Re: [PATCH] mips-tdep.c: Fix bug introduced at time of NUM_REGS multiarch

[Andrew Cagney <ac131313@redhat.com>]

> I've just committed the patch below.  As noted in the subject, it fixes
> a bug that was introduced when I multiarched NUM_REGS a short time ago.
> 
> 	* mips-tdep.c (heuristic_proc_desc): Clear memory associated with
> 	``temp_saved_regs'', not the pointer or other storage contiguous
> 	to this pointer.
> 
> Index: mips-tdep.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/mips-tdep.c,v
> retrieving revision 1.147
> diff -u -p -r1.147 mips-tdep.c
> --- mips-tdep.c	13 Dec 2002 18:09:30 -0000	1.147
> +++ mips-tdep.c	21 Dec 2002 06:31:58 -0000
> @@ -2142,7 +2142,7 @@ heuristic_proc_desc (CORE_ADDR start_pc,
>      return NULL;
>    memset (&temp_proc_desc, '\0', sizeof (temp_proc_desc));
>    temp_saved_regs = xrealloc (temp_saved_regs, SIZEOF_FRAME_SAVED_REGS);
> -  memset (&temp_saved_regs, '\0', SIZEOF_FRAME_SAVED_REGS);
> +  memset (temp_saved_regs, '\0', SIZEOF_FRAME_SAVED_REGS);
>    PROC_LOW_ADDR (&temp_proc_desc) = start_pc;
>    PROC_FRAME_REG (&temp_proc_desc) = SP_REGNUM;
>    PROC_PC_REG (&temp_proc_desc) = RA_REGNUM;

Outch!

BTW, it isn't as bad as:

	memcpy (&frame->saved_regs, new_regs, sizeof (frame->saved_regs));

which has been cloned a number of times and now appears in multiple 
targets.  (frame->saved_regs is a pointer to the buffer).

good catch,
Andrew