Re: [Qemu-devel] [PATCH 3/3] i386/kvm: initialize struct at full before ioctl call

[Christian Borntraeger]

On 30.07.19 19:14, Philippe Mathieu-Daudé wrote:
> On 7/30/19 7:05 PM, Christian Borntraeger wrote:
>> On 30.07.19 18:44, Philippe Mathieu-Daudé wrote:
>>> On 7/30/19 6:01 PM, Andrey Shinkevich wrote:
>>>> Not the whole structure is initialized before passing it to the KVM.
>>>> Reduce the number of Valgrind reports.
>>>>
>>>> Signed-off-by: Andrey Shinkevich <address@hidden>
>>>> ---
>>>>  target/i386/kvm.c | 3 +++
>>>>  1 file changed, 3 insertions(+)
>>>>
>>>> diff --git a/target/i386/kvm.c b/target/i386/kvm.c
>>>> index dbbb137..ed57e31 100644
>>>> --- a/target/i386/kvm.c
>>>> +++ b/target/i386/kvm.c
>>>> @@ -190,6 +190,7 @@ static int kvm_get_tsc(CPUState *cs)
>>>>          return 0;
>>>>      }
>>>>  
>>>> +    memset(&msr_data, 0, sizeof(msr_data));
>>>
>>> I wonder the overhead of this one...
>>
>> Cant we use designated initializers like in
>>
>> commit bdfc8480c50a53d91aa9a513d23a84de0d5fbc86
>> Author:     Christian Borntraeger <address@hidden>
>> AuthorDate: Thu Oct 30 09:23:41 2014 +0100
>> Commit:     Paolo Bonzini <address@hidden>
>> CommitDate: Mon Dec 15 12:21:01 2014 +0100
>>
>>     valgrind/i386: avoid false positives on KVM_SET_XCRS ioctl
>>
>> and others?
> 
> Is the compiler smart enough to figure out it doesn't need to zeroes in
> case env->tsc_valid is true and the function returns?

Good question, we would need to double check with objdump.