Re: [PATCH 1/3] x86/HVM: fold hvm_io_assist() into its only caller

[Jan Beulich <jbeulich@xxxxxxxx>]

On 15.07.2020 14:40, Paul Durrant wrote:
>> From: Jan Beulich <jbeulich@xxxxxxxx>
>> Sent: 15 July 2020 13:04
>>
>>  static bool hvm_wait_for_io(struct hvm_ioreq_vcpu *sv, ioreq_t *p)
>>  {
>>      unsigned int prev_state = STATE_IOREQ_NONE;
>> +    uint64_t data = ~0;
>>
>> -    while ( sv->pending )
>> -    {
>> +    do {
>>          unsigned int state = p->state;
> 
> I guess this is beneficial from the point of view of restricting cope and...
> 
>>
>>          smp_rmb();
>> @@ -132,7 +121,6 @@ static bool hvm_wait_for_io(struct hvm_i
>>               * emulator is dying and it races with an I/O being
>>               * requested.
>>               */
>> -            hvm_io_assist(sv, ~0ul);
>>              break;
> 
> ...(as you say) allowing this early break, but a forward jump to an 'out' 
> label would be more consistent with other code. It works though so...

Since this gets restructured by subsequent patches I thought I'd
avoid the introduction of a disliked by me "goto".

> Reviewed-by: Paul Durrant <paul@xxxxxxx>

Thanks, Jan