gecko-dev/nsprpub/lib/tests/arena.c

402 строки
11 KiB
C

/* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
/* ***** BEGIN LICENSE BLOCK *****
* Version: MPL 1.1/GPL 2.0/LGPL 2.1
*
* The contents of this file are subject to the Mozilla Public License Version
* 1.1 (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
* http://www.mozilla.org/MPL/
*
* Software distributed under the License is distributed on an "AS IS" basis,
* WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
* for the specific language governing rights and limitations under the
* License.
*
* The Original Code is the Netscape Portable Runtime (NSPR).
*
* The Initial Developer of the Original Code is
* Netscape Communications Corporation.
* Portions created by the Initial Developer are Copyright (C) 1998-2000
* the Initial Developer. All Rights Reserved.
*
* Contributor(s):
*
* Alternatively, the contents of this file may be used under the terms of
* either the GNU General Public License Version 2 or later (the "GPL"), or
* the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
* in which case the provisions of the GPL or the LGPL are applicable instead
* of those above. If you wish to allow use of your version of this file only
* under the terms of either the GPL or the LGPL, and not to allow others to
* use your version of this file under the terms of the MPL, indicate your
* decision by deleting the provisions above and replace them with the notice
* and other provisions required by the GPL or the LGPL. If you do not delete
* the provisions above, a recipient may use your version of this file under
* the terms of any one of the MPL, the GPL or the LGPL.
*
* ***** END LICENSE BLOCK ***** */
/*
** File: arena.c
** Description: Testing arenas
**
*/
#include <string.h>
#include <time.h>
#include <stdlib.h>
#include "nspr.h"
#include "plarena.h"
#include "plgetopt.h"
PRLogModuleInfo *tLM;
PRIntn threadCount = 0;
PRMonitor *tMon;
PRBool failed_already = PR_FALSE;
/* Arguments from the command line with default values */
PRIntn debug_mode = 0;
PRIntn poolMin = 4096;
PRIntn poolMax = (100 * 4096);
PRIntn arenaMin = 40;
PRIntn arenaMax = (100 * 40);
PRIntn stressIterations = 15;
PRIntn maxAlloc = (1024 * 1024);
PRIntn stressThreads = 4;
void DumpAll( void )
{
return;
}
/*
** Test Arena allocation.
*/
static void ArenaAllocate( void )
{
PLArenaPool ap;
void *ptr;
PRInt32 i;
PL_InitArenaPool( &ap, "AllocArena", 2048, sizeof(double));
PR_LOG( tLM, PR_LOG_DEBUG, ("AA, InitPool -- Pool: %p. first: %p, current: %p, size: %d",
&ap, ap.first, ap.current, ap.arenasize ));
for( i = 0; i < 150; i++ )
{
PL_ARENA_ALLOCATE( ptr, &ap, 512 );
PR_LOG( tLM, PR_LOG_DEBUG,("AA, after alloc -- Pool: %p. first: %p, current: %p, size: %d",
&ap, ap.first, ap.current, ap.arenasize ));
PR_LOG( tLM, PR_LOG_DEBUG,(
"AA -- Pool: %p. alloc: %p ", &ap, ptr ));
}
PL_FreeArenaPool( &ap );
for( i = 0; i < 221; i++ )
{
PL_ARENA_ALLOCATE( ptr, &ap, 512 );
PR_LOG( tLM, PR_LOG_DEBUG,("AA, after alloc -- Pool: %p. first: %p, current: %p, size: %d",
&ap, ap.first, ap.current, ap.arenasize ));
PR_LOG( tLM, PR_LOG_DEBUG,(
"AA -- Pool: %p. alloc: %p ", &ap, ptr ));
}
PL_FreeArenaPool( &ap );
return;
} /* end ArenaGrow() */
/*
** Test Arena grow.
*/
static void ArenaGrow( void )
{
PLArenaPool ap;
void *ptr;
PRInt32 i;
PL_InitArenaPool( &ap, "TheArena", 4096, sizeof(double));
PL_ARENA_ALLOCATE( ptr, &ap, 512 );
PR_LOG( tLM, PR_LOG_DEBUG, ("Before growth -- Pool: %p. alloc: %p ", &ap, ptr ));
for( i = 0; i < 10; i++ )
{
PL_ARENA_GROW( ptr, &ap, 512, 7000 );
PR_LOG( tLM, PR_LOG_DEBUG, ("After growth -- Pool: %p. alloc: %p ", &ap, ptr ));
}
return;
