Re: [PATCH v13 for-xen-4.5 09/21] x86/VPMU: Add public xenpmu.h

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

On 10/13/2014 09:31 AM, Jan Beulich wrote:
>
>> --- a/xen/include/public/arch-x86/xen-x86_32.h
>> +++ b/xen/include/public/arch-x86/xen-x86_32.h
>> @@ -136,6 +136,16 @@ struct cpu_user_regs {
>>   typedef struct cpu_user_regs cpu_user_regs_t;
>>   DEFINE_XEN_GUEST_HANDLE(cpu_user_regs_t);
>>   
>> +struct xen_pmu_regs {
>> +    uint32_t eip;
>> +    uint32_t esp;
>> +    uint32_t eflags;
>> +    uint16_t cs;
>> +    uint16_t ss;
>> +};
> I still don't see why this and ...
>
>> --- a/xen/include/public/arch-x86/xen-x86_64.h
>> +++ b/xen/include/public/arch-x86/xen-x86_64.h
>> @@ -174,6 +174,16 @@ struct cpu_user_regs {
>>   typedef struct cpu_user_regs cpu_user_regs_t;
>>   DEFINE_XEN_GUEST_HANDLE(cpu_user_regs_t);
>>   
>> +struct xen_pmu_regs {
>> +    __DECL_REG(ip);
>> +    __DECL_REG(sp);
>> +    __DECL_REG(flags);
>> +    uint16_t cs;
>> +    uint16_t ss;
>> +};
> aren't folded and aren't placed in the PMU-specific header.

Because it follows cpu_user_regs() definitions. I thought you agreed to 
that but if not I can move it to PMU code.

>
> Furthermore I'm sure you realized (the latest when adding back
> the eflags field for VM86 mode recognition) that to the consumer of
> this data things are still ambiguous: You can't tell protected from
> real mode, yet HVM guests definitely use that mode.

I can set a bit in the sample's flag field if the guest is in real mode.

-boris
-boris