Re: [PATCH] SCSI: don't get target/host busy_count in scsi_mq_get_budget()

[Jens Axboe <axboe@kernel.dk>]

On 11/08/2017 11:22 AM, Laurence Oberman wrote:
> On Wed, 2017-11-08 at 10:57 -0700, Jens Axboe wrote:
>> On 11/08/2017 09:41 AM, Bart Van Assche wrote:
>>> On Tue, 2017-11-07 at 20:06 -0700, Jens Axboe wrote:
>>>> At this point, I have no idea what Bart's setup looks like. Bart,
>>>> it
>>>> would be REALLY helpful if you could tell us how you are
>>>> reproducing
>>>> your hang. I don't know why this has to be dragged out.
>>>
>>> Hello Jens,
>>>
>>> It is a disappointment to me that you have allowed Ming to evaluate
>>> other
>>> approaches than reverting "blk-mq: don't handle TAG_SHARED in
>>> restart". That
>>> patch namely replaces an algorithm that is trusted by the community
>>> with an
>>> algorithm of which even Ming acknowledged that it is racy. A quote
>>> from [1]:
>>> "IO hang may be caused if all requests are completed just before
>>> the current
>>> SCSI device is added to shost->starved_list". I don't know of any
>>> way to fix
>>> that race other than serializing request submission and completion
>>> by adding
>>> locking around these actions, which is something we don't want.
>>> Hence my
>>> request to revert that patch.
>>
>> I was reluctant to revert it, in case we could work out a better way
>> of
>> doing it. As I mentioned in the other replies, it's not exactly the
>> prettiest or most efficient. However, since we currently don't have
>> a good solution for the issue, I'm fine with reverting that patch.
>>
>>> Regarding the test I run, here is a summary of what I mentioned in
>>> previous
>>> e-mails:
>>> * I modified the SRP initiator such that the SCSI target queue
>>> depth is
>>>   reduced to one by setting starget->can_queue to 1 from inside
>>>   scsi_host_template.target_alloc.
>>> * With that modified SRP initiator I run the srp-test software as
>>> follows
>>>   until something breaks:
>>>   while ./run_tests -f xfs -d -e deadline -r 60; do :; done
>>
>> What kernel options are needed? Where do I download everything I
>> need?
>>
>> In other words, would it be possible to do a fuller guide for getting
>> this setup and running?
>>
>> I'll run my simple test case as well, since it's currently breaking
>> basically everywhere.
>>
>>> Today a system with at least one InfiniBand HCA is required to run
>>> that test.
>>> When I have the time I will post the SRP initiator and target
>>> patches on the
>>> linux-rdma mailing list that make it possible to run that test
>>> against the
>>> SoftRoCE driver (drivers/infiniband/sw/rxe). The only hardware
>>> required to
>>> use that driver is an Ethernet adapter.
>>
>> OK, I guess I can't run it then... I'll have to rely on your testing.
> 
> Hello 
> 
> I agree with Bart in this case, we should revert this.
> My test-bed is tied up and I have not been able to give it back to Ming
> so he could follow up on Bart's last update.
> 
> Right now its safer to revert.

I had already reverted it when sending out that email, so we should be
all set (hopefully).

-- 
Jens Axboe