Re: [kaffe] Simple program shows strange results
hi,
In message Re: [kaffe] Simple program shows strange results
on 03/03/05, Ito Kazumitsu [EMAIL PROTECTED] writes:
I have made a simpler test program:
This strange problem occurs with float and double, but not
with int or byte.
$ cat Test.java.m4
public class Test {
public static void main(String[] args) {
T x = C1;
if (x C2) {
System.err.println(xC2: + x);
x = C2;
}
System.err.println(x);
}
}
CVS update and try this again. (Thanks for the test case!)
If you're interested, the causes were:
I modified move_float() in icode.c to look like
move_int/move_ref/etc since it was generating inefficient code on
the PowerPC.
Unfortunately, the x86 jitter depended on the way the old
move_float() worked, so it was not working properly when a slot
was aliased. For example, in the following bytecode:
ldc 10.0
fstore_1
fload_1
Kaffe optimizes away the store and subsequent load since it is
potentially redundant. Unfortunately, because of the bizarro way
the x86 FPU works, this does not work and the 10.0 value never
gets written to the proper place on the stack. Therefore, the
floating point number that gets printed is just whatever garbage
is on the stack at the time.
Solution:
Modify slotAlias() in machine.c to aggressively spill Rreadonce
registers (the x86 float register). This method works and seems
to generate the same code as before on x86 (as far as I know),
without screwing over CPUs with more sensible FPUs.
thanks,
tim stack
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] Simple program shows strange results
In message Re: [kaffe] Simple program shows strange results on 03/03/07, Ito Kazumitsu [EMAIL PROTECTED] writes: I also tried rebuilding kaffe after replacing kaffe/kaffevm/jit/machine.c by Revision 1.51 and without using --with-engine=intrp. The result was also sastisfactory. I am sorry that I reported a wrong thing. The result was NOT sastisfactory. And kaffe/kaffevm/jit/machine.c had nothing to do with my environment, but kaffe/kaffevm/jit3 was used for building kaffe. ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] Simple program shows strange results
In message Re: [kaffe] Simple program shows strange results on 03/03/07, Ito Kazumitsu [EMAIL PROTECTED] writes: I rebuilt CVS version of kaffe on a machine: $ uname -a Linux pub2.hitachi-cable.co.jp 2.0.38 #2 Sun Jan 2 11:44:53 JST 2000 i686 unknown using ./configure --with-engine=intrp --with-rt-jar=/usr/local/kaffe/jre/lib/rt.jar and got sastisfactory results. I reconfirmed this fact. ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] Simple program shows strange results
Interestingly enough, if I run either an interpretive kaffe 1.0.7
plus my MIPS patches, or an interpretive MIPS build of the
current kaffe.org CVS sources on a little-endian MIPS/Linux box
(a MIPS Malta board with a 5Kc processor on it), I get::
[EMAIL PROTECTED] BUGS]$ javac Test.java
[EMAIL PROTECTED] BUGS]$ ls
Hello.class Hello.java Test.class Test.java
[EMAIL PROTECTED] BUGS]$ java Test
9.99
xpf: 9.99 9.0
x pe pf =9.0 7.0 9.0
which looks to me to be your correct result.
So the problem may be unique to x86 platforms, and if so,
that vastly reduces the number of places one needs to look
for the problem. It may be an x86 JIT bug, for example.
Could someone run the test on an interpretive-only x86
kaffe?
Kevin K.
- Original Message -
From: Ito Kazumitsu [EMAIL PROTECTED]
To: [EMAIL PROTECTED]
Sent: Monday, March 03, 2003 5:59 PM
Subject: [kaffe] Simple program shows strange results
The following program:
public class Test {
public static void main(String[] args) {
float x = 0f;
float pb = 15.0f;
float pc = 2.0f;
float pd = 6.0f;
float pe = 7.0f;
float pf = 9.0f;
x = (1.0f - (pc / pd)) * pb;
System.err.println(x);
if (x pe) {
System.err.println(xpe: + x + + pe);
x = pe;
}
else if (x pf) {
System.err.println(xpf: + x + + pf);
x = pf;
}
System.err.println(x pe pf = + x + + pe + + pf);
}
}
shows the following output when run with j2sdk1.4.0:
bash$ java Test
9.99
xpf: 9.99 9.0
x pe pf =9.0 7.0 9.0
which is what I expect. But when run with kaffe, it shows:
bash-2.05b$ java Test
9.99
xpe: 0.0 7.0
x pe pf =7.0 7.0 9.0
The kaffe version is
Engine: Just-in-time v3 Version: 1.1.x-cvs Java Version: 1.1
Configuration/Compilation options:
Compile date : Sun Feb 23 11:17:00 JST 2003
Compile host : ph.maczuka.gcd.org
Install prefix: /usr/local/kaffe
Thread system : unix-jthreads
CC: gcc
CFLAGS: -g -O2 -Wall -Wstrict-prototypes
LDFLAGS :
ChangeLog head: 2003-02-12 Dalibor Topic [EMAIL PROTECTED]
By the way, changing the said program as follows, that is
changing the System.err.println(x) to System.err.println(x= + x),
I get expected output.
