Re: [RFC PATCH 0/7] support multi-phtread per lcore

[Bruce Richardson <bruce.richardson-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>]

On Fri, Dec 19, 2014 at 01:28:47AM +0000, Liang, Cunming wrote:
> 
> 
> > -----Original Message-----
> > From: Walukiewicz, Miroslaw
> > Sent: Thursday, December 18, 2014 8:20 PM
> > To: Liang, Cunming; dev-VfR2kkLFssw@public.gmane.org
> > Subject: RE: [dpdk-dev] [RFC PATCH 0/7] support multi-phtread per lcore
> > 
> > I have another question regarding your patch.
> > 
> >  Could we extend values returned by rte_lcore_id() to set them per thread (really
> > the DPDK lcore is a pthread but started on specific core) instead of creating linear
> > thread id.
> [Liang, Cunming] As you said, __lcore_id is already per thread. 
> Per the semantic meaning, it stands for logic cpu id. 
> When multi-thread running on the same lcore, they should get the same value return by rte_lcore_id().
> The same effective like 'schedu_getcpu()', but less using cost.
> > 
> > The patch would be much simpler and will work same way. The only change
> > would be extending rte_lcore_id when rte_pthread_create() is called.
> [Liang, Cunming] I ever think about it which using rte_lcore_id() to get unique id per pthread rather than have a new API.
> But the name lcore actually no longer identify for cpu id. It may impact all existing user application who use the exact meaning of it.
> How do you think ?
> > 

I'm conflicted on this one. However, I think far more applications would be broken
to start having to use thread_id in place of an lcore_id than would be broken
by having the lcore_id no longer actually correspond to a core.
I'm actually struggling to come up with a large number of scenarios where it's
important to an app to determine the cpu it's running on, compared to the large
number of cases where you need to have a data-structure per thread. In DPDK libs
alone, you see this assumption that lcore_id == thread_id a large number of times.

Despite the slight logical inconsistency, I think it's better to avoid introducing
a thread-id and continue having lcore_id representing a unique thread.

/Bruce

> > The value __lcore_id has really an attribute __thread that means it is valid not
> > only per CPU core but also per thread.
> > 
> > The mempools, timers, statistics would work without any modifications in that
> > environment.
> > 
> >  I do not see any reason why old legacy DPDK applications would not work in that
> > model.
> > 
> > Mirek
> > 
> > > -----Original Message-----
> > > From: Liang, Cunming
> > > Sent: Monday, December 15, 2014 12:53 PM
> > > To: Walukiewicz, Miroslaw; dev-VfR2kkLFssw@public.gmane.org
> > > Subject: RE: [dpdk-dev] [RFC PATCH 0/7] support multi-phtread per lcore
> > >
> > > Hi Mirek,
> > >
> > > That sounds great.
> > > Looking forward to it.
> > >
> > > -Cunming
> > >
> > > > -----Original Message-----
> > > > From: Walukiewicz, Miroslaw
> > > > Sent: Monday, December 15, 2014 7:11 PM
> > > > To: Liang, Cunming; dev-VfR2kkLFssw@public.gmane.org
> > > > Subject: RE: [dpdk-dev] [RFC PATCH 0/7] support multi-phtread per lcore
> > > >
> > > > Hi Cunming,
> > > >
> > > > The timers could be used by any application/library started as a standard
> > > > pthread.
> > > > Each pthread needs to have assigned some identifier same way as you are
> > > doing
> > > > it for mempools (the rte_linear_thread_id and rte_lcore_id are good
> > > examples)
> > > >
> > > > I made series of patches extending the rte timers API to use with such kind
> > > of
> > > > identifier keeping existing API working also.
> > > >
> > > > I will send it soon.
> > > >
> > > > Mirek
> > > >
> > > >
> > > > > -----Original Message-----
> > > > > From: Liang, Cunming
> > > > > Sent: Friday, December 12, 2014 6:45 AM
> > > > > To: Walukiewicz, Miroslaw; dev-VfR2kkLFssw@public.gmane.org
> > > > > Subject: RE: [dpdk-dev] [RFC PATCH 0/7] support multi-phtread per lcore
> > > > >
> > > > > Thanks Mirek. That's a good point which wasn't mentioned in cover
> > > letter.
> > > > > For 'rte_timer', I only expect it be used within the 'legacy per-lcore'
> > > pthread.
> > > > > I'm appreciate if you can give me some cases which can't use it to fit.
> > > > > In case have to use 'rte_timer' in multi-pthread, there are some
> > > > > prerequisites and limitations.
> > > > > 1. Make sure thread local variable 'lcore_id' is set correctly (e.g. do
> > > pthread
> > > > > init by rte_pthread_prepare)
> > > > > 2. As 'rte_timer' is not preemptable, when using
> > > rte_timer_manager/reset in
> > > > > multi-pthread, make sure they're not on the same core.
> > > > >
> > > > > -Cunming
> > > > >
> > > > > > -----Original Message-----
> > > > > > From: Walukiewicz, Miroslaw
> > > > > > Sent: Thursday, December 11, 2014 5:57 PM
> > > > > > To: Liang, Cunming; dev-VfR2kkLFssw@public.gmane.org
> > > > > > Subject: RE: [dpdk-dev] [RFC PATCH 0/7] support multi-phtread per
> > > lcore
> > > > > >
> > > > > > Thank you Cunming for explanation.
> > > > > >
> > > > > > What about DPDK timers? They also depend on rte_lcore_id() to avoid
> > > > > spinlocks.
> > > > > >
> > > > > > Mirek
> > > > > >
> > > > > > > -----Original Message-----
> > > > > > > From: dev [mailto:dev-bounces-VfR2kkLFssw@public.gmane.org] On Behalf Of Cunming
> > > Liang
> > > > > > > Sent: Thursday, December 11, 2014 3:05 AM
> > > > > > > To: dev-VfR2kkLFssw@public.gmane.org
> > > > > > > Subject: [dpdk-dev] [RFC PATCH 0/7] support multi-phtread per lcore
> > > > > > >
> > > > > > >
> > > > > > > Scope & Usage Scenario
> > > > > > > ========================
> > > > > > >
> > > > > > > DPDK usually pin pthread per core to avoid task switch overhead. It
> > > gains
> > > > > > > performance a lot, but it's not efficient in all cases. In some cases, it
> > > may
> > > > > > > too expensive to use the whole core for a lightweight workload. It's a
> > > > > > > reasonable demand to have multiple threads per core and each
> > > threads
> > > > > > > share CPU
> > > > > > > in an assigned weight.
> > > > > > >
> > > > > > > In fact, nothing avoid user to create normal pthread and using cgroup
> > > to
> > > > > > > control the CPU share. One of the purpose for the patchset is to clean
> > > the
> > > > > > > gaps of using more DPDK libraries in the normal pthread. In addition, it
> > > > > > > demonstrates performance gain by proactive 'yield' when doing idle
> > > loop
> > > > > > > in packet IO. It also provides several 'rte_pthread_*' APIs to easy life.
> > > > > > >
> > > > > > >
> > > > > > > Changes to DPDK libraries
> > > > > > > ==========================
> > > > > > >
> > > > > > > Some of DPDK libraries must run in DPDK environment.
> > > > > > >
> > > > > > > # rte_mempool
> > > > > > >
> > > > > > > In rte_mempool doc, it mentions a thread not created by EAL must
> > > not
> > > > > use
> > > > > > > mempools. The root cause is it uses a per-lcore cache inside
> > > mempool.
> > > > > > > And 'rte_lcore_id()' will not return a correct value.
> > > > > > >
> > > > > > > The patchset changes this a little. The index of mempool cache won't
> > > be a
> > > > > > > lcore_id. Instead of it, using a linear number generated by the
> > > allocator.
> > > > > > > For those legacy EAL per-lcore thread, it apply for an unique linear id
> > > > > > > during creation. For those normal pthread expecting to use
> > > > > rte_mempool, it
> > > > > > > requires to apply for a linear id explicitly. Now the mempool cache
> > > looks
> > > > > like
> > > > > > > a per-thread base. The linear ID actually identify for the linear thread
> > > id.
> > > > > > >
> > > > > > > However, there's another problem. The rte_mempool is not
> > > > > preemptable.
> > > > > > > The
> > > > > > > problem comes from rte_ring, so talk together in next section.
> > > > > > >
> > > > > > > # rte_ring
> > > > > > >
> > > > > > > rte_ring supports multi-producer enqueue and multi-consumer
> > > > > dequeue.
> > > > > > > But it's
> > > > > > > not preemptable. There's conversation talking about this before.
> > > > > > > http://dpdk.org/ml/archives/dev/2013-November/000714.html
> > > > > > >
> > > > > > > Let's say there's two pthreads running on the same core doing
> > > enqueue
> > > > > on
> > > > > > > the
> > > > > > > same rte_ring. If the 1st pthread is preempted by the 2nd pthread
> > > while
> > > > > it
> > > > > > > has
> > > > > > > already modified the prod.head, the 2nd pthread will spin until the 1st
> > > > > one
> > > > > > > scheduled agian. It causes time wasting. In addition, if the 2nd
> > > pthread
> > > > > has
> > > > > > > absolutely higer priority, it's more terrible.
> > > > > > >
> > > > > > > But it doesn't means we can't use. Just need to narrow down the
> > > > > situation
> > > > > > > when
> > > > > > > it's used by multi-pthread on the same core.
> > > > > > > - It CAN be used for any single-producer or single-consumer situation.
> > > > > > > - It MAY be used by multi-producer/consumer pthread whose
> > > scheduling
> > > > > > > policy
> > > > > > > are all SCHED_OTHER(cfs). User SHOULD aware of the performance
> > > > > penalty
> > > > > > > befor
> > > > > > > using it.
> > > > > > > - It MUST not be used by multi-producer/consumer pthread, while
> > > some
> > > > > of
> > > > > > > their
> > > > > > > scheduling policies is SCHED_FIFO or SCHED_RR.
> > > > > > >
> > > > > > >
> > > > > > > Performance
> > > > > > > ==============
> > > > > > >
> > > > > > > It loses performance by introducing task switching. On packet IO
> > > > > perspective,
> > > > > > > we can gain some back by improving IO effective rate. When the
> > > pthread
> > > > > do
> > > > > > > idle
> > > > > > > loop on an empty rx queue, it should proactively yield. We can also
> > > slow
> > > > > > > down
> > > > > > > rx for a bit while to take more advantage of the bulk receiving in the
> > > next
> > > > > > > loop. In practice, increase the rx ring size also helps to improve the
> > > > > overrall
> > > > > > > throughput.
> > > > > > >
> > > > > > >
> > > > > > > Cgroup Control
> > > > > > > ================
> > > > > > >
> > > > > > > Here's a simple example, there's four pthread doing packet IO on the
> > > > > same
> > > > > > > core.
> > > > > > > We expect the CPU share rate is 1:1:2:4.
> > > > > > > > mkdir /sys/fs/cgroup/cpu/dpdk
> > > > > > > > mkdir /sys/fs/cgroup/cpu/dpdk/thread0
> > > > > > > > mkdir /sys/fs/cgroup/cpu/dpdk/thread1
> > > > > > > > mkdir /sys/fs/cgroup/cpu/dpdk/thread2
> > > > > > > > mkdir /sys/fs/cgroup/cpu/dpdk/thread3
> > > > > > > > cd /sys/fs/cgroup/cpu/dpdk
> > > > > > > > echo 256 > thread0/cpu.shares
> > > > > > > > echo 256 > thread1/cpu.shares
> > > > > > > > echo 512 > thread2/cpu.shares
> > > > > > > > echo 1024 > thread3/cpu.shares
> > > > > > >
> > > > > > >
> > > > > > > -END-
> > > > > > >
> > > > > > > Any comments are welcome.
> > > > > > >
> > > > > > > Thanks
> > > > > > >
> > > > > > > *** BLURB HERE ***
> > > > > > >
> > > > > > > Cunming Liang (7):
> > > > > > >   eal: add linear thread id as pthread-local variable
> > > > > > >   mempool: use linear-tid as mempool cache index
> > > > > > >   ring: use linear-tid as ring debug stats index
> > > > > > >   eal: add simple API for multi-pthread
> > > > > > >   testpmd: support multi-pthread mode
> > > > > > >   sample: add new sample for multi-pthread
> > > > > > >   eal: macro for cpuset w/ or w/o CPU_ALLOC
> > > > > > >
> > > > > > >  app/test-pmd/cmdline.c                    |  41 +++++
> > > > > > >  app/test-pmd/testpmd.c                    |  84 ++++++++-
> > > > > > >  app/test-pmd/testpmd.h                    |   1 +
> > > > > > >  config/common_linuxapp                    |   1 +
> > > > > > >  examples/multi-pthread/Makefile           |  57 ++++++
> > > > > > >  examples/multi-pthread/main.c             | 232
> > > > > > ++++++++++++++++++++++++
> > > > > > >  examples/multi-pthread/main.h             |  46 +++++
> > > > > > >  lib/librte_eal/common/include/rte_eal.h   |  15 ++
> > > > > > >  lib/librte_eal/common/include/rte_lcore.h |  12 ++
> > > > > > >  lib/librte_eal/linuxapp/eal/eal_thread.c  | 282
> > > > > > > +++++++++++++++++++++++++++---
> > > > > > >  lib/librte_mempool/rte_mempool.h          |  22 +--
> > > > > > >  lib/librte_ring/rte_ring.h                |   6 +-
> > > > > > >  12 files changed, 755 insertions(+), 44 deletions(-)
> > > > > > >  create mode 100644 examples/multi-pthread/Makefile
> > > > > > >  create mode 100644 examples/multi-pthread/main.c
> > > > > > >  create mode 100644 examples/multi-pthread/main.h
> > > > > > >
> > > > > > > --
> > > > > > > 1.8.1.4
>