On 07/29/2016 02:57 AM, Peter Maydell wrote:
> On 28 July 2016 at 17:38, Wei Huang <w...@redhat.com> wrote:
>> This patch adds a pmu=[on/off] option to enable/disable vpmu support
>> in guest vm. There are several reasons to justify this option. First
>> vpmu can be problematic for cross-migration between different SoC as
>> perf counters is architecture-dependent. It is more flexible to
>> have an option to turn it on/off. Secondly it matches the -cpu pmu
>> option in libivrt. This patch has been tested on both DT/ACPI modes.
>
>
> What particular two systems are you trying to migrate between?
One example: APM's Mustang has 5 perf counters while AMD's Seattle has 7
counters.
> In general we don't support migrating between different CPU
> types at the moment, so the PMU sholud be the same on both ends.
>
> (If we ever do get to supporting cross-cpu-type migration
> then it will probably involve a very long and detailed command
> line to specify exactly a whole lot of things like pmu yes/no,
> number of hw breakpoints/watchpoints, and everything else that
> can differ between implementations.)
>
> That said, I don't have any objection to making the PMU
> presence controllable (especially if we have similar
> control on x86).
>
>> --- a/target-arm/cpu.h
>> +++ b/target-arm/cpu.h
>> @@ -579,8 +579,9 @@ struct ARMCPU {
>> bool powered_off;
>> /* CPU has security extension */
>> bool has_el3;
>> - /* CPU has PMU (Performance Monitor Unit) */
>> - bool has_pmu;
>> +
>> + /* CPU has vPMU (Performance Monitor Unit) support */
>> + bool enable_pmu;
>
> Why rename the flag? has_foo is what we use for other features,
> as you can see in the context of this bit of the patch.
I will fix it. Maybe follow the suggestion Drew's suggestion, keeping
has_pmu and add another option for turning it on/off.
>
>>
>> /* CPU has memory protection unit */
>> bool has_mpu;
>
> thanks
> -- PMM
>
