RE: [PATCH 2/2 ver 3] PowerPC, fix support for printing the function return value for non-trivial values.

[Carl Love <cel@us.ibm.com>]

Ulrich, Simon, Tom, GDB maintainers:

> 
> Just so you can test properly, I also see these failures that are
> likely
> due to the same bug:
> 
> UNRESOLVED: gdb.base/foll-vfork.exp: exec: vfork parent follow,
> finish after tcatch vfork: finish
> UNRESOLVED: gdb.base/foll-vfork.exp: exec: vfork child follow, finish
> after tcatch vfork: finish
> UNRESOLVED: gdb.base/foll-vfork.exp: exit: vfork parent follow,
> finish after tcatch vfork: finish
> UNRESOLVED: gdb.base/foll-vfork.exp: exit: vfork child follow, finish
> after tcatch vfork: finish
> UNRESOLVED: gdb.base/sigaltstack.exp: finish to throw INNER
> UNRESOLVED: gdb.base/sigstep.exp: finish from handleri: leave signal
> trampoline

I have dug into why I didn't catch the issue earlier.  Looking
carefully at the regression test output, the foll-vfork.exp,
sigaltstack.exp, sigstep.exp and asm-source.exp tests are all failing
similarly.  In the GDB regression test output I see:

Running /home/carll/GDB/build-non-trivial/gdb/testsuite/../../../binutils-gdb-n\
on-trivial/gdb/testsuite/gdb.base/foll-vfork.exp ...
ERROR: GDB process no longer exists
ERROR: GDB process no longer exists
ERROR: GDB process no longer exists
ERROR: GDB process no longer exists

Running .../binutils-gdb-non-trivial/gdb/testsuite/gdb.asm/asm-source.exp ...
ERROR: GDB process no longer exists

Running ../binutils-gdb-non-trivial/gdb/testsuite/gdb.base/sigaltstack.exp ...
ERROR: GDB process no longer exists
ERROR: Couldn't send finish to GDB.
ERROR: can not find channel named "exp17"

Running .../binutils-gdb-non-trivial/gdb/testsuite/gdb.base/sigstep.exp ...
ERROR: GDB process no longer exists

The failures then end up being recorded in the summary as "# of
unresolved testcases" not as "# of unexpected failures".  I can see
with the current mainline code base there are 8 unresolved  testcases. 
I updated the fix that I sent to Tom earlier per the comments from
Ulrich.  With that fix applied, the number of unresolved test cases
drops to 1.  If I run the regression tests with the tree backed up the
the commit just before my commits from Monday, I see one unresolved
test case.  So, clearly my commit added 7 unresolved test cases, the
six that Simon mentioned above and the one from Tom.  The original
unresolved test case is:

Running .../binutils-gdb-non-trivial/gdb/testsuite/gdb.base/gdb-sigterm.exp ...
FAIL: gdb.base/gdb-sigterm.exp: pass=0: expect eof (GDB internal error)
ERROR: Could not resync from internal error (recursive internal problem)

I always check that "# of unexpected failures" did not change with my
patch applied.  I didn't notice that the number of unresolved testcases
changed and hence missed the breakage my patch caused.  My bad.  I will
be sure to watch the other stats in the gdb summary a lot closer in the
future.

Thanks for pointing out the failures and the for help resolving the
issue. 

I will post my updated fix for Tom to test/review.  I think we have
this all figured out and fixed.  Sorry for breaking things.

                      Carl Love