wireguard-openbsd/sys/lib/libkern/arch, branch jd/queueboosts WireGuard implementation for the OpenBSD kernel https://git.zx2c4.com/wireguard-openbsd/atom/sys/lib/libkern/arch?h=jd%2Fqueueboosts 2020-06-10T20:19:29Z Provide an optimized implementation of ffs(3) in the kernel on 2020-06-10T20:19:29Z naddy naddy@openbsd.org 2020-06-10T20:19:29Z urn:sha1:d41240fab998b3c42f736af0544339ebedba7e73 arm64/powerpc/powerpc64, making use of the count leading zeros instruction. powerpc testing by cwen@; ok kettenis@ deraadt@ Give some END()s to assembly symbols. 2019-11-05T08:16:43Z mpi mpi@openbsd.org 2019-11-05T08:16:43Z urn:sha1:6abd9ea796a8bef365c616b03174fbc2051daf73 ok kettenis@, jca@ Add movstrSI12_i4 for linking -Os/-Oz kernels. Sadly however it 2019-04-25T21:47:51Z deraadt deraadt@openbsd.org 2019-04-25T21:47:51Z urn:sha1:b473154e22060a3008dfa99fb077cc8885f89747 fails and locks early in boot. Use _ALIGN_TRAPS instead of _ALIGN_TEXT where nops aren't needed. 2018-07-13T05:56:57Z guenther guenther@openbsd.org 2018-07-13T05:56:57Z urn:sha1:b8dcfd0b0b268c3e88ec6096e5493ada637efb37 ok deraadt@ In asm.h ensure NENTRY uses the old-school nop-sled align, but change standard 2018-07-10T16:01:26Z deraadt deraadt@openbsd.org 2018-07-10T16:01:26Z urn:sha1:a5910b4fbcaa3dd48faff38aeadcb22c953af71e ENTRY is a trapsled. Fix a few functions which fall-through into an ENTRY macro. amd64 binaries now are free of double+-nop sequences (except for one assember nit in aes-586.pl). Previous changes by guenther got us here. ok mortimer kettenis Add retguard macros for libkern 2018-07-03T23:22:48Z mortimer mortimer@openbsd.org 2018-07-03T23:22:48Z urn:sha1:1d66f0a004efa322e2047d0371a4cfe69dca0a66 ok deraadt Add aliases for __aeabi_idivmod and __aeabi_uidivmod calls clang emits. 2017-10-26T15:23:59Z jsg jsg@openbsd.org 2017-10-26T15:23:59Z urn:sha1:e20faded7ddad8c25e311d87284042fc4f8d89df Remainder of the division is already stored in r1 in __divsi3/__udivsi3. ok kettenis@ Fix the __sync_* functions to make them work properly 2017-08-20T11:12:42Z visa visa@openbsd.org 2017-08-20T11:12:42Z urn:sha1:ec223ac0f2a20d15b0aa7691596d53012f339a5a as full memory barriers. Enable radeondrm(4) on loongson to get accelerated graphics 2017-05-21T13:00:53Z visa visa@openbsd.org 2017-05-21T13:00:53Z urn:sha1:77e35563a4c16e39b4b3a3bf8bdfcf7309c73848 with the RS780E chipset. OK kettenis@, jsg@ Copy data by using 64-bit loads and stores rather than 32-bit 2016-11-14T15:02:54Z visa visa@openbsd.org 2016-11-14T15:02:54Z urn:sha1:731d7aef1d1f7c62999ecf3015de68f1f64e6d5f operations. This about doubles the routine's throughput. No binary change on the 32-bit bootblocks Discussed with miod@ long ago