvm.org/D5896>

-Matt
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/libclc-dev/attachments/20160624/5d0b9c93/attachment.html>