Backtrace_symbols and .symtab

[Yury Gribov]

On 10/24/2014 06:28 PM, Carlos O'Donell wrote:
> On 10/24/2014 09:44 AM, Yury Gribov wrote:
>> Sure, this will be a separate API because we don't want to slow
>> existing users. So either add some flags parameter or a completely
>> new function.
>
> I don't oppose a new backtrace function does a more complete job
> if .symtab is present.
>
> I agree with Steven Munroe that the existing API performance should
> remain unchanged in the default case.
>
> However, the proof is in the pudding. What does the design look
> like, how complicated is it, does it work, etc.

Got it. I'll get back with a draft patch which illustrates my ideas then 
(not sure about timing though).

-Y