This is the mail archive of the gcc-regression@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: libjava failures between 115092 and 115112


Geoff Keating wrote:
Hi All,

Between revisions 115092 and 115112, hundreds of libjava tests started failing. I'm guessing it's due to this change:

------------------------------------------------------------------------
r115097 | fitzsim | 2006-06-30 14:31:48 -0400 (Fri, 30 Jun 2006) | 18 lines

2006-06-30 Thomas Fitzsimmons <fitzsim@redhat.com>

        * classpath/tools/Makefile.am: Comment out tools wrapper creation.
        (TOOLS_ZIP): Rename tools.zip tools.jar.
        (noinst_DATA): Do not install tools.jar.
        * classpath/tools/Makefile.in: Regenerate.
        * classpath/Makefile.am (SUBDIRS): Add tools.
        (DIST_SUBDIRS): Likewise.
        * classpath/Makefile.in: Regenerate.
        * Makefile.am (toolexeclib_LTLIBRARIES): Add libgcj-tools.la.
        (jar_DATA): Add libgcj-tools-$(gcc_version).jar.
        (bin_PROGRAMS): Add gappletviewer, gjarsigner and gkeytool.
        Add build logic for libgcj-tools.la,
        libgcj-tools-$(gcc_version).jar and gappletviewer, gjarsigner and
        gkeytool.
        * Makefile.in: Regenerate.

it could also be Jason's changes in 115095 or 115105, but it seems very much less likely. A typical failure is:

/Users/regress/tbox/svn-gcc/libjava/testsuite/libjava.compile/ArrayClass.java:0: error: Can't find default package 'java.lang'. Check the CLASSPATH environment variable and the access to the archives

although all kinds of other things happen including


AssertBug.java:4: internal compiler error: vector VEC(tree,base) index domain error, in class_depth at java/class.c:516

Out of stack space.

ArrayClass.java:6: internal compiler error: in push_type, at java/expr.c:281

Case.java:0: error: cannot find file for class java.lang.Object

and so on. Probably the error handling for missing classes could be substantially improved...


You can find a dejagnu log at

<http://gcc.gnu.org/regtest/HEAD/native-logsum/powerpc-apple-darwin8.5.0/libjava/testsuite/libjava.log>

I ran "make check" successfully in the libjava build directory on powerpc-apple-darwin7.9.0 before committing my patch, so I don't see how it could have introduced these failures. I'll try again to see if I can reproduce this locally though.


Tom


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]