Re: gcj segmentation fault

Re: gcj segmentation fault

[Giovanni Visciano]

>-- Messaggio originale --
>From: Andrew Haley <aph@redhat.com>
>Date: Wed, 22 Oct 2003 19:46:59 +0100
>To: Jeff Sturm <jsturm@one-point.com>
>Cc: Andrew Pinski <apinski@apple.com>,
>   Giovanni Visciano <giovannivisciano@virgilio.it>, <tromey@redhat.com>,
>   <gcc@sources.redhat.com>
>Subject: Re: gcj segmentation fault
>
>
>
>Jeff Sturm writes:
> > On Wed, 22 Oct 2003, Andrew Pinski wrote:
> > > > my binutils:
> > > > binutils       2.12.90.0.1-4
> > >
> > >  From <http://gcc.gnu.org/install/specific.html#ix86-*-linux*>:
> > > As of GCC 3.3, binutils 2.13.1 or later is required for this platform.
> > > See bug 10877 for more information.
> >
> > The backtrace doesn't look right for 10877, which is related to
> > relocations, not EH.
>
>Yeah, but one truly evil binutils bug *is* related to EH.
>
>http://sources.redhat.com/ml/binutils/2003-02/msg00352.html
>http://sources.redhat.com/ml/binutils/2003-02/msg00471.html
>
>Andrew.

with binutils 2.13.90, it works
thanks Giovanni

Re: gcj segmentation fault

[Andrew Haley]

Jeff Sturm writes:
 > On Wed, 22 Oct 2003, Andrew Pinski wrote:
 > > > my binutils:
 > > > binutils       2.12.90.0.1-4
 > >
 > >  From <http://gcc.gnu.org/install/specific.html#ix86-*-linux*>:
 > > As of GCC 3.3, binutils 2.13.1 or later is required for this platform.
 > > See bug 10877 for more information.
 > 
 > The backtrace doesn't look right for 10877, which is related to
 > relocations, not EH.

Yeah, but one truly evil binutils bug *is* related to EH.

http://sources.redhat.com/ml/binutils/2003-02/msg00352.html
http://sources.redhat.com/ml/binutils/2003-02/msg00471.html

Andrew.

Re: gcj segmentation fault

[Jeff Sturm]

On Wed, 22 Oct 2003, Andrew Pinski wrote:
> > my binutils:
> > binutils       2.12.90.0.1-4
>
>  From <http://gcc.gnu.org/install/specific.html#ix86-*-linux*>:
> As of GCC 3.3, binutils 2.13.1 or later is required for this platform.
> See bug 10877 for more information.

The backtrace doesn't look right for 10877, which is related to
relocations, not EH.  And only manifests at -O3.

Regardless, the OP should definitely upgrade binutils and see if the
problem goes away.

Jeff

Re: gcj segmentation fault

[Andrew Pinski]

> my binutils:
> binutils       2.12.90.0.1-4

 From <http://gcc.gnu.org/install/specific.html#ix86-*-linux*>:
As of GCC 3.3, binutils 2.13.1 or later is required for this platform. 
See bug 10877 for more information.

The bug (in binutils) also happens when using libgcj so that is what 
you are seeing.


Thanks,
Andrew Pinski

Re: gcj segmentation fault

[Giovanni Visciano]

>Try "bt".  That should tell us something.
>
>I suspect that you have the wrong version of binutils.  We've seen
>crashes in exception handling when ld is too old.  That's just a guess
>though.
>
>Tom

this is the output:

(gdb) bt
#0  0x79730065 in ?? ()
#1  0x40393ba2 in _Jv_Throw (value=0x808bfa0) at exception.cc:100
#2  0x403a3e12 in java::io::FileDescriptor::open(java::lang::String*, int)
() at gcj/cni.h:64
#3  0x403d6590 in java.io.FileDescriptor.FileDescriptor(java.lang.String,
int) (this=0x808bfa0, path=0x4a415641,
    mode=1245795905) at java/io/FileDescriptor.java:60
#4  0x403d66d1 in java.io.FileInputStream.FileInputStream(java.lang.String)
(this=0x80db3f0, name=0x80db400)
    at java/io/FileInputStream.java:35
#5  0x403d672e in java.io.FileInputStream.FileInputStream(java.io.File)
(this=0x4a415641, file=0x4a415641)
    at java/io/FileInputStream.java:39
#6  0x40422964 in gnu.gcj.protocol.file.Connection.connect() (this=0x80d8f90)
    at gnu/gcj/protocol/file/Connection.java:56
#7  0x40484f19 in java.net.JarURLConnection.connect() (this=0x80db3f0) at
java/net/JarURLConnection.java:91
#8  0x40423fd7 in gnu.gcj.protocol.jar.Connection.getJarFile() (this=0x80db3f0)
    at gnu/gcj/protocol/jar/Connection.java:48
#9  0x4048c8a1 in java.net.URLClassLoader$JarURLLoader.URLClassLoader$JarURLLoader(java.net.URLClassLoader,
java.ne          t.URL) (this=0x809ff30, classloader=0x80d8fc0, baseURL=0x808bfa0)
at java/net/URLClassLoader.java:240
#10 0x4048cfc5 in java.net.URLClassLoader.addURL(java.net.URL) (this=0x80d8fc0,
newUrl=0x8095e38)
    at java/net/URLClassLoader.java:760
#11 0x4048d03f in java.net.URLClassLoader.addURLs(java.net.URL[]) (this=0x80d8fc0,
newUrls=0x8095eb0)
    at java/net/URLClassLoader.java:780
#12 0x4048cbc7 in java.net.URLClassLoader.URLClassLoader(java.net.URL[])
(this=0x80d8fc0, urls=0x4a415641)
    at java/net/URLClassLoader.java:658
#13 0x40426658 in gnu.gcj.runtime.VMClassLoader.VMClassLoader() (this=0x4a415641)
    at gnu/gcj/runtime/VMClassLoader.java:20
#14 0x404269a8 in gnu.gcj.runtime.VMClassLoader.__U3c_clinit__U3e_() ()
at gnu/gcj/runtime/VMClassLoader.java:71
#15 0x403a6569 in java::lang::Class::initializeClass() (this=0x40687b00)
at java/lang/natClass.cc:814
#16 0x40550f7c in _Jv_InitClass (klass=0x4a415641) at java/lang/Class.h:265
#17 0x403bd138 in java.lang.ClassLoader.getSystemClassLoader() () at java/lang/ClassLoader.java:203
#18 0x403a4f95 in java::lang::Class::getClassLoader() (this=0x4a415641)
at java/lang/natClass.cc:134
#19 0x403a4f19 in java::lang::Class::forName(java::lang::String*) (className=0x4a415641)
---Type <return> to continue, or q <return> to quit---
    at java/lang/natClass.cc:104
#20 0x4036f0cb in gnu.gcj.convert.UnicodeToBytes.getDefaultEncoder() ()
at gnu/gcj/convert/UnicodeToBytes.java:49
#21 0x403e64fc in java.io.PrintStream.PrintStream(java.io.OutputStream,
boolean) (this=0x80d8fc0, out=0x4a415641)
    at java/io/PrintStream.java:241
#22 0x403cc925 in java.lang.System.__U3c_clinit__U3e_() () at java/lang/System.java:135
#23 0x403a6569 in java::lang::Class::initializeClass() (this=0x8049d10)
at java/lang/natClass.cc:814
#24 0x40550f7c in _Jv_InitClass (klass=0x4a415641) at java/lang/Class.h:265
#25 0x403cc1e8 in java.lang.System.getProperty(java.lang.String) (key=0x1)
at java/lang/System.java:393
#26 0x403cf5a2 in java.lang.Throwable.__U3c_clinit__U3e_() () at java/lang/Throwable.java:403
#27 0x403a6569 in java::lang::Class::initializeClass() (this=0x8049d10)
at java/lang/natClass.cc:814
#28 0x403a65a5 in java::lang::Class::initializeClass() (this=0x8049d10)
at java/lang/Class.h:265
#29 0x403a65a5 in java::lang::Class::initializeClass() (this=0x8049d10)
at java/lang/Class.h:265
#30 0x403a65a5 in java::lang::Class::initializeClass() (this=0x8049d10)
at java/lang/Class.h:265
#31 0x40386961 in _Jv_AllocObjectNoFinalizer (klass=0x8049d10, size=20)
at java/lang/Class.h:265
#32 0x4038699c in _Jv_AllocObject (klass=0x1, size=1245795905) at prims.cc:431
#33 0x40387849 in _Jv_CreateJavaVM(void*) () at prims.cc:937
#34 0x40387a4f in _Jv_RunMain(java::lang::Class*, char const*, int, char
const**, bool) (klass=0x8049ae0,
    name=0x0, argc=1, argv=0xbffff9a4, is_jar=false) at prims.cc:989
#35 0x40387c09 in JvRunMain (klass=0x4a415641, argc=1245795905, argv=0x4a415641)
at prims.cc:1027
#36 0x08048904 in main ()

my binutils:
binutils       2.12.90.0.1-4

thanks Giovanni

Re: gcj segmentation fault

[Tom Tromey]

>>>>> "Giovanni" == Giovanni Visciano <giovannivisciano@virgilio.it> writes:

Giovanni> (gdb) run
Giovanni> Starting program: /home/gio/Desktop/a.out
Giovanni> [New Thread 1024 (LWP 488)]
Giovanni> Program received signal SIGSEGV, Segmentation fault.
Giovanni> [Switching to Thread 1024 (LWP 488)]
Giovanni> 0x79730065 in ?? ()

Try "bt".  That should tell us something.

I suspect that you have the wrong version of binutils.  We've seen
crashes in exception handling when ld is too old.  That's just a guess
though.

Tom

Re: gcj segmentation fault

[Giovanni Visciano]

>On Tue, 21 Oct 2003, Giovanni Visciano wrote:
>> ./configure --prefix=/home/gio/Gcc3.3.2 --program-suffix=-3.3.2
>> make
>> make install
>>
>> added to my PATH /home/gio/Gcc3.3.2/bin,
>> and to LD_LIBRARY_PATH /home/gio/Gcc-3.3.2/lib
>>
>> gcc-3.3.2 woks without problems.
>> gcj-3.3.2 compiles my java programs but when I run them:
>>
>> ex.
>> gcj-3.3.2 --main=HelloWorld HelloWorld.java
>> (a.out created)
>> ./a.out
>> (segmentation fault)
>
>There aren't enough clues in your message to guess what may have gone
>wrong.
>
>Try running "ldd a.out" to verify it is linking to the libgcj.so in
>$prefix/lib.  Try running under gdb to get a stack trace.

Thanks for your reply Jeff; this is the output:

gio@debian:~$ ldd a.out
libgcc_s.so.1 => /home/gio/Gcc-3.3.2/lib/libgcc_s.so.1 (0x40014000)
        libgcj.so.4 => /home/gio/Gcc-3.3.2/lib/libgcj.so.4 (0x4001c000)
        libm.so.6 => /lib/libm.so.6 (0x40805000)
        libpthread.so.0 => /lib/libpthread.so.0 (0x40826000)
        libdl.so.2 => /lib/libdl.so.2 (0x4083a000)
        libc.so.6 => /lib/libc.so.6 (0x4083d000)
        /lib/ld-linux.so.2 => /lib/ld-linux.so.2 (0x40000000)

runnig a.out with gdb:

(gdb) run
Starting program: /home/gio/Desktop/a.out
[New Thread 1024 (LWP 488)]

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 1024 (LWP 488)]
0x79730065 in ?? ()

Giovanni.

Re: gcj segmentation fault

[Jeff Sturm]

On Tue, 21 Oct 2003, Giovanni Visciano wrote:
> ./configure --prefix=/home/gio/Gcc3.3.2 --program-suffix=-3.3.2
> make
> make install
>
> added to my PATH /home/gio/Gcc3.3.2/bin,
> and to LD_LIBRARY_PATH /home/gio/Gcc-3.3.2/lib
>
> gcc-3.3.2 woks without problems.
> gcj-3.3.2 compiles my java programs but when I run them:
>
> ex.
> gcj-3.3.2 --main=HelloWorld HelloWorld.java
> (a.out created)
> ./a.out
> (segmentation fault)

There aren't enough clues in your message to guess what may have gone
wrong.

Try running "ldd a.out" to verify it is linking to the libgcj.so in
$prefix/lib.  Try running under gdb to get a stack trace.

Jeff

gcj segmentation fault

[Giovanni Visciano]

hi,
I have compiled and installed gcc3.3.2:

./configure --prefix=/home/gio/Gcc3.3.2 --program-suffix=-3.3.2
make
make install

added to my PATH /home/gio/Gcc3.3.2/bin,
and to LD_LIBRARY_PATH /home/gio/Gcc-3.3.2/lib

gcc-3.3.2 woks without problems.
gcj-3.3.2 compiles my java programs but when I run them:

ex.
gcj-3.3.2 --main=HelloWorld HelloWorld.java
(a.out created)
./a.out
(segmentation fault)

no problems with gcc3.2.3.

I am using Debian Woody
thanks Giovanni.