[rfa] Add some flags to tramp frame unwinder

Andrew Cagney cagney@gnu.org
Mon Dec 13 21:34:00 GMT 2004


Randolph Chung wrote:
>>Instead of this, delete the test and in its place add a comment 
>>explaining what was there.  I just hit another similar case - signal 
>>trampolines on altstacks - where there's either/or a valid symbol or 
>>segment.
> 
> 
> ok, checked in the following
> 
> 
> 
> 2004-12-12  Randolph Chung  <tausq@debian.org>
> 
> 	* tramp-frame.c (tramp_frame_sniffer): Allow frames with names or 
> 	sections to be trampolines too.

Thanks.

Andrew

> Index: tramp-frame.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/tramp-frame.c,v
> retrieving revision 1.7
> diff -u -p -r1.7 tramp-frame.c
> --- tramp-frame.c	7 Nov 2004 12:54:58 -0000	1.7
> +++ tramp-frame.c	13 Dec 2004 01:35:42 -0000
> @@ -122,19 +122,12 @@ tramp_frame_sniffer (const struct frame_
>    const struct tramp_frame *tramp = self->unwind_data->tramp_frame;
>    CORE_ADDR pc = frame_pc_unwind (next_frame);
>    CORE_ADDR func;
> -  char *name;
>    struct tramp_frame_cache *tramp_cache;
>  
> -  /* If the function has a valid symbol name, it isn't a
> -     trampoline.  */
> -  find_pc_partial_function (pc, &name, NULL, NULL);
> -  if (name != NULL)
> -    return 0;
> -  /* If the function lives in a valid section (even without a starting
> -     point) it isn't a trampoline.  */
> -  if (find_pc_section (pc) != NULL)
> -    return 0;
> -  /* Finally, check that the trampoline matches at PC.  */
> +  /* tausq/2004-12-12: We used to assume if pc has a name or is in a valid 
> +     section, then this is not a trampoline.  However, this assumption is
> +     false on HPUX which has a signal trampoline that has a name; it can
> +     also be false when using an alternative signal stack.  */
>    func = tramp_frame_start (tramp, next_frame, pc);
>    if (func == 0)
>      return 0;
> 
> randolph



More information about the Gdb-patches mailing list