Source-Changes-HG archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

[src/trunk]: src/share/mk Build aarch64 modules without fp or simd instructions.



details:   https://anonhg.NetBSD.org/src/rev/ca7a478fa4d5
branches:  trunk
changeset: 846801:ca7a478fa4d5
user:      jmcneill <jmcneill%NetBSD.org@localhost>
date:      Sun Dec 01 20:24:47 2019 +0000

description:
Build aarch64 modules without fp or simd instructions.

diffstat:

 share/mk/bsd.kmodule.mk |  6 ++++--
 1 files changed, 4 insertions(+), 2 deletions(-)

diffs (20 lines):

diff -r da73cce824f7 -r ca7a478fa4d5 share/mk/bsd.kmodule.mk
--- a/share/mk/bsd.kmodule.mk   Sun Dec 01 19:21:13 2019 +0000
+++ b/share/mk/bsd.kmodule.mk   Sun Dec 01 20:24:47 2019 +0000
@@ -1,4 +1,4 @@
-#      $NetBSD: bsd.kmodule.mk,v 1.62 2019/07/05 08:28:16 hannken Exp $
+#      $NetBSD: bsd.kmodule.mk,v 1.63 2019/12/01 20:24:47 jmcneill Exp $
 
 # We are not building this with PIE
 MKPIE=no
@@ -42,7 +42,9 @@
 # The real solution to this involves generating trampolines for those
 # relocations inside the loader and removing this workaround, as the
 # resulting code would be much faster.
-.if ${MACHINE_CPU} == "arm"
+.if ${MACHINE_CPU} == "aarch64"
+CFLAGS+=       -march=armv8-a+nofp+nosimd
+.elif ${MACHINE_CPU} == "arm"
 CFLAGS+=       -fno-common -fno-unwind-tables
 .elif ${MACHINE_CPU} == "hppa"
 CFLAGS+=       -mlong-calls



Home | Main Index | Thread Index | Old Index