DPDK patches and discussions
 help / color / mirror / Atom feed
From: Stephen Hemminger <stephen@networkplumber.org>
To: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH 2/2] eal/x86: Use lock-prefixed instructions to reduce cost of rte_smp_mb()
Date: Thu, 8 Mar 2018 13:15:10 -0800	[thread overview]
Message-ID: <20180308131510.677ff22d@xeon-e3> (raw)
In-Reply-To: <2601191342CEEE43887BDE71AB9772585FAC39B0@irsmsx105.ger.corp.intel.com>

On Fri, 1 Dec 2017 23:08:39 +0000
"Ananyev, Konstantin" <konstantin.ananyev@intel.com> wrote:

> Hi Stephen,
> 
> > -----Original Message-----
> > From: Stephen Hemminger [mailto:stephen@networkplumber.org]
> > Sent: Friday, December 1, 2017 6:04 PM
> > To: Ananyev, Konstantin <konstantin.ananyev@intel.com>
> > Cc: dev@dpdk.org
> > Subject: Re: [dpdk-dev] [PATCH 2/2] eal/x86: Use lock-prefixed instructions to reduce cost of rte_smp_mb()
> > 
> > On Fri,  1 Dec 2017 11:12:51 +0000
> > Konstantin Ananyev <konstantin.ananyev@intel.com> wrote:
> >   
> > > On x86 it  is possible to use lock-prefixed instructions to get
> > > the similar effect as mfence.
> > > As pointed by Java guys, on most modern HW that gives a better
> > > performance than using mfence:
> > > https://shipilev.net/blog/2014/on-the-fence-with-dependencies/
> > > That patch adopts that technique for rte_smp_mb() implementation.
> > > On BDW 2.2 mb_autotest on single lcore reports 2X cycle reduction,
> > > i.e. from ~110 to ~55 cycles per operation.
> > >
> > > Signed-off-by: Konstantin Ananyev <konstantin.ananyev@intel.com>
> > > ---
> > >  .../common/include/arch/x86/rte_atomic.h           | 45 +++++++++++++++++++++-
> > >  1 file changed, 43 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/lib/librte_eal/common/include/arch/x86/rte_atomic.h b/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> > > index 4eac66631..07b7fa7f7 100644
> > > --- a/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> > > +++ b/lib/librte_eal/common/include/arch/x86/rte_atomic.h
> > > @@ -55,12 +55,53 @@ extern "C" {
> > >
> > >  #define	rte_rmb() _mm_lfence()
> > >
> > > -#define rte_smp_mb() rte_mb()
> > > -
> > >  #define rte_smp_wmb() rte_compiler_barrier()
> > >
> > >  #define rte_smp_rmb() rte_compiler_barrier()
> > >
> > > +/*
> > > + * From Intel Software Development Manual; Vol 3;
> > > + * 8.2.2 Memory Ordering in P6 and More Recent Processor Families:
> > > + * ...
> > > + * . Reads are not reordered with other reads.
> > > + * . Writes are not reordered with older reads.
> > > + * . Writes to memory are not reordered with other writes,
> > > + *   with the following exceptions:
> > > + *   . streaming stores (writes) executed with the non-temporal move
> > > + *     instructions (MOVNTI, MOVNTQ, MOVNTDQ, MOVNTPS, and MOVNTPD); and
> > > + *   . string operations (see Section 8.2.4.1).
> > > + *  ...
> > > + * . Reads may be reordered with older writes to different locations but not
> > > + * with older writes to the same location.
> > > + * . Reads or writes cannot be reordered with I/O instructions,
> > > + * locked instructions, or serializing instructions.
> > > + * . Reads cannot pass earlier LFENCE and MFENCE instructions.
> > > + * . Writes ... cannot pass earlier LFENCE, SFENCE, and MFENCE instructions.
> > > + * . LFENCE instructions cannot pass earlier reads.
> > > + * . SFENCE instructions cannot pass earlier writes ...
> > > + * . MFENCE instructions cannot pass earlier reads, writes ...
> > > + *
> > > + * As pointed by Java guys, that makes possible to use lock-prefixed
> > > + * instructions to get the same effect as mfence and on most modern HW
> > > + * that gives a better perfomarnce than using mfence:
> > > + * https://shipilev.net/blog/2014/on-the-fence-with-dependencies/
> > > + * So below we use that technique for rte_smp_mb() implementation.
> > > + */
> > > +
> > > +#ifdef RTE_ARCH_I686
> > > +#define	RTE_SP	RTE_STR(esp)
> > > +#else
> > > +#define	RTE_SP	RTE_STR(rsp)
> > > +#endif
> > > +
> > > +#define RTE_MB_DUMMY_MEMP	"-128(%%" RTE_SP ")"
> > > +
> > > +static __rte_always_inline void
> > > +rte_smp_mb(void)
> > > +{
> > > +	asm volatile("lock addl $0," RTE_MB_DUMMY_MEMP "; " ::: "memory");
> > > +}
> > > +
> > >  #define rte_io_mb() rte_mb()
> > >
> > >  #define rte_io_wmb() rte_compiler_barrier()  
> > 
> > The lock instruction is a stronger barrier than the compiler barrier
> > and has worse performance impact. Are you sure it is necessary to use it in DPDK.
> > Linux kernel has successfully used simple compiler reodering barrier for years.  
> 
> Where do you see compiler barrier?
> Right now for x86 rte_smp_mb()==rte_mb()==mfence.
> So I am replacing mfence with 'lock add'.
> As comment above says - on most modern x86 systems it is faster,
> while allow to preserve memory ordering.

