Date: Tue, 11 Apr 1995 22:11:13 +0300 From: Kai Vorma <vode@snakemail.hut.fi> To: hackers@FreeBSD.org Subject: A better malloc() for FreeBSD Message-ID: <199504111911.WAA08335@vinkku.hut.fi>
next in thread | raw e-mail | index | archive | help
As I wrote about two weeks ago the BSD malloc isn't very
space-efficient but wastes memory quite liberally. This is because of
two main reasons
1) It allocates blocks that are 2^n-4 bytes long. So if you say
malloc(1024) you actually get 2044 bytes.
2) It cannot coalesc freed blocks:
break is 0x1fbffc (2072032 bytes sbrked) after 1000 allocations of 1024
break is 0x1fbffc (2072032 bytes sbrked) after freeing all allocations
break is 0x27cffc (2600416 bytes sbrked) after allocating 512000
I searched for better malloc implementations and found two good
candidates:
o CSRI malloc v1.17 by Mark Moraes
o malloc-2.5.3b by Doug Lea
Both mallocs are very space-efficient and performs coalescing on
frees. The CSRI malloc is a bit slow and somewhat bloated (it does
have hooks for debugging and some helper functions like strdup,
emalloc, evalloc etc.). Doug Lea's malloc is very fast and simple so I
choosed it.
I replaced the BSD malloc in libc with Doug Lea's malloc about week
ago and recompiled all static programs in /bin and /sbin. So far I
have had only two problems:
1) sed crashes in some situations. At the end of this message is a
patch (commited to freebsd-current a few days ago) that fixes sed.
2) Originally I installed Doug Lea's malloc as stdlib/dlmalloc.c
and just commented out old malloc (gen/valloc.c, stdlib/malloc.c
and stdlib/calloc.c) from makefiles. Everything worked smoothly
until I tried to link something with static gnumalloc (linker
complained about multiple defined symbols _malloc etc. Why?) so I had
to replace the original stdlib/malloc.c with dlmalloc.c and comment
out calloc() and valloc() from it.
Does it help? Perhaps :-) I have 16MB of memory so things go smoothly
anyway. However, the virtual size of XF_86 dropped from 7MB to 4MB and
xv no longer grows until it hits datasize limit. I would suppose that
the new malloc causes less TLB and page faults, too.
If you want to try your luck then fetch malloc-2.5.3b.c from
gee.cs.oswego.edu:/pub/misc, apply the following patch to it and
install it as /usr/src/lib/libc/stdlib/malloc.c. Then recompile and
install libc (take a backup of your old libc.so first! Actually take a
full backup of your system first - I managed to destroy my cnews
installation before I found and fixed a bug in sed).
I'd like to see Doug Lea's malloc integrated into libc at some
point. I don't think it should go into 2.1 but how about -current
after 2.1 is out?
..vode
PS. The comments says that it is a preliminary working version and
unreleased. I asked the author and he said that it is stable and
he has just forgotten to remove those comments.
---------------------------------------------------------------------------
*** usr/src/usr.bin/sed/compile.c.old Tue Apr 11 10:44:45 1995
--- usr/src/usr.bin/sed/compile.c Thu Apr 6 05:15:37 1995
***************
*** 217,222 ****
--- 217,224 ----
if (!*p)
p = NULL;
cmd2 = xmalloc(sizeof(struct s_command));
+ cmd2->nonsel = cmd2->inrange = 0;
+ cmd2->a1 = cmd2->a2 = NULL;
cmd2->code = '}';
*compile_stream("}", &cmd->u.c, p) = cmd2;
cmd->next = cmd2;
---------------------------------------------------------------------------
*** malloc-2.5.3b.c Sun Mar 26 10:50:54 1995
--- malloc.c Fri Apr 7 19:21:29 1995
***************
*** 271,276 ****
--- 271,278 ----
/* preliminaries */
+ #include <sys/param.h>
+
#ifndef __STD_C
#ifdef __STDC__
#define __STD_C 1
***************
*** 1365,1373 ****
}
-
/* Derivatives */
#if __STD_C
Void_t* valloc(size_t bytes)
#else
--- 1367,1375 ----
}
/* Derivatives */
+ #if 0
#if __STD_C
Void_t* valloc(size_t bytes)
#else
***************
*** 1394,1400 ****
while (sz-- > 0) *q++ = 0;
return p;
}
!
#if __STD_C
void cfree(Void_t *mem)
#else
--- 1396,1402 ----
while (sz-- > 0) *q++ = 0;
return p;
}
! #endif
#if __STD_C
void cfree(Void_t *mem)
#else
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199504111911.WAA08335>