Dummy frames broken on PowerPC

[Adam Fedor]

 >> (gdb) finish
 >> Run till exit from #0  0x100004c8 in marker2 (a=99) at 
 >/mnt/loc-computed/src-noindent/src/gdb/testsuite/gdb.base/break.c:49
 >> /mnt/loc-computed/src-noindent/src/gdb/dummy-frame.c:295: 
 >internal-error: dummy_frame_pop: Assertion `get_frame_type (fi) == 
 >DUMMY_FRAME' failed.
 >> A problem internal to GDB has been detected.  Further debugging may 
 >prove unreliable.
 >> Quit this debugging session? (y or n)
 >
 >What does the backtrace look like?
 >

I get

(gdb) bt
#0  0x1000037c in _start ()

 >I suspect frame_type_from_pc() may have some of all the logic needed to
 >spot a dummy frame.  Contrast it with dummy_frame_p().  However, I've no
 >basis for this - the ppc doesn't use the new frame code and the only
 >other time it could affect things is with create_new_frame() - that,
 >ignoring MIPS and SPARC, has only one call.
 >
 >Anyway, try changing the logic to test (dummy_frame_p() != NULL) when
 >DEPRECATED_USE_GENERIC_DUMMY_FRAMES.

I don't know if this is what you meant, but this patch  fixes it (Not 
that I'm actually going to submit it. It looks ugly).

Index: dummy-frame.c
===================================================================
RCS file: /cvs/src/src/gdb/dummy-frame.c,v
retrieving revision 1.10
diff -u -p -r1.10 dummy-frame.c
--- dummy-frame.c       19 Jan 2003 17:39:16 -0000      1.10
+++ dummy-frame.c       18 Feb 2003 03:11:21 -0000
@@ -292,7 +292,11 @@ dummy_frame_pop (struct frame_info *fi,
    struct dummy_frame *dummy = cached_find_dummy_frame (fi, cache);

    /* If it isn't, what are we even doing here?  */
+#ifdef DEPRECATED_USE_GENERIC_DUMMY_FRAMES
+  gdb_assert (dummy_frame_p(fi->pc) != NULL);
+#else
    gdb_assert (get_frame_type (fi) == DUMMY_FRAME);
+#endif

    if (dummy == NULL)
      error ("Can't pop dummy frame!");