On 2019/11/16 0:37, Igor Mammedov wrote:
>> +
>> + /* zero means OSPM does not acknowledge the error */
>> + if (!read_ack_register) {
>> + if (loop < 3) {
>> + usleep(100 * 1000);
>> + loop++;
>> + goto retry;
> as minimum this loop can stall guest repeatedly for 0.3s if guest triggers BQL,
> until it handles error.
I think reparations for 0.3s is reasonable.
1. 0.3s is the worst case to repeat, if guest acknowledge it in before 0.3s, the guest can not stall
2. if the previous error is not acknowledged, the next error will be lost, error handling(safety) is more important than others.
>
> (not sure what to suggest here though)
>
> (not sure what to suggest here though)
>
