linux-kernel - Re: [PATCH 3/4] KVM: VMX: Macrofy the MSR bitmap getters and setters

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: <YFIwsP9gF19MyCm7@google.com>
Date:   Wed, 17 Mar 2021 09:39:12 -0700
From:   Sean Christopherson <seanjc@...gle.com>
To:     Paolo Bonzini <pbonzini@...hat.com>
Cc:     Vitaly Kuznetsov <vkuznets@...hat.com>,
        Wanpeng Li <wanpengli@...cent.com>,
        Jim Mattson <jmattson@...gle.com>,
        Joerg Roedel <joro@...tes.org>, kvm@...r.kernel.org,
        linux-kernel@...r.kernel.org, Alexander Graf <graf@...zon.com>,
        Yuan Yao <yaoyuan0329os@...il.com>
Subject: Re: [PATCH 3/4] KVM: VMX: Macrofy the MSR bitmap getters and setters

On Wed, Mar 17, 2021, Paolo Bonzini wrote:
> On 16/03/21 19:44, Sean Christopherson wrote:
> > +	return (ret)true;						      \
> 
> I'm not sure if (void)true is amazing or disgusting, but anyway...

Definitely both.

> > +BUILD_VMX_MSR_BITMAP_HELPER(bool, test, read)
> > +BUILD_VMX_MSR_BITMAP_HELPER(bool, test, write)
> > +BUILD_VMX_MSR_BITMAP_HELPER(void, clear, read, __)
> > +BUILD_VMX_MSR_BITMAP_HELPER(void, clear, write, __)
> > +BUILD_VMX_MSR_BITMAP_HELPER(void, set, read, __)
> > +BUILD_VMX_MSR_BITMAP_HELPER(void, set, write, __)
> 
> ... I guess we have an armed truce where you let me do my bit manipulation
> magic and I let you do your macro magic.

Ha, mutually assured destruction.

> Still, I think gluing the variadic arguments with ## is a bit too much.

Heh, I don't disagree at all.  Honestly, I was surprised it worked, and couldn't
resist throwing it in because it's so absurd.

> This would be slightly less mysterious:
> 
> +BUILD_VMX_MSR_BITMAP_HELPER(bool, vmx_test_msr_bitmap_, read, test_bit)
> +BUILD_VMX_MSR_BITMAP_HELPER(bool, vmx_test_msr_bitmap_, write, test_bit)
> +BUILD_VMX_MSR_BITMAP_HELPER(void, vmx_clear_msr_bitmap_, read, __clear_bit)
> +BUILD_VMX_MSR_BITMAP_HELPER(void, vmx_clear_msr_bitmap_, write,
> __clear_bit)
> +BUILD_VMX_MSR_BITMAP_HELPER(void, vmx_set_msr_bitmap_, read, __set_bit)
> +BUILD_VMX_MSR_BITMAP_HELPER(void, vmx_set_msr_bitmap_, write, __set_bit)
> 
> And I also wonder if we really need to expand all six functions one at a
> time.  You could remove the third argument and VMX_MSR_BITMAP_BASE_*, at the
> cost of expanding the inline functions' body twice in
> BUILD_VMX_MSR_BITMAP_HELPER.

I'll play around with the macros to see if I can make them less obnoxious.  I
found it easier to differentiate between the read/write offset and the high/low
offset by building them one at a time.  I'll see if I can find a compromise.