wireguard-openbsd/lib/libcompiler_rt, branch jd/queueboosts WireGuard implementation for the OpenBSD kernel https://git.zx2c4.com/wireguard-openbsd/atom/lib/libcompiler_rt?h=jd%2Fqueueboosts 2020-04-04T22:11:36Z Implement __atomic_is_lock_free for powerpc. Needed because the 2020-04-04T22:11:36Z kettenis kettenis@openbsd.org 2020-04-04T22:11:36Z urn:sha1:004e15d6a27248383965a270534db07a47fde1ae architecture doesn't implement 64-bit atomic operations. This implementation is pessimistic and only flags naturally aligned operations up to and including 32-bit as lock free. tested by cwen@ ok gkoehler@ Update compiler-rt to 8.0.1. This is mostly a cosmetic update for us. 2019-09-29T17:53:57Z patrick patrick@openbsd.org 2019-09-29T17:53:57Z urn:sha1:220151357c8ea1d11791300a3ebd505967bb5615 Tested in snaps ok kettenis@ Make compiler-rt's __clear_cache() work on 32-bit ARM by making 2019-09-28T13:19:19Z patrick patrick@openbsd.org 2019-09-28T13:19:19Z urn:sha1:46dc7675d379b942787d1ea50b1017273002af8a it use sysarch(). From Josh Elsasser ok kettenis@ Update compiler-rt to 7.0.1. This is mostly a cosmetic update since 2019-01-29T22:35:43Z patrick patrick@openbsd.org 2019-01-29T22:35:43Z urn:sha1:64bb0428c690670660f6397e66ca194e512556ae most changes are guarded behind WIN32 or GCC ifdefs. The functional change is in emutls and AMD model checking. ok kettenis@ Use the same additional arch specific files for sparc64 as is done for mips 2018-12-25T09:20:49Z claudio claudio@openbsd.org 2018-12-25T09:20:49Z urn:sha1:e37d018de997f24864e475a305434b79e3e2b38b or aarch64. Needed to build userland with clang. OK visa@ kettenis@ Add quad-precision floating-point support routines for mips64. 2018-10-21T05:13:49Z visa visa@openbsd.org 2018-10-21T05:13:49Z urn:sha1:e0fb7a9cc24d8d3744a2055b87717f2bb8e0d238 OK kettenis@ Update to compiler-rt 6.0.0. 2018-09-18T20:21:40Z patrick patrick@openbsd.org 2018-09-18T20:21:40Z urn:sha1:3d56ebce3ebb9c8a172fe01306f9897260997ec2 tested by naddy@ ok kettenis@ On armv7, remove symbols that are also present in our libc. Because of the 2018-01-16T23:04:42Z kettenis kettenis@openbsd.org 2018-01-16T23:04:42Z urn:sha1:7448b33d2f632aeb82fb0db488c7848468c2061e EABI aliases these can lead to duplicate symbol errors, especially when mixing code compiled with gcc and clang. The intention is to revert this once we convert armv7 to hardfloat. ok patrick@, guenther@ Update to compiler-rt 5.0.1. 2017-12-26T20:59:44Z patrick patrick@openbsd.org 2017-12-26T20:59:44Z urn:sha1:c72a1c0960669bfd1eb87723f902bc05ee727707 ok kettenis@ In compiler-rt, a few assembler implementations for i386 floating point 2017-09-08T19:04:00Z naddy naddy@openbsd.org 2017-09-08T19:04:00Z urn:sha1:81c9d1cab52ead7d467511330c500b2cc3789b76 conversion functions use SSE2 instructions, but these are not guarded by #ifdef __SSE2__, and there is no implementation using general purpose registers. For these functions, use the generic C variants instead, otherwise they will cause SIGILL on older processors. From FreeBSD; ok deraadt@ mlarkin@ patrick@