All of lore.kernel.org
 help / color / mirror / Atom feed
From: "H. Peter Anvin" <hpa@zytor.com>
To: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Steven Rostedt <rostedt@goodmis.org>,
	Thomas Gleixner <tglx@linutronix.de>,
	Koki Sanagi <sanagi.koki@jp.fujitsu.com>,
	Peter Zijlstra <peterz@infradead.org>,
	Ingo Molnar <mingo@elte.hu>,
	Frederic Weisbecker <fweisbec@gmail.com>,
	nhorman@tuxdriver.com, scott.a.mcmillan@intel.com,
	laijs@cn.fujitsu.com, LKML <linux-kernel@vger.kernel.org>,
	eric.dumazet@gmail.com, kaneshige.kenji@jp.fujitsu.com,
	David Miller <davem@davemloft.net>,
	izumi.taku@jp.fujitsu.com, kosaki.motohiro@jp.fujitsu.com,
	Heiko Carstens <heiko.carstens@de.ibm.com>,
	"Luck, Tony" <tony.luck@intel.com>,
	Jason Baron <jbaron@redhat.com>
Subject: Re: [PATCH] tracing: Cleanup the convoluted softirq tracepoints
Date: Tue, 19 Oct 2010 15:49:14 -0700	[thread overview]
Message-ID: <4CBE206A.20702@zytor.com> (raw)
In-Reply-To: <20101019224126.GD3519@Krystal>

On 10/19/2010 03:41 PM, Mathieu Desnoyers wrote:
>>
>> OK, first of all, there are some serious WTFs here:
>>
>> # define JUMP_LABEL_INITIAL_NOP ".byte 0xe9 \n\t .long 0\n\t"
>>
>> A jump instruction is one of the worst possible NOPs.  Why are we doing
>> this?
> 
> This code is dynamically patched at boot time (and module load time) with a
> better nop, just like the function tracer does.
> 

That's just ridiculous... start out with something sane and you at least
have the chance of not having to patch it.

> Intel's manual "Intel 64 and IA-32 Architectures Optimization Reference Manual"
> 
> http://www.intel.com/Assets/PDF/manual/248966.pdf
> 
> Page C-33 (or 577 in the pdf)
> 
> "7. Selection of conditional jump instructions should be based on the
>     recommendation of section Section 3.4.1, “Branch Prediction Optimization,” to
>     improve the predictability of branches. When branches are predicted
>     successfully, the latency of jcc is effectively zero."
> 
> So it mentions "jcc", but not jmp. Is there any reason for jmp to have a higher
> latency than jcc ?
> 
> In this manual, the latency of predicted jcc is therefore 0 cycle, and its
> throughput is 0.5 cycle/insn.
> 
> NOP (page C-29) is stated to have a latency of 0.5 to 1 cycle/insn (depending on
> the exact HW), and throughput of 0.5 cycle/insn.
> 
> However, I have not found "jmp" explicitly in this listing.
> 
> So if we were executing tracepoints in a maze of jumps, we could argue that
> instruction throughput is the most important there. However, if we expect the
> common case to be surrounded by some non-ALU instructions, latency tends to
> become the most important criterion.
> 
> But I feel I might be missing something important that distinguish "jcc" from
> "jmp".

NOP has a latency of 0.5-1.0 cycle/insns, *but has no consumers*.

JMP/Jcc does have a consumer -- the IP -- and actually measuring shows
that it is much, much worse than NOP and other dummy instructions.

	-hpa

  reply	other threads:[~2010-10-19 22:50 UTC|newest]

Thread overview: 93+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-08-23  9:41 [PATCH v4 0/5] netdev: show a process of packets Koki Sanagi
2010-08-23  9:42 ` [PATCH v4 1/5] irq: add tracepoint to softirq_raise Koki Sanagi
2010-09-03 15:29   ` Frederic Weisbecker
2010-09-03 15:39     ` Steven Rostedt
2010-09-03 15:42       ` Frederic Weisbecker
2010-09-03 15:43     ` Steven Rostedt
2010-09-03 15:50       ` Frederic Weisbecker
2010-09-06  1:46         ` Koki Sanagi
2010-09-08  8:33   ` [tip:perf/core] irq: Add " tip-bot for Lai Jiangshan
2010-09-08 11:25     ` [sparc build bug] " Ingo Molnar
2010-09-08 12:26       ` [PATCH] irq: Fix circular headers dependency Frederic Weisbecker
2010-09-09 19:54         ` [tip:perf/core] " tip-bot for Frederic Weisbecker
2010-10-18  9:44       ` [sparc build bug] Re: [tip:perf/core] irq: Add tracepoint to softirq_raise Peter Zijlstra
2010-10-18 10:11         ` Peter Zijlstra
2010-10-18 10:26           ` Heiko Carstens
2010-10-18 10:48         ` Peter Zijlstra
2010-10-19 10:58           ` Koki Sanagi
2010-10-19 11:25             ` Peter Zijlstra
2010-10-19 13:00             ` [PATCH] tracing: Cleanup the convoluted softirq tracepoints Thomas Gleixner
2010-10-19 13:08               ` Peter Zijlstra
2010-10-19 13:22               ` Mathieu Desnoyers
2010-10-19 13:41                 ` Thomas Gleixner
2010-10-19 13:54                   ` Steven Rostedt
2010-10-19 14:07                     ` Thomas Gleixner
2010-10-19 14:28                       ` Mathieu Desnoyers
2010-10-19 19:49                         ` Thomas Gleixner
2010-10-19 20:55                           ` Steven Rostedt
2010-10-19 21:07                             ` Thomas Gleixner
2010-10-19 21:23                               ` Steven Rostedt
2010-10-19 21:48                                 ` H. Peter Anvin
2010-10-19 22:23                                   ` Steven Rostedt
2010-10-19 22:26                                     ` H. Peter Anvin
2010-10-19 22:27                                     ` Peter Zijlstra
2010-10-19 23:39                                       ` H. Peter Anvin
2010-10-19 23:45                                         ` Steven Rostedt
2010-10-20  0:43                                         ` Jason Baron
2010-10-19 22:41                                   ` Mathieu Desnoyers
2010-10-19 22:49                                     ` H. Peter Anvin [this message]
2010-10-19 23:05                                       ` Steven Rostedt
2010-10-19 23:09                                         ` H. Peter Anvin
2010-10-20 15:27                                         ` Jason Baron
2010-10-20 15:41                                           ` Mathieu Desnoyers
2010-10-25 21:54                                           ` H. Peter Anvin
2010-10-25 22:01                                             ` Mathieu Desnoyers
2010-10-25 22:12                                               ` H. Peter Anvin
2010-10-25 22:19                                                 ` H. Peter Anvin
2010-10-25 22:55                                                 ` Mathieu Desnoyers
2010-10-26  0:39                                                   ` Steven Rostedt
2010-10-26  1:14                                                     ` Mathieu Desnoyers
2010-10-19 22:04                                 ` Thomas Gleixner
2010-10-19 22:33                                   ` Steven Rostedt
2010-10-21 16:18                                     ` Thomas Gleixner
2010-10-21 17:05                                       ` Steven Rostedt
2010-10-21 19:56                                         ` Thomas Gleixner
2010-10-25 22:31                                           ` H. Peter Anvin
2010-10-19 21:45                             ` Thomas Gleixner
2010-10-19 22:14                               ` Steven Rostedt
2010-10-19 21:16                           ` David Daney
2010-10-19 21:32                             ` Jason Baron
2010-10-19 21:38                               ` David Daney
2010-10-19 21:47                             ` Steven Rostedt
2010-10-19 21:28                           ` Jason Baron
2010-10-19 21:55                             ` Thomas Gleixner
2010-10-19 22:17                               ` Thomas Gleixner
2010-10-20  1:36                                 ` Steven Rostedt
2010-10-20  1:52                                   ` Jason Baron
2010-10-25 22:32                                     ` H. Peter Anvin
2010-10-19 22:38                               ` Jason Baron
2010-10-19 22:44                                 ` H. Peter Anvin
2010-10-19 22:56                                   ` Steven Rostedt
2010-10-19 22:57                                     ` H. Peter Anvin
2010-10-19 14:46                       ` Steven Rostedt
2010-10-19 14:00                   ` Mathieu Desnoyers
2010-10-21 14:52               ` [tip:perf/core] " tip-bot for Thomas Gleixner
2010-08-23  9:43 ` [PATCH v4 2/5] napi: convert trace_napi_poll to TRACE_EVENT Koki Sanagi
2010-08-24  3:52   ` David Miller
2010-09-08  8:34   ` [tip:perf/core] napi: Convert " tip-bot for Neil Horman
2010-08-23  9:45 ` [PATCH v4 3/5] netdev: add tracepoints to netdev layer Koki Sanagi
2010-08-24  3:53   ` David Miller
2010-09-08  8:34   ` [tip:perf/core] netdev: Add " tip-bot for Koki Sanagi
2010-08-23  9:46 ` [PATCH v4 4/5] skb: add tracepoints to freeing skb Koki Sanagi
2010-08-24  3:53   ` David Miller
2010-09-08  8:35   ` [tip:perf/core] skb: Add " tip-bot for Koki Sanagi
2010-08-23  9:47 ` [PATCH v4 5/5] perf:add a script shows a process of packet Koki Sanagi
2010-08-24  3:53   ` David Miller
2010-09-07 16:57   ` Frederic Weisbecker
2010-09-08  8:35   ` [tip:perf/core] perf: Add a script to show packets processing tip-bot for Koki Sanagi
2010-08-30 23:50 ` [PATCH v4 0/5] netdev: show a process of packets Steven Rostedt
2010-09-03  2:10   ` Koki Sanagi
2010-09-03  2:17     ` David Miller
2010-09-03  2:55       ` Koki Sanagi
2010-09-03  4:46         ` Frederic Weisbecker
2010-09-03  5:12           ` Koki Sanagi

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=4CBE206A.20702@zytor.com \
    --to=hpa@zytor.com \
    --cc=davem@davemloft.net \
    --cc=eric.dumazet@gmail.com \
    --cc=fweisbec@gmail.com \
    --cc=heiko.carstens@de.ibm.com \
    --cc=izumi.taku@jp.fujitsu.com \
    --cc=jbaron@redhat.com \
    --cc=kaneshige.kenji@jp.fujitsu.com \
    --cc=kosaki.motohiro@jp.fujitsu.com \
    --cc=laijs@cn.fujitsu.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mathieu.desnoyers@efficios.com \
    --cc=mingo@elte.hu \
    --cc=nhorman@tuxdriver.com \
    --cc=peterz@infradead.org \
    --cc=rostedt@goodmis.org \
    --cc=sanagi.koki@jp.fujitsu.com \
    --cc=scott.a.mcmillan@intel.com \
    --cc=tglx@linutronix.de \
    --cc=tony.luck@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.