Re: [patch] Fix `return' of long/long-long results with no debuginfo

[Joel Brobecker <brobecker@adacore.com>]

> 2009-03-13  Jan Kratochvil  <jan.kratochvil@redhat.com>
> 
> 	* stack.c (return_command <retval_exp>): New variables retval_expr
> 	and old_chain.  Inline parse_and_eval to initialize retval_expr.  Check
> 	RETVAL_EXPR for UNOP_CAST and set RETURN_TYPE to the RETURN_VALUE type
> 	if RETURN_TYPE is NULL.

This part looks good to me.

> gdb/testsuite/
> 2009-03-13  Jan Kratochvil  <jan.kratochvil@redhat.com>
> 
> 	* gdb.base/return-nodebug.exp, gdb.base/return-nodebug.S: New.

This part looks good to me too.  Just a couple of things:
  - The name of the C file has a small typo in the extension....
  - A comment seems malformed:

> +	# Test return from a function with no debug info (symbol; still it may
> +	# have a minimal-symbol) if it does not crash.

The part in between parens sound strange. I can suggest:

    # Verify that we do not crash when using "return" from a function
    # with no debugging info (it may still have an associated minimal
    # symbol).

I'm not sure that the side note about having minimal symbols is really
necessary, though, but it's definitely not harmful in this case...

(thanks for teaching me about "prepare_for_testing" ;-)

-- 
Joel