[Bug java/13024] New: [3.4 regression] gcj can't build current rhug

[Bug java/13024] New: [3.4 regression] gcj can't build current rhug

[tromey at gcc dot gnu dot org]

Using gcj cvs head, I can't build rhug cvs head.

gcj --classpath=upstream/src:../../rhug/xerces/upstream/src --encoding=8859_1
-fassume-compiled -g -c
../../rhug/xerces/upstream/src/org/apache/xerces/impl/xs/traversers/XSDComplexTypeTraverser.java
-MT
upstream/src/org/apache/xerces/impl/xs/traversers/lib_org_apache_xerces_la-XSDComplexTypeTraverser.lo
-MD -MP -MF
.deps/upstream/src/org/apache/xerces/impl/xs/traversers/lib_org_apache_xerces_la-XSDComplexTypeTraverser.TPlo
 -fPIC -o
upstream/src/org/apache/xerces/impl/xs/traversers/.libs/lib_org_apache_xerces_la-XSDComplexTypeTraverser.o
../../rhug/xerces/upstream/src/org/apache/xerces/impl/xs/traversers/XSDComplexTypeTraverser.java:
In class `org.apache.xerces.impl.xs.traversers.XSDComplexTypeTraverser':
../../rhug/xerces/upstream/src/org/apache/xerces/impl/xs/traversers/XSDComplexTypeTraverser.java:
In method
`org.apache.xerces.impl.xs.traversers.XSDComplexTypeTraverser.traverseComplexTypeDecl(org.w3c.dom.Element,java.lang.String,java.lang.Object[],org.apache.xerces.impl.xs.traversers.XSDocumentInfo,org.apache.xerces.impl.xs.SchemaGrammar)':
../../rhug/xerces/upstream/src/org/apache/xerces/impl/xs/traversers/XSDComplexTypeTraverser.java:212:
internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.


It is very important that we be able to build rhug; it is one of our
release criteria packages.

-- 
           Summary: [3.4 regression] gcj can't build current rhug
           Product: gcc
           Version: 3.4
            Status: UNCONFIRMED
          Severity: normal
          Priority: P1
         Component: java
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: tromey at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug java/13024] [3.4 regression] gcj can't build current rhug

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-12 18:08 -------
What target are you compiling for, i686-pc-linux?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |critical
           Keywords|                            |ice-on-valid-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug java/13024] [3.4 regression] gcj can't build current rhug

[tromey at gcc dot gnu dot org]

------- Additional Comments From tromey at gcc dot gnu dot org  2003-11-12 18:30 -------
Sorry... this is x86 Red Hat Linux 9.
I'm doing a native build.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug java/13024] [3.4 regression] gcj can't build current rhug

[tromey at gcc dot gnu dot org]

------- Additional Comments From tromey at gcc dot gnu dot org  2003-11-14 19:18 -------
Should fix for 3.4


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |3.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug java/13024] [3.4 regression] gcj can't build current rhug

[jsturm at gcc dot gnu dot org]

------- Additional Comments From jsturm at gcc dot gnu dot org  2003-11-19 04:52 -------
This is a buffer overflow error in prepare_eh_table_type.  Patch coming soon.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2003-11-19 04:52:05
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug java/13024] [3.4 regression] gcj can't build current rhug

[jsturm at gcc dot gnu dot org]

------- Additional Comments From jsturm at gcc dot gnu dot org  2003-11-19 21:11 -------
http://gcc.gnu.org/ml/java-patches/2003-q4/msg00419.html

With this patch my rhug build gets further, but:

gcj --encoding=UTF-8 -fassume-compiled
--classpath=upstream/src/share:../../jakarta-tomcat/upstream/src/share:../jakarta-servletapi/upstream/src/share:../xerces/upstream/src
-O2 -g -O2 -c
../../jakarta-tomcat/upstream/src/share/org/apache/tomcat/loader/AdaptiveClassLoader.java
-MT upstream/src/share/org/apache/tomcat/loader/AdaptiveClassLoader.lo -MD -MP
-MF .deps/upstream/src/share/org/apache/tomcat/loader/AdaptiveClassLoader.TPlo
-o upstream/src/share/org/apache/tomcat/loader/AdaptiveClassLoader.o >/dev/null 2>&1
make[1]: ***
[upstream/src/share/org/apache/tomcat/loader/AdaptiveClassLoader.lo] Error 1


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug java/13024] [3.4 regression] gcj can't build current rhug

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2003-11-20 03:35 -------
Subject: Bug 13024

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	jsturm@gcc.gnu.org	2003-11-20 03:34:59

Modified files:
	gcc/java       : ChangeLog except.c 

Log message:
	Fix PR java/13024.
	* except.c (prepare_eh_table_type): Allocate variable-sized
	buffer `buf' with alloca.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/ChangeLog.diff?cvsroot=gcc&r1=1.1295&r2=1.1296
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/java/except.c.diff?cvsroot=gcc&r1=1.40&r2=1.41



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug java/13024] [3.4 regression] gcj can't build current rhug

[tromey at gcc dot gnu dot org]

------- Additional Comments From tromey at gcc dot gnu dot org  2003-11-20 17:55 -------
I don't see Jeff's failure, but I do see a different failure
while building rhug.  This one is extra-mysterious, since
it seems to also include a bug in printing the error message
(I know Frank Eigler looked at this a little, but I don't think
a patch was forthcoming)


source='built/scm/AutoInit.java' object='built/scm/AutoInit.o' libtool=no \
depfile='.deps/built/scm/AutoInit.Po' tmpdepfile='.deps/built/scm/AutoInit.TPo' \
depmode=gcc3 /bin/sh ../../rhug/depcomp \
gcj -Ibuilt -I../../rhug/jasmin/upstream/lib/cup
-I../../rhug/jasmin/upstream/lib/jas/src -I../../rhug/jasmin/upstream/src -g -c
-o built/scm/AutoInit.o `test -f built/scm/AutoInit.java || echo
'../../rhug/jasmin/'`built/scm/AutoInit.java
built/scm/AutoInit.java:43: error: .
   e.definevar(Symbol.intern("jas-code-var-size"), new scmsetVarSize());
                                                                       ^
built/scm/AutoInit.java:8: error: .
   e.definevar(Symbol.intern("make-ascii-cpe"), new scmAsciiCP());
   ^
built/scm/AutoInit.java:9: error: .
   e.definevar(Symbol.intern("make-class-cpe"), new scmClassCP());
   ^
built/scm/AutoInit.java:10: error: .
   e.definevar(Symbol.intern("make-name-type-cpe"), new scmNameTypeCP());
   ^
built/scm/AutoInit.java:11: error: .
   e.definevar(Symbol.intern("make-field-cpe"), new scmFieldCP());
   ^
built/scm/AutoInit.java:12: error: .
   e.definevar(Symbol.intern("make-interface-cpe"), new scmInterfaceCP());
   ^
built/scm/AutoInit.java:13: error: .
   e.definevar(Symbol.intern("make-method-cpe"), new scmMethodCP());
   ^
built/scm/AutoInit.java:14: error: .
   e.definevar(Symbol.intern("make-integer-cpe"), new scmIntegerCP());
   ^
built/scm/AutoInit.java:15: error: .
   e.definevar(Symbol.intern("make-float-cpe"), new scmFloatCP());
   ^
built/scm/AutoInit.java:16: error: .
   e.definevar(Symbol.intern("make-long-cpe"), new scmLongCP());
   ^
built/scm/AutoInit.java:17: error: .
   e.definevar(Symbol.intern("make-double-cpe"), new scmDoubleCP());
   ^
built/scm/AutoInit.java:18: error: .
   e.definevar(Symbol.intern("make-string-cpe"), new scmStringCP());
   ^
built/scm/AutoInit.java:19: error: .
   e.definevar(Symbol.intern("make-field"), new scmVar());
   ^
built/scm/AutoInit.java:20: error: .
   e.definevar(Symbol.intern("make-const"), new scmConstAttr());
   ^
built/scm/AutoInit.java:21: error: .
   e.definevar(Symbol.intern("make-outputstream"), new scmscmOutputStream());
   ^
built/scm/AutoInit.java:22: error: .
   e.definevar(Symbol.intern("make-label"), new scmLabel());
   ^
built/scm/AutoInit.java:23: error: .
   e.definevar(Symbol.intern("make-class-env"), new scmClassEnv());
   ^
built/scm/AutoInit.java:24: error: .
   e.definevar(Symbol.intern("make-code"), new scmCodeAttr());
   ^
built/scm/AutoInit.java:25: error: .
   e.definevar(Symbol.intern("make-exception"), new scmExceptAttr());
   ^
built/scm/AutoInit.java:26: error: .
   e.definevar(Symbol.intern("make-catchtable"), new scmCatchtable());
   ^
built/scm/AutoInit.java:27: error: .
   e.definevar(Symbol.intern("make-catch-entry"), new scmCatchEntry());
   ^
built/scm/AutoInit.java:28: error: .
   e.definevar(Symbol.intern("iinc"), new scmIincInsn());
   ^
built/scm/AutoInit.java:29: error: .
   e.definevar(Symbol.intern("multianewarray"), new scmMultiarrayInsn());
   ^
built/scm/AutoInit.java:30: error: .
   e.definevar(Symbol.intern("invokeinterface"), new scmInvokeinterfaceInsn());
   ^
built/scm/AutoInit.java:31: error: .
   e.definevar(Symbol.intern("jas-class-addcpe"), new scmaddCPItem());
   ^
built/scm/AutoInit.java:32: error: .
   e.definevar(Symbol.intern("jas-class-addfield"), new scmaddField());
   ^
built/scm/AutoInit.java:33: error: .
   e.definevar(Symbol.intern("jas-class-addinterface"), new scmaddInterface());
   ^
built/scm/AutoInit.java:34: error: .
   e.definevar(Symbol.intern("jas-class-setclass"), new scmsetClass());
   ^
built/scm/AutoInit.java:35: error: .
   e.definevar(Symbol.intern("jas-class-setsuperclass"), new scmsetSuperClass());
   ^
built/scm/AutoInit.java:36: error: .
   e.definevar(Symbol.intern("jas-class-addmethod"), new scmaddMethod());
   ^
built/scm/AutoInit.java:37: error: .
   e.definevar(Symbol.intern("jas-class-setaccess"), new scmsetClassAccess());
   ^
built/scm/AutoInit.java:38: error: .
   e.definevar(Symbol.intern("jas-class-setsource"), new scmsetSource());
   ^
built/scm/AutoInit.java:39: error: .
   e.definevar(Symbol.intern("jas-class-write"), new scmwrite());
   ^
built/scm/AutoInit.java:40: error: .
   e.definevar(Symbol.intern("jas-exception-add"), new scmaddException());
   ^
built/scm/AutoInit.java:41: error: .
   e.definevar(Symbol.intern("jas-code-addinsn"), new scmaddInsn());
   ^
built/scm/AutoInit.java:42: error: .
   e.definevar(Symbol.intern("jas-code-stack-size"), new scmsetStackSize());
   ^
36 errors
make[2]: *** [built/scm/AutoInit.o] Error 1
make[2]: Leaving directory `/home/tromey/gnu/Nightly/rhug/build/jasmin'


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug java/13024] [3.4 regression] gcj can't build current rhug

[jsturm at gcc dot gnu dot org]

------- Additional Comments From jsturm at gcc dot gnu dot org  2003-11-21 02:53 -------
Compile  the test case with -O2:

Bug.java:17: internal compiler error: in make_edges, at cfgbuild.c:382
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.

Tom, I'm guessing it works for you because you built with --disable-shared.  The
test case passes if you use -fPIC.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug optimization/13024] [3.4 regression] gcj can't build current rhug

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-21 08:58 -------
Now this fulls under the optimization componet as cfg is part of optimizating the code.
Returning to the pool as Jeff will most likely not fix this bug.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|jsturm at gcc dot gnu dot   |unassigned at gcc dot gnu
                   |org                         |dot org
             Status|ASSIGNED                    |NEW
          Component|java                        |optimization


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug optimization/13024] [3.4 regression] gcj can't build current rhug

[tromey at gcc dot gnu dot org]

------- Additional Comments From tromey at gcc dot gnu dot org  2003-11-21 16:21 -------
Actually, I configure rhug with --disable-static, not --disable-shared.
I think static rhug builds might be known not to work for some reason,
but I forget why.  (It might not be our problem, e.g. rhug might not
take into account the oddities of static linking.)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug optimization/13024] [3.4 regression] gcj can't build current rhug

[jsturm at gcc dot gnu dot org]

------- Additional Comments From jsturm at gcc dot gnu dot org  2003-11-22 20:19 -------
(In reply to comment #10)
> Now this fulls under the optimization componet as cfg is part of optimizating
the code.

Sounds good.

> Returning to the pool as Jeff will most likely not fix this bug.

I'll take a stab at it.  I'm not sure this is still a java-only bug, but java
does emit a large number of indirect jumps due to TRY_FINALLY_EXPR.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug optimization/13024] [3.4 regression] gcj can't build current rhug

[jsturm at gcc dot gnu dot org]

------- Additional Comments From jsturm at gcc dot gnu dot org  2003-11-22 20:35 -------
(In reply to comment #11)
> Actually, I configure rhug with --disable-static, not --disable-shared.
> I think static rhug builds might be known not to work for some reason,
> but I forget why.  (It might not be our problem, e.g. rhug might not
> take into account the oddities of static linking.)

It was a libtool problem IIRC.  I posted a workaround once to the rhug list:

http://sources.redhat.com/ml/rhug-rhats/2002-12/msg00031.html

I still use this for my local builds.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug optimization/13024] [3.4 regression] gcj can't build current rhug

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2003-12-01 17:53 -------
Subject: Bug 13024

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	jsturm@gcc.gnu.org	2003-12-01 17:53:11

Modified files:
	gcc            : ChangeLog toplev.c 

Log message:
	PR optimization/13024
	* toplev.c (rest_of_handle_new_regalloc): Remove rebuild_notes
	parameter.
	(rest_of_handle_old_regalloc): Likewise.  Add rebuild_notes
	declaration.  Rebuild jump labels following local_alloc if necessary.
	(rest_of_compilation): Remove rebuild_label_notes_after_reload
	declaration.  Don't pass rebuild_notes parameter to
	rest_of_handle_new_regalloc and rest_of_handle_old_regalloc.
	Don't rebuild jump labels.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.1877&r2=2.1878
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/toplev.c.diff?cvsroot=gcc&r1=1.845&r2=1.846



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug optimization/13024] [3.4 regression] gcj can't build current rhug

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2003-12-01 23:11 -------
Fixed for 3.4.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024

[Bug rtl-optimization/13024] [3.4 regression] gcj can't build current rhug

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2005-03-08 17:31 -------
Hmm, this fails on the mainline again:
PR13024.java: In class 'PR13024':^MPR13024.java: In method 'PR13024.isZipOrJarArchive(java.io.File)':
^MPR13024.java:0: error: dominator of 3 should be 2, not 0^M
PR13024.java:0: internal compiler error: in verify_dominators, at dominance.c:854^M
Please submit a full bug report,^Mwith preprocessed source if appropriate.^M
See <URL:http://gcc.gnu.org/bugs.html> for instructions.^M


I think this was caused by Jeff's patch.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13024