mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: kumon@flab.fujitsu.co.jp
To: Andrew Morton <andrewm@uow.edu.au>
Cc: kumon@flab.fujitsu.co.jp, linux-kernel@vger.kernel.org
Cc: kumon@flab.fujitsu.co.jp
Subject: Re: [PATCH] Re: Negative scalability by removal of lock_kernel()?(Was:  Strange performance behavior of 2.4.0-test9)
Date: Thu, 2 Nov 2000 20:09:07 +0900	[thread overview]
Message-ID: <200011021109.UAA24021@asami.proc.flab.fujitsu.co.jp> (raw)
In-Reply-To: <39FEE701.CAC21AE5@uow.edu.au>
In-Reply-To: <39F957BC.4289FF10@uow.edu.au> <39F92187.A7621A09@timpanogas.org> <Pine.GSO.4.21.0010270257550.18660-100000@weyl.math.psu.edu> <20001027094613.A18382@gruyere.muc.suse.de> <200010271257.VAA24374@asami.proc.flab.fujitsu.co.jp> <39FEE701.CAC21AE5@uow.edu.au>

Andrew Morton writes:
 > This patch is a moderate rewrite of __wake_up_common.  I'd be
 > interested in seeing how much difference it makes to the
 > performance of Apache when the file-locking serialisation is
 > disabled.
 > - It implements last-in/first-out semantics for waking
 >   TASK_EXCLUSIVE tasks.
 > - It fixes what was surely a bug wherein __wake_up_common
 >   scans the entire wait queue even when it has found the
 >   task which it wants to run.

We've measured Apache w/ and w/o serialize_accept on
several kernel configurations.

Apache compilation settings are those two:
	* without option (conventional setting)
	  (w/ serialize)
	* with -DSINGLE_LISTEN_UNSERIALIZED_ACCEPT
	  (w/o serialize)

We compared the performance of distributed binary and our binary with
default setting, the performance is almost equivalent. All following
data are based on our binaries.

			w/ serialize	w/o serialize
2.40-t10-pre5		2237		5358
2.40-t10-pre5+P2	5253		5355**
2.40-t10-pre5+P3	---		NG

** with this configuration, once we observed the machine completely
deadlocked with the following message:

Unable to handle kernel NULL pointer dereferenceNMI watchdog detected LOCKUP on CPU1.

Actually, the NMI message followed few seconds after the NULL
dereference message.
The virtual address nor eip are not displayed.

 > It's stable, but it's a work-in-progress.
 > - __wake_up_common does a fair amount of SMP-specific
 >   stuff when compiled for UP which needs sorting out
 > - it seems there's somebody in the networking code who
 >   changes a task state incorrectly when it's on a wait
 >   queue. This used to be OK, but it's not OK now that
 >   I'm relying upon the wait queue being in the state
 >   which it should be.

Your last patch makes the problem very clear.

2.4.0-test10-pre5 with the LIFO patch (P3), we can't get the values.
It always deadlock with same manner.  Perhaps, it failed to get
console lock then deadlock.

What we can now is combining several printk's in the null-pointer
dereference message into one printk and make it display much usefull
data.

--
Computer Systems Laboratory, Fujitsu Labs.
kumon@flab.fujitsu.co.jp
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
Please read the FAQ at http://www.tux.org/lkml/

  parent reply	other threads:[~2000-11-02 11:09 UTC|newest]

Thread overview: 41+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <200010261405.XAA19135@asami.proc.flab.fujitsu.co.jp>
     [not found] ` <200010250736.QAA12373@asami.proc.flab.fujitsu.co.jp>
     [not found]   ` <Pine.LNX.4.21.0010251242050.943-100000@duckman.distro.conectiva>
     [not found]     ` <200010260138.KAA17028@asami.proc.flab.fujitsu.co.jp>
2000-10-27  6:24       ` Negative scalability by removal of lock_kernel()? (Was: " kumon
2000-10-27  6:32         ` Negative scalability by removal of lock_kernel()?(Was: " Jeff V. Merkey
2000-10-27  7:13           ` Alexander Viro
2000-10-27  7:46             ` Andi Kleen
2000-10-27 10:23               ` Andrew Morton
2000-10-27 10:25                 ` Andi Kleen
2000-10-27 12:57               ` [PATCH] " kumon
2000-10-28 15:46                 ` Andrew Morton
2000-10-28 15:58                   ` Andi Kleen
2000-10-28 16:05                   ` Jeff Garzik
2000-10-28 16:20                   ` [PATCH] Re: Negative scalability by removal of lock_kernel()?(Was: Alan Cox
2000-10-29 19:45                     ` dean gaudet
2000-10-30  6:29                       ` Andi Kleen
2000-10-30 15:28                         ` Andrea Arcangeli
2000-10-30 16:36                           ` Rik van Riel
2000-10-30 18:02                             ` Andrea Arcangeli
2000-10-28 16:46                   ` [PATCH] Re: Negative scalability by removal of lock_kernel()?(Was: Strange performance behavior of 2.4.0-test9) Andrew Morton
2000-10-30  9:27                   ` kumon
2000-10-30 15:00                     ` Andrew Morton
2000-10-30 23:24                       ` dean gaudet
2000-11-04  5:08                         ` [PATCH] Re: Negative scalability by removal of lock_kernel()?(Was:Strange " Andrew Morton
2000-11-04  6:23                           ` Linus Torvalds
2000-11-04 10:54                             ` [PATCH] Re: Negative scalability by removal of Alan Cox
2000-11-04 17:22                               ` Linus Torvalds
2000-11-05 16:22                                 ` Andrea Arcangeli
2000-11-05 20:21                               ` dean gaudet
2000-11-05 22:43                                 ` Alan Cox
2000-11-04 20:03                             ` [PATCH] Re: Negative scalability by removal of lock_kernel()?(Was:Strange performance behavior of 2.4.0-test9) dean gaudet
2000-11-04 20:42                               ` [PATCH] Re: Negative scalability by removal of lock_kernel()?(Was:Strange Alan Cox
2000-11-04 20:11                           ` [PATCH] Re: Negative scalability by removal of lock_kernel()?(Was:Strange performance behavior of 2.4.0-test9) dean gaudet
2000-11-04 20:43                             ` [PATCH] Re: Negative scalability by removal of lock_kernel()?(Was:Strange Alan Cox
2000-11-05  4:52                               ` dean gaudet
2000-10-31 15:36                 ` [PATCH] Re: Negative scalability by removal of lock_kernel()?(Was: Strange performance behavior of 2.4.0-test9) Andrew Morton
2000-11-02 12:50                   ` kumon
2000-11-01  1:02                 ` kumon
2000-11-02 11:09                 ` kumon [this message]
2000-11-04  5:07                   ` Andrew Morton
2000-10-27  8:17             ` Jeff V. Merkey
2000-11-04  5:55             ` Preemptive scheduling of woken-up processes kumon
2000-10-27 10:11           ` Negative scalability by removal of lock_kernel()?(Was: Strange performance behavior of 2.4.0-test9) kumon
2000-11-05  4:19 [PATCH] Re: Negative scalability by removal of lock_kernel()?(Was:Strange " Dave Wagner

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=200011021109.UAA24021@asami.proc.flab.fujitsu.co.jp \
    --to=kumon@flab.fujitsu.co.jp \
    --cc=andrewm@uow.edu.au \
    --cc=linux-kernel@vger.kernel.org \
    /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

all inboxes | Powered by JetHome®