handle_wrmsr behavior under kvm-83

handle_wrmsr behavior under kvm-83

[Steven Stovall]

When looking at the EXIT_REASON_MSR_WRITE events that trigger handle_wrmsr() invocations, why do I only seem to see indices for MSR_MTRRfix* (0x250 - 0x2ff)? This behavior seems to have changed since kvm-60, where I would see for example SYSENTER MSR writes.

Steven

Re: handle_wrmsr behavior under kvm-83

[Avi Kivity]

Steven Stovall wrote:
> When looking at the EXIT_REASON_MSR_WRITE events that trigger handle_wrmsr() invocations, why do I only seem to see indices for MSR_MTRRfix* (0x250 - 0x2ff)? This behavior seems to have changed since kvm-60, where I would see for example SYSENTER MSR writes.
>   

In vmx_init():

    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_FS_BASE);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_GS_BASE);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_CS);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_ESP);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_EIP);


-- 
error compiling committee.c: too many arguments to function

Re: handle_wrmsr behavior under kvm-83

[Steven Stovall]

Right, but I've tried commenting those calls out, even disabling CPU_BASED_USE_MSR_BITMAPS, allocation of vmx_msr_bitmap and so on -- without effect. Ive started trying to interrogate the (now-reinstituted) vmx_msr_bitmap from vmx_vcpu_setup, since the controls can be rewritten there...

Steven

----- Original Message -----
From: "Avi Kivity" <avi@redhat.com>
To: "Steven Stovall" <sstovall@neuraliq.com>
Cc: "kvm" <kvm@vger.kernel.org>
Sent: Monday, February 2, 2009 8:27:07 AM GMT -08:00 US/Canada Pacific
Subject: Re: handle_wrmsr behavior under kvm-83

Steven Stovall wrote:
> When looking at the EXIT_REASON_MSR_WRITE events that trigger handle_wrmsr() invocations, why do I only seem to see indices for MSR_MTRRfix* (0x250 - 0x2ff)? This behavior seems to have changed since kvm-60, where I would see for example SYSENTER MSR writes.
>   

In vmx_init():

    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_FS_BASE);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_GS_BASE);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_CS);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_ESP);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_EIP);


-- 
error compiling committee.c: too many arguments to function

Re: handle_wrmsr behavior under kvm-83

[Steven Stovall]

Well the bits for the sysenter msrs (0x174 - 0x176) are definitely set -- but I still only see exits for 0x200 - 0x2ff. And I definitely wrmsr(l)(MSR_IA32_SYSENTER_*) at vmx_vcpu_setup() time. Ive been going over Ch. 20, and still cant figure out what could be going on here.

Steven

>Right, but I've tried commenting those calls out, even disabling CPU_BASED_USE_MSR_BITMAPS, allocation >of vmx_msr_bitmap and so on -- without effect. Ive started trying to interrogate the (now-reinstituted) >vmx_msr_bitmap from vmx_vcpu_setup, since the controls can be rewritten there...

>Steven

----- Original Message -----
From: "Avi Kivity" <avi@redhat.com>
To: "Steven Stovall" <sstovall@neuraliq.com>
Cc: "kvm" <kvm@vger.kernel.org>
Sent: Monday, February 2, 2009 8:27:07 AM GMT -08:00 US/Canada Pacific
Subject: Re: handle_wrmsr behavior under kvm-83

Steven Stovall wrote:
> When looking at the EXIT_REASON_MSR_WRITE events that trigger handle_wrmsr() invocations, why do I only seem to see indices for MSR_MTRRfix* (0x250 - 0x2ff)? This behavior seems to have changed since kvm-60, where I would see for example SYSENTER MSR writes.
>   

In vmx_init():

    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_FS_BASE);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_GS_BASE);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_CS);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_ESP);
    vmx_disable_intercept_for_msr(vmx_msr_bitmap, MSR_IA32_SYSENTER_EIP);


-- 
error compiling committee.c: too many arguments to function


--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html