linux-kernel - Re: [PATCH 15/17] KVM: retpolines: x86: eliminate retpoline from vmx.c exit 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 for Android: free password hash cracker in your pocket

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

Message-ID: <20190923210838.GA23063@redhat.com>
Date:   Mon, 23 Sep 2019 17:08:38 -0400
From:   Andrea Arcangeli <aarcange@...hat.com>
To:     Sean Christopherson <sean.j.christopherson@...el.com>
Cc:     Paolo Bonzini <pbonzini@...hat.com>,
        Vitaly Kuznetsov <vkuznets@...hat.com>,
        "Dr. David Alan Gilbert" <dgilbert@...hat.com>,
        Marcelo Tosatti <mtosatti@...hat.com>,
        Peter Xu <peterx@...hat.com>, kvm@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH 15/17] KVM: retpolines: x86: eliminate retpoline from
 vmx.c exit handlers

Hello,

On Mon, Sep 23, 2019 at 01:23:49PM -0700, Sean Christopherson wrote:
> The attached patch should do the trick.

The two most attractive options to me remains what I already have
implemented under #ifdef CONFIG_RETPOLINE with direct calls
(optionally replacing the "if" with a small "switch" still under
CONFIG_RETPOLINE if we give up the prioritization of the checks), or
the replacement of kvm_vmx_exit_handlers with a switch() as suggested
by Vitaly which would cleanup some code.

The intermediate solution that makes "const" work, has the cons of
forcing to parse EXIT_REASON_VMCLEAR and the other vmx exit reasons
twice, first through a pointer to function (or another if or switch
statement) then with a second switch() statement.

If we'd use a single switch statement per Vitaly's suggestion, the "if
nested" would better be more simply implemented as:

	switch (exit_reason) {
		case EXIT_REASON_VMCLEAR:
			if (nested)
				return handle_vmclear(vcpu);
			else
				return handle_vmx_instruction(vcpu);
		case EXIT_REASON_VMCLEAR:
			if (nested)
		[..]

This also removes the compiler dependency to auto inline
handle_vmclear in the added nested_vmx_handle_vmx_instruction extern
call.

VMREAD/WRITE/RESUME are the most frequent vmexit in l0 while nested
runs in l2.

Thanks,
Andrea