Re: gcj darwin test results for 3.3

[Andrew Haley <aph at redhat dot com>]

Andreas Tobler writes:
 > Andrew Haley wrote:
 > > Firstly, the interpreter doesn't work.
 > 
 > Why does the interpreter not work? I enabled it and I can build class 
 > files and execute them? You mean something specific with the interpreter?

I don't know why it doesn't work.  You do, however, have about a
bazillion gij failures.

 > > 
 > > Ignoring that:
 > > 
 > > 
 > >>FAIL: pr8823 compilation from bytecode
 > >>FAIL: pr8823 -O compilation from bytecode
 > > 
 > > 
 > 
 > > 
 > > 
 > > These tests failed to compile.  Did gcj say what the problem was?
 > 
 > -shared unkown linker option. Darwin is not able to build a .so.
 > > 
 > > 
 > >>FAIL: invoke.c compilation
 > >>FAIL: martin.c compilation
 > >>FAIL: noclass.c compilation
 > >>FAIL: overload.c compilation
 > >>FAIL: register.c compilation
 > >>FAIL: simple_int.c compilation
 > >>FAIL: throwit.c compilation
 > >>FAIL: virtual.c compilation
 > 
 > Same as above.
 > > 
 > > 
 > > Looks like JNI doesn't compile.  Probably a linker problem.
 > > 
 > > 
 > >>FAIL: FileHandleGcTest execution - source compiled test
 > >>FAIL: FileHandleGcTest execution - bytecode->native test
 > >>FAIL: FileHandleGcTest -O execution - source compiled test
 > > 
 > > 
 > > Don't worry about this.
 > 
 > The above?
Yes.

 > >>FAIL: SyncTest execution - bytecode->native test
 > >>FAIL: SyncTest -O execution - bytecode->native test
 > Doesn't say something useful:
 > PASS: SyncTest compilation from source
 > ok
 > PASS: SyncTest execution - source compiled test
 > PASS: SyncTest output - source compiled test
 > PASS: SyncTest byte compilation
 > SyncTest^M
 > SyncTest ^M
 > FAIL: SyncTest execution - gij test
 > UNTESTED: SyncTest output - gij test
 > 
 > Do I get some more output on --verbose?
 > 
 > >>FAIL: Thread_Interrupt -O output - source compiled test
 > Same here
 > >>FAIL: Throw_2 execution - source compiled test
 > >>FAIL: Throw_2 execution - bytecode->native test
 > >>FAIL: Throw_2 -O execution - source compiled test
 > >>FAIL: Throw_2 -O execution - bytecode->native test
 > 
 > Same here
 > > 
 > > 
 > > This worries me.
 > The above?

Yes.

 > >>FAIL: invokethrow execution - source compiled test
 > >>FAIL: invokethrow execution - bytecode->native test
 > >>FAIL: invokethrow -O execution - source compiled test
 > >>FAIL: invokethrow -O execution - bytecode->native test
 > > 
 > > 
 > > Throwing an exception doesn't work through libffi.
 > 
 > Ok.
 > > 
 > > 
 > >>FAIL: pr8823 compilation from bytecode
 > 
 > pr8823.java: In class `pr8823':^M
 > pr8823.java: In method `pr8823.main(java.lang.String[])':^M
 > pr8823.java:17: error: class 'A' has no method named 'm' matching 
 > signature '()V
 > '^M
 > pr8823.java: In method `pr8823.m()':^M
 > pr8823.java:22: internal compiler error: tree check: accessed elt 0 of 
 > tree_vec
 > with 3 elts in find_local_variable, at java/decl.c:210^M
 > 
 > >>FAIL: pr8823 -O compilation from bytecode
 > The same as pr8823
 > 
 > >>FAIL: TestEarlyGC compilation from source
 > 
 > /usr/bin/ld: 
 > /Volumes/reserved1/gcchfs/objdir/powerpc-apple-darwin6.0/libjava/te
 > stsuite/dummy.o malformed object (bad indirect symbol table entry (2) 
 > (symbol at
 >   index 21 is not external)^M
 > collect2: ld returned 1 exit status^M
 > compiler exited with status 1
 > 
 > >>FAIL: TestLeak compilation from source
 > The same as TestEarlyGC
 > 
 > >>FAIL: TestMultiple compilation from source
 > The same as TestEarlyGC
 > >>FAIL: TestParent compilation from source
 > 
 > The same as TestEarlyGC

Assembler bug, probably.

Andrew.