On 25/01/2022 22:15, Glenn Washburn wrote:
Is amd64 is another non-working/unused build target? Here's the link error.
ar cru liblibc.a target/libc/ctype.o target/libc/diskio.o target/libc/extra.o target/libc/misc.o target/libc/string.o target/libc/vsprintf.o; ranlib liblibc.a ar: `u' modifier ignored since `D' is the default (see `U') gcc -fno-builtin -O0 -g -DNATIVE_BITWIDTH_EQUALS_HOST_BITWIDTH -USWAP_ENDIANNESS -Wall -Wredundant-decls -Wshadow -Wpointer-arith -Wstrict-prototypes -Wmissing-declarations -Wundef -Wendif-labels -Wstrict-aliasing -Wwrite-strings -Wmissing-prototypes -Wnested-externs -Werror -MMD -MP -MT openbios-unix -MF '/.d' -rdynamic -o openbios-unix target/arch/unix/unix.o target/arch/unix/boot.o target/arch/unix/blk.o libbootstrap.a libpackages.a libopenbios.a libdrivers.a libfs.a liblibc.a /usr/bin/ld: liblibc.a(misc.o):/home/g10/openbios.git/libc/misc.c:20: multiple definition of `errno_int'; target/arch/unix/unix.o:/home/g10/openbios.git/arch/unix/unix.c:71: first defined here
Indeed errno_int is declared in both those locations. It would be easy enough for me to fix for amd64 Linux, but I can't test that it would work on OS X (per the comment in ./arch/unix/unix.c).
amd64 did build for me when I was experimenting with GitHub workflows and it would be useful to keep it around. I haven't tested OpenBIOS with coreboot for a few years now, but it did work at last time I tried with a very small tweak in drivers/pci.c to raise the IO base address enough on x86 to avoid clobbering the fw_cfg port.
These days there is enough CI available on Linux that I think it is worth keeping the build working there as a priority (well, at least until someone using MacOS complains!) so feel free to send a patch.
For reference my experimental Github workflow can be found here: https://github.com/mcayland/openbios/blob/master/.github/workflows/main.yml. I would certainly like to find some time to finish this so it can form the basis of a more official release process.
ATB,
Mark.