Re: [PATCH] KVM: x86: Check for host supported fields in shadow vmcs

[Bandan Das]

Paolo Bonzini <pbonzini@xxxxxxxxxx> writes:

> Il 19/04/2014 19:34, Bandan Das ha scritto:
>>
>> We track shadow vmcs fields through two static lists,
>> one for read only fields and another for r/w. However, with
>> addition of new vmcs fields, not all fields may be supported on
>> all hosts. If so, copy_vmcs12_to_shadow() trying to vmwrite on older
>> hosts will result in a vmwrite error. For example, commit
>> 36be0b9deb23161 introduced GUEST_BNDCFGS, which is not supported
>> for all processors. Create new lists based out of intersection of
>> static lists and host support and use them for tracking
>> shadow fields instead
>>
>> Signed-off-by: Bandan Das <bsd@xxxxxxxxxx>
>> ---
>>  arch/x86/kvm/vmx.c | 98 +++++++++++++++++++++++++++++++++++++++++++-----------
>>  1 file changed, 79 insertions(+), 19 deletions(-)
>>
>> diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c
>> index 7bed3e3..ffc2232 100644
>> --- a/arch/x86/kvm/vmx.c
>> +++ b/arch/x86/kvm/vmx.c
>> @@ -502,7 +502,10 @@ static inline struct vcpu_vmx *to_vmx(struct kvm_vcpu *vcpu)
>>  #define FIELD64(number, name)	[number] = VMCS12_OFFSET(name), \
>>  				[number##_HIGH] = VMCS12_OFFSET(name)+4
>>
>> -
>> +/*
>> + * Do not use the two lists below directly
>> + * Use vmcs_shadow_fields instead
>> + */
>>  static const unsigned long shadow_read_only_fields[] = {
>>  	/*
>>  	 * We do NOT shadow fields that are modified when L0
>> @@ -526,8 +529,6 @@ static const unsigned long shadow_read_only_fields[] = {
>>  	GUEST_LINEAR_ADDRESS,
>>  	GUEST_PHYSICAL_ADDRESS
>>  };
>> -static const int max_shadow_read_only_fields =
>> -	ARRAY_SIZE(shadow_read_only_fields);
>>
>>  static const unsigned long shadow_read_write_fields[] = {
>>  	GUEST_RIP,
>> @@ -558,8 +559,18 @@ static const unsigned long shadow_read_write_fields[] = {
>>  	HOST_FS_SELECTOR,
>>  	HOST_GS_SELECTOR
>>  };
>> -static const int max_shadow_read_write_fields =
>> -	ARRAY_SIZE(shadow_read_write_fields);
>
> Can we just remove the "const" here, and compress the arrays down
> similar to what kvm_init_msr_list does.

Yep, agreed. That makes sense. Thanks for the pointer.

>> +/* If new shadow fields are added, these should be modified appropriately */
>> +#define VMCS_MAX_RO_FIELDS 10
>> +#define VMCS_MAX_RW_FIELDS 30
>> +
>> +struct vmcs_shadow_fields_data {
>> +	int shadow_ro_fields_len;
>> +	int shadow_rw_fields_len;
>> +	unsigned long shadow_read_only_fields[VMCS_MAX_RO_FIELDS];
>> +	unsigned long shadow_read_write_fields[VMCS_MAX_RW_FIELDS];
>> +};
>> +static struct vmcs_shadow_fields_data vmcs_shadow_fields;
>>
>>  static const unsigned short vmcs_field_to_offset_table[] = {
>>  	FIELD(VIRTUAL_PROCESSOR_ID, virtual_processor_id),
>> @@ -3027,6 +3038,56 @@ static __init int alloc_kvm_area(void)
>>  	return 0;
>>  }
>>
>> +static void cleanup_vmcs_shadow_fields(void)
>> +{
>> +	memset(&vmcs_shadow_fields, 0,
>> +	       sizeof(struct vmcs_shadow_fields_data));
>> +}
>> +
>> +static void init_vmcs_shadow_fields(void)
>> +{
>> +	struct vmcs_shadow_fields_data *vmcs_ptr = &vmcs_shadow_fields;
>> +	int max_shadow_read_write_fields = ARRAY_SIZE(shadow_read_write_fields);
>> +	int max_shadow_read_only_fields = ARRAY_SIZE(shadow_read_only_fields);
>> +	int i, j;
>> +
>> +	for (i = 0, j = 0; i < max_shadow_read_write_fields; i++) {
>> +		if (i >= VMCS_MAX_RW_FIELDS) {
>> +			WARN(1, "Shadow RW fields index out of bounds\n");
>> +			break;
>> +		}
>> +		if ((shadow_read_write_fields[i] == GUEST_BNDCFGS) &&
>> +		    !vmx_mpx_supported())
>> +			continue;
>
> Please code this as a "switch" statement for easier future
> extensibility.  Again, this would be similar to kvm_init_msr_list.

Ok.

> How did you find this?  Do you have access to a machine with shadow
> VMCS?  Is that Ivy Bridge Xeon E5 or does some lower-end Haswell have
> it?

Yeah, it seems the DIY grade Haswells do indeed have shadow vmcs support.
The system I am using has i7-4770 and I found this while testing my Xen 
as L1 changes on top of 3.15-rc1.

> Paolo
>
>> +		vmcs_ptr->shadow_read_write_fields[j++] =
>> +			shadow_read_write_fields[i];
>> +		vmcs_ptr->shadow_rw_fields_len++;
>> +	}
>> +
>> +	for (i = 0, j = 0; i < max_shadow_read_only_fields; i++) {
>> +		if (i >= VMCS_MAX_RO_FIELDS) {
>> +			WARN(1, "Shadow RO fields index out of bounds\n");
>> +			break;
>> +		}
>> +		vmcs_ptr->shadow_read_only_fields[j++] =
>> +			shadow_read_only_fields[i];
>> +		vmcs_ptr->shadow_ro_fields_len++;
>> +	}
>> +
>> +	/* shadowed read/write fields */
>> +	for (i = 0; i < vmcs_ptr->shadow_rw_fields_len; i++) {
>> +		clear_bit(vmcs_ptr->shadow_read_write_fields[i],
>> +			  vmx_vmwrite_bitmap);
>> +		clear_bit(vmcs_ptr->shadow_read_write_fields[i],
>> +			  vmx_vmread_bitmap);
>> +	}
>> +	/* shadowed read only fields */
>> +	for (i = 0; i < vmcs_ptr->shadow_ro_fields_len; i++)
>> +		clear_bit(vmcs_ptr->shadow_read_only_fields[i],
>> +			  vmx_vmread_bitmap);
>> +
>> +}
>> +
>>  static __init int hardware_setup(void)
>>  {
>>  	if (setup_vmcs_config(&vmcs_config) < 0)
>> @@ -3039,6 +3100,8 @@ static __init int hardware_setup(void)
>>  		enable_vpid = 0;
>>  	if (!cpu_has_vmx_shadow_vmcs())
>>  		enable_shadow_vmcs = 0;
>> +	if (enable_shadow_vmcs)
>> +		init_vmcs_shadow_fields();
>>
>>  	if (!cpu_has_vmx_ept() ||
>>  	    !cpu_has_vmx_ept_4levels()) {
>> @@ -3084,6 +3147,8 @@ static __init int hardware_setup(void)
>>
>>  static __exit void hardware_unsetup(void)
>>  {
>> +	if (enable_shadow_vmcs)
>> +		cleanup_vmcs_shadow_fields();
>>  	free_kvm_area();
>>  }
>>
>> @@ -6159,8 +6224,9 @@ static void copy_shadow_to_vmcs12(struct vcpu_vmx *vmx)
>>  	unsigned long field;
>>  	u64 field_value;
>>  	struct vmcs *shadow_vmcs = vmx->nested.current_shadow_vmcs;
>> -	const unsigned long *fields = shadow_read_write_fields;
>> -	const int num_fields = max_shadow_read_write_fields;
>> +	const unsigned long *fields =
>> +		vmcs_shadow_fields.shadow_read_write_fields;
>> +	const int num_fields = vmcs_shadow_fields.shadow_rw_fields_len;
>>
>>  	vmcs_load(shadow_vmcs);
>>
>> @@ -6189,13 +6255,15 @@ static void copy_shadow_to_vmcs12(struct vcpu_vmx *vmx)
>>
>>  static void copy_vmcs12_to_shadow(struct vcpu_vmx *vmx)
>>  {
>> +	struct vmcs_shadow_fields_data *ptr = &vmcs_shadow_fields;
>> +
>>  	const unsigned long *fields[] = {
>> -		shadow_read_write_fields,
>> -		shadow_read_only_fields
>> +		ptr->shadow_read_write_fields,
>> +		ptr->shadow_read_only_fields
>>  	};
>>  	const int max_fields[] = {
>> -		max_shadow_read_write_fields,
>> -		max_shadow_read_only_fields
>> +		ptr->shadow_rw_fields_len,
>> +		ptr->shadow_ro_fields_len
>>  	};
>>  	int i, q;
>>  	unsigned long field;
>> @@ -8817,14 +8885,6 @@ static int __init vmx_init(void)
>>
>>  	memset(vmx_vmread_bitmap, 0xff, PAGE_SIZE);
>>  	memset(vmx_vmwrite_bitmap, 0xff, PAGE_SIZE);
>> -	/* shadowed read/write fields */
>> -	for (i = 0; i < max_shadow_read_write_fields; i++) {
>> -		clear_bit(shadow_read_write_fields[i], vmx_vmwrite_bitmap);
>> -		clear_bit(shadow_read_write_fields[i], vmx_vmread_bitmap);
>> -	}
>> -	/* shadowed read only fields */
>> -	for (i = 0; i < max_shadow_read_only_fields; i++)
>> -		clear_bit(shadow_read_only_fields[i], vmx_vmread_bitmap);
>>
>>  	/*
>>  	 * Allow direct access to the PC debug port (it is often used for I/O
>>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/