Re: [net-next PATCH v2 5/8] net: Track start of busy loop instead of when it should end

[Alexander Duyck]

On Thu, Mar 23, 2017 at 9:27 PM, Eric Dumazet <eric.dumazet@xxxxxxxxx> wrote:
> On Thu, 2017-03-23 at 20:42 -0700, Alexander Duyck wrote:
>> On Thu, Mar 23, 2017 at 6:24 PM, Eric Dumazet <eric.dumazet@xxxxxxxxx> wrote:
>> > On Thu, 2017-03-23 at 14:37 -0700, Alexander Duyck wrote:
>> >> From: Alexander Duyck <alexander.h.duyck@xxxxxxxxx>
>> >>
>> >
>> >> The last bit I changed is to move from using a shift by 10 to just using
>> >> NSEC_PER_USEC and using multiplication for any run time calculations and
>> >> division for a few compile time ones.  This should be more accurate and
>> >> perform about the same on most architectures since modern CPUs typically
>> >> handle multiplication without too much overhead.
>> >
>> >
>> > busy polling thread can be preempted for more than 2 seconds.
>>
>> If it is preempted is the timing value even valid anymore?  I was
>> wondering about that.  Also when preemption is enabled is there
>> anything to prevent us from being migrated to a CPU?  If so what do we
>> do about architectures that allow drift between the clocks?
>>
>> > Using usec instead of nanoseconds gave us 3 orders of magnitude cushion.
>>
>> Yes, but the problem is we also opened up an issue where if the clock
>> was approaching a roll-over we could add a value to it that would put
>> us in a state where we would never time out.
>
> If you believe there is a bug, send a fix for net tree ?
>
> I really do not see a bug, given we use time_after(now, end_time) which
> handles roll-over just fine.

Right, but time_after assumes roll over.  When you are using a time
value based off of local_clock() >> 10, you don't ever roll over when
you do addition.  Just the clock rolls over.  At least on 64 bit
systems.

So if local time approaches something like all 1's, and we have
shifted it by 10 it is then the max it can ever reach is
0x003FFFFFFFFFFFFF.  I can add our loop time to that and it won't roll
over.  In the mean time the busy_loop_us_ can never exceed whatever I
added to that so we are now locked into a loop.  I realize I am
probably being pedantic, and it will have an exceedingly small rate of
occurrence, but it is still an issue.

>>
>> > We do not need nsec accuracy for busy polling users, if this restricts
>> > range and usability under stress.
>>
>> Yes and no.  So the standard use cases suggest using values of 50 to
>> 100 microseconds.  I suspect that for most people that is probably
>> what they are using.  The addition of preemption kind of threw a
>> wrench in the works because now instead of spending that time busy
>> polling you can get preempted and then are off doing something else
>> for the entire period of time.
>
> That is fine. Busy polling heavy users are pinning threads to cpu, and
> the re-schedule check is basically a way to make sure ksoftirqd will get
> a chance to service BH.
>
>>
>> What would you think of changing this so that instead of tracking the
>> total time this function is active, instead we tracked the total time
>> we spent with preemption disabled?  What I would do is move the
>> start_time configuration to just after the preempt_disable() call.
>> Then if we end up yielding to another thread we would just reset the
>> start_time when we restarted instead of trying to deal with all the
>> extra clock nonsense that we would have to deal with otherwise since I
>> don't know if we actually want to count time where we aren't actually
>> doing anything.  In addition this would bring us closer to how NAPI
>> already works since it essentially will either find an event, or if we
>> time out we hand it off to the softirq which in turn can handle it or
>> hand it off to softirqd.  The only item that might be a bit more
>> difficult to deal with then would be the way the times are used in
>> fs/select.c but I don't know if that is really the right way to go
>> anyway. With the preemption changes and such it might just make sense
>> to drop those bits and rely on just the socket polling alone.
>>
>> The other option is to switch over everything from using unsigned long
>> to using uint64_t and time_after64.  Then we can guarantee the range
>> needed and then some, but then we are playing with a u64 time value on
>> 32b architectures which might be a bit more expensive.  Even with that
>> though I still need to clean up the sysctl since it doesn't make sense
>> to allow negative values for the busy_poll usec to be used which is
>> currently the case.
>>
>> Anyway let me know what you think and I can probably spin out a new
>> set tomorrow.
>
>
> Leave usec resolution, I see no point switching to nsec, as long we
> support 32bit kernels.

I wonder how much it really matters in the grand scheme of things.
The question I would have is which is more expensive.  We are either
having to do a set of double precision shifts, or a multiplication,
double precision addition, and double precision compare.  I need to
take a look at some instruction tables to see which is more expensive.

My preference would be to switch things over to a u64 at this point as
it fixes the time_after problem, and has no additional cost on 64b and
bumps up the maximum usable value on 32b systems by a power of 2, but
I will have to think about it and see if there isn't a better way to
deal with that issue.

> If you believe min/max values should be added to the sysctls, because we
> do not trust root anymore, please send patches only addressing that.

It isn't so much a min/max as just the wrong function used.  I will
switch it over to using proc_douintvec().

Also it isn't that I don't trust root, I don't like us giving root the
option of configuring things wrongly without giving them some sort of
warning.  So switching things over to using unsigned int will match
what we are actually storing but the problem is anything over INT_MAX
on 32 bit systems will give you busy polling with an immediate time
out since the signed bit is what is being used in "time_after" to
determine if we crossed the boundary or not.

- Alex