On 06/03/2018 21:50, Radim Krčmář wrote: > GCC 8 emitted MOVDQU when compiling vmx.elf, but we do not enable > CR4.OSFXSR in that test, so the instruction throws #UD. > This patch forbids all sse2 instructions, instead of enabling > CR4.OSFXSR, as I think it's better to keep the environment minimal. > > Signed-off-by: Radim Krčmář <rkrcmar@xxxxxxxxxx> > --- > This probably hit a minor KVM bug: the instruction was "f3 0f 6f 41 28", > but KVM reported that #UD happened on "f3 0f 6f 41". That's probably just because "no opcode" does not decode mod/rm/sib. > --- > x86/Makefile.x86_64 | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/x86/Makefile.x86_64 b/x86/Makefile.x86_64 > index 623fc5b37726..6caa3a8863f1 100644 > --- a/x86/Makefile.x86_64 > +++ b/x86/Makefile.x86_64 > @@ -1,7 +1,7 @@ > cstart.o = $(TEST_DIR)/cstart64.o > bits = 64 > ldarch = elf64-x86-64 > -COMMON_CFLAGS += -mno-red-zone > +COMMON_CFLAGS += -mno-red-zone -mno-sse2 Why not -mno-sse? Paolo > cflatobjs += lib/x86/setjmp64.o > cflatobjs += lib/x86/intel-iommu.o
