Re: [PATCH 3/4] x86,module: Detect VMX vs SLD conflicts

[Masami Hiramatsu]

On Wed, 8 Apr 2020 01:51:24 +0900
Masami Hiramatsu <mhiramat@xxxxxxxxxx> wrote:

> Hi Peter,
> 
> On Tue, 07 Apr 2020 13:02:39 +0200
> Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> 
> > +static bool insn_is_vmx(struct insn *insn)
> > +{
> > +	u8 modrm = insn->modrm.bytes[0];
> > +	u8 modrm_mod = X86_MODRM_MOD(modrm);
> > +	u8 modrm_reg = X86_MODRM_REG(modrm);
> > +
> > +	u8 prefix = insn->prefixes.bytes[0];
> > +
> > +	if (insn->opcode.bytes[0] != 0x0f)
> > +		return false;
> > +
> > +	switch (insn->opcode.bytes[1]) {
> > +	case 0x01:
> > +		switch (insn->opcode.bytes[2]) {
> > +		case 0xc1: /* VMCALL */
> > +		case 0xc2: /* VMLAUNCH */
> > +		case 0xc3: /* VMRESUME */
> > +		case 0xc4: /* VMXOFF */
> > +			return true;
> > +
> > +		default:
> > +			break;
> > +		}
> > +		break;
> > +
> > +	case 0x78: /* VMREAD */
> > +	case 0x79: /* VMWRITE */
> > +		return true;
> > +
> > +	case 0xc7:
> > +		/* VMPTRLD, VMPTRST, VMCLEAR, VMXON */
> > +		if (modrm_mod == 0x03)
> > +			break;
> > +
> > +		if ((modrm_reg == 6 && (!prefix || prefix == 0x66 || prefix == 0xf3)) ||
> > +		    (modrm_reg == 7 && (!prefix || prefix == 0xf3)))
> > +			return true;
> > +
> > +		break;
> > +
> > +	default:
> > +		break;
> > +	}
> > +
> > +	return false;
> > +}
> 
> OK, so here is what you need ;)
> 
> From 36f4f6aec623b0190fde95c8630a6a1d8c23ffc9 Mon Sep 17 00:00:00 2001
> From: Masami Hiramatsu <mhiramat@xxxxxxxxxx>
> Date: Wed, 8 Apr 2020 01:04:41 +0900
> Subject: [PATCH] x86: insn: Add insn_is_vmx()
> 
> Add insn_is_vmx() to identify the given instruction is
> for VMX or not. This is simply identifying those instructions
> by mnemonic pattern.
> 

Hmm, I found that this is still not enough... since the inat_tables
mixes the instruction attributes on same entry in group tables.
It distinguishes opcodes by last-prefix variants, but not by
MOD & R/M bits, since it is designed only for decoding instructions.

Thank you,

-- 
Masami Hiramatsu <mhiramat@xxxxxxxxxx>