linux-kernel - Re: [PATCH] KVM: x86: Avoid emulating instructions on #UD mistakenly

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:	Mon, 18 Aug 2014 21:53:11 +0300
From:	Nadav Amit <nadav.amit@...il.com>
To:	Paolo Bonzini <pbonzini@...hat.com>
Cc:	gleb@...nel.org, Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...hat.com>,
	"H. Peter Anvin" <hpa@...or.com>,
	the arch/x86 maintainers <x86@...nel.org>,
	KVM <kvm@...r.kernel.org>,
	Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
	Nadav Amit <namit@...technion.ac.il>
Subject: Re: [PATCH] KVM: x86: Avoid emulating instructions on #UD mistakenly


On Aug 18, 2014, at 11:31 AM, Paolo Bonzini <pbonzini@...hat.com> wrote:

> Il 13/08/2014 16:21, Nadav Amit ha scritto:
>> Correction: the word “never” in the message is too harsh.
>> Nonetheless, there is a regression bug. I encountered it with “wrfsbase” instruction.
> 
> So KVM is emulating wrfsbase even if the host doesn't support it?
KVM doesn’t know wrfsbase - wrfsbase is encoded like clflush with rep-prefix.
Therefore, the emulator thinks it can emulate it as clflush, and eliminates the #UD.

> 
> I'm swapping the order of the two operands of &&, since the first one will almost
> always be true and the second one will almost always be false.
> 
> Also, there's now no need to test EmulateOnUD in the condition below.  Does the
> below look good to you?
> 
> diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/emulate.c
> index 37a83b24e040..ef117b842334 100644
> --- a/arch/x86/kvm/emulate.c
> +++ b/arch/x86/kvm/emulate.c
> @@ -4394,11 +4394,11 @@ done_prefixes:
> 
>  	ctxt->execute = opcode.u.execute;
> 
> -	if (!(ctxt->d & EmulateOnUD) && ctxt->ud)
> +	if (unlikely(ctxt->ud) && likely(!(ctxt->d & EmulateOnUD)))
>  		return EMULATION_FAILED;
> 
>  	if (unlikely(ctxt->d &
> -		     (NotImpl|EmulateOnUD|Stack|Op3264|Sse|Mmx|Intercept|CheckPerm))) {
> +		     (NotImpl|Stack|Op3264|Sse|Mmx|Intercept|CheckPerm))) {
>  		/*
>  		 * These are copied unconditionally here, and checked unconditionally
>  		 * in x86_emulate_insn.
> 

Sure. Until I find some bugs in it. ;-)

Thanks,
Nadav

Download attachment "signature.asc" of type "application/pgp-signature" (496 bytes)