Re: NetBSD-7 build issues with earmv7hf

[Sad Clouds <cryintothebluesky%googlemail.com@localhost>]

On Fri, 8 Jul 2016 07:35:04 +0000 (UTC)
John Klos <john%ziaspace.com@localhost> wrote:

> > /bin/sh ./build.sh -j 1 -u -m evbarm -a earmv7hf
> > and this fails with the errors below. Any ideas what may be wrong
> > here?
> >
> > #   compile  BCM5301X/vfp_init.o
> > /opt/tools.evbarm/bin/armv7--netbsdelf-eabihf-gcc -mfpu=vfp
> > -mfloat-abi=soft -ffreestanding -fno-zero-initialized-in-bss -pipe
> > -fno-strict-aliasing -fno-common -mfpu=vfp -std=gnu99 -Werror -Wall
> > -Wno-main -Wno-format-zero-length -Wpointer-arith
> > -Wmissing-prototypes -Wstrict-prototypes -Wold-style-definition
> > -Wswitch -Wshadow -Wcast-qual -Wwrite-strings -Wno-unreachable-code
> > -Wno-pointer-sign -Wno-attributes -Wno-sign-compare
> > --sysroot=/opt/obj.evbarm/destdir.evbarm -mcpu=cortex-a9 -mfpu=neon
> > -DKERNEL_BASES_EQUAL -DKERNEL_BASE_VOFFSET=0 -I.
> > -I/opt/src/sys/../common/include -I/opt/src/sys/arch -I/opt/src/sys
> > -nostdinc -DCHILD_MAX=1024 -DOPEN_MAX=1024
> > -D__HAVE_MM_MD_DIRECT_MAPPED_PHYS -D__HAVE_MM_MD_CACHE_ALIASING
> > -D__HAVE_FAST_SOFTINTS -D__HAVE_CPU_UAREA_ALLOC_IDLELWP
> > -D__HAVE_CPU_COUNTER -D__HAVE_PCI_CONF_HOOK
> > -DKERNEL_BASE_EXT=0x80000000 -DARM_GENERIC_TODR -DRNDVERBOSE
> > -DDIAGNOSTIC -DDEBUG -DMAXUSERS=32 -D_KERNEL -D_KERNEL_OPT
> > -std=gnu99
> > -I/opt/src/sys/lib/libkern/../../../common/lib /libc/quad
> > -I/opt/src/sys/lib/libkern/../../../common/lib/libc/string
> > -I/opt/src/sys/lib/libkern/../../../common/lib/libc/arch/arm/string
> > -c /opt/src/sys/arch/arm/vfp/vfp_init.c {standard input}: Assembler
> > messages:
> 
> Where did the -mfpu=vfp get introduced? Just specifying earmv7hf
> won't set that.
> 
> John

I guess this is set by NetBSD build system. My mk.conf definitely has
any CPUFLAGS commented out:

# tail -n 5 $MAKECONF
.else
COPTS+=-pipe
#CPUFLAGS+=-mcpu=cortex-a8
MKX11=yes
.endif