From mboxrd@z Thu Jan  1 00:00:00 1970
From: Andrew Cagney <ac131313@cygnus.com>
To: Daniel Jacobowitz <dmj+@andrew.cmu.edu>
Cc: gdb-patches@sources.redhat.com
Subject: Re: [rfa] mips heuristic_proc_start fix
Date: Thu, 12 Jul 2001 00:47:00 -0000
Message-id: <3B4D4FC6.6090003@cygnus.com>
References: <20010706112010.A5578@nevyn.them.org> <3B46030B.2010007@cygnus.com> <20010706113232.A6209@nevyn.them.org> <20010706114028.A6366@nevyn.them.org>
X-SW-Source: 2001-07/msg00283.html

>> 
>> Glarg.  Yes, it's signed, of course!  Let me peer at this for a moment
>> and see why the math was coming out wrong; the patch is correct, but
>> now it's not quite as obvious to me why :)
> 
> 
> No, I stand uncorrected:


Sorry, yes, my bad.

> +  if (start_pc < fence)
> +    {
> +      warning ("Warning: GDB can't find the start of the function at 0x%s (wraparound).",
> +	       paddr_nz (pc));
> +      return 0;
> +    }


Hmm, with this patch UINT_MAX would give a fence of VM_MIN_ADDRESS yet 
UINT_MAX-1 would give the above warning.

Could I suggest instead something like:

   pc = ADDR_BITS_REMOVE (pc);
   start_pc = pc;
   if (start_pc >= heuristic_fence_post)
     fence = start_pc - heuristic_fence_post;
   else
     fence = 0;
   if (start_pc == 0)
     return 0;

   if (heuristic_fence_post == UINT_MAX
       || fence < VM_MIN_ADDRESS)
     fence = VM_MIN_ADDRESS;

The test being moved to before the assignment to make what is happening 
more transparent.

	enjoy,
		Andrew