> -----Original Message-----
> From: Petre Pircalabu [mailto:ppircal...@bitdefender.com]
> Sent: 21 September 2017 06:12
> To: xen-devel@lists.xen.org
> Cc: Ian Jackson <ian.jack...@citrix.com>; Wei Liu <wei.l...@citrix.com>;
> Andrew Cooper <andrew.coop...@citrix.com>; George Dunlap
> <george.dun...@citrix.com>; jbeul...@suse.com; konrad.w...@oracle.com;
> sstabell...@kernel.org; Tim (Xen.org) <t...@xen.org>; Paul Durrant
> <paul.durr...@citrix.com>; rcojoc...@bitdefender.com;
> ta...@tklengyel.com; jun.nakaj...@intel.com; Kevin Tian
> <kevin.t...@intel.com>; Petre Pircalabu <ppircal...@bitdefender.com>
> Subject: [PATCH v12 4/4] x86emul: Raise #UD when emulating an
> unrecognized instruction.
>
> Modified the behavior of hvm_emulate_one_insn and
> vmx_realmode_emulate_one to generate an Invalid Opcode trap when
> X86EMUL_UNRECOGNIZED is returned by the emulator instead of just
> crashing the domain.
>
> Signed-off-by: Petre Pircalabu <ppircal...@bitdefender.com>
> Reviewed-by: Kevin Tian <kevin.t...@intel.com>
> ---
> xen/arch/x86/hvm/io.c | 6 +++++-
> xen/arch/x86/hvm/vmx/realmode.c | 11 ++++++++++-
> 2 files changed, 15 insertions(+), 2 deletions(-)
>
> diff --git a/xen/arch/x86/hvm/io.c b/xen/arch/x86/hvm/io.c
> index 7152c28..c7b1c53 100644
> --- a/xen/arch/x86/hvm/io.c
> +++ b/xen/arch/x86/hvm/io.c
> @@ -96,10 +96,14 @@ bool
> hvm_emulate_one_insn(hvm_emulate_validate_t *validate, const char
> *descr)
> switch ( rc )
> {
> case X86EMUL_UNHANDLEABLE:
> - case X86EMUL_UNIMPLEMENTED:
> hvm_dump_emulation_state(XENLOG_G_WARNING, descr, &ctxt, rc);
> return false;
>
> + case X86EMUL_UNRECOGNIZED:
> + hvm_dump_emulation_state(XENLOG_G_WARNING, descr, &ctxt, rc);
> + hvm_inject_hw_exception(TRAP_invalid_op, X86_EVENT_NO_EC);
> + break;
> +
> case X86EMUL_EXCEPTION:
> hvm_inject_event(&ctxt.ctxt.event);
> break;
> diff --git a/xen/arch/x86/hvm/vmx/realmode.c
> b/xen/arch/x86/hvm/vmx/realmode.c
> index b93792d..03dea6c 100644
> --- a/xen/arch/x86/hvm/vmx/realmode.c
> +++ b/xen/arch/x86/hvm/vmx/realmode.c
> @@ -106,12 +106,21 @@ void vmx_realmode_emulate_one(struct
> hvm_emulate_ctxt *hvmemul_ctxt)
> if ( hvm_vcpu_io_need_completion(vio) || vio->mmio_retry )
> vio->io_completion = HVMIO_realmode_completion;
>
> - if ( rc == X86EMUL_UNHANDLEABLE || rc == X86EMUL_UNIMPLEMENTED
> )
> + if ( rc == X86EMUL_UNHANDLEABLE )
I don't quite understand this change. Why has it become unnecessary to deal
with X86EMUL_UNIMPLEMENTED? Patch #1 added this change so it seems odd that
patch #4 would then revert it.
Paul
> {
> gdprintk(XENLOG_ERR, "Failed to emulate insn.\n");
> goto fail;
> }
>
> + if ( rc == X86EMUL_UNRECOGNIZED )
> + {
> + gdprintk(XENLOG_ERR, "Unrecognized insn.\n");
> + if ( curr->arch.hvm_vcpu.guest_cr[0] & X86_CR0_PE )
> + goto fail;
> +
> + realmode_deliver_exception(TRAP_invalid_op, 0, hvmemul_ctxt);
> + }
> +
> if ( rc == X86EMUL_EXCEPTION )
> {
> if ( unlikely(curr->domain->debugger_attached) &&
> --
> 2.7.4
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel
