[Bug java/18091] Valgrind errors building libjava

2005-01-20 Thread pinskia at gcc dot gnu dot org 


-- 
   What|Removed |Added

   Target Milestone|--- |4.0.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18091

[Bug java/18091] Valgrind errors building libjava

2005-01-20 Thread tromey at gcc dot gnu dot org 

--- Additional Comments From tromey at gcc dot gnu dot org  2005-01-21 
02:39 ---
I checked in Andrew Pinski's fix.


-- 
   What|Removed |Added

 Status|REOPENED|RESOLVED
 Resolution||FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18091

[Bug java/18091] Valgrind errors building libjava

2005-01-20 Thread cvs-commit at gcc dot gnu dot org 

--- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-01-21 
02:38 ---
Subject: Bug 18091

CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED]   2005-01-21 02:38:24

Modified files:
gcc/java   : ChangeLog jcf-write.c 

Log message:
2005-01-20  Andrew Pinski  <[EMAIL PROTECTED]>

PR java/18091:
* jcf-write.c (perform_relocations): Don't call memcpy if source
and destination are the same.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/ChangeLog.diff?cvsroot=gcc&r1=1.1530&r2=1.1531
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/jcf-write.c.diff?cvsroot=gcc&r1=1.158&r2=1.159



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18091

[Bug java/18091] Valgrind errors building libjava

2005-01-03 Thread pinskia at gcc dot gnu dot org 

--- Additional Comments From pinskia at gcc dot gnu dot org  2005-01-03 
22:25 ---
Not totally fixed.
The following is not fixed yet:
==21511== Source and destination overlap in memcpy(0x1BE08FEC, 0x1BE08FEC, 5)
==21511==at 0x1B904BCE: memcpy (mac_replace_strmem.c:113)
==21511==by 0x80D37F2: write_classfile (jcf-write.c:2800)

Yes this is harmless but we should not do memcpy at all.
The easy fix would be the following:
  if (n > 0)
memcpy (new_ptr, old_ptr, n);
Changed to:
  if (n > 0 && new_ptr != old_ptr)
memcpy (new_ptr, old_ptr, n);

The comment above explains how this can happen.

-- 
   What|Removed |Added

 Status|RESOLVED|REOPENED
 Resolution|FIXED   |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18091

[Bug java/18091] Valgrind errors building libjava

2004-12-16 Thread aph at gcc dot gnu dot org 

--- Additional Comments From aph at gcc dot gnu dot org  2004-12-16 14:29 
---
comment

-- 
   What|Removed |Added

 Status|NEW |RESOLVED
 Resolution||FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18091

[Bug java/18091] Valgrind errors building libjava

2004-10-21 Thread cvs-commit at gcc dot gnu dot org 

--- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-10-21 14:38 
---
Subject: Bug 18091

CVSROOT:/cvs/gcc
Module name:gcc
Changes by: [EMAIL PROTECTED]   2004-10-21 14:38:17

Modified files:
gcc/java   : jcf-parse.c ChangeLog 

Log message:
2004-10-21  Andrew Haley  <[EMAIL PROTECTED]>

PR java/18091:
* jcf-parse.c (set_source_filename): Add code to build new sfname.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/jcf-parse.c.diff?cvsroot=gcc&r1=1.176&r2=1.177
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/ChangeLog.diff?cvsroot=gcc&r1=1.1489&r2=1.1490



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18091

[Bug java/18091] Valgrind errors building libjava

2004-10-20 Thread pinskia at gcc dot gnu dot org 

--- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-20 21:04 
---
I think the "unitialized conditional move" will get fixed when:
==21620== Conditional jump or move depends on uninitialised value(s)
==21620==at 0x80C4165: get_attribute (jcf-parse.c:160)
is fixed.

See comment #1 where I show the problem.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18091

[Bug java/18091] Valgrind errors building libjava

2004-10-20 Thread mckinlay at redhat dot com 

--- Additional Comments From mckinlay at redhat dot com  2004-10-20 20:49 ---
For the memcpy() thing, in the error given we seem to be relocating something to
the exact same position. In this case the memcpy() should be harmless. But,
maybe it is possible to get real overlapping relocations. Its probably best just
to replace the memcpy() with a memmove() here.

I'm not sure the "unitialized conditional move" errors are really bugs. I've
seen this a bit when running valgrind on code compiled with recent GCCs ?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18091

[Bug java/18091] Valgrind errors building libjava

2004-10-20 Thread pinskia at gcc dot gnu dot org 

--- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-20 20:27 
---
memcpy(0x1BE08FEC, 0x1BE08FEC, 5)

  new_ptr -= n;
  old_ptr -= n;
  if (n > 0)
memcpy (new_ptr, old_ptr, n);

Someone needs to look into this part though because I don't understand the code at all.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18091

[Bug java/18091] Valgrind errors building libjava

2004-10-20 Thread pinskia at gcc dot gnu dot org 

--- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-20 20:24 
---
This is definitely a bug:
  /* Concatenate current package prefix with new sfname. */
  char *buf = xmalloc (i+new_len+3);
  /* Replace '.' by DIR_SEPARATOR. */
  for (; i >= 0;  i--)
{
  if (buf[i] == '.')
buf[i] = DIR_SEPARATOR;
}

This could cause problems.

-- 
   What|Removed |Added

 Status|UNCONFIRMED |NEW
 Ever Confirmed||1
   Last reconfirmed|-00-00 00:00:00 |2004-10-20 20:24:29
   date||


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18091