linux-kernel - RE: [PATCH 6/7] KVM: VMX: Provide separate subroutines for invoking NMI vs. IRQ handlers

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

Message-ID: <BN6PR1101MB2161F8ACB46A3C027E0553ECA8E19@BN6PR1101MB2161.namprd11.prod.outlook.com>
Date:   Thu, 15 Dec 2022 05:18:30 +0000
From:   "Li, Xin3" <xin3.li@...el.com>
To:     "Li, Xin3" <xin3.li@...el.com>,
        "Christopherson,, Sean" <seanjc@...gle.com>
CC:     Paolo Bonzini <pbonzini@...hat.com>,
        "kvm@...r.kernel.org" <kvm@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        Peter Zijlstra <peterz@...radead.org>,
        "Lutomirski, Andy" <luto@...nel.org>,
        Thomas Gleixner <tglx@...utronix.de>
Subject: RE: [PATCH 6/7] KVM: VMX: Provide separate subroutines for invoking
 NMI vs. IRQ handlers

> > > > +	 * "Restore" RSP from RBP, even though IRET has already unwound
> > > > RSP to
> > > > +	 * the correct value.  objtool doesn't know the callee will IRET and,
> > > > +	 * without the explicit restore, thinks the stack is getting walloped.
> > > > +	 * Using an unwind hint is problematic due to x86-64's dynamic
> > > > alignment.
> > > > +	 */
> > > > +	mov %_ASM_BP, %_ASM_SP
> > > > +	pop %_ASM_BP
> > > > +	RET
> > >
> > > For NMI, after this RET instruction, we continue to block NMIs. IRET
> > instead?
> >
> > No, IRET has already been done by the kernel-provided handler, e.g.
> > asm_exc_nmi_kvm_vmx()
> > by way of error_return().  That's why the CALL above (that got snipped) is
> > preceded
> > by the creation of a synthetic NMI/IRQ stack frame: the target returns from
> > the CALL
> > via IRET, not RET.
> 
> 
> You're right, this assembly makes another call into the asm entry point, which
> returns here with IRET.

Like IRET for IDT, we _need_ ERETS for FRED to unblock NMI ASAP. However
a FRED stack frame has way more information than an IDT stack frame,
thus it's complicated to create a FRED stack frame with assembly.

So I prefer "INT $2" for FRED now.

I will post the FRED patch set soon, lets sync up on this afterwards.

Xin