From owner-freebsd-threads@FreeBSD.ORG  Thu Mar  3 07:30:19 2005
Return-Path: <owner-freebsd-threads@FreeBSD.ORG>
Delivered-To: freebsd-threads@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id 8073C16A4CE
	for <threads@freebsd.org>; Thu,  3 Mar 2005 07:30:19 +0000 (GMT)
Received: from mail.emict.com (brig.emict.com [212.90.172.42])
	by mx1.FreeBSD.org (Postfix) with ESMTP id 8194A43D1F
	for <threads@freebsd.org>; Thu,  3 Mar 2005 07:30:18 +0000 (GMT)
	(envelope-from andrit@ukr.net)
Received: from BORJA (unknown [203.199.120.221])
	by mail.emict.com (Postfix) with ESMTP id 73D9333649
	for <threads@freebsd.org>; Thu,  3 Mar 2005 09:30:15 +0200 (EET)
Message-ID: <005201c51fc2$d8676b60$090210ac@BORJA>
From: "Andriy Tkachuk" <andrit@ukr.net>
To: <threads@freebsd.org>
References: <Pine.GSO.4.43.0503021042310.26125-100000@sea.ntplx.net>
	<000b01c51fbb$4189ea30$090210ac@BORJA>
Date: Thu, 3 Mar 2005 13:00:12 +0530
MIME-Version: 1.0
Content-Type: multipart/mixed;
	boundary="----=_NextPart_000_004F_01C51FF0.EF8D15A0"
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 6.00.2800.1158
x-mimeole: Produced By Microsoft MimeOLE V6.00.2800.1165
Subject: 
	Re: patch for threads/76690 - critical - fork hang in child for-lc_r
X-BeenThere: freebsd-threads@freebsd.org
X-Mailman-Version: 2.1.1
Precedence: list
List-Id: Threading on FreeBSD <freebsd-threads.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-threads>,
	<mailto:freebsd-threads-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-threads>
List-Post: <mailto:freebsd-threads@freebsd.org>
List-Help: <mailto:freebsd-threads-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-threads>,
	<mailto:freebsd-threads-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Thu, 03 Mar 2005 07:30:19 -0000

This is a multi-part message in MIME format.

------=_NextPart_000_004F_01C51FF0.EF8D15A0
Content-Type: text/plain;
	charset="ISO-8859-1"
Content-Transfer-Encoding: 7bit

But if one wants to use pure user threads
on his UP system, what he will chose if not libc_r ?

And i have some test program with shows
the better results for libc_r than for libpthreads.
Take a look.

The program is the 500 threads, each of them allocate
memory in loop and then free it in another loop.
Program outputs the time consumed for this two loops.

See the results.

> ./a.out m 1000 10        # 1000 iterations, malloc by 10 bytes memory
chanks
thread 0 created
thread 1 created
thread 2 created
thread 3 created
...
thread 499 created
0.000870
0.000544
0.000695
0.000594
0.000913
...    (after some time, say 10-20 seconds, there is stable picture: )
0.001056
0.000756
0.000485
0.000476
0.000528
0.000479
0.000550
0.000481
0.000561
0.000541
0.000483
0.000558
0.000482
0.000561
0.000532
0.000479
0.000551
0.000484
0.000923
0.001180
0.000981
0.000492
0.000996
0.000536
0.000975
0.000534

this is for libc_r. Let's see what about lpthreads.

> c++ -lpthread test2.cc
> ./a.out m 1000 10
thread 0 created
thread 1 created
...
thread 498 created
thread 499 created
0.001704
0.001707
0.001671
0.001683
0.001691
0.001664
0.001661
0.001683
0.001709
...
0.001643
0.002018
0.001668
0.001672
0.001744
0.001689
0.001672
0.001682
0.001643
0.001687
0.001670
0.001692
0.001677
0.001732
0.001650
0.001685
0.001678
0.001685
1.303800
5.266314
5.268821
5.268158
5.268539
5.268644
5.268515
5.268963
5.269147
5.268968
5.268928
...
5.341573
5.341051
5.341417
5.341327
5.341084
5.340510
5.340974
5.340914
5.341118
...


