On Thu, Jan 31 2019, George Koehler <kern...@gmail.com> wrote: > My OpenBSD/macppc machine has built devel/llvm 7.0.1p0, but a pair of > bugs in clang 6 are still in clang 7. A moment ago, I reported these > bugs to llvm: > > - https://bugs.llvm.org/show_bug.cgi?id=40553 > wrong sizeof(long double) in 32-bit PowerPC NetBSD, OpenBSD > > - https://bugs.llvm.org/show_bug.cgi?id=40554 > misaligned float in PowerPC code causes SIGBUS in OpenBSD/macppc > > Each report includes an example program that reproduces the bug, and > a patch that tries to fix it. The long double bug is breaking > graphics/kdiagram on powerpc. (Most C++ ports use ports-gcc, but > x11/qt5/qt5.port.mk prefers ports-clang before ports-gcc, so ports > using Qt5 may suffer from clang bugs.) > > I am now using the attached diff in devel/llvm. I have packaged > kdiagram with this. My REVISION-main bump (from 0 to 1) conflicts > with recent changes in CVS. My patches might not be correct; someone > who knows llvm might be able to write better patches. > > I am using lang/gcc/8 as my ports-gcc; I have edited > - infrastructure/mk/gcc4.port.mk to use MODGCC4_VERSION?=8 > - devel/llvm/Makefile to use GCC_VER = 8.2.0 > These edits are not in the attached diff.
fwiw, you may have noticed that the "long double diff" has already been added both to base and ports-clang. For the other diff I don't know yet what to do. I don't have a powerpc machine at hand at the moment, maybe soon (tm). I see that you've sent a follow-up on https://bugs.llvm.org/show_bug.cgi?id=40554 POWER8 is not relevant for us, but may indeed be relevant for upstream since their PowerPC target seems to handle what look like different ISA to me. -- jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
signature.asc
Description: PGP signature
