Re: java/6071: GCJ does not compile a large (21417 lines) java source file

Re: java/6071: GCJ does not compile a large (21417 lines) java source file

[steven]

Synopsis: GCJ does not compile a large (21417 lines) java source file

State-Changed-From-To: open->closed
State-Changed-By: steven
State-Changed-When: Tue May 13 06:25:49 2003
State-Changed-Why:
    Should have been in "feedback" state.  Not heard of the reporter for over a year.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=6071

Re: java/6071: GCJ does not compile a large (21417 lines) java source file

[Anthony Green]

The following reply was made to PR java/6071; it has been noted by GNATS.

From: Anthony Green <green@redhat.com>
To: shroff@transeda.com
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: java/6071: GCJ does not compile a large (21417 lines) java
	source file
Date: 26 Apr 2002 15:52:51 -0700

 On Wed, 2002-03-27 at 09:31, shroff@transeda.com wrote:
 > 
 > >Number:         6071
 > >Category:       java
 > >Synopsis:       GCJ does not compile a large (21417 lines) java source file
 >
 
 GCJ in the latest 3.1 snapshot should build on Solaris 2.7 now.  Could
 you please give this a go?
 
 Thanks!
 
 AG
 
 

Re: java/6071: GCJ does not compile a large (21417 lines) java source file

[Alexandre Petit-Bianco]

The following reply was made to PR java/6071; it has been noted by GNATS.

From: Alexandre Petit-Bianco <apbianco@cygnus.com>
To: shroff@transeda.com, gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: java/6071: GCJ does not compile a large (21417 lines) java source 
 file
Date: Thu, 28 Mar 2002 10:15:53 -0800 (PST)

 Saurin B. Shroff writes:
 
 > Unfortunately I have not been successful in downloading the source
 > tree and compiling it on my machine (Solaris 2.7). Compile always
 > fails due to some error in one of the include file. My current
 > distribution (3.0.1) was downloaded as precompiled binaries.
 
 OK, so you're really on 2.7 (I thought it was a typo as your 3.0.1
 invokations indicates sparc-sun-solaris2.8; hence I thought you would
 have been able to build 3.1, see http://gcc.gnu.org/java/gcj-3.1-status.html)
 
 I think you should report the 2.7 build failure.
 
 ./A

Re: java/6071: GCJ does not compile a large (21417 lines) java source file

[Tom Tromey]

The following reply was made to PR java/6071; it has been noted by GNATS.

From: Tom Tromey <tromey@redhat.com>
To: shroff@transeda.com
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: java/6071: GCJ does not compile a large (21417 lines) java source file
Date: 27 Mar 2002 15:02:56 -0700

 >>>>> "Saurin" == Saurin B Shroff <shroff@transeda.com> writes:
 
 Saurin> Due to proprietary reason, I cannot supply the java file, but
 Saurin> is there anything else I can do to help you debug the problem?
 
 I suppose you could try to run jc1 in the debugger and track down the
 actual infinite loop (assuming one exists) or other bug.
 
 Saurin> It seems that gcj is stuck in infinite loop in
 Saurin> "generate_bytecode_insns" routine.
 
 It could be.  But note that generate_bytecode_insns is recursive.  So
 it can appear on the stack many times without this indicating a bug.
 
 Tom

Re: java/6071: GCJ does not compile a large (21417 lines) java source file

[Alexandre Petit-Bianco]

The following reply was made to PR java/6071; it has been noted by GNATS.

From: Alexandre Petit-Bianco <apbianco@cygnus.com>
To: shroff@transeda.com
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: java/6071: GCJ does not compile a large (21417 lines) java source file
Date: Wed, 27 Mar 2002 09:38:51 -0800 (PST)

 shroff@transeda.com writes:
 
 > I have a rather large .java file (21417 lines, 950 static methods) that
 > does not compile with gcj. I let gcj run for about 60 minutes, and it
 > still could not generate the class file. I killed the process after
 > 60 minutes of cpu time. Sun "javac" compiles this file in 90 seconds.
 
 You could try a newer version of gcc, like the 3.1 branch (which builds
 on Solaris 2.8. If the symptoms persist, you can run gdb on the jc1
 invocation and interrupt the process once you suspect it entered the
 generate_bytecode_insns loop and then to step through an iteration to
 find out what's going on.
 