} /* end ArenaGrow() */
/*
** Test arena Mark and Release.
*/
static void MarkAndRelease( void )
{
PLArenaPool ap;
void *ptr = NULL;
void *mark0, *mark1;
PRIntn i;
PL_InitArenaPool( &ap, "TheArena", 4096, sizeof(double));
mark0 = PL_ARENA_MARK( &ap );
PR_LOG( tLM, PR_LOG_DEBUG,
("mark0. ap: %p, ap.f: %p, ap.c: %p, ap.siz: %d, alloc: %p, m0: %p",
&ap, ap.first.next, ap.current, ap.arenasize, ptr, mark0 ));
for( i = 0; i < 201; i++ )
{
PL_ARENA_ALLOCATE( ptr, &ap, 512 );
PR_LOG( tLM, PR_LOG_DEBUG,
("mr. ap: %p, ap.f: %p, ap.c: %p, ap.siz: %d, alloc: %p",
&ap, ap.first.next, ap.current, ap.arenasize, ptr ));
}
mark1 = PL_ARENA_MARK( &ap );
PR_LOG( tLM, PR_LOG_DEBUG,
("mark1. ap: %p, ap.f: %p, ap.c: %p, ap.siz: %d, alloc: %p, m1: %p",
&ap, ap.first.next, ap.current, ap.arenasize, ptr, mark1 ));
for( i = 0; i < 225; i++ )
{
PL_ARENA_ALLOCATE( ptr, &ap, 512 );
PR_LOG( tLM, PR_LOG_DEBUG,
("mr. ap: %p, ap.f: %p, ap.c: %p, ap.siz: %d, alloc: %p",
&ap, ap.first.next, ap.current, ap.arenasize, ptr ));
}
PL_ARENA_RELEASE( &ap, mark1 );
PR_LOG( tLM, PR_LOG_DEBUG,
("Release-1: %p -- Pool: %p. first: %p, current: %p, size: %d",
mark1, &ap, ap.first, ap.current, ap.arenasize ));
for( i = 0; i < 20; i++ )
{
PL_ARENA_ALLOCATE( ptr, &ap, 512 );
PR_LOG( tLM, PR_LOG_DEBUG,
("mr. ap: %p, ap.f: %p, ap.c: %p, ap.siz: %d, alloc: %p",
&ap, ap.first.next, ap.current, ap.arenasize, ptr ));
}
PL_ARENA_RELEASE( &ap, mark1 );
PR_LOG( tLM, PR_LOG_DEBUG,
("Release-1. ap: %p, ap.f: %p, ap.c: %p, ap.siz: %d, alloc: %p",
&ap, ap.first.next, ap.current, ap.arenasize, ptr ));
PL_ARENA_RELEASE( &ap, mark0 );
PR_LOG( tLM, PR_LOG_DEBUG,
("Release-0. ap: %p, ap.f: %p, ap.c: %p, ap.siz: %d, alloc: %p",
&ap, ap.first.next, ap.current, ap.arenasize, ptr ));
PL_FreeArenaPool( &ap );
PR_LOG( tLM, PR_LOG_DEBUG,
("Free. ap: %p, ap.f: %p, ap.c: %p, ap.siz: %d, alloc: %p",
&ap, ap.first.next, ap.current, ap.arenasize, ptr ));
PL_FinishArenaPool( &ap );
PR_LOG( tLM, PR_LOG_DEBUG,
("Finish. ap: %p, ap.f: %p, ap.c: %p, ap.siz: %d, alloc: %p",
&ap, ap.first.next, ap.current, ap.arenasize, ptr ));
return;
} /* end MarkAndRelease() */
/*
** RandSize() returns a random number in the range
** min..max, rounded to the next doubleword
**
*/
static PRIntn RandSize( PRIntn min, PRIntn max )
{
PRIntn sz = (rand() % (max -min)) + min + sizeof(double);
sz &= ~sizeof(double)-1;
return(sz);