There are cases like virtio/vhost where we could be using compiler_barrier.
The mfence to lock add conversion makes sense.

  reply	other threads:[~2018-03-08 21:15 UTC|newest]

Thread overview: 26+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-12-01 11:12 [dpdk-dev] [PATCH 1/2] test/test: introduce new test-case for rte_smp_mb() Konstantin Ananyev
2017-12-01 11:12 ` [dpdk-dev] [PATCH 2/2] eal/x86: Use lock-prefixed instructions to reduce cost of rte_smp_mb() Konstantin Ananyev
2017-12-01 18:04   ` Stephen Hemminger
2017-12-01 23:08     ` Ananyev, Konstantin
2018-03-08 21:15       ` Stephen Hemminger [this message]
2017-12-11 17:11   ` Bruce Richardson
2017-12-11 17:30     ` Ananyev, Konstantin
2017-12-18 15:34   ` [dpdk-dev] [PATCH v2 0/2] eal/x86: Optimize rte_smp_mb() and create a new test case for it Konstantin Ananyev
2017-12-18 15:34   ` [dpdk-dev] [PATCH v2 1/2] test/test: introduce new test-case for rte_smp_mb() Konstantin Ananyev
2018-01-12 17:23     ` Thomas Monjalon
2018-01-12 17:58       ` Ananyev, Konstantin
2018-01-13 13:54     ` Wiles, Keith
2018-01-13 13:54     ` Wiles, Keith
2018-01-15 15:04     ` [dpdk-dev] [PATCH v3 0/2] eal/x86: Optimize rte_smp_mb() and create a new test case for it Konstantin Ananyev
2018-01-15 15:04     ` [dpdk-dev] [PATCH v3 1/2] test/test: introduce new test-case for rte_smp_mb() Konstantin Ananyev
2018-01-16  0:16       ` Thomas Monjalon
2018-01-15 15:04     ` [dpdk-dev] [PATCH v3 2/2] eal/x86: Use lock-prefixed instructions to reduce cost of rte_smp_mb() Konstantin Ananyev
2018-01-15 15:09       ` Konstantin Ananyev
     [not found]         ` <8b05f533-d146-7f97-48f4-82ddcfc3613b@redhat.com>
2018-01-16  1:54           ` [dpdk-dev] Fwd: " Michael S. Tsirkin
2018-01-29  9:29             ` Ananyev, Konstantin
2018-01-29 17:29               ` Michael S. Tsirkin
2018-01-29 15:47         ` [dpdk-dev] " Thomas Monjalon
2018-01-30 19:33           ` Stephen Hemminger
2017-12-18 15:34   ` [dpdk-dev] [PATCH v2 " Konstantin Ananyev
2017-12-18 15:46     ` Bruce Richardson
2017-12-11 17:08 ` [dpdk-dev] [PATCH 1/2] test/test: introduce new test-case for rte_smp_mb() Bruce Richardson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20180308131510.677ff22d@xeon-e3 \
    --to=stephen@networkplumber.org \
    --cc=dev@dpdk.org \
    --cc=konstantin.ananyev@intel.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).