Re: [PATCH v12 for-xen-4.5 12/20] x86/VPMU: Initialize PMU for PV(H) guests

[Boris Ostrovsky <boris.ostrovsky@oracle.com>]

On 09/29/2014 11:42 AM, Jan Beulich wrote:
>>>> On 29.09.14 at 17:41, <boris.ostrovsky@oracle.com> wrote:
>> On 09/29/2014 11:25 AM, Jan Beulich wrote:
>>>>>> On 25.09.14 at 21:28, <boris.ostrovsky@oracle.com> wrote:
>>>> @@ -389,14 +390,26 @@ static int amd_vpmu_initialise(struct vcpu *v)
>>>>    	 }
>>>>        }
>>>>    
>>>> -    ctxt = xzalloc_bytes(sizeof(struct xen_pmu_amd_ctxt) +
>>>> -                         2 * sizeof(uint64_t) * AMD_MAX_COUNTERS);
>>>> -    if ( !ctxt )
>>>> +    regs_size = 2 * sizeof(uint64_t) * AMD_MAX_COUNTERS;
>>>> +    if ( is_hvm_domain(v->domain) )
>>>>        {
>>>> -        gdprintk(XENLOG_WARNING, "Insufficient memory for PMU, "
>>>> -            " PMU feature is unavailable on domain %d vcpu %d.\n",
>>>> -            v->vcpu_id, v->domain->domain_id);
>>>> -        return -ENOMEM;
>>>> +        ctxt = xzalloc_bytes(sizeof(struct xen_pmu_amd_ctxt) + regs_size);
>>>> +        if ( !ctxt )
>>>> +        {
>>>> +            gdprintk(XENLOG_WARNING, "Insufficient memory for PMU, "
>>>> +                "PMU feature is unavailable\n");
>>>> +            return -ENOMEM;
>>>> +        }
>>>> +    }
>>>> +    else
>>>> +    {
>>>> +        if ( sizeof(struct xen_pmu_data) + regs_size > PAGE_SIZE )
>>> This is a compile time constant condition - no reason to issue a
>>> message and return failure at runtime, just BUILD_BUG_ON() instead.
>> It will not be if I replace AMD_MAX_COUNTERS with runtime register
>> count, as you asked in an earlier comment.
> For which case see the respective VMX side comment.

Not sure I understand what you mean here. it will be almost exactly the 
same.

-boris