This is a prerequisite for proper runtime support on FreeBSD as it will
allow such a runtime to be compiled.
Details
- Reviewers
arichardson emaste dim pkubaj nemanjai
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
Could you add a RUN: line to `clang/test/CodeGenCXX/float128-declarations.cpp? Code LGTM.
// RUN: %clang_cc1 -no-opaque-pointers -emit-llvm -triple x86_64-unknown-freebsd -std=c++11 \ // RUN: %s -o - | FileCheck %s -check-prefix=CHECK-X86
OK. I think the switch will become unwieldy if additional arch-dependent things are added but could be dealt with then.
Could you modify it appropriately to also enable ppc, ppc64 and ppc64le? I checked whether it works on FreeBSD 13.1-RELEASE and it builds and works fine.
EDIT: or better just go for now with ppc64le, since enabling for older variants should probably be done depending on the hardware.
My comments are minor nits that don't require another review, so LGTM.
clang/lib/Basic/Targets/OSTargets.h | ||
---|---|---|
245–250 | Seems a bit odd to have a separate switch on the same value for __float128 rather than using fallthrough. | |
clang/test/CodeGenCXX/float128-declarations.cpp | ||
5–6 | Now that you added ppc64le, did you mean to also add a line with that triple without the explicit -target-feature? |
Looks like this still needs someone to push it -- @brooks if you rebase and send to GitHub I cna pick it up
The submitted patch https://github.com/llvm/llvm-project/commit/23c47eba879769a29772c999be2991201c2fe399 was not the same since it omitted ppc64. So I guess this should remain open
Seems a bit odd to have a separate switch on the same value for __float128 rather than using fallthrough.