Re: [avr-gcc-list] [AVR] RTL prologue/epilogue ver.3

[Weddington, Eric]

> -----Original Message-----
> From: address@hidden
> [mailto:address@hidden On
> Behalf Of Boyapati, Anitha
> Sent: Thursday, October 13, 2011 1:48 AM
> To: address@hidden
> Subject: Re: [avr-gcc-list] [AVR] RTL prologue/epilogue ver.3
> 
> 
> Comment -2:
> 
> I think this patch severely affects the step in/step out feature of
> debugger like AVR Studio (here after referred to as AS). AS depends on
> 'rcall'/'ret' instruction sequences to recognize a function call and
> step into. If a matching 'ret' is not found, the debugger gets
confused.
> Currently, the problem is worked around but has much lower
performance.
> 
> Also it is misleading to use 'rcall' for purposes it is not meant for
> (like stack allocation). In my opinion, the hack is not desirable by
> default. I would therefore recommend making it optional using a switch
> like 'mrcall'.
> 
> Thoughts?

Anitha,

That optimization was put in so many years ago, that it would be silly
to now take it out just because the AVR Studio debugger does not know
how to handle it. We've taken advantage of side-effects of the rcall
instruction to improve the code size.

It would be better if the debugger could be made to be smarter, not dumb
the compiler down more just because the debugger hasn't been improved in
this area.

I've had this same conversation internally at least 2 years ago (or
more) now, and I'm surprised that it hasn't gotten any further...