So what? The test2.cc is attached.

(of course all was done on current:
> uname -a
FreeBSD fbsd 6.0-CURRENT FreeBSD 6.0-CURRENT #0: Tue Mar  1 16:03:09 IST
2005     ant@fbsd:/usr/obj/usr/src/sys/FBSD  i386
)


> > Is there a reason you are still using libc_r?
>
> I don't use it at all, but there was critical bug
> i threads-pr - so i just decided to fix it since i
> have had some time )
>
> If this library is needless, why is it still there
> and this pr in criticals?

------=_NextPart_000_004F_01C51FF0.EF8D15A0
Content-Type: application/octet-stream;
	name="test2.cc"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: attachment;
	filename="test2.cc"

#include <pthread.h>=0A=
#include <stdio.h>=0A=
#include <unistd.h>=0A=
#include <sys/time.h>=0A=
#include <string>=0A=
=0A=
using std::string;=0A=
=0A=
#define COUNT_THREADS	500=0A=
=0A=
pthread_t   thread[COUNT_THREADS];=0A=
=0A=
char mode;=0A=
int iters;=0A=
int chanksz;=0A=
=0A=
void f()=0A=
{=0A=
  struct timeval t1,t2;=0A=
  void* p[iters];=0A=
  string s;=0A=
=0A=
  sleep(2);=0A=
  =0A=
  while (1)=0A=
  {=0A=
=0A=
    gettimeofday(&t1, NULL);=0A=
=0A=
    if (mode =3D=3D 's') {=0A=
	    s =3D "";=0A=
	    for (int i=3D0; i<iters; i++)=0A=
		    s +=3D 'a';=0A=
    }=0A=
    if (mode =3D=3D 'S') {=0A=
	    string s;=0A=
	    for (int i=3D0; i<iters; i++)=0A=
		    s =3D 'a';=0A=
    }=0A=
=0A=
    if (mode =3D=3D 'f')=0A=
	    for (int i=3D0; i<iters; i++)=0A=
		    gettimeofday(&t2, NULL);=0A=
=0A=
    if (mode =3D=3D 'm') {=0A=
	    for (int i=3D0; i<iters; i++)=0A=
		    p[i] =3D malloc(chanksz);=0A=
	    for (int i=3D0; i<iters; i++)=0A=
		    free(p[i]);=0A=
    }=0A=
    if (mode =3D=3D 'n') {=0A=
	    for (int i=3D0; i<iters; i++)=0A=
		    p[i] =3D  (char*) new char(chanksz);=0A=
	    for (int i=3D0; i<iters; i++)=0A=
		    delete (char*) p[i];=0A=
    }=0A=
=0A=
    gettimeofday(&t2, NULL);=0A=
=0A=
    int usec_d =3D t2.tv_usec-t1.tv_usec;=0A=
    int sec_d =3D t2.tv_sec-t1.tv_sec;=0A=
    if (usec_d<0) {=0A=
    	sec_d--;=0A=
	usec_d +=3D 1000000;=0A=
    }=0A=
=0A=
    printf( "%u.%06u \n",  sec_d, usec_d);=0A=
  }=0A=
}=0A=
=0A=
=0A=
int main(int argc, char *argv[])=0A=
{=0A=
=0A=
   mode =3D argv[1][0];=0A=
   iters =3D atoi(argv[2]);=0A=
=0A=
  if (mode =3D=3D 'n' || mode =3D=3D 'm')=0A=
    chanksz =3D atoi(argv[3]);=0A=
  =0A=
  for (int i=3D0; i<COUNT_THREADS; i++) {=0A=
     pthread_create ( &(thread[i]), NULL, (void*(*)(void*))f, NULL );=0A=
     printf("thread %i created\n", i);=0A=
  }=0A=
  while(1) { sleep(1); }=0A=
}=0A=

------=_NextPart_000_004F_01C51FF0.EF8D15A0--