> Il giorno 28 giu 2017, alle ore 14:42, Jens Axboe <axboe@xxxxxxxxx> ha scritto: > > On 06/27/2017 11:39 PM, Paolo Valente wrote: >> >>> Il giorno 27 giu 2017, alle ore 20:29, Jens Axboe <axboe@xxxxxxxxx> ha scritto: >>> >>> On 06/27/2017 12:27 PM, Paolo Valente wrote: >>>> >>>>> Il giorno 27 giu 2017, alle ore 16:41, Jens Axboe <axboe@xxxxxxxxx> ha scritto: >>>>> >>>>> On 06/27/2017 12:09 AM, Paolo Valente wrote: >>>>>> >>>>>>> Il giorno 19 giu 2017, alle ore 13:43, Paolo Valente <paolo.valente@xxxxxxxxxx> ha scritto: >>>>>>> >>>>>>> This commit fixes a bug triggered by a non-trivial sequence of >>>>>>> events. These events are briefly described in the next two >>>>>>> paragraphs. The impatiens, or those who are familiar with queue >>>>>>> merging and splitting, can jump directly to the last paragraph. >>>>>>> >>>>>>> On each I/O-request arrival for a shared bfq_queue, i.e., for a >>>>>>> bfq_queue that is the result of the merge of two or more bfq_queues, >>>>>>> BFQ checks whether the shared bfq_queue has become seeky (i.e., if too >>>>>>> many random I/O requests have arrived for the bfq_queue; if the device >>>>>>> is non rotational, then random requests must be also small for the >>>>>>> bfq_queue to be tagged as seeky). If the shared bfq_queue is actually >>>>>>> detected as seeky, then a split occurs: the bfq I/O context of the >>>>>>> process that has issued the request is redirected from the shared >>>>>>> bfq_queue to a new non-shared bfq_queue. As a degenerate case, if the >>>>>>> shared bfq_queue actually happens to be shared only by one process >>>>>>> (because of previous splits), then no new bfq_queue is created: the >>>>>>> state of the shared bfq_queue is just changed from shared to non >>>>>>> shared. >>>>>>> >>>>>>> Regardless of whether a brand new non-shared bfq_queue is created, or >>>>>>> the pre-existing shared bfq_queue is just turned into a non-shared >>>>>>> bfq_queue, several parameters of the non-shared bfq_queue are set >>>>>>> (restored) to the original values they had when the bfq_queue >>>>>>> associated with the bfq I/O context of the process (that has just >>>>>>> issued an I/O request) was merged with the shared bfq_queue. One of >>>>>>> these parameters is the weight-raising state. >>>>>>> >>>>>>> If, on the split of a shared bfq_queue, >>>>>>> 1) a pre-existing shared bfq_queue is turned into a non-shared >>>>>>> bfq_queue; >>>>>>> 2) the previously shared bfq_queue happens to be busy; >>>>>>> 3) the weight-raising state of the previously shared bfq_queue happens >>>>>>> to change; >>>>>>> the number of weight-raised busy queues changes. The field >>>>>>> wr_busy_queues must then be updated accordingly, but such an update >>>>>>> was missing. This commit adds the missing update. >>>>>>> >>>>>> >>>>>> Hi Jens, >>>>>> any idea of the possible fate of this fix? >>>>> >>>>> I sort of missed this one. It looks trivial enough for 4.12, or we >>>>> can defer until 4.13. What do you think? >>>>> >>>> >>>> It should actually be something trivial, and hopefully correct, >>>> because a further throughput improvement (for BFQ), which depends on >>>> this fix, is now working properly, and we didn't see any regression so >>>> far. In addition, since this improvement is virtually ready for >>>> submission, further steps may be probably easier if this fix gets in >>>> sooner (whatever the luck of the improvement will be). >>> >>> OK, let's queue it up for 4.13 then. >>> >> >> My arguments was in favor of 4.12 actually. Maybe you did mean 4.12 >> here? > > You were talking about further improvements and new development on top > of this, so I assumed you meant 4.13. However, further development is > not the main criteria or concern for whether this fix should go into > 4.12 or not. Ok, thanks for your explanation and patience. > The main concern is if this fixes something that is crucial > to have in 4.12. It's late in the cycle, I'd rather not push anything > that isn't a regression fix at this point. > Hard to assess precisely how crucial this is. Certainly it fixes a regression. The practical, negative effects of this regression are systematic when one tries to add the throughput improvement I mentioned: the improvement almost never works. If BFQ is used as it is, then negative effects on throughput are less likely to happen. I hope that this piece of information is somehow useful for your decision. Thanks, Paolo > -- > Jens Axboe