 ./A

java/6071: GCJ does not compile a large (21417 lines) java source file

[shroff]

>Number:         6071
>Category:       java
>Synopsis:       GCJ does not compile a large (21417 lines) java source file
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Wed Mar 27 09:36:00 PST 2002
>Closed-Date:
>Last-Modified:
>Originator:     Saurin Shroff
>Release:        unknown-1.0
>Organization:
>Environment:
Sun/Solaris 2.7
GCJ 3.0.1
>Description:
I have a rather large .java file (21417 lines, 950 static methods) that
does not compile with gcj. I let gcj run for about 60 minutes, and it
still could not generate the class file. I killed the process after
60 minutes of cpu time. Sun "javac" compiles this file in 90 seconds.

Due to proprietary reason, I cannot supply the java file, but is there anything
else I can do to help you debug the problem? Here is the command line and the output:

I used "proctool" to view the stack trace of the process at various
intervals; here the stack trace that I saw at all times after initial 1 minute:

 00050640 generate_bytecode_insns (35be30, 86bbf28, ffbeed18, 86b27f8, 1, 86bbd60) + 202c
 0005040c generate_bytecode_insns (4719968, 2, ffbeed18, 2d9440, 86bbd28, ffffffff) + 1df8
 000504e8 generate_bytecode_insns (35fb20, 2, ffbeed18, 86bbeb8, 86bbd68, ffffffff) + 1ed4
 0004e6dc generate_bytecode_insns (35f7c0, 2, ffbeed18, 2d9440, 4b, 86bbd20) + c8
 0005040c generate_bytecode_insns (35bd70, 2, ffbeed18, 86bbeb8, 86bbcf8, ffffffff) + 1df8
 0004eb44 generate_bytecode_insns (35bd88, 2, ffbeed18, 1, c97450, 86b2878) + 530
 0004eb34 generate_bytecode_insns (35bda0, 2, ffbeed18, 86bbeb8, 86bbce0, ffffffff) + 520
 000503c0 generate_bytecode_insns (35bdb8, 2, ffbeed18, 2d9440, 86bbeb8, ffffffff) + 1dac
 0005040c generate_bytecode_insns (35bd40, 2, ffbeed18, 86bbeb8, 86bbc88, ffffffff) + 1df8
 0004e6a4 generate_bytecode_insns (35f6c0, 2, ffbeed18, 2d9440, 86bbc70, ffffffff) + 90
 00050130 generate_bytecode_insns (35f6c0, 2, ffbeed18, bf5840, ffbeed3c, 86b2878) + 1b1c
 0004e6a4 generate_bytecode_insns (35f660, 2, ffbeed18, 86b27f8, 32eaa0, 0) + 90
 00051c1c generate_classfile (0, ffbeed18, 2e2fd8, ffff, ffbeed3c, 27b000) + 440
 00052cf0 write_classfile (2e2fd8, 2de4c0, 28f3b48, 1, 0, c8) + 54
 000238e0 java_expand_classes (27b000, 2f9f87, 3000080, ff23d4bc, 0, ff2359c8) + 1d8
 0004a3e0 yyparse  (25c000, 46bc8, 7fffffff, 27b400, 255c00, 504b0304) + 480
 00056b00 compile_file (2f9f87, 255c00, 256000, ffffffff, 256000, 2d) + 2ec
 0005a72c main     (c, ffbeefe4, 255c00, 25b08c, 0, 0) + 5bc
 00011450 _start   (0, 0, 0, 0, 0, 0) + 5c

It seems that gcj is stuck in infinite loop in "generate_bytecode_insns" routine.

gcj -v -C --encoding=UTF-8 -CLASSPATH /home/shroff/projects/gcj file.java
Reading specs from /usr/local/lib/gcc-lib/sparc-sun-solaris2.8/3.0.1/specs
Configured with: ../configure --with-as=/usr/local/bin/as --with-ld=/usr/local/bin/ld --enable-libgcj
Thread model: posix
gcc version 3.0.1
 /usr/local/lib/gcc-lib/sparc-sun-solaris2.8/3.0.1/jc1 file.java -quiet -dumpbase file.java -g1 -version -fencoding=UTF-8 -fencoding=UTF-8 -fCLASSPATH=/home/shroff/projects/gcj -fsyntax-only -femit-class-files -o /dev/null
GNU Java version 3.0.1 (sparc-sun-solaris2.8)
        compiled by GNU C version 3.0.1.

>How-To-Repeat:

>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: