On Mon, 22 May 2017 22:23:17 PDT (-0700), o...@lixom.net wrote: > (new top-level subthread here since this is a separate topic): > > On Mon, May 22, 2017 at 5:41 PM, Palmer Dabbelt <pal...@dabbelt.com> wrote: > >> diff --git a/arch/riscv/Makefile b/arch/riscv/Makefile >> new file mode 100644 >> index 000000000000..07ef200e0675 >> --- /dev/null >> +++ b/arch/riscv/Makefile >> @@ -0,0 +1,64 @@ >> +# This file is included by the global makefile so that you can add your own >> +# architecture-specific flags and dependencies. Remember to do have actions >> +# for "archclean" and "archdep" for cleaning up and making dependencies for >> +# this architecture >> +# >> +# This file is subject to the terms and conditions of the GNU General Public >> +# License. See the file "COPYING" in the main directory of this archive >> +# for more details. >> +# >> + >> +LDFLAGS := >> +OBJCOPYFLAGS := -O binary >> +LDFLAGS_vmlinux := >> +KBUILD_AFLAGS_MODULE += -fPIC >> +KBUILD_CFLAGS_MODULE += -fPIC >> + >> +ifeq ($(ARCH),riscv) >> + KBUILD_DEFCONFIG = riscv64_spike >> +else >> + KBUILD_DEFCONFIG = $(ARCH)_spike >> +endif >> + >> +export BITS >> +ifeq ($(CONFIG_64BIT),y) >> + BITS := 64 >> + UTS_MACHINE := riscv64 >> + >> + KBUILD_CFLAGS += -mabi=lp64 >> + KBUILD_AFLAGS += -mabi=lp64 >> + KBUILD_MARCH = rv64im >> + LDFLAGS += -melf64lriscv >> +else >> + BITS := 32 >> + UTS_MACHINE := riscv32 >> + >> + KBUILD_CFLAGS += -mabi=ilp32 >> + KBUILD_AFLAGS += -mabi=ilp32 >> + KBUILD_MARCH = rv32im >> + LDFLAGS += -melf32lriscv >> +endif >> + >> +ifeq ($(CONFIG_RV_ATOMIC),y) >> + KBUILD_RV_ATOMIC = a >> +endif >> + >> +KBUILD_CFLAGS += -Wall >> + >> +ifeq ($(CONFIG_RVC),y) >> + KBUILD_RVC = c >> +endif >> + >> +KBUILD_AFLAGS += -march=$(KBUILD_MARCH)$(KBUILD_RV_ATOMIC)fd$(KBUILD_RVC) >> + >> +KBUILD_CFLAGS += -march=$(KBUILD_MARCH)$(KBUILD_RV_ATOMIC)$(KBUILD_RVC) >> +KBUILD_CFLAGS += -mno-save-restore >> +KBUILD_CFLAGS += -mstrict-align > > I built a vanilla gcc-7.1.0 here, with 'riscv64-linux' as target, and I get: > > riscv64-linux-gcc: error: unrecognized command line option > '-mstrict-align'; did you mean '-Wstrict-aliasing'? > > > The suggestion seems completely bogus, but the error is real. Looking > at the gcc sources, I only see strict-align plumbed up on rs6000, > aarch64, m68k(!) and v850. Or am I missing something here?
We didn't get the "-mstrict-align" patch for RISC-V into the 7.1.0 release. The boot loader install machine-mode unaligned access handlers, so in theory you shouldn't need this at all -- it snuck its way in as a debugging attempt because the Radeon driver was exhibiting some weird behavior. I'll drop this from the next patch set.