}
/*
** StressThread()
** A bunch of these beat on individual arenas
** This tests the free_list protection.
**
*/
static void PR_CALLBACK StressThread( void *arg )
{
PLArenaPool ap;
PRIntn i;
PRIntn sz;
void *ptr;
PRThread *tp = PR_GetCurrentThread();
PR_LOG( tLM, PR_LOG_DEBUG, ("Stress Thread %p started\n", PR_GetCurrentThread()));
PL_InitArenaPool( &ap, "TheArena", RandSize( poolMin, poolMax), sizeof(double));
for ( i = 0; i < stressIterations; i++ )
{
PRIntn allocated = 0;
while ( allocated < maxAlloc )
{
sz = RandSize( arenaMin, arenaMax );
PL_ARENA_ALLOCATE( ptr, &ap, sz );
if ( ptr == NULL )
{
PR_LOG( tLM, PR_LOG_ERROR, ("ARENA_ALLOCATE() returned NULL\n\tAllocated: %d\n", allocated));
break;
}
allocated += sz;
}
PR_LOG( tLM, PR_LOG_DEBUG, ("Stress thread %p finished one iteration\n", tp));
PL_FreeArenaPool( &ap );
}
PR_LOG( tLM, PR_LOG_DEBUG, ("Stress thread %p finished all iteration\n", tp));
PL_FinishArenaPool( &ap );
PR_LOG( tLM, PR_LOG_DEBUG, ("Stress thread %p after FinishArenaPool()\n", tp));
/* That's all folks! let's quit */
PR_EnterMonitor(tMon);
threadCount--;
PR_Notify(tMon);
PR_ExitMonitor(tMon);
return;
}
/*
** Stress()
** Flog the hell out of arenas multi-threaded.
** Do NOT pass an individual arena to another thread.
**
*/
static void Stress( void )
{
PRThread *tt;
PRIntn i;
tMon = PR_NewMonitor();
for ( i = 0 ; i < stressThreads ; i++ )
{
PR_EnterMonitor(tMon);
tt = PR_CreateThread(PR_USER_THREAD,
StressThread,
NULL,
PR_PRIORITY_NORMAL,
PR_GLOBAL_THREAD,
PR_UNJOINABLE_THREAD,
0);
threadCount++;
PR_ExitMonitor(tMon);
}
/* Wait for all threads to exit */
PR_EnterMonitor(tMon);
while ( threadCount != 0 )
{
PR_Wait(tMon, PR_INTERVAL_NO_TIMEOUT);
}
PR_ExitMonitor(tMon);
PR_DestroyMonitor(tMon);
return;
} /* end Stress() */
/*
** EvaluateResults()
** uses failed_already to display results and set program
** exit code.
*/
static PRIntn EvaluateResults(void)
{
PRIntn rc = 0;
if ( failed_already == PR_TRUE )
{
PR_LOG( tLM, PR_LOG_DEBUG, ("FAIL\n"));
rc =1;
}
else
{
PR_LOG( tLM, PR_LOG_DEBUG, ("PASS\n"));
}
return(rc);
} /* EvaluateResults() */
void Help( void )
{
printf("arena [options]\n");
printf("where options are:\n");
printf("-p <n> minimum size of an arena pool. Default(%d)\n", poolMin);
printf("-P <n> maximum size of an arena pool. Default(%d)\n", poolMax);
printf("-a <n> minimum size of an arena allocation. Default(%d)\n", arenaMin);
printf("-A <n> maximum size of an arena allocation. Default(%d)\n", arenaMax);
printf("-i <n> number of iterations in a stress thread. Default(%d)\n", stressIterations);
printf("-s <n> maximum allocation for a single stress thread. Default(%d)\n", maxAlloc);
printf("-t <n> number of stress threads. Default(%d)\n", stressThreads );
printf("-d enable debug mode\n");
printf("\n");
exit(1);
}
PRIntn main(PRIntn argc, char *argv[])
{
PLOptStatus os;
PLOptState *opt = PL_CreateOptState(argc, argv, "dhp:P:a:A:i:s:t:");
while (PL_OPT_EOL != (os = PL_GetNextOpt(opt)))
{
if (PL_OPT_BAD == os) continue;
switch (opt->option)
{
case 'a': /* arena Min size */
arenaMin = atol( opt->value );
break;
case 'A': /* arena Max size */
arenaMax = atol( opt->value );
break;
case 'p': /* pool Min size */
poolMin = atol( opt->value );
break;
case 'P': /* pool Max size */
poolMax = atol( opt->value );
break;
case 'i': /* Iterations in stress tests */
stressIterations = atol( opt->value );
break;
case 's': /* storage to get per iteration */
maxAlloc = atol( opt->value );
break;
case 't': /* Number of stress threads to create */
stressThreads = atol( opt->value );
break;
case 'd': /* debug mode */
debug_mode = 1;
break;
case 'h': /* help */
default:
Help();
} /* end switch() */
} /* end while() */
PL_DestroyOptState(opt);
srand( (unsigned)time( NULL ) ); /* seed random number generator */
tLM = PR_NewLogModule("testcase");
#if 0
ArenaAllocate();
ArenaGrow();
#endif
MarkAndRelease();
Stress();
return(EvaluateResults());
} /* end main() */
/* arena.c */