On Tue, Dec 25, 2012 at 07:51:54AM -0800, H. Peter Anvin wrote: > Right, it's the ljmpq issue. > > Borislav Petkov <[email protected]> wrote: > > >On Mon, Dec 24, 2012 at 05:06:01PM -0800, H. Peter Anvin wrote: > >> Could this be the ljmpq problem that Borislav reported > >> (Intel implemented ljmpq, AMD didn't, and I was tempted by a > >> micro-optimization which broke AMD which made it into the patchset)? > > > >It has to be. Just booted Yinghai's -v8 in kvm on an AMD host and it > >worked fine. > > > >With the change below it keeps rebooting like I reported earlier. I'd > >go > >out on a limb here and guess that the guest is triple-faulting due to > >an > >unhandled #GP caused by an invalid opcode or similar. > > > >--- > >diff --git a/arch/x86/kernel/head_64.S b/arch/x86/kernel/head_64.S > >index d94f6d68be2a..1842d30c96a2 100644 > >--- a/arch/x86/kernel/head_64.S > >+++ b/arch/x86/kernel/head_64.S > >@@ -279,11 +279,8 @@ ENTRY(secondary_startup_64) > > * REX.W + FF /5 JMP m16:64 Jump far, absolute indirect, > > * address given in m16:64. > > */ > >- movq initial_code(%rip),%rax > > pushq $0 # fake return address to stop unwinder > >- pushq $__KERNEL_CS # set correct cs > >- pushq %rax # target address in negative space > >- lretq > >+ rex64 ljmp *initial_code(%rip) > > > > #ifdef CONFIG_HOTPLUG_CPU
Btw, I'm being told that we could also do a JMP FAR through a call gate in 64-bit mode. Basically, the insn operand contains a CS:OFFSET whereas the OFFSET is ignored but only CS is looked at. It has to point to a call gate descriptor which contains two pieces of a Code-Segment Offset[0:63] and this points into the virtual address space where we want to jump to. Here's a picture to explain it better (figure 4-31 on page 105): http://support.amd.com/us/Processor_TechDocs/24593_APM_v2.pdf It looks like Intel should support that too, AFAICT. -- Regards/Gruss, Boris. Sent from a fat crate under my desk. Formatting is fine. -- -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