public class Test {
public static void main(String[] args) {
float x = 0f;
float pb = 15.0f;
float pc = 2.0f;
float pd = 6.0f;
float pe = 7.0f;
float pf = 9.0f;
x = (1.0f - (pc / pd)) * pb;
System.err.println(x= + x);
if (x pe) {
System.err.println(xpe: + x + + pe);
x = pe;
}
else if (x pf) {
System.err.println(xpf: + x + + pf);
x = pf;
}
System.err.println(x pe pf = + x + + pe + + pf);
}
}
bash-2.05b$ java Test
x=9.99
xpf: 9.99 9.0
x pe pf =9.0 7.0 9.0
I wonder what is happening.
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
___
kaffe mailing list
[EMAIL PROTECTED]
http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] Simple program shows strange results
In message Re: [kaffe] Simple program shows strange results on 03/03/06, Kevin D. Kissell [EMAIL PROTECTED] writes: So the problem may be unique to x86 platforms, and if so, that vastly reduces the number of places one needs to look for the problem. It may be an x86 JIT bug, for example. Could someone run the test on an interpretive-only x86 kaffe? I rebuilt CVS version of kaffe on a machine: $ uname -a Linux pub2.hitachi-cable.co.jp 2.0.38 #2 Sun Jan 2 11:44:53 JST 2000 i686 unknown using ./configure --with-engine=intrp --with-rt-jar=/usr/local/kaffe/jre/lib/rt.jar and got sastisfactory results. ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] Simple program shows strange results
In message Re: [kaffe] Simple program shows strange results on 03/03/05, Ito Kazumitsu [EMAIL PROTECTED] writes: Additional Information: bash-2.05b$ uname -a FreeBSD ph.maczuka.gcd.org 4.7-RELEASE FreeBSD 4.7-RELEASE #0: Wed Oct 9 15:08:34 GMT 2002 [EMAIL PROTECTED]:/usr/obj/usr/src/sys/GENERIC i386 This problem occurs on Linux, too. $ java Test; uname -a; kaffe -fullversion 4.7761093E-34 Linux pub2.hitachi-cable.co.jp 2.0.38 #2 Sun Jan 2 11:44:53 JST 2000 i686 unknown Kaffe Virtual Machine Copyright (c) 1996-2002 Kaffe.org project contributors (please see the source code for a full list of contributors). All rights reserved. Portions Copyright (c) 1996-2002 Transvirtual Technologies, Inc. The Kaffe virtual machine is free software, licensed under the terms of the GNU General Public License. Kaffe.org is a an independent, free software community project, not directly affiliated with Transvirtual Technologies, Inc. Kaffe is a Trademark of Transvirtual Technologies, Inc. Kaffe comes with ABSOLUTELY NO WARRANTY. Engine: Just-in-time v3 Version: 1.1.x-cvs Java Version: 1.1 Configuration/Compilation options: Compile date : Wed Mar 5 21:02:05 JST 2003 Compile host : pub2.hitachi-cable.co.jp Install prefix: /usr/local/kaffe Thread system : unix-jthreads CC: gcc CFLAGS: -g -O2 -Wall -Wstrict-prototypes LDFLAGS : ChangeLog head: 2003-03-01 Dalibor Topic [EMAIL PROTECTED] ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] Simple program shows strange results
: == Dalibor Topic [EMAIL PROTECTED] writes: : Does the problem occurs when the test is compiled with : jikes or Sun's javac, too? Yes, I tried with the class file generated by Sun's javac and also by GCJ. ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
Re: [kaffe] Simple program shows strange results
In message Re: [kaffe] Simple program shows strange results on 03/03/06, Ito Kazumitsu [EMAIL PROTECTED] writes: These are the results: [EMAIL PROTECTED] ito]$ for i in kjc gcj Sun-javac; do ln -s -f Test.class.$i Test.class ; java Test; done 4.7761093E-34 4.7761093E-34 4.7761203E-34 On the other hand, all of them show good results when run with Sun's java or compiled by GCJ. bash$ for i in kjc gcj Sun-javac; do cp Test.class.$i Test.class; java Test; done 10.0 10.0 10.0 bash$ for i in kjc gcj Sun-javac; do cp Test.class.$i Test.class; gcj --main=Test Test.class; ./a.exe; rm a.exe; done 10.0 10.0 10.0 ___ kaffe mailing list [EMAIL PROTECTED] http://kaffe.org/cgi-bin/mailman/listinfo/kaffe